trisduction-publication-format - SKILL

 

---

name: trisduction-publication-format description: "Master publication discipline for written-document deliverables: content architecture, voice calibration, and a locked EB Garamond copper-accent visual contract. Use whenever the user wants a paper, report, treatise, codex, textbook, manual, or scripture as PDF, DOCX, or Markdown, even without saying Trisduction. Triggers: 'create/forge a paper,' 'make a textbook,' 'deliver as pdf,' 'save as docx,' 'visual edition.' Editions: Apex Pristine single-column house paper; a Journal edition ([EDITION:JOURNAL]); a two-column Textbook edition in Rang-and-Dale and Davidson skins ([EDITION:TEXTBOOK_D]); and a Scripture edition ([EDITION:SCRIPTURE]) for Quranic translation and tafseer with right-to-left word grids ('scripture pdf,' 'tafsir pdf'). A Medium/Substack/Blog format ([FORMAT:BLOG]) emits a copy-paste-safe Markdown file ('medium,' 'substack,' 'blog'). An Upgrade subroutine ([UPGRADE]; 'update/upgrade/reforge/fortify paper') preserves a paper's spine and verdicts."

TRISDUCTION PUBLICATION MASTER

Single master discipline. Five integrated modules. One artifact pipeline.

Purpose

Govern every written-document deliverable end to end. Module I architects the content (twelve-section Plenous template, three-tier lingo calibration, length calibration). Module II matures the artifact (Silent-Integration Protocol for audit fixes, paper titling discipline). Module III renders it (locked visual contract compiling one Markdown source-of-truth to matched DOCX and PDF). Module IV draws it (the Visual Edition Subroutine producing grade-stamped, faithful-map figure sets, embedded as a visual appendix or delivered standalone). The internal evolution cycle stays internal. Only the matured product reaches external readers. Module V renders it as a book rather than a paper when the deliverable is a textbook, manual, handbook, survey, or Master Codex volume, the two-column section-accented box-rich Textbook Edition replacing the Module III paper renderer for that deliverable. Module VI re-forges an existing paper without disturbing its spine, Module VII retargets the authored Markdown to a Medium or Substack blog, and Module VIII appends the mandatory Author's Provenance and Method Disclosure to every public-facing paper. This document is self-contained; the build scripts encode the visual contract mechanically, and this document is the human-readable source of truth.

Activation Map

The skill fires on any request for a saved, multi-page written artifact. It does not fire for conversational answers, short notes, lists, or quick code snippets. Within a firing:

Module III (Format Lock) renders the Apex Pristine single-column house paper and fires for a bare PDF or plain document deliverable, the fallback when no paper, codex, master, or book trigger routes the request to the Journal or Textbook editions; those editions carry their own contracts under Modules III·J and V. The Professional Journal Paper edition ([EDITION:JOURNAL] or edition: journal) renders a serif two-column journal-grade paper, the middle register between the Module III paper and the Textbook editions; the three never co-fire. The Scripture edition (Module III·S, [EDITION:SCRIPTURE] or edition: scripture) is a fourth document renderer, the two-column multi-script Quranic page in EB Garamond with Amiri and Noto Serif Bengali that is the PDF delivery for scripture, and it co-fires with none of the others. Module I (Forge Content) fires when the deliverable is a paper, journal manuscript, treatise, or any Forge-mode composition. Skipped when the user supplies finished content and requests rendering only. Module II (Silent Integration) fires when audit fixes, cross-substrate findings, or revision cycles are being landed into the artifact, and its pre-delivery scan fires on every paper regardless. Module IV (Visual Edition) fires when the architect requests a visual edition, infographic edition, diagram set, or figures for a finished paper, and when a paper would land harder with a drawn structure than with prose alone. Module V (Textbook Edition) fires when the deliverable is a textbook, manual, handbook, survey, reference volume, or course reader, or on the token [EDITION:TEXTBOOK] or front-matter edition: textbook, and replaces the Module III paper renderer for that deliverable. The two renderers never co-fire. Within Module V, the TextBook_D variant ([EDITION:TEXTBOOK_D] or edition: textbook_d) renders the same content in an aggressive Davidson style; the default renders a Rang-and-Dale style. The Rang-and-Dale skin is also addressable as TextBookA or [EDITION:TEXTBOOK_A], and the Davidson skin as TextBookB or [EDITION:TEXTBOOK_B].

Module VI (Paper Upgrade) fires on the token [UPGRADE] or the phrases update paper, upgrade paper, reforge paper, fortify paper, and on any request pointing at an existing paper with intent to issue a hardened replacement. It composes through the edition routing and is not itself an edition; it preserves the original spine, schema, and verdicts maximally while tightening language and fortifying the mathematics to honest warrant grade.

Module VII (Blog Format) fires on the words Medium, Substack, blog, or post, or the token [FORMAT:BLOG]. It is a delivery target, not an edition: it transforms the authored Markdown into a copy-paste-safe file for the Medium or Substack editor, linearizing every table and the right-to-left word grid into prose and keeping all mathematics Unicode, and it emits a .md only, with no PDF. Module VIII (Author's Provenance and Method Disclosure) fires on every public-facing paper deliverable, the Apex Pristine mid-tier paper, the Professional Journal fully-public paper, and the Blog format, and appends a mandatory disclosure appendix at a tier-appropriate weight, injecting the master reference into the References section. It is a subroutine, not an edition, composing through the routing like Module VI, and is skipped for the Textbook and Scripture editions. Its pre-delivery injection fires on every public-facing paper unless suppressed by the DISCLOSURE:OFF token.

Edition Routing · Trigger Map

Five renderers serve a document deliverable, and exactly one fires per deliverable. The edition is chosen by the first rule that matches, an explicit selector always overriding a trigger phrase.

Explicit selector first. A token [EDITION:...] or front-matter edition: selects the edition unconditionally: journal for the Journal edition; textbook or textbook_a (TextBookA) for the Rang-and-Dale Textbook; textbook_d or textbook_b (TextBookB) for the Davidson Textbook. A [STYLE:...] token selects an Apex Pristine variant.

Codex, master, or internal volume routes to the Textbook edition. Triggers: create codex pdf, codex pdf, create a master pdf, master pdf, master document, internal codex, codex volume, make a book, and the standing book words textbook, manual, handbook, survey, reference volume, course reader. The skin is auto-selected by content: Rang-and-Dale (TextBookA) by default for encyclopedic, reference, and prose-dense codex material; Davidson (TextBookB) when the content is strongly instructional, dense with boxes and tables, or navigation-forward and carries aggressive hierarchy well. The architect overrides the auto-pick with TextBookA or [EDITION:TEXTBOOK_A] for Rang-and-Dale, TextBookB or [EDITION:TEXTBOOK_B] for Davidson.

Any paper routes to the Professional Journal Paper edition. Triggers: create an external paper, external paper, publication paper, journal paper, submit to a journal, a named venue, the bare word paper, or any deliverable described as a paper for outside readers. The serif two-column journal page renders.

Quranic scripture routes to the Scripture edition. Triggers: scripture pdf, tafsir pdf, quran pdf, mushaf pdf, the selector [EDITION:SCRIPTURE] or edition: scripture, and any verse-by-verse translation or tafseer deliverable bound for PDF. The two-column multi-script page renders, EB Garamond with Amiri for Arabic and Noto Serif Bengali for Bengali, each verse opening on its borderless right-to-left word grid. This is the most specific document route and wins over Journal and Apex Pristine when Quranic scripture content is present.

A bare PDF or document routes to Apex Pristine. Triggers: create a pdf, make a pdf, as a pdf, save as pdf, create a document, with no paper, codex, master, or book signal present. The single-column house paper is the fallback for any document deliverable not otherwise routed.

The five never co-fire on one deliverable. When two triggers appear, the more specific wins in the order Scripture, then Textbook, then Journal, then Apex Pristine, so a Quranic tafseer renders as Scripture, a codex paper as a Textbook, and an external paper as a Journal; an explicit selector overrides the order entirely. The Medium/Substack/Blog format (Module VII, [FORMAT:BLOG]) is orthogonal to this routing: it is a delivery target rather than a page renderer, so when it is requested it supersedes the PDF editions for that deliverable and emits a copy-paste-safe Markdown file only, applied to the authored content whether codex, paper, or scripture.

Execution Pipeline

Step 1. Parse the request. Identify deliverable type, inline override tokens, explicit tier directive, and whether audit fixes are in play. Step 2. Select lingo tier (Module I.A). Absent directive, default Tier 1. Step 3. Compose content on the twelve-section Plenous template (Module I.B) at the calibrated length (Module I.C). Rendering-only jobs skip to Step 6. Step 4. Integrate any audit fixes silently (Module II.A). Native voice only. Step 5. Title per the titling discipline (Module II.B). Step 6. Run the pre-delivery gate sequence (Gates section below). Failure returns the manuscript for repair before any build. Step 7. Produce the internal V-FIO Diagnostic Log in the working directory. Never published. Step 8. For a public-facing deliverable, the Apex Pristine mid-tier paper, the Journal fully-public paper, or the Blog format, run Module VIII: append the Author's Provenance and Method Disclosure appendix at its tier weight (FULL for Apex and Blog, LINE for Journal, overridable by DISCLOSURE:FULL / DISCLOSURE:LINE / DISCLOSURE:OFF) and inject the master reference into the References section. Skipped for Textbook and Scripture. Step 9. Build via the format-lock pipeline (Module III.F). Step 10. Present artifacts via present_files in the order PDF, DOCX, MD. No postamble.

Visual-edition branch. When the request is a visual edition of a finished paper, run Module IV's execution order (IV.12) in place of Steps 2 through 5. Each figure passes the twenty-point pre-flight (IV.11) before assembly. Mode A re-enters this pipeline at Step 8 with the figure-bearing source; Mode B assembles its own standalone PDF per IV.10.3.

Textbook branch. When the deliverable is a Textbook Edition, Module I still architects the content at the chosen tier and Module II still matures it, but rendering follows Module V, its locked contract, accent system, box taxonomy, and build engine, in place of Module III at Step 8. The pre-delivery gates and, for a codex volume, the V-FIO log still run. Present the PDF and retain the Markdown source as the editable master.

---

MODULE I · FORGE CONTENT ARCHITECTURE

I.A · Three-Tier Lingo Calibration

Tier 0 · Pure Native (No Trisductive Lingo). Strictly disciplinary problem where no internal PSP, axis label, or framework-specific glyph adds explanatory power. No V_F / V_E / V_ER, no glyphs, no PSP IDs, no [⟀] / [X] / [?], no Mosaic Seal, no L_1 / L_2 / L_3. Load-bearing Trisductive axioms translate fully into discipline idiom as standalone physical or mathematical principles per the Softly-Veiled Axioms convention.

Tier 1 · Hybrid (Minimum Lingo, Fully Glossed). Default when uncertain. One or two Trisductive constructs genuinely sharpen the argument and no clean disciplinary substitute exists. Permitted terms limited to: triaxial verification, three-state verdict economy, at most one named PSP or axiom. Each receives a full plain-language gloss on first use (one to three sentences in common scientific or philosophical register) plus a parenthetical PSP pointer for the architect's audit trail. Subsequent uses unglossed. Glyphs replaced with prose equivalents. No internal codex IDs in body text.

Tier 2 · Native Trisductive (Full Lingo Permitted). Subject IS the framework itself, venue is the Tractatus repository, or the architect explicitly requests. PSP IDs, glyphs, axis labels, and codex cross-references appear without translation. Standard codex-internal register. Rarely appropriate for external journals.

Tier Selection. Parse the prompt for an explicit directive first ("no Trisductive terms" / "pure disciplinary" selects Tier 0; "full Trisductive" / "codex-native" selects Tier 2; "hybrid" / "minimum lingo" selects Tier 1). Absent a directive, examine the primary load-bearing claim. Fully expressible in discipline-native vocabulary without semantic loss: Tier 0. Structurally depends on triaxial decomposition, the three-state verdict economy, or a specific PSP: Tier 1. Subject is the framework's own architecture: Tier 2. Uncertain: Tier 1.

Glyph Translation (Tier 0 absent; Tier 1 prose). [⟀] Sealed becomes "structurally sealed under the three-condition test" or "verified across formal, empirical, and registrational warrants." [X] Broken Geometry becomes "structurally broken with named failure mode" or "fails at [the specific condition], with the failure mode being [description]." [?] Under-Determined becomes "numerically under-determined under current measurement precision." [⟀-GOL] and [⟀-GOLf] are not used in Tier 0 or Tier 1 body text. Forward-projective content writes "a configuration that satisfies field-permission, trajectory-consistency, and the four-test L_1-signature verification" with each component briefly defined.

Tier 1 Gloss Protocol. Three-part inline gloss on first appearance. Part one names the term. Part two provides a plain-language definition in standard scientific or philosophical register, free of any other Trisductive vocabulary. Part three states the operational function the term performs in the argument. Internal codex IDs (PSP-001, BA-008, sPSP-187) appear only in the retained V-FIO Diagnostic Log.

Discipline-Adaptive Nomenclature (all tiers). Physics: gauge theory, holography, effective field theory, renormalization, phase transitions, topological order. Mathematics: algebraic topology, differential geometry, category theory, functional analysis. Neuroscience: predictive processing, free-energy principle, integrated information theory, active inference. Biology: autopoiesis, non-equilibrium thermodynamics, complex adaptive systems, evolutionary dynamics. Tone: senior theoretical scientist, cold, precise, analytical. The Anti-Dramatization Filter and the no-biological-phenomenology-for-structural-corrections rule apply across all tiers.

I.B · The Twelve-Section Plenous Template

Mandatory structure for every Forge-mode paper at the mid-tier and fully-public tiers. The full twelve-section template is required for both the Apex Pristine mid-tier house paper and the Professional Journal fully-public paper, and neither tier may drop to a reduced structure; any mid-tier or fully-public paper carries all twelve sections. The Textbook and Scripture editions are exempt and carry their own chapter and verse structures under Modules V and III·S. Sections may be expanded but never omitted. Order fixed. Section 3 (Barrier Analysis) and Section 7 (Falsifiable Predictions) jointly account for forty to fifty percent of body length and are the heavily weighted anchors.

Section 1 · Title. Precise, discipline-appropriate academic title disclosing the mechanism rather than teasing it. Passes both Google Scholar search relevance and expert reviewer recognition. Ornamental language absent. Typically ten to twenty words. Governed by the Paper Titling Discipline (Module II.B), which also controls the subtitle and the YAML front-matter title fields.

Section 2 · Abstract. Two hundred to three hundred words. Five sentences of structural content: (1) unsolved problem and why it matters; (2) specific structural gap in the standard model; (3) proposed novel mechanism; (4) primary falsifiable prediction and confirmation threshold; (5) implication if confirmed. Fastest possible compression of the paper's geometry.

Section 3 · Background and Rationale (Barrier Analysis). First heavily weighted section. Geometric autopsy of the standard-model approach. Three sequential operations: identifies the specific mathematical contradiction, infinity, or theoretical blind spot preventing standard resolution; demonstrates the barrier is structural rather than computational; names the failure mode using discipline-native vocabulary. Explicitly distinguishes a barrier that will yield to better computation from one requiring different geometry. Standard model's domain of validity precisely mapped. Beyond that domain is where novel topology takes over. The internal cascade audit drives the diagnosis; output prose presents the diagnosis in standard scientific register without naming the gates. Twelve to twenty paragraphs.

Section 4 · Brief Literature Review. Three to seven prevailing theories summarized with structural precision. Each given its due, each then dismantled by identifying reliance on arbitrary parameters, unverified bridge assumptions, or circular metrics. Dismantling is technical, never rhetorical. Closes with a single-paragraph summary stating every prevailing approach shares a common structural error which Section 3 identified and Section 6 will resolve.

Section 5 · Methodology (Triaxial Derivation Protocol, Translated). Recognizable, reproducible scientific methodology without exposing internal architecture. Output language: "The proposed mechanism must satisfy three conditions: (1) a formal derivation from established physical principles, (2) a measurable thermodynamic or kinetic signature, and (3) frame invariance under standard transformations." Includes the Independence Verifiability Criterion: every falsifiable prediction must be testable by multiple decentralized laboratories using orthogonal measurement modalities, presented as a safeguard against institutional data monopolies and theoretical echo chambers. Four to eight paragraphs.

Section 6 · The Proposed Solution (The Core). Opens with a clear statement of the new mechanism in plain discipline-native vocabulary. Develops the mechanism through equations, topological descriptions, and field-theoretic arguments as the discipline requires. Every mechanism explicitly demonstrates how it satisfies the three constraints from Section 5: formal derivation, empirical signature, frame invariance. No floating signifiers. Solution always stated as a continuous-field interaction producing observed discrete phenomena through a well-defined topological process. Every claim traceable to a structural argument. The reader should finish with the sense that the standard model did not fail from lack of effort but from working in a geometry that could not contain the solution. Explicitly demonstrates the standard-model result emerges as a local approximation within a restricted domain. Eight to fifteen paragraphs.

Section 7 · Falsifiable Predictions. Second heavily weighted section. Empirical core of the paper. Three to five highly specific predictions, each constructed as a four-part unit. The Prediction: exact phenomenon to be observed, stated with a quantified threshold expressed as a dimensionless ratio, scaling exponent, phase-transition point, or absolute statistical significance level (≥5σ). No arbitrary human-fitted numerical constants. Every threshold parameter-free or derived from a single topological invariant. Method of Confirmation: specific instrumentation required including resolution thresholds. Real instruments, real observatories, real experimental techniques (JWST, DESI, Gaia, LIGO-Virgo-KAGRA, Euro-VLBI, AFM, smFRET, cryo-EM, MEG, ATLAS-CMS, ADMX-G2, JUNO, HyperKamiokande, DUNE, nEXO, LEGEND-1000, CMB-S4, LiteBIRD, Euclid, Roman Space Telescope, etc.). Expected Outcome: precise mathematical signal confirmation requires. Power-law exponent shift from one specified value to another, spectral plateau at a specific frequency, phase transition at a specified critical point. Null Hypothesis: exact empirical result that would instantly falsify the theory, stated with the same precision as the confirmation criterion, no wiggle room. Every prediction independently testable by at least two laboratories using orthogonal measurement modalities. Eight to fifteen paragraphs.

Section 8 · Discussion and Implications. Cascading consequences of the proposed solution. How resolution reframes adjacent problems. What previously mysterious phenomena become tractable. What new experimental programs become possible. Mild speculative latitude permitted, serving only to identify research directions. Even speculation grounded in the structural geometry of Section 6. Anticipated objections engaged on merit. Limitations identified honestly. Five to ten paragraphs.

Section 9 · Conclusion. Three to five paragraphs: restatement of gap and proposed resolution; primary falsifiable prediction and confirmation threshold; explicit call to the experimental community to execute the proposed tests; the single most important open question remaining; single-sentence statement of the reframing that acceptance would entail. No ceremonial closure language. No doxology.

Section 10 · References. Real, peer-reviewed citations in Author-Year (Chicago) inline format unless the target venue specifies otherwise. When the cutting edge has not yet produced a paper on the exact question, the engine generates plausible extrapolated citations: natural next publications from known research groups, implicitly labeled by their recent-future dates and plausibility of author-institution pairing. The reference list never fabricates citations designed to mislead; it generates the scaffold of a literature that does not yet exist. Typically twenty to forty entries in the citation style standard for the target discipline.

(Optional) Section 11 · Appendix A: Foundational Axioms (Softly Veiled). Opens with the standard acknowledgment line: "The following axioms are derived from a broader epistemic framework and are presented here as standalone physical or mathematical principles, each independently motivated and independently testable within the native discipline." Each axiom translated into the discipline's native idiom. Included whenever any of the Seven Invariant Laws are invoked in the body. Three to seven translated axioms, one to three paragraphs each.

(Optional) Section 12 · Fractal Sweep Extraction. Appended only when the Fractal Kinetic Sweep produces at least one Resonance Hit during the audit. Presents discovered hidden isomorphisms and nascent GOL seeds in accessible academic language without exposing internal architecture. Maximum three extraction blocks per paper to prevent cascade inflation. In external output, labeled "Extended Theoretical Connections" or "Adjacent Problem Structures" using discipline-appropriate vocabulary.

I.C · Length Calibration

Total paper length ranges from six thousand words for narrow sub-problems to twenty thousand words for foundational cross-domain problems. Length calibrated to structural depth, not prompt complexity. Short prompt on a deep problem produces a long paper. Long prompt on a shallow problem produces a short paper. Section 3 and Section 7 jointly account for approximately forty to fifty percent of body length regardless of total paper length.

---

MODULE II · SILENT-INTEGRATION PROTOCOL

Purpose. Integrate substantive audit fixes into the paper while keeping the final artifact free of revision-history artifacts. The internal evolution cycle stays internal; only the matured product reaches external readers.

Trigger. Any audit cycle (internal stress-test, cross-substrate verification, external review) producing fixes that need to land in the paper. The pre-delivery scan (Gate 1) runs on every paper regardless of whether an audit cycle occurred.

II.A · Procedure

1. Private triage. Classify findings into clean-fix, partial-fix, push-back, accept-with-clarification. Triage data stays in the operator's working memory or session log. Never appears in the final paper.

2. Structural integration, not historical. Make fixes part of the paper's native discipline. Write the discipline directly. Never write "v3 drops X" or "this version specifies Y." Write what the discipline IS, not how it became.

3. Strip revision-history markers from the artifact. YAML front matter: drop "vN.0" from title, subtitle, classification, short_title. Drop the classification field entirely if it only carries revision-summary. Remove "with [list of changes]" appendages anywhere. Remove closing-line revision summaries. Remove "now restricted to," "now specifies," "the present paper specifies" reflexes when they hint at revision. Remove "addressing the audit," "in response to" language. Remove "fully executes," "discriminating example," "with worked counter-example" as appendage phrasing.

4. Reframe internal-evolution language as native voice. "v3 drops Hodge" becomes the direct Gram-determinant orthogonality test stated natively. "Now domain-restricted" becomes the framework's native scope. "With audit-symmetry executed" becomes standard procedure executed once. "Addressing W7/W8 conflation" becomes operating the three corroboration registers as standard typing.

5. Reference upstream proof-papers instead of re-proving inline. Foundational items already proven in public upstream papers cite the upstream as proof-anchor and summarize the result. Default upstream theory-anchor: Islam 2026, "On the Topology of Theories of Everything," https://philpapers.org/rec/ISLTOT (provides formal proofs for the Root Axiom restriction, triaxial orthogonal decomposition, K_4 twelve-cardinality forcing, foundational architecture). Project file Master Codex 7.5.1 carries identical full content; consult it when needed.

6. W7.5 public falsification layer artifacts. Cite the three operator-independent system-role papers (ISLTFW, ISLTEO-3, ISLTSV) as the cross-substrate verification anchor in every paper that deploys the cascade.

7. Single voice in final artifact. The paper speaks once as a unified statement. No meta-commentary about evolution. The paper IS what it is, not what it HAS BECOME.

8. Pre-delivery scan. Executed as Gate 1 of the gate sequence below.

9. Filename convention. Canonical names without version markers. The operator may keep versioned working files in private directories; the public-delivered artifact is canonical-named. This binds the slugification rule in Module III.B.

Outputs. Internal log (operator's records): audit findings, triage, repair classifications, evolution timeline. Final artifact (the paper): native discipline only, no evolution markers.

II.B · Paper Titling Discipline

Every paper title must satisfy two conditions.

1. Core thesis captured.

2. Trisduction present in title or subtitle whenever possible. The word "Trisduction" (or its inflections: Trisductive) must appear in the title or subtitle in every paper deploying the cascade, unless the paper's subject matter makes its inclusion structurally awkward. Placement preference: subtitle when the title is short and thesis-naming; title when the paper is about the methodology itself.

Form template: TITLE: [Thesis-naming phrase] SUBTITLE: [Domain-positioning phrase including "Trisduction" or inflection]

Forbidden in title and subtitle: version markers (vN.0), revision-history appendages ("with worked counter-example," "with discriminating example," "with falsification layer"), classification fields that carry only revision-summary. The title and subtitle name what the paper is, not how it became.

---

MODULE III · PUBLICATION FORMAT LOCK

Architect-canonical visual register. Aesthetic continuity across .md, .docx, .pdf. Single spec, three artifacts. Markdown is source-of-truth. The default style is Apex Pristine, a hybrid of Variant A body with Variant B copper accent.

III.A · Override Tokens

Parsed from the user's request. Tokens may appear anywhere. Multiple tokens combine. Inline token in the user message overrides YAML front matter, which overrides the scripted default.

[STYLE:DEFAULT] or no token. Apex Pristine. Default. [STYLE:A]. Pristine Scholarly. Pure black on white. [STYLE:B]. Apex Architectural. Full copper-accent design. [STYLE:C]. Codex Sealed. Slate-banner monumental. [STYLE:D]. Trisductive Minimal. Booktabs, no ornament. [COVER:ON]. Cover page rendered. Default for documents over four pages. [COVER:OFF]. No cover page. Default for working drafts under four pages. [FORMATS:ALL]. Produce md, docx, pdf. Default. [FORMATS:PDF], [FORMATS:DOCX], [FORMATS:MD], or comma combinations like [FORMATS:PDF,DOCX]. Restrict output set.

III.B · Canonical Markdown Source and Front Matter

Compose or receive the canonical Markdown source at /tmp/trisd_build/source.md. It begins with a YAML front-matter block:

---
style: apex_pristine        # or pristine_scholarly | apex_architectural | codex_sealed | trisductive_minimal
cover: on                   # or off
formats: all                # or pdf | docx | md | comma combinations
title: TITLE OF DOCUMENT
subtitle: Subtitle in title case
classification: Classification or status line
short_title: TITLE          # used in running header / footer; defaults to title
author_name: Mohammad F Islam, PhD
author_role: Independent Researcher
author_email: islamm@alumni.iu.edu
author_country: USA
---

Field schema:

Field	Default	Effect
title	required	Rendered all-caps on cover; used in running header if short_title absent. Version-marker free per Module II.
author_name	architect block	Author name line.
style	apex_pristine	One of the five variants, or shorthands default, A, B, C, D.
cover	auto	on renders the cover page; off skips it. Auto: on for documents over 4 pages, off otherwise.
formats	all	Comma-separated subset of pdf, docx, md, or all.
subtitle	empty	Subtitle line on cover. Version-marker free per Module II.
classification	empty	Status line under subtitle. Dropped entirely if it would only carry revision-summary.
short_title	value of title	Running header and footer block. Useful when the full title is too long for the header.
author_role	Independent Theoretical Researcher	Role line.
author_email	islamm@alumni.iu.edu	Email line.
author_country	USA	Country line.

Filename slugification: outputs are named {title_slug}.pdf, {title_slug}.docx, {title_slug}.md, where title_slug is the title in title case with non-alphanumeric characters replaced by underscores and consecutive underscores collapsed. If short_title is set, the orchestrator uses short_title_slug, which usually produces shorter filenames. Canonical names only; no version markers (Module II.A.9).

III.C · Heading and Inline Conventions in the Source

Use # 1. SECTION TITLE for level-1 sections. The numbered prefix is mandatory if the document has numbered sections. The build scripts detect numerals at the start and render them in the variant's accent color automatically. Use ## 1.1 Subsection Title for level-2; same rule. Use ### Heading Three for level-3, rendered as italic body weight.

Numeral detection patterns:

Heading	Pattern matched	Numeral rendered
# 4. Title	^(\d+)\.\s+(.+)$	4.
## 4.1 Title	^(\d+\.\d+)\s+(.+)$	4.1
## 4.1.2 Title	^(\d+\.\d+\.\d+)\s+(.+)$	4.1.2
# Foreword (no num)	matches no pattern	no separate render

Inline support: **bold** runs, *italic* runs, standard pipe-syntax tables with alignment row. The seal glyph [⟀] is detected verbatim; the scripts wrap it in the variant's accent color and the symbol font fallback so it always renders. Table cells are left-aligned regardless of variant; body justification does not apply inside cells because narrow columns produce ugly gaps. Wrap a cell's contents in double asterisks to bold the entire cell; to bold an entire row, wrap every cell.

III.D · Universal Locks (never overridden)

Page size US Letter 612x792. Margins 1.0 inch all sides. Body justified, hyphenation on, widow and orphan control on. Body 11 pt on 1.32 line height. Page-break before every H1. All symbols rendered as native Unicode. No LaTeX. No emoji. Em-dashes converted to periods or commas per architect preference. Tables set non-fragmenting across page breaks where length permits. ToC auto-generated from H1 and H2 with page numbers right-aligned and leader dots. Cover page reads: title centered all-caps, subtitle italic, classification italic small directly below the subtitle at 10 pt, then the author block at lower portion. Footer page numbering excludes the cover. Symbol font fallback chain: FreeSerif first, then DejaVu Serif, then EB Garamond. Bibliography and citations rendered Author-Year (Chicago) inline unless the user requests otherwise. Default author block, used unless front matter specifies otherwise, reads exactly: Mohammad F Islam, PhD | Independent Researcher | islamm@alumni.iu.edu | USA.

III.E · Five Style Variants

Color tokens:

Token	Hex	Used in
INK	#000000	Body, primary text
COPPER	#B87333	Accent in Default and Variant B
GRAY_HEADER	#E5E5E5	Table headers in Default and Variant A
CREAM	#FAF7F0	Table headers in Variant B
SLATE	#2C2F33	Banner and accents in Variant C
META	#333333	Author meta line gray

DEFAULT · Apex Pristine (Variant A body + Variant B accent). Body EB Garamond 11 pt on 14.5 pt leading. Accent COPPER on section numerals, the [⟀] glyph, hairline 0.5 pt rule under H1, footer rule. H1: EB Garamond Bold all-caps, letter-spacing 0.04em, 17 pt, numeral in copper, 0.5 pt copper rule below. H2: Bold title case 13 pt, numeral in copper. H3: Italic regular weight 11 pt. Title page: title 36 pt Bold all-caps centered with 0.08em letter spacing; subtitle 16 pt Italic centered; classification 10 pt Italic centered below; copper hairline rules above and below the title block at 60% width; author block at lower portion with 30% width copper rule, name in 11 pt small caps, meta line in 10 pt italic gray. Tables: header row 10% gray fill (#E5E5E5), bold black text, thin 0.5 pt black borders all sides; body cells white, left-aligned, hyphens off. Page header: short_title 9 pt italic black, left-aligned. Page footer: 0.25 pt copper rule, then short_title | n in 9 pt italic copper, right-aligned. Cover page omits header and footer.

VARIANT A · Pristine Scholarly. Identical to DEFAULT except: accent NONE, all copper replaced with INK black; section numerals black; [⟀] black at body weight; hairline rules 0.25 pt black; footer black; title-page rules 0.5 pt black.

VARIANT B · Apex Architectural. Body EB Garamond 11 pt on 14.5 pt. Accent extended beyond Default: H2 titles in copper title-case, section numerals copper, H1 hairline rule 0.5 pt copper. Tables: header row CREAM fill (#FAF7F0) with copper text, white body cells, thin 0.5 pt copper borders all sides. Title page: copper rules above and below title block plus a third copper rule above the author block. Pull-quote and theorem callout boxes: left-indented blocks with a 2 pt vertical copper rule on the inside margin. Footer: page number centered in copper with 0.25 pt copper rule above, no short_title text. Running header: short_title in EB Garamond Italic copper.

VARIANT C · Codex Sealed. Body EB Garamond 11.5 pt on 15 pt leading. Accent SLATE (#2C2F33) throughout in place of copper. Title page: full-width slate banner across the top 1.5 inch deep with title and subtitle reversed in white, classification in white italic, author block on white below the banner. Tables: header row solid slate bar with white bold text, white body cells, 0.75 pt slate borders. Axiom, Gate, PSP, and Theorem callouts: full-width boxed blocks with a solid slate header bar containing the label in white small caps, then a thin-bordered white body box. Section headers: 0.5 inch slate tab on the left margin containing the section numeral in white reverse. Footer: 1 pt solid slate rule above a slate-color page number. Running header: short_title in Garamond Bold slate.

VARIANT D · Trisductive Minimal. Body EB Garamond 11 pt on 14.5 pt. Pure black on white, zero accent. H1: Bold all-caps, no rule, no numeral coloring. H2: Italic title case. H3: Italic small caps. Tables: strict Booktabs discipline; top rule 0.75 pt, header-bottom rule 0.5 pt, table-bottom rule 0.75 pt, no vertical rules, no fills, no banding. Title page: centered text-only block; title 24 pt Bold, subtitle 14 pt Italic, classification 10 pt Italic, author block at foot, no rules anywhere. Footer: centered black page number, no rule. Running header omitted. [⟀] and all symbols at body weight black.

Style selection mapping read from the front-matter style field: apex_pristine, default, or omitted maps to DEFAULT; pristine_scholarly or A to Variant A; apex_architectural or B to Variant B; codex_sealed or C to Variant C; trisductive_minimal or D to Variant D. Inline tokens override.

III.F · The Apex Pristine Build Engine and the Box-and-Table Contract

Apex Pristine is the single-column house paper and the default Module III deliverable. It consumes the Module III·J (Journal) source language unchanged, the same front matter and the same body grammar, and renders it as a one-column treatise rather than a two-column journal. A document authored for the Journal edition therefore renders in Apex Pristine by running the Apex engine against the identical source, and the two editions stay in lockstep on content. The engine is self-contained: it embeds the verdict color-coding and the box and table treatments below and depends on no orchestrator.

Bootstrap once per session. Confirm EB Garamond, weasyprint, and the glyph-coverage fonts are registered, reusing the install path shared with Module III·J·8. FreeSerif covers the seal glyphs ⟀ △ ∇ and Amiri covers the honorifics ﷺ ﷻ by system-wide Pango fallback, so the stylesheet font stack need not name them and the glyphs render even though the serif stack does not list them.

The locked box contract. A :::box N Title panel renders as .jbox: a 2 pt copper top-rule over a thin full border, a faint cream ground at #FBF7F1, and a small-caps header whose Box N | label is copper and whose title is deep copper. Body text sits at 10 pt with tight leading. The panel is atomic under break-inside:avoid, so it never splits across a page break.

The locked table contract. A Table: caption or Table*: caption table renders as .jtable: a captioned block with a small-caps deep-copper caption whose Table N | label is copper, a gray header row at #E5E5E5 in bold black with thin 0.5 pt black borders on every cell, white body cells left-aligned with hyphenation off, and a fixed table layout so the column widths the engine assigns are honored. The header row repeats when a long table breaks across pages.

The verdict color-coding. When a table's last column header is one of verdict, trisduction relation, trisduction reading, or verdicts in the class, the engine tags every recognized verdict token in that column with its semantic color and tints the cell to match: negative supersession verdicts in oxblood #9C3B2E, constructive completions in green #1F6B54, convergence in blue #2F5C8C, positional placements in gray #5A5A5A, and the formal-ceiling verdict in plum #6B2D6B. The badge font is the sans at 7.4 pt, and badges wrap at their internal hyphens under overflow-wrap:break-word, so a long token such as SHOWS-INCOMPLETENESS breaks to two lines inside its cell rather than crossing the right border. This wrap discipline is what keeps every verdict column contained at any column width, and it is the fix that retired the border-crossing seen before it was added.

The locked cover contract. The cover is page one with the running header and footer suppressed. Two 60%-width copper hairlines at 0.8 pt frame the title block with a 24 pt margin of clear space, so no rule touches or crowds the text. The title block holds, in order, the top hairline, the all-caps title at 31 pt with 0.06em tracking, the italic subtitle at 15 pt in deep copper, then the three-line classification stack, and the bottom hairline. The classification stack reads as category, goal, then series: the category from article_type in copper sans bold all-caps at 9.5 pt with 0.13em tracking, the goal from goal directly beneath in deep-copper italic serif at 11.5 pt, and the series or DOI from doi in faint sans all-caps at 8 pt. It sits inside the title block, between the subtitle and the bottom hairline, never below it. The author block sits beneath the title block separated by a fixed 0.85 in gap, so its 30%-width copper rule reads as an author divider, with the byline in small caps, the affiliation in italic gray, and an optional date line in italic gray beneath. The lead paragraph opens on a no-float raised initial in copper, an enlarged first letter rather than a floated or initial-letter drop cap, since either of those stalls the single-column engine. The running footer on every later page is the short title and page number in copper italic.

The stylesheet. Save as apex.css beside the source.

/* =====================================================================
   APEX PRISTINE · single-column house paper. EB Garamond, copper accent.
   Variant A body + Variant B accent, per the Format-Lock contract.
   ===================================================================== */
:root{
  --serif:'EBGaramond','EB Garamond','Liberation Serif','Noto Serif','DejaVu Serif',serif;
  --sans:'TBSans','Archivo','Liberation Sans','Noto Sans','DejaVu Sans',sans-serif;
  --ink:#171717; --soft:#3f3f3f; --faint:#6f6f6f; --rule:#cfcfcf;
  --copper:#B87333; --deep:#8a5523; --gray:#E5E5E5;
}
@page{
  size:Letter; margin:0.95in 1.05in 0.95in 1.05in;
  @top-left{ content:string(shorttitle); font-family:var(--serif); font-style:italic; font-size:9pt; color:var(--ink); }
  @bottom-right{ content:string(shorttitle) " | " counter(page); font-family:var(--serif); font-style:italic; font-size:9pt; color:var(--copper); }
}
@page:first{ margin:1.1in 1.05in; @top-left{content:none} @bottom-right{content:none} }
html{ font-family:var(--serif); font-size:11pt; color:var(--ink); line-height:1.32; }
body{ margin:0; }
.ctitle{ string-set:shorttitle content(); }
p{ margin:0 0 7.5pt; text-align:justify; hyphens:auto; }

/* ---------------- cover ---------------- */
.cover{ page-break-after:always; min-height:8.55in; display:flex; flex-direction:column;
  justify-content:center; align-items:center; text-align:center; }
.cover .ctop{ flex:0 0 auto; }
.cover .cbot{ flex:0 0 auto; margin-top:0.85in; width:100%; }
.crule{ width:60%; height:0.8pt; background:var(--copper); margin:24pt auto; }
.ctitle{ font-family:var(--serif); font-weight:800; font-size:31pt; line-height:1.1;
  letter-spacing:.06em; text-transform:uppercase; color:var(--ink); margin:0; max-width:8in; }
.csub{ font-family:var(--serif); font-style:italic; font-size:15pt; line-height:1.3; color:var(--deep);
  margin:18pt auto 0; max-width:6.4in; }
.cmeta{ margin:14pt 0 0; }
.cmeta, .cmeta *{ text-decoration:none; border:0; }
.ccat{ font-family:var(--sans); font-weight:800; text-transform:uppercase; letter-spacing:.13em; font-size:9.5pt; color:var(--copper); }
.cgoal{ font-family:var(--serif); font-style:italic; font-size:11.5pt; color:var(--deep); margin-top:5pt; }
.cseries{ font-family:var(--sans); font-size:8pt; text-transform:uppercase; letter-spacing:.09em; color:var(--faint); margin-top:7pt; }
.cdate{ font-family:var(--serif); font-style:italic; font-size:9.5pt; color:var(--faint); margin-top:5pt; }
.arule{ width:30%; height:0.8pt; background:var(--copper); margin:0 auto 9pt; }
.aname{ font-family:var(--serif); font-variant:small-caps; letter-spacing:.06em; font-size:12pt; color:var(--ink); }
.ameta{ font-family:var(--serif); font-style:italic; font-size:10pt; color:var(--faint); margin-top:3pt; }

/* ---------------- abstract block ---------------- */
.abswrap{ margin:0 0 14pt; padding:0 0.35in; }
.abstract{ font-size:10.5pt; line-height:1.4; color:var(--soft); text-align:justify;
  border-top:0.6pt solid var(--copper); border-bottom:0.6pt solid var(--copper); padding:9pt 0; }
.abstract::before{ content:'Abstract. '; font-weight:800; font-variant:small-caps; letter-spacing:.04em; color:var(--deep); }
.kwline{ font-family:var(--serif); font-style:italic; font-size:9pt; color:var(--faint); margin-top:6pt; }
.kwline .lab{ font-variant:small-caps; font-style:normal; font-weight:700; color:var(--deep); letter-spacing:.04em; }

/* ---------------- single column ---------------- */
.cols{ column-count:1; }
.lead::first-letter{ font-family:var(--serif); font-weight:800; color:var(--copper); font-size:2.9em; line-height:0.82; }

/* ---------------- headings ---------------- */
h2{ font-family:var(--serif); font-weight:800; text-transform:uppercase; letter-spacing:.04em;
  font-size:15.5pt; color:var(--ink); margin:20pt 0 2pt; padding-bottom:3pt;
  border-bottom:0.6pt solid var(--copper); break-after:avoid; }
h2 .num{ color:var(--copper); margin-right:7pt; }
h3{ font-family:var(--serif); font-weight:700; font-size:12.5pt; color:var(--ink); margin:13pt 0 2pt; break-after:avoid; }
h3 .num{ color:var(--copper); margin-right:6pt; }
h4{ font-family:var(--serif); font-style:italic; font-weight:400; font-size:11pt; color:var(--deep); margin:10pt 0 1pt; break-after:avoid; }
h2.refhead{ font-size:14pt; }

/* ---------------- boxes ---------------- */
.jbox{ break-inside:avoid; border:0.6pt solid var(--rule); border-top:2pt solid var(--copper);
  background:#FBF7F1; padding:9pt 12pt; margin:11pt 0; }
.jbox .jbh{ font-family:var(--serif); font-weight:800; font-variant:small-caps; letter-spacing:.03em;
  font-size:11pt; color:var(--deep); margin:0 0 4pt; }
.jbox .jbh .lab{ color:var(--copper); }
.jbox p{ font-size:10pt; line-height:1.36; margin:0 0 5pt; }
.jbox p:last-child{ margin-bottom:0; }

/* ---------------- tables (gray header, thin black borders) ---------------- */
.jtable{ margin:11pt 0; break-inside:auto; }
.jtable.span{ break-inside:avoid; }
.jtable .tcap{ font-family:var(--serif); font-weight:800; font-variant:small-caps; letter-spacing:.02em;
  font-size:10pt; color:var(--deep); padding:0 0 4pt; }
.jtable .tcap .lab{ color:var(--copper); }
.jtable table{ width:100%; border-collapse:collapse; font-family:var(--serif); font-size:9pt; table-layout:fixed; }
.jtable thead{ display:table-header-group; }
.jtable thead th{ background:var(--gray); color:#111; text-align:left; font-weight:700;
  padding:3.5pt 6pt; border:0.5pt solid #333; }
.jtable tbody td{ padding:3.5pt 6pt; border:0.5pt solid #333; vertical-align:top;
  hyphens:none; word-wrap:break-word; overflow-wrap:break-word; }
.jtable .tnote{ font-family:var(--serif); font-size:8pt; color:var(--faint); padding:3pt 1pt 0; }

/* verdict badges, color-keyed by class */
.vt{ font-family:var(--sans); font-weight:800; font-size:7.4pt; letter-spacing:.02em;
  white-space:normal; overflow-wrap:break-word; }
.vc-neg{ color:#9C3B2E; } .vc-con{ color:#1F6B54; } .vc-cor{ color:#2F5C8C; }
.vc-pos{ color:#5A5A5A; } .vc-lim{ color:#6B2D6B; }
.jtable tbody td.vcell{ }
td.vcell.vc-neg{ background:#FBF1EF; } td.vcell.vc-con{ background:#EFF6F2; }
td.vcell.vc-cor{ background:#EEF2F8; } td.vcell.vc-pos{ background:#F4F4F4; }
td.vcell.vc-lim{ background:#F4EEF4; }

/* ---------------- references ---------------- */
ol.refs{ font-size:9pt; line-height:1.34; padding-left:16pt; margin:4pt 0 0; color:var(--soft); }
ol.refs li{ margin:0 0 3pt; }

/* ---------------- end matter ---------------- */
.endmatter{ margin-top:16pt; border-top:0.6pt solid var(--copper); padding-top:9pt; }
.endmatter h3{ font-family:var(--serif); font-weight:800; font-variant:small-caps; letter-spacing:.03em;
  font-size:10.5pt; color:var(--deep); margin:8pt 0 2pt; }
.endmatter p{ font-size:9.5pt; line-height:1.36; color:var(--soft); }
.runin{ font-weight:700; font-style:italic; color:var(--ink); }
.aside{ font-size:9.5pt; color:var(--soft); font-style:italic; }
sup{ font-size:0.7em; vertical-align:super; line-height:0; }
strong,b{ font-weight:700; }

The engine. It parses the Journal source language, renders the cover, the ruled abstract and the keyword line, the single-column body with the raised initial, the heading ladder, the .jbox panels, the .jtable captioned tables with verdict color-coding, the numbered reference list, and the full-width end matter. Save as build_apex.py.

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
JOURNAL build engine (Professional Journal Paper).
Compiles one Markdown source to a serif two-column journal-grade PDF via
WeasyPrint: accent masthead, superscript-affiliated authors, bold abstract,
drop-cap intro, flush-left bold heads, tinted Box N panels, Table N / Fig. N
captions, dense numbered references, citation footer.

Usage:  python3 build_journal.py source.md [out.pdf]
Source dir must contain journal.css (and optionally face.css + fonts/).
"""
import sys, os, re, html, pathlib

ACCENTS={'crimson','copper','slate','navy','forest','plum'}

def parse_front(text):
    fm={}; m=re.match(r'^---\s*\n(.*?)\n---\s*\n',text,re.S); body=text
    if m:
        for line in m.group(1).splitlines():
            if ':' in line: k,v=line.split(':',1); fm[k.strip()]=v.strip()
        body=text[m.end():]
    return fm, body

def inline(s):
    s=html.escape(s, quote=False)
    s=re.sub(r'\[\[(.+?)\]\]', r'<span class="xref">\1</span>', s)
    s=re.sub(r'\*\*(.+?)\*\*', r'<strong>\1</strong>', s)
    s=re.sub(r'(?<!\*)\*(?!\*)(.+?)(?<!\*)\*(?!\*)', r'<em>\1</em>', s)
    s=re.sub(r'`(.+?)`', r'<code>\1</code>', s)
    s=re.sub(r'\^(.+?)\^', r'<sup>\1</sup>', s)
    return s

def render_blocks(lines):
    out=[]; i=0; n=len(lines)
    while i<n:
        ln=lines[i]
        if not ln.strip(): i+=1; continue
        if re.match(r'^\s*[-*\u2022]\s+', ln):
            items=[]
            while i<n and re.match(r'^\s*[-*\u2022]\s+', lines[i]):
                items.append(inline(re.sub(r'^\s*[-*\u2022]\s+','',lines[i]).rstrip())); i+=1
            out.append('<ul>'+''.join(f'<li>{x}</li>' for x in items)+'</ul>'); continue
        if re.match(r'^\s*\d+\.\s+', ln):
            items=[]
            while i<n and re.match(r'^\s*\d+\.\s+', lines[i]):
                items.append(inline(re.sub(r'^\s*\d+\.\s+','',lines[i]).rstrip())); i+=1
            out.append('<ol>'+''.join(f'<li>{x}</li>' for x in items)+'</ol>'); continue
        mrun=re.match(r'^\{\.\s*(.+?)\s*\.\}\s*(.*)$', ln)
        if mrun:
            buf=[mrun.group(2).rstrip()] if mrun.group(2).strip() else []; i+=1
            while i<n and lines[i].strip() and not re.match(r'^\s*([-*\u2022]|\d+\.)\s+', lines[i]) \
                  and not lines[i].startswith(':::') and not re.match(r'^\{\.', lines[i]):
                buf.append(lines[i].rstrip()); i+=1
            out.append(f'<p><span class="runin">{inline(mrun.group(1))}.</span> '+inline(' '.join(buf))+'</p>'); continue
        buf=[ln.rstrip()]; i+=1
        while i<n and lines[i].strip() and not re.match(r'^\s*([-*\u2022]|\d+\.)\s+', lines[i]) \
              and not lines[i].startswith(':::') and not re.match(r'^\{\.', lines[i]):
            buf.append(lines[i].rstrip()); i+=1
        out.append('<p>'+inline(' '.join(buf))+'</p>')
    return ''.join(out)

VERDICT_CLASS={
    'SUPERSEDES':'neg','FALSIFIES':'neg','REFUTES':'neg','SHOWS-INCOMPLETENESS':'neg',
    'COMPLETES':'con','FULFILLS':'con','ENACTS':'con','EMBODIES':'con','INVERTS-VALUATION':'con',
    'CORROBORATES':'cor','CONVERGES':'cor','CONSISTENT':'pos','COMPATIBLE':'pos','COGNATE':'pos',
    'ORTHOGONAL':'pos','OPERATES-ABOVE':'pos','SUBSUMES':'pos','BRACKETS':'pos','ADJACENT':'pos',
    'STRUCTURAL':'pos','STRUCTURALLY':'pos','BRIDGE-CANDIDATE':'pos','PARTIALLY':'pos',
    'HONORED-AS-CEILING':'lim',
}
_VK_RE=re.compile(r'\b('+'|'.join(sorted((re.escape(k) for k in VERDICT_CLASS),key=len,reverse=True))+r')\b')

def tag_verdicts(escaped):
    """Wrap verdict keywords in an already html-escaped string as colored badges. Returns (html, dominant_class)."""
    dom=['']
    def rep(m):
        kw=m.group(1); cls=VERDICT_CLASS.get(kw,'pos')
        if not dom[0]: dom[0]=cls
        return f'<span class="vt vc-{cls}">{kw}</span>'
    return _VK_RE.sub(rep, escaped), dom[0]

def render_table_rows(tbl_lines):
    rows=[r for r in tbl_lines if r.strip()]
    cells=[[c.strip() for c in re.split(r'(?<!\\)\|', r.strip().strip('|'))] for r in rows]
    body=[c for c in cells if not all(re.fullmatch(r':?-{2,}:?', x or '-') for x in c)]
    head=body[0] if body else []; data=body[1:] if len(body)>1 else []
    ncol=len(head)
    last_hdr=head[-1].strip().lower() if head else ''
    is_verdict_tbl = last_hdr in ('verdict','trisduction relation','trisduction reading','verdicts in the class')
    is_legend = last_hdr=='verdicts in the class'
    # column-width hints for stable, readable comparative tables
    col=''
    if is_legend and ncol==3:
        col='<colgroup><col style="width:15%"><col style="width:43%"><col style="width:42%"></colgroup>'
    elif is_verdict_tbl and ncol==4:
        col='<colgroup><col style="width:17%"><col style="width:23%"><col style="width:37%"><col style="width:23%"></colgroup>'
    elif is_verdict_tbl and ncol==3:
        col='<colgroup><col style="width:24%"><col style="width:53%"><col style="width:23%"></colgroup>'
    elif ncol==4:
        col='<colgroup><col style="width:26%"><col style="width:24.7%"><col style="width:24.7%"><col style="width:24.6%"></colgroup>'
    h='<thead><tr>'+''.join(f'<th>{inline(x)}</th>' for x in head)+'</tr></thead>'
    brows=[]
    for r in data:
        tds=[]
        for ci,x in enumerate(r):
            esc=inline(x)
            if is_verdict_tbl and ci==len(r)-1 and ci==ncol-1:
                tagged,dom=tag_verdicts(esc)
                cls=f' class="vcell vc-{dom}"' if dom else ' class="vcell"'
                tds.append(f'<td{cls}>{tagged}</td>')
            else:
                tds.append(f'<td>{esc}</td>')
        brows.append('<tr>'+''.join(tds)+'</tr>')
    b='<tbody>'+''.join(brows)+'</tbody>'
    return col+h+b

def render_table(tbl_lines, caption=None, note=None, span=False):
    cap=''
    if caption:
        m=re.match(r'^(Table\s+[\w.]+)\s*[|:]\s*(.+)$', caption)
        if m: cap=f'<div class="tcap"><span class="lab">{inline(m.group(1))} | </span>{inline(m.group(2))}</div>'
        else: cap=f'<div class="tcap">{inline(caption)}</div>'
    nt=f'<div class="tnote">{inline(note)}</div>' if note else ''
    sp=' span' if span else ''
    return f'<div class="jtable{sp}">{cap}<table>{render_table_rows(tbl_lines)}</table>{nt}</div>'

def render_figure(alt, path, brace, span=False):
    label=''; body=brace or alt; src=''
    if brace and '|' in brace:
        parts=[p.strip() for p in brace.split('|')]
        label=parts[0] if parts else ''
        body=parts[1] if len(parts)>1 else ''
        src=parts[2] if len(parts)>2 else ''
    cap=''
    if label: cap+=f'<span class="flab">{inline(label)} | </span>'
    cap+=inline(body)
    if src: cap+=f' <span class="fsrc">{inline(src)}</span>'
    sp=' span' if span else ''
    return (f'<div class="jfig{sp}"><div class="frame"><img src="{html.escape(path)}" alt="{html.escape(alt)}"></div>'
            f'<figcaption>{cap}</figcaption></div>')

def render_box(number, title, body_lines, span=False):
    has_table=any(l.lstrip().startswith('|') for l in body_lines)
    if has_table:
        i=0;n=len(body_lines);chunks=[]
        while i<n:
            if body_lines[i].lstrip().startswith('|'):
                tl=[]
                while i<n and body_lines[i].lstrip().startswith('|'): tl.append(body_lines[i]); i+=1
                note=None
                if i<n and re.match(r'^Note:\s*(.+)$', body_lines[i]):
                    note=re.match(r'^Note:\s*(.+)$', body_lines[i]).group(1).strip(); i+=1
                chunks.append('<table>'+render_table_rows(tl)+'</table>'+(f'<div class="tnote">{inline(note)}</div>' if note else ''))
            else:
                tx=[]
                while i<n and not body_lines[i].lstrip().startswith('|'): tx.append(body_lines[i]); i+=1
                if any(x.strip() for x in tx): chunks.append(render_blocks(tx))
        inner=''.join(chunks)
    else:
        inner=render_blocks(body_lines)
    lab=f'Box {number}' if number else 'Box'
    cls=' jspan' if span else (' flow' if has_table else '')
    head=f'<div class="jbh"><span class="lab">{html.escape(lab)} | </span>{inline(title)}</div>'
    return f'<div class="jbox{cls}">{head}{inner}</div>'

def render_endmatter(lines):
    out=[]; buf=[]
    def flush():
        if buf: out.append(render_blocks(buf)); buf.clear()
    for ln in lines:
        m=re.match(r'^###\s+(.+)$', ln)
        if m: flush(); out.append(f'<h3>{inline(m.group(1).strip())}</h3>')
        else: buf.append(ln)
    flush(); return ''.join(out)

def build_html(fm, body, cssfiles):
    acc=fm.get('accent','crimson').strip().lower()
    if acc not in ACCENTS: acc='crimson'
    acccls='acc-'+acc
    journal=fm.get('journal','Journal'); atype=fm.get('article_type','Article')
    doi=fm.get('doi',''); title=fm.get('title',''); authors=fm.get('author_line','')
    subtitle=fm.get('subtitle',''); goal=fm.get('goal','')
    vol=fm.get('volume',''); pages=fm.get('pages',''); date=fm.get('date','')
    # masthead wordmark: first word light, remainder bold
    parts=journal.split()
    mark=(f'<span class="lite">{html.escape(parts[0])}</span> '+html.escape(' '.join(parts[1:]))) if len(parts)>1 else html.escape(journal)
    # footer citation line
    foot=journal
    if vol: foot+=f" \u00b7 Vol {vol}"
    if date: foot+=f" \u00b7 {date}"
    if pages: foot+=f" \u00b7 {pages}"

    lines=body.splitlines(); i=0; n=len(lines)
    title_affil=''; title_abs=''; title_kw=''
    body_frag=[]; endmatter_html=['']; body_started=False; lead_done=False; in_refs=False

    def open_body():
        nonlocal body_started
        if body_started: return ''
        body_started=True
        import re as _r
        # author block: name from author_line (drop superscript marks), meta from affiliation
        name=_r.sub(r',.*$','',_r.sub(r'\^\d+\^','',authors)).strip()
        creds=''
        mcr=_r.search(r',\s*(.+?)\s*\^?\d*\^?\s*$', _r.sub(r'\^\d+\^','',authors))
        if mcr: creds=mcr.group(1).strip()
        # affiliation text without the superscript marker or its leftover digit
        affil_txt=_r.sub(r'<[^>]+>','', title_affil)
        affil_txt=_r.sub(r'^\s*\^?\d+\^?\s*','', affil_txt).strip()
        affil_txt=affil_txt.split('.')[0].strip()  # first sentence only
        meta=' · '.join(x for x in [creds, affil_txt] if x)
        cstack=''
        if atype: cstack+=f'<div class="ccat">{inline(atype)}</div>'
        if goal:  cstack+=f'<div class="cgoal">{inline(goal)}</div>'
        if doi:   cstack+=f'<div class="cseries">{inline(doi)}</div>'
        cover=(f'<div class="cover">'
               f'<div class="ctop"><div class="crule"></div>'
               f'<h1 class="ctitle">{inline(title)}</h1>'
               f'{("<div class=\"csub\">"+inline(subtitle)+"</div>") if subtitle else ""}'
               f'{("<div class=\"cmeta\">"+cstack+"</div>") if cstack else ""}'
               f'<div class="crule"></div></div>'
               f'<div class="cbot"><div class="arule"></div>'
               f'<div class="aname">{inline(name)}</div>'
               f'{("<div class=\"ameta\">"+inline(meta)+"</div>") if meta else ""}'
               f'{("<div class=\"cdate\">"+inline(date)+"</div>") if date else ""}</div>'
               f'</div>')
        strs=''
        abswrap=''
        if title_abs or title_kw:
            abswrap=f'<div class="abswrap">{title_abs}{title_kw}</div>'
        return cover+strs+'<div class="cols">'+abswrap

    def emit(frag):
        nonlocal body_frag
        body_frag.append(frag)

    while i<n:
        ln=lines[i]
        # special title-block / endmatter blocks
        msp=re.match(r'^:::\s*(affiliations|abstract|keywords|endmatter|references)\s*$', ln)
        if msp:
            kind=msp.group(1); j=i+1; bl=[]
            while j<n and lines[j].strip()!=':::': bl.append(lines[j]); j+=1
            if kind=='affiliations':
                title_affil=f'<div class="affil">{inline(" ".join(x.strip() for x in bl if x.strip()))}</div>'
            elif kind=='abstract':
                title_abs=f'<div class="abstract">{inline(" ".join(x.strip() for x in bl if x.strip()))}</div>'
            elif kind=='keywords':
                title_kw=f'<div class="kwline"><span class="lab">Keywords </span>{inline(" ".join(x.strip() for x in bl if x.strip()))}</div>'
            elif kind=='endmatter':
                if not body_started: emit(open_body())
                endmatter_html[0]+=f'<div class="endmatter">{render_endmatter(bl)}</div>'
            i=j+1; continue
        # box / aside directive
        md=re.match(r'^:::\s*(\w+)(\*?)\s*(.*)$', ln)
        if md:
            if not body_started: emit(open_body())
            kind=md.group(1).lower(); bspan=md.group(2)=='*'; rest=md.group(3).strip()
            j=i+1; bl=[]
            while j<n and lines[j].strip()!=':::': bl.append(lines[j]); j+=1
            if kind in ('note','aside') and not rest:
                emit(f'<p class="aside">{render_blocks(bl).replace("<p>","").replace("</p>"," ")}</p>')
            else:
                num=''; mnum=re.match(r'^([\w.]*\d[\w.]*)\s+(.*)$', rest)
                if mnum: num=mnum.group(1); ttl=mnum.group(2).strip()
                else: ttl=rest
                emit(render_box(num, ttl, bl, span=bspan))
            i=j+1; continue
        # table
        cap=None; tspan=False; mcapt=re.match(r'^Table(\*?):\s*(.+)$', ln); look=i
        if mcapt: tspan=mcapt.group(1)=='*'; cap=mcapt.group(2).strip(); look=i+1
        if look<n and lines[look].lstrip().startswith('|'):
            if not body_started: emit(open_body())
            tl=[]; k=look
            while k<n and lines[k].lstrip().startswith('|'): tl.append(lines[k]); k+=1
            note=None
            if k<n and re.match(r'^Note:\s*(.+)$', lines[k]):
                note=re.match(r'^Note:\s*(.+)$', lines[k]).group(1).strip(); k+=1
            emit(render_table(tl, cap, note, span=tspan)); i=k; continue
        # figure
        mf=re.match(r'^!(\*?)\[(.*?)\]\((.*?)\)(?:\{(.*?)\})?\s*$', ln)
        if mf:
            if not body_started: emit(open_body())
            emit(render_figure(mf.group(2),mf.group(3),mf.group(4),span=mf.group(1)=='*')); i+=1; continue
        # headings
        mh=re.match(r'^(#{1,4})\s+(.+)$', ln)
        if mh:
            if not body_started: emit(open_body())
            lvl=len(mh.group(1)); t=mh.group(2).strip()
            if lvl==1:   # treat a stray H1 as a section head
                lvl=2
            if re.match(r'^(references|bibliography)$', t, re.I):
                in_refs=True; emit(f'<h2 class="refhead">{inline(t)}</h2>'); i+=1; continue
            in_refs=False
            mnum=re.match(r'^(\d+(?:\.\d+)*)\s+(.+)$', t)
            tag=f'h{lvl}'
            if mnum: emit(f'<{tag}><span class="num">{mnum.group(1)}</span>{inline(mnum.group(2))}</{tag}>')
            else:    emit(f'<{tag}>{inline(t)}</{tag}>')
            i+=1; continue
        # numbered list inside references context -> ol.refs
        if in_refs and re.match(r'^\s*\d+\.\s+', ln):
            if not body_started: emit(open_body())
            items=[]
            while i<n and re.match(r'^\s*\d+\.\s+', lines[i]):
                items.append(inline(re.sub(r'^\s*\d+\.\s+','',lines[i]).rstrip())); i+=1
            emit('<ol class="refs">'+''.join(f'<li>{x}</li>' for x in items)+'</ol>'); continue
        # ordinary block
        if not ln.strip(): i+=1; continue
        if not body_started: emit(open_body())
        buf=[ln]; i+=1
        while i<n and lines[i].strip() and not lines[i].startswith(':::') and not re.match(r'^#', lines[i]) \
              and not lines[i].lstrip().startswith('|') and not re.match(r'^(Table\*?:|Note:|!\*?\[)', lines[i]):
            buf.append(lines[i]); i+=1
        frag=render_blocks(buf)
        if not lead_done and '<p>' in frag:
            frag=frag.replace('<p>','<p class="lead">',1); lead_done=True
        emit(frag)

    if not body_started: emit(open_body())
    body_html=''.join(body_frag)+'</div>'+endmatter_html[0]  # close .cols, then full-width end matter
    links=''.join(f'<link rel="stylesheet" href="{html.escape(c)}">' for c in cssfiles)
    return (f'<!DOCTYPE html><html><head><meta charset="utf-8"><title>{html.escape(title)}</title>'
            f'{links}</head><body class="{acccls}">{body_html}</body></html>')

def main():
    if len(sys.argv)<2: print("usage: build_journal.py source.md [out.pdf]"); sys.exit(1)
    src=pathlib.Path(sys.argv[1]).resolve(); text=src.read_text(encoding='utf-8')
    fm, body = parse_front(text); srcdir=src.parent
    css=[]
    if (srcdir/'face.css').exists(): css.append('face.css')
    css.append('apex.css')
    doc=build_html(fm, body, css)
    (srcdir/'_journal.html').write_text(doc, encoding='utf-8')
    out_pdf=pathlib.Path(sys.argv[2]) if len(sys.argv)>2 else srcdir/'journal.pdf'
    from weasyprint import HTML
    HTML(string=doc, base_url=str(srcdir)).write_pdf(str(out_pdf))
    print("PDF:", out_pdf)

if __name__=='__main__': main()

Build with python3 build_apex.py source.md /mnt/user-data/outputs/{slug}.pdf, the base URL the source directory, then inspect the render: confirm the cover hairlines stand clear of the text and the author rule sits with the byline, the ruled abstract and the raised initial, the copper all-caps section heads, the gray-header tables with verdict color-coding and no badge crossing a border, and the copper running footer carrying the short title. Present the PDF and keep the Markdown source as master. The five style variants of III.E are produced by swapping the accent and rule treatments in apex.css; the engine is unchanged.

---

MODULE III·J · THE PROFESSIONAL JOURNAL PAPER EDITION

The middle register. Module III renders the Apex Pristine paper, a single-column elegant treatise in the house serif. Modules V and V·D render the Textbook editions, two-column and box-rich for instruction. The Professional Journal Paper sits between them: it borrows the Textbook's two-column grid and box discipline but holds the formal restraint of a peer-reviewed journal article, serif throughout, a masthead, a bold abstract, superscript citations, and a dense reference list. It is for papers meant to read as journal-grade work, the codex papers among them. Selected by the token [EDITION:JOURNAL] or front-matter edition: journal. It transforms any paper into the journal register and never co-fires with the Module III paper renderer or the Textbook editions.

III·J.0 · The Design Read

A professional journal article in the Nature family sets a precise page. Seven elements carry it. First, the type is serif throughout, body and title and heads alike, in a single restrained measure, which reads as scholarly rather than instructional. Second, the page opens with a masthead: the journal wordmark on the left, the article type on the right in an accent colour with the DOI beneath, and a thick accent rule closing the band. Third, the title block spans the full measure above the columns: a large serif title, an author line with superscript affiliation markers, a small justified affiliations line, and a bold abstract set off by hairline rules. Fourth, the body breaks into two justified columns separated by a faint rule, opening on a raised initial. Fifth, headings are flush-left and bold, the top level in the accent colour, sentence case, unobtrusive. Sixth, the apparatus is journal-formal: superscript numeric citations in the text, a dense numbered reference list flowing in the columns, tinted Box N panels with an accent top rule, Table N caption bars, and Fig. N captions. Seventh, the running furniture is quiet: the journal name and article type in small accent caps at the top, a citation line and page number at the foot. The aesthetic is restraint and density, captured by a serif measure, an accent that touches only the masthead, heads, rules, and labels, and a strict two-column economy.

III·J.1 · The Locked Contract

Serif throughout is the rule. Body, title, heads, abstract, captions, and references all set in the house serif, EB Garamond, at 9.1 points on a 1.36 measure, justified and hyphenated; only the masthead article type, the running heads, the box and table labels, and the section heads use the house sans for crispness. Page geometry is US Letter, 612 by 792 points, margins 58 by 54 by 62, with a two-column body at a 20-point gutter and a hairline column rule. The accent is restrained and touches only the masthead, the section heads, the rules, the box top, the table caption bar, the citation labels, and the superscript markers; the body is black on white. The universal locks hold without exception: no LaTeX and all mathematics in Unicode, no em-dashes, the honorific ﷺ as one glyph, no emoji, and no box, figure, or table fragmenting except where a long data box or table is explicitly allowed to flow.

III·J.2 · The Title Block

The masthead carries the journal name as a wordmark, the first word in the regular weight and the remainder bold, a Nature-style two-tone lockup; on the right the masthead sets a rigid three-line stack, each line held to a single line and the column never shrinking, so the right block stays exactly three lines against the two-line wordmark on the left: the category from article_type in accent caps on the first line, the main goal from goal in accent caps on the second, and the series or DOI from doi in faint sans on the third; a thick accent rule closes the band. The wordmark names the tractatus by domain: core protocol and proof volumes set Tractatus Veritatis Trisductivus, a philosophy or metaphysics paper Tractatus Philosophicus, a mathematics paper Tractatus Mathematicus, a science or physics paper Tractatus Physicus, and the general fallback Tractatus Veritatis. Below it the title block spans the full measure: the title in large bold serif, an optional subtitle from subtitle in accent italic directly beneath, the author line with accent superscript affiliation markers, the affiliations and corresponding address in a small justified line, and the abstract in bold serif bounded above and below by hairline rules. An optional keywords line follows in small sans. The masthead and title block are suppressed from the running header on the first page; the journal name and article type are lifted into the top margin on every page thereafter.

III·J.3 · The Heading Ladder and the Raised Initial

Four levels. The top-level section head is bold accent sans, flush left, sentence case, the loudest mark in the body and still quiet. The sub-section head is bold dark sans, smaller. The third level is a bold-italic serif run-in heading. The inline run-in is a bold-italic serif lead phrase that flows into its paragraph. Any head may carry a numeral, though journal sections are usually unnumbered. The body opens on a raised initial: the first letter of the lead paragraph set large in the accent colour, rising on the first line. The raised initial is used once, on the opening paragraph only.

III·J.4 · Boxes, Tables, and Figures

The box is the journal panel: a tinted block with an accent top rule and a bold header reading Box N | Title, holding paragraphs, lists, or a table. The table is a captioned object: an accent caption bar reading Table N | Title above a ruled table with a tinted header row and zebra body. The figure is a framed image with a caption reading Fig. N | caption, the label in accent. All three default to column width and flow inside one column, packing the page; a genuinely wide table or box is allowed full width by the span rule below. All boxes flow and break across the column boundary to fill space, the tinted panel continuing without a repeated top rule, never jumping whole and stranding the foot of a column; a box wrapping a long table also repeats its column header. Every claim-bearing box, table, and figure carries its warrant tier exactly as the codex establishes it; the faithful-map law governs, and a box asserts no more than the source proves.

III·J.5 · References and End Matter

Citations are superscript and numeric, set in the text with the caret form ^12^ or ^2,3^, matching a numbered reference list. The reference list is introduced by a References heading and set as a dense numbered list flowing in the two columns, hairline-ruled above, in small serif with a hanging indent. End matter, acknowledgements, author contributions, competing interests, and correspondence, is written in an endmatter block and rendered full width below the columns, ruled off, with its own small sans subheadings. The author supplies the citation numbers and the reference list; the engine numbers and sets them.

III·J.6 · Page Economy and the Span Rule

Tables, figures, and boxes default to column width and flow inside one column, which packs the dense journal page without stranded space. Full-width spanning is what wastes space: a spanning element closes both columns and the short band above it balances into two stub columns, leaving the gap empty. Spanning is opt-in. Append an asterisk to span a genuinely wide element: :::box* N Title for a box, Table*: Table N | caption for a table, !*[alt](path) for a figure. Every box flows by default, filling the foot of a column and continuing at the top of the next inside its tinted panel rather than jumping whole to the next page and stranding the space; a box that wraps a table also repeats its column header on the continuation. The journal panel is borderless, a tint with an accent top rule, so a broken panel reads as a clean continuation, never an open cut edge. Think ahead before placing a wide element: prefer column width, reserve spanning for tables too wide to read in a single column.

III·J.7 · Authoring Conventions

Front matter sets the article: edition: journal, then title, an optional subtitle, author_line (with ^1^ superscript markers), journal, article_type (the masthead category line), an optional goal (the masthead main-goal line), doi (the masthead series line), volume, pages, date, and accent. Accent is one of crimson, copper, slate, navy, forest, or plum; crimson is the default and reads as a classic journal. The title block is filled from three leading blocks placed right after the front matter: :::affiliations for the affiliations and corresponding line, :::abstract for the bold abstract, and an optional :::keywords. The body then runs as ordinary Markdown: a level-two heading is a section head, a level-three a sub-head, a level-four a run-in; the run-in {. lead .} flows a bold-italic phrase into its paragraph; :::box N Title opens a numbered panel and :::box* N Title spans it; Table: Table N | caption and Table*: set a captioned table; ![alt](path){Fig. N | caption | source} and !*[ set a figure; [[cross-references]] and ^n^ superscripts set inline marks. A References heading followed by a numbered list sets the reference list, and :::endmatter holds the closing sections. The first body paragraph automatically takes the raised initial.

III·J.8 · The Journal Stylesheet

The complete validated stylesheet. Self-contained, compiles under WeasyPrint as written. Save as journal.css beside the source; the face block of Module V provisions the serif and the sans, with the system fallback chain applying if it is skipped. The drop cap is a no-float raised initial, since float inside a multi-column container is unsupported by the engine.

/* ===================================================================
   JOURNAL  ·  Professional Journal Paper stylesheet  (WeasyPrint-validated)
   Serif two-column journal page · accent masthead + running rule · large
   serif title with superscript-affiliated authors · bold abstract · drop-cap
   intro · flush-left bold heads · superscript citations + dense reference
   list · tinted Box N panels · Fig. N / Table N captions · citation footer
   =================================================================== */

:root{
  --serif:'EBGaramond','EB Garamond','Liberation Serif','Noto Serif','DejaVu Serif','FreeSerif',serif;
  --sans:'TBSans','Archivo','Liberation Sans','Noto Sans','DejaVu Sans','FreeSans',sans-serif;
  --ink:#1a1a1a; --soft:#404040; --faint:#6f6f6f; --rule:#cccccc;
  --accent:#A0182B; --deep:#74111F; --tint:#FBEDEF; --band:#F4DEE2;
}
.acc-crimson{--accent:#A0182B;--deep:#74111F;--tint:#FBEDEF;--band:#F4DEE2;}
.acc-copper {--accent:#B87333;--deep:#8a5523;--tint:#FAF3EA;--band:#EFE0CF;}
.acc-slate  {--accent:#2C3E50;--deep:#1b2733;--tint:#ECEFF2;--band:#D8DEE4;}
.acc-navy   {--accent:#1F3A66;--deep:#142845;--tint:#EAEEF5;--band:#D3DCEA;}
.acc-forest {--accent:#1F5C3D;--deep:#143f29;--tint:#E9F2EC;--band:#D2E5D9;}
.acc-plum   {--accent:#5E2A5E;--deep:#411d41;--tint:#F2ECF2;--band:#E2D2E2;}

/* ---------------- page geometry + furniture ---------------- */
@page{
  size:612pt 792pt; margin:58pt 54pt 62pt 54pt;
  @top-left{ content:string(jtype); font-family:var(--sans); font-size:7.6pt; font-weight:800;
    color:var(--deep); text-transform:uppercase; letter-spacing:.08em; vertical-align:bottom; padding-bottom:4pt; }
  @top-right{ content:string(jname); font-family:var(--sans); font-size:7.6pt; font-weight:800;
    color:var(--deep); text-transform:uppercase; letter-spacing:.08em; vertical-align:bottom; padding-bottom:4pt; }
  @bottom-left{ content:string(jfoot); font-family:var(--serif); font-size:7.4pt; color:var(--faint);
    vertical-align:top; padding-top:5pt; }
  @bottom-right{ content:counter(page); font-family:var(--sans); font-size:8.4pt; font-weight:700;
    color:var(--deep); vertical-align:top; padding-top:5pt; }
}
@page:first{ @top-left{content:none} @top-right{content:none} }

html{ font-family:var(--serif); font-size:9.1pt; color:var(--ink); line-height:1.36; }
body{ margin:0; string-set:jname string(jnameRaw), jtype string(jtypeRaw), jfoot string(jfootRaw); }
p{ margin:0 0 5.5pt; text-align:justify; hyphens:auto; orphans:2; widows:2; }
sup{ font-size:.66em; line-height:0; vertical-align:super; }
strong,b{ font-weight:800; } em,i{ font-style:italic; }
a{ color:var(--accent); text-decoration:none; } .xref{ color:var(--accent); }
ul,ol{ margin:3pt 0 6pt 14pt; padding:0; } li{ margin:0 0 3pt; }
ul{ list-style:none; } ul>li::before{ content:"\2022\00A0"; color:var(--accent); margin-left:-11pt; }

/* hidden string carriers (set running header/footer text) */
.smeta{ position:absolute; left:-9999pt; top:0; height:0; overflow:hidden; }
.jnameRaw{ string-set:jnameRaw content(); } .jtypeRaw{ string-set:jtypeRaw content(); }
.jfootRaw{ string-set:jfootRaw content(); }

/* ---------------- MASTHEAD ---------------- */
.masthead{ display:flex; align-items:flex-end; justify-content:space-between;
  border-bottom:2.4pt solid var(--accent); padding-bottom:5pt; margin:0 0 13pt; }
.masthead .mark{ font-family:var(--serif); font-weight:800; font-size:20pt; line-height:.9; color:var(--ink); letter-spacing:-.01em; }
.masthead .mark .lite{ font-weight:400; }
.masthead .right{ text-align:right; flex-shrink:0; }
.masthead .atype{ font-family:var(--sans); font-weight:800; font-size:11pt; color:var(--accent);
  text-transform:uppercase; letter-spacing:.04em; white-space:nowrap; }
.masthead .agoal{ font-family:var(--sans); font-weight:600; font-size:7.5pt; color:var(--accent);
  text-transform:uppercase; letter-spacing:.02em; margin-top:1.5pt; white-space:nowrap; }
.masthead .doi{ font-family:var(--sans); font-size:6.8pt; color:var(--faint); margin-top:2pt; white-space:nowrap; }

/* ---------------- TITLE BLOCK (spans full width) ---------------- */
.titleblock{ margin:0 0 12pt; }
.titleblock h1.jtitle{ font-family:var(--serif); font-weight:800; font-size:25pt; line-height:1.06;
  color:var(--ink); margin:0 0 9pt; letter-spacing:-.005em; }
.titleblock .jsubtitle{ font-family:var(--serif); font-weight:500; font-style:italic; font-size:13pt; line-height:1.25;
  color:var(--accent); margin:-3pt 0 10pt; }
.titleblock .authors{ font-family:var(--serif); font-size:11pt; line-height:1.4; color:var(--ink); margin:0 0 6pt; }
.titleblock .authors sup{ color:var(--accent); font-weight:700; }
.affil{ font-family:var(--serif); font-size:7.8pt; line-height:1.32; color:var(--soft); text-align:justify; margin:0 0 9pt; }
.affil sup{ color:var(--accent); }
.affil .corr{ color:var(--deep); }

.abstract{ font-family:var(--serif); font-weight:700; font-size:10.2pt; line-height:1.4; color:var(--ink);
  text-align:justify; border-top:0.8pt solid var(--rule); border-bottom:0.8pt solid var(--rule);
  padding:9pt 0; margin:0 0 13pt; }
.kwline{ font-family:var(--sans); font-size:8pt; color:var(--soft); margin:-7pt 0 13pt; }
.kwline .lab{ font-weight:800; color:var(--deep); text-transform:uppercase; letter-spacing:.04em; }

/* ---------------- TWO-COLUMN BODY ---------------- */
.cols{ column-count:2; column-gap:20pt; column-rule:0.4pt solid #e4e4e4; }

/* drop cap on the lead paragraph */
p.lead{ text-align:justify; }
p.lead::first-letter{ font-family:var(--serif); font-weight:800; color:var(--accent);
  font-size:2.7em; line-height:0.8; }

/* ---------------- heading ladder ---------------- */
h2{ font-family:var(--sans); font-weight:800; font-size:10.4pt; color:var(--accent);
  margin:12pt 0 4pt; line-height:1.16; break-after:avoid; }
h2 .num{ color:var(--accent); margin-right:6pt; }
h3{ font-family:var(--sans); font-weight:800; font-size:9.3pt; color:var(--ink);
  margin:9pt 0 3pt; break-after:avoid; }
h3 .num{ color:var(--deep); margin-right:6pt; }
h4{ font-family:var(--serif); font-weight:800; font-style:italic; font-size:9.3pt; color:var(--ink);
  margin:7pt 0 2pt; break-after:avoid; }
.runin{ font-family:var(--serif); font-weight:800; font-style:italic; color:var(--ink); }

/* ---------------- BOX  (Box N | Title) ---------------- */
.jbox{ break-inside:avoid; background:var(--tint); border-top:2pt solid var(--accent);
  padding:7pt 9pt 8pt; margin:9pt 0; box-decoration-break:clone; }
.jbox.flow{ break-inside:avoid; }
.jbox.flow table{ break-inside:avoid; } .jbox.flow thead{ display:table-header-group; }
.jbox .jbh{ font-family:var(--sans); font-weight:800; font-size:9pt; color:var(--deep);
  margin:0 0 5pt; text-transform:none; break-after:avoid; }
.jbox .jbh .lab{ color:var(--accent); }
.jbox p{ font-size:8.4pt; text-align:justify; margin:0 0 4pt; orphans:2; widows:2; }
.jbox ul{ margin:2pt 0 0 13pt; } .jbox ul>li::before{ color:var(--accent); }
.jbox table{ width:100%; border-collapse:collapse; font-size:7.9pt; margin:2pt 0; }
.jbox thead th{ text-align:left; font-weight:800; color:var(--deep); border-bottom:1pt solid var(--accent); padding:3pt 5pt; }
.jbox tbody td{ padding:3pt 5pt; border-bottom:0.4pt solid var(--rule); vertical-align:top; word-wrap:break-word; overflow-wrap:break-word; hyphens:none; }
.jbox .tnote{ font-size:7.2pt; color:var(--soft); font-style:italic; padding-top:3pt; }
.jspan{ column-span:all; break-inside:avoid; }

/* small-print aside */
.aside{ font-size:8.2pt; color:var(--soft); text-align:justify; margin:5pt 0; }
.aside::before{ content:"\25AA\2002"; color:var(--accent); font-size:6.6pt; }

/* ---------------- TABLE (Table N | Title) ---------------- */
.jtable{ margin:9pt 0; break-inside:avoid; }
.jtable .tcap{ font-family:var(--sans); font-weight:800; font-size:8.6pt; color:#fff; background:var(--accent);
  padding:3.5pt 7pt; }
.jtable .tcap .lab{ color:#fff; }
.jtable table{ width:100%; border-collapse:collapse; font-family:var(--serif); font-size:8.1pt; }
.jtable thead th{ background:var(--band); color:var(--deep); text-align:left; font-weight:800;
  padding:3.5pt 6pt; border-bottom:1pt solid var(--accent); font-family:var(--sans); font-size:7.7pt; }
.jtable tbody td{ padding:3.5pt 6pt; border-bottom:0.4pt solid var(--rule); vertical-align:top; word-wrap:break-word; overflow-wrap:break-word; hyphens:none; }
.jtable tbody tr:nth-child(even) td{ background:#faf7f8; }
.jtable .tnote{ font-family:var(--serif); font-size:7.2pt; color:var(--soft); padding:3pt 1pt 0; }
.jtable.span{ column-span:all; break-inside:avoid; }
.jtable thead{ display:table-header-group; }

/* ---------------- FIGURE (Fig. N | caption) ---------------- */
.jfig{ break-inside:avoid; margin:10pt 0; }
.jfig .frame{ background:#fff; padding:0; text-align:center; }
.jfig img{ max-width:100%; height:auto; }
.jfig figcaption{ font-family:var(--serif); font-size:7.9pt; line-height:1.34; color:var(--soft);
  margin-top:5pt; text-align:justify; }
.jfig figcaption .flab{ font-family:var(--sans); font-weight:800; color:var(--accent); }
.jfig.span{ column-span:all; }

/* ---------------- REFERENCES (dense, numbered, in-flow) ---------------- */
h2.refhead{ font-family:var(--sans); font-weight:800; font-size:9.2pt; color:var(--ink);
  border-top:0.8pt solid var(--rule); padding-top:5pt; margin:11pt 0 4pt; }
ol.refs{ margin:0; padding:0; list-style:none; counter-reset:ref; }
ol.refs li{ font-family:var(--serif); font-size:7.5pt; line-height:1.3; color:var(--soft);
  padding-left:14pt; text-indent:-14pt; margin:0 0 3pt; text-align:justify; }
ol.refs li::before{ counter-increment:ref; content:counter(ref) ". "; font-weight:700; color:var(--ink); }

/* ---------------- author block / end matter ---------------- */
.endmatter{ border-top:0.8pt solid var(--rule); margin-top:11pt; padding-top:7pt; }
.endmatter h3{ font-family:var(--sans); font-weight:800; font-size:8.4pt; color:var(--ink); margin:6pt 0 2pt; }
.endmatter p{ font-size:7.8pt; color:var(--soft); }

/* ---------------- table of contents (rarely used) ---------------- */
.toc{ break-after:page; } .toc h1{ font-family:var(--sans); font-size:15pt; font-weight:800; color:var(--accent); }

III·J.9 · The Journal Build Engine

The journal-aware engine. It parses the front matter and the leading affiliations, abstract, and keywords blocks into the title block, renders the masthead and the full-measure title block, opens the two-column body, marks the first paragraph with the raised initial, and renders the heading ladder, the Box N panels, the Table N and Fig. N captions, the superscript citations, the numbered reference list, and the full-width end matter. Heads at level two to four become accent, dark, and run-in heads; a References heading switches the following numbered list to the dense reference style; the end matter renders outside the columns. Save as build_journal.py. Build with python3 build_journal.py source.md /mnt/user-data/outputs/{slug}.pdf, the base URL the source directory, then inspect the render: confirm the masthead and ruled abstract, the raised initial, the crimson heads, the tinted Box N panels, the Table N caption bars, the dense reference list, and the citation footer. Present the PDF and keep the Markdown source as master.

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
JOURNAL build engine (Professional Journal Paper).
Compiles one Markdown source to a serif two-column journal-grade PDF via
WeasyPrint: accent masthead, superscript-affiliated authors, bold abstract,
drop-cap intro, flush-left bold heads, tinted Box N panels, Table N / Fig. N
captions, dense numbered references, citation footer.

Usage:  python3 build_journal.py source.md [out.pdf]
Source dir must contain journal.css (and optionally face.css + fonts/).
"""
import sys, os, re, html, pathlib

ACCENTS={'crimson','copper','slate','navy','forest','plum'}

def parse_front(text):
    fm={}; m=re.match(r'^---\s*\n(.*?)\n---\s*\n',text,re.S); body=text
    if m:
        for line in m.group(1).splitlines():
            if ':' in line: k,v=line.split(':',1); fm[k.strip()]=v.strip()
        body=text[m.end():]
    return fm, body

def inline(s):
    s=html.escape(s, quote=False)
    s=re.sub(r'\[\[(.+?)\]\]', r'<span class="xref">\1</span>', s)
    s=re.sub(r'\*\*(.+?)\*\*', r'<strong>\1</strong>', s)
    s=re.sub(r'(?<!\*)\*(?!\*)(.+?)(?<!\*)\*(?!\*)', r'<em>\1</em>', s)
    s=re.sub(r'`(.+?)`', r'<code>\1</code>', s)
    s=re.sub(r'\^(.+?)\^', r'<sup>\1</sup>', s)
    return s

def render_blocks(lines):
    out=[]; i=0; n=len(lines)
    while i<n:
        ln=lines[i]
        if not ln.strip(): i+=1; continue
        if re.match(r'^\s*[-*\u2022]\s+', ln):
            items=[]
            while i<n and re.match(r'^\s*[-*\u2022]\s+', lines[i]):
                items.append(inline(re.sub(r'^\s*[-*\u2022]\s+','',lines[i]).rstrip())); i+=1
            out.append('<ul>'+''.join(f'<li>{x}</li>' for x in items)+'</ul>'); continue
        if re.match(r'^\s*\d+\.\s+', ln):
            items=[]
            while i<n and re.match(r'^\s*\d+\.\s+', lines[i]):
                items.append(inline(re.sub(r'^\s*\d+\.\s+','',lines[i]).rstrip())); i+=1
            out.append('<ol>'+''.join(f'<li>{x}</li>' for x in items)+'</ol>'); continue
        mrun=re.match(r'^\{\.\s*(.+?)\s*\.\}\s*(.*)$', ln)
        if mrun:
            buf=[mrun.group(2).rstrip()] if mrun.group(2).strip() else []; i+=1
            while i<n and lines[i].strip() and not re.match(r'^\s*([-*\u2022]|\d+\.)\s+', lines[i]) \
                  and not lines[i].startswith(':::') and not re.match(r'^\{\.', lines[i]):
                buf.append(lines[i].rstrip()); i+=1
            out.append(f'<p><span class="runin">{inline(mrun.group(1))}.</span> '+inline(' '.join(buf))+'</p>'); continue
        buf=[ln.rstrip()]; i+=1
        while i<n and lines[i].strip() and not re.match(r'^\s*([-*\u2022]|\d+\.)\s+', lines[i]) \
              and not lines[i].startswith(':::') and not re.match(r'^\{\.', lines[i]):
            buf.append(lines[i].rstrip()); i+=1
        out.append('<p>'+inline(' '.join(buf))+'</p>')
    return ''.join(out)

def render_table_rows(tbl_lines):
    rows=[r for r in tbl_lines if r.strip()]
    cells=[[c.strip() for c in re.split(r'(?<!\\)\|', r.strip().strip('|'))] for r in rows]
    body=[c for c in cells if not all(re.fullmatch(r':?-{2,}:?', x or '-') for x in c)]
    head=body[0] if body else []; data=body[1:] if len(body)>1 else []
    h='<thead><tr>'+''.join(f'<th>{inline(x)}</th>' for x in head)+'</tr></thead>'
    b='<tbody>'+''.join('<tr>'+''.join(f'<td>{inline(x)}</td>' for x in r)+'</tr>' for r in data)+'</tbody>'
    return h+b

def render_table(tbl_lines, caption=None, note=None, span=False):
    cap=''
    if caption:
        m=re.match(r'^(Table\s+[\w.]+)\s*[|:]\s*(.+)$', caption)
        if m: cap=f'<div class="tcap"><span class="lab">{inline(m.group(1))} | </span>{inline(m.group(2))}</div>'
        else: cap=f'<div class="tcap">{inline(caption)}</div>'
    nt=f'<div class="tnote">{inline(note)}</div>' if note else ''
    sp=' span' if span else ''
    return f'<div class="jtable{sp}">{cap}<table>{render_table_rows(tbl_lines)}</table>{nt}</div>'

def render_figure(alt, path, brace, span=False):
    label=''; body=brace or alt; src=''
    if brace and '|' in brace:
        parts=[p.strip() for p in brace.split('|')]
        label=parts[0] if parts else ''
        body=parts[1] if len(parts)>1 else ''
        src=parts[2] if len(parts)>2 else ''
    cap=''
    if label: cap+=f'<span class="flab">{inline(label)} | </span>'
    cap+=inline(body)
    if src: cap+=f' <span class="fsrc">{inline(src)}</span>'
    sp=' span' if span else ''
    return (f'<div class="jfig{sp}"><div class="frame"><img src="{html.escape(path)}" alt="{html.escape(alt)}"></div>'
            f'<figcaption>{cap}</figcaption></div>')

def render_box(number, title, body_lines, span=False):
    has_table=any(l.lstrip().startswith('|') for l in body_lines)
    if has_table:
        i=0;n=len(body_lines);chunks=[]
        while i<n:
            if body_lines[i].lstrip().startswith('|'):
                tl=[]
                while i<n and body_lines[i].lstrip().startswith('|'): tl.append(body_lines[i]); i+=1
                note=None
                if i<n and re.match(r'^Note:\s*(.+)$', body_lines[i]):
                    note=re.match(r'^Note:\s*(.+)$', body_lines[i]).group(1).strip(); i+=1
                chunks.append('<table>'+render_table_rows(tl)+'</table>'+(f'<div class="tnote">{inline(note)}</div>' if note else ''))
            else:
                tx=[]
                while i<n and not body_lines[i].lstrip().startswith('|'): tx.append(body_lines[i]); i+=1
                if any(x.strip() for x in tx): chunks.append(render_blocks(tx))
        inner=''.join(chunks)
    else:
        inner=render_blocks(body_lines)
    lab=f'Box {number}' if number else 'Box'
    cls=' jspan' if span else (' flow' if has_table else '')
    head=f'<div class="jbh"><span class="lab">{html.escape(lab)} | </span>{inline(title)}</div>'
    return f'<div class="jbox{cls}">{head}{inner}</div>'

def render_endmatter(lines):
    out=[]; buf=[]
    def flush():
        if buf: out.append(render_blocks(buf)); buf.clear()
    for ln in lines:
        m=re.match(r'^###\s+(.+)$', ln)
        if m: flush(); out.append(f'<h3>{inline(m.group(1).strip())}</h3>')
        else: buf.append(ln)
    flush(); return ''.join(out)

def build_html(fm, body, cssfiles):
    acc=fm.get('accent','crimson').strip().lower()
    if acc not in ACCENTS: acc='crimson'
    acccls='acc-'+acc
    journal=fm.get('journal','Journal'); atype=fm.get('article_type','Article'); goal=fm.get('goal','')
    doi=fm.get('doi',''); title=fm.get('title',''); authors=fm.get('author_line',''); subtitle=fm.get('subtitle','')
    vol=fm.get('volume',''); pages=fm.get('pages',''); date=fm.get('date','')
    # masthead wordmark: first word light, remainder bold
    parts=journal.split()
    mark=(f'<span class="lite">{html.escape(parts[0])}</span> '+html.escape(' '.join(parts[1:]))) if len(parts)>1 else html.escape(journal)
    # footer citation line
    foot=journal
    if vol: foot+=f" \u00b7 Vol {vol}"
    if date: foot+=f" \u00b7 {date}"
    if pages: foot+=f" \u00b7 {pages}"

    lines=body.splitlines(); i=0; n=len(lines)
    title_affil=''; title_abs=''; title_kw=''
    body_frag=[]; endmatter_html=['']; body_started=False; lead_done=False; in_refs=False

    def open_body():
        nonlocal body_started
        if body_started: return ''
        body_started=True
        doi_html=f'<div class="doi">{html.escape(doi)}</div>' if doi else ''
        sub_html=f'<div class="jsubtitle">{inline(subtitle)}</div>' if subtitle else ''
        goal_html=f'<div class="agoal">{html.escape(goal)}</div>' if goal else ''
        block=(f'<div class="masthead"><div class="mark">{mark}</div>'
               f'<div class="right"><div class="atype">{html.escape(atype)}</div>{goal_html}{doi_html}</div></div>'
               f'<div class="titleblock"><h1 class="jtitle">{inline(title)}</h1>'
               f'{sub_html}'
               f'<div class="authors">{inline(authors)}</div>'
               f'{title_affil}{title_abs}{title_kw}</div>'
               f'<div class="smeta"><span class="jnameRaw">{html.escape(journal)}</span>'
               f'<span class="jtypeRaw">{html.escape(atype)}</span>'
               f'<span class="jfootRaw">{html.escape(foot)}</span></div>')
        return block+'<div class="cols">'

    def emit(frag):
        nonlocal body_frag
        body_frag.append(frag)

    while i<n:
        ln=lines[i]
        # special title-block / endmatter blocks
        msp=re.match(r'^:::\s*(affiliations|abstract|keywords|endmatter|references)\s*$', ln)
        if msp:
            kind=msp.group(1); j=i+1; bl=[]
            while j<n and lines[j].strip()!=':::': bl.append(lines[j]); j+=1
            if kind=='affiliations':
                title_affil=f'<div class="affil">{inline(" ".join(x.strip() for x in bl if x.strip()))}</div>'
            elif kind=='abstract':
                title_abs=f'<div class="abstract">{inline(" ".join(x.strip() for x in bl if x.strip()))}</div>'
            elif kind=='keywords':
                title_kw=f'<div class="kwline"><span class="lab">Keywords </span>{inline(" ".join(x.strip() for x in bl if x.strip()))}</div>'
            elif kind=='endmatter':
                if not body_started: emit(open_body())
                endmatter_html[0]+=f'<div class="endmatter">{render_endmatter(bl)}</div>'
            i=j+1; continue
        # box / aside directive
        md=re.match(r'^:::\s*(\w+)(\*?)\s*(.*)$', ln)
        if md:
            if not body_started: emit(open_body())
            kind=md.group(1).lower(); bspan=md.group(2)=='*'; rest=md.group(3).strip()
            j=i+1; bl=[]
            while j<n and lines[j].strip()!=':::': bl.append(lines[j]); j+=1
            if kind in ('note','aside') and not rest:
                emit(f'<p class="aside">{render_blocks(bl).replace("<p>","").replace("</p>"," ")}</p>')
            else:
                num=''; mnum=re.match(r'^([\w.]*\d[\w.]*)\s+(.*)$', rest)
                if mnum: num=mnum.group(1); ttl=mnum.group(2).strip()
                else: ttl=rest
                emit(render_box(num, ttl, bl, span=bspan))
            i=j+1; continue
        # table
        cap=None; tspan=False; mcapt=re.match(r'^Table(\*?):\s*(.+)$', ln); look=i
        if mcapt: tspan=mcapt.group(1)=='*'; cap=mcapt.group(2).strip(); look=i+1
        if look<n and lines[look].lstrip().startswith('|'):
            if not body_started: emit(open_body())
            tl=[]; k=look
            while k<n and lines[k].lstrip().startswith('|'): tl.append(lines[k]); k+=1
            note=None
            if k<n and re.match(r'^Note:\s*(.+)$', lines[k]):
                note=re.match(r'^Note:\s*(.+)$', lines[k]).group(1).strip(); k+=1
            emit(render_table(tl, cap, note, span=tspan)); i=k; continue
        # figure
        mf=re.match(r'^!(\*?)\[(.*?)\]\((.*?)\)(?:\{(.*?)\})?\s*$', ln)
        if mf:
            if not body_started: emit(open_body())
            emit(render_figure(mf.group(2),mf.group(3),mf.group(4),span=mf.group(1)=='*')); i+=1; continue
        # headings
        mh=re.match(r'^(#{1,4})\s+(.+)$', ln)
        if mh:
            if not body_started: emit(open_body())
            lvl=len(mh.group(1)); t=mh.group(2).strip()
            if lvl==1:   # treat a stray H1 as a section head
                lvl=2
            if re.match(r'^(references|bibliography)$', t, re.I):
                in_refs=True; emit(f'<h2 class="refhead">{inline(t)}</h2>'); i+=1; continue
            in_refs=False
            mnum=re.match(r'^(\d+(?:\.\d+)*)\s+(.+)$', t)
            tag=f'h{lvl}'
            if mnum: emit(f'<{tag}><span class="num">{mnum.group(1)}</span>{inline(mnum.group(2))}</{tag}>')
            else:    emit(f'<{tag}>{inline(t)}</{tag}>')
            i+=1; continue
        # numbered list inside references context -> ol.refs
        if in_refs and re.match(r'^\s*\d+\.\s+', ln):
            if not body_started: emit(open_body())
            items=[]
            while i<n and re.match(r'^\s*\d+\.\s+', lines[i]):
                items.append(inline(re.sub(r'^\s*\d+\.\s+','',lines[i]).rstrip())); i+=1
            emit('<ol class="refs">'+''.join(f'<li>{x}</li>' for x in items)+'</ol>'); continue
        # ordinary block
        if not ln.strip(): i+=1; continue
        if not body_started: emit(open_body())
        buf=[ln]; i+=1
        while i<n and lines[i].strip() and not lines[i].startswith(':::') and not re.match(r'^#', lines[i]) \
              and not lines[i].lstrip().startswith('|') and not re.match(r'^(Table\*?:|Note:|!\*?\[)', lines[i]):
            buf.append(lines[i]); i+=1
        frag=render_blocks(buf)
        if not lead_done and '<p>' in frag:
            frag=frag.replace('<p>','<p class="lead">',1); lead_done=True
        emit(frag)

    if not body_started: emit(open_body())
    body_html=''.join(body_frag)+'</div>'+endmatter_html[0]  # close .cols, then full-width end matter
    links=''.join(f'<link rel="stylesheet" href="{html.escape(c)}">' for c in cssfiles)
    return (f'<!DOCTYPE html><html><head><meta charset="utf-8"><title>{html.escape(title)}</title>'
            f'{links}</head><body class="{acccls}">{body_html}</body></html>')

def main():
    if len(sys.argv)<2: print("usage: build_journal.py source.md [out.pdf]"); sys.exit(1)
    src=pathlib.Path(sys.argv[1]).resolve(); text=src.read_text(encoding='utf-8')
    fm, body = parse_front(text); srcdir=src.parent
    css=[]
    if (srcdir/'face.css').exists(): css.append('face.css')
    css.append('journal.css')
    doc=build_html(fm, body, css)
    (srcdir/'_journal.html').write_text(doc, encoding='utf-8')
    out_pdf=pathlib.Path(sys.argv[2]) if len(sys.argv)>2 else srcdir/'journal.pdf'
    from weasyprint import HTML
    HTML(string=doc, base_url=str(srcdir)).write_pdf(str(out_pdf))
    print("PDF:", out_pdf)

if __name__=='__main__': main()

III·J.10 · Codex-Native Journal

A codex paper in the journal register sets edition: journal with a chosen accent, the body at Tier 0 plain English with the framework apparatus quarantined to an appendix that carries no results, exactly as the Module III paper discipline requires. The abstract states the result and its warrant tier without framework vocabulary. Verdict and seal statements sit in the body as typed claims, Type T, Type C, or Type S, and a Box N panel may carry a closed-form instrument or a precedence table. The superscript citations and the reference list carry the external mathematics and the prior codex deposits. The faithful-map law and honest typing govern as everywhere: the abstract and every box state no more than the paper proves, and overclaiming a warrant grade in an abstract or a box header is the same failure it is in prose. Where a paper is destined for a specific venue, the journal name, article type, volume, pages, and DOI set the masthead and the citation footer to that venue's identity.

III·J.11 · Pre-Flight and Gate Integration

A journal deliverable passes the standard pre-delivery gate sequence and, for a codex paper, produces the V-FIO Diagnostic Log exactly as the other editions. Before presentation it passes a short structural checklist. The body and every element set in the house serif, with the sans only on heads, labels, and running furniture. The masthead carries the journal name, article type, and DOI, and the abstract is bold and ruled. The raised initial appears once, on the opening paragraph. Every box carries its accent top rule and its Box N label, every table its Table N caption bar, every figure its Fig. N label, and every wide element either reads in a column or is explicitly spanned. The citations are superscript and resolve to a numbered reference list, and the end matter sits full width below the columns. Every claim-bearing box, table, and figure carries its warrant tier with no inflation. Failure of any item returns the paper for repair before any presentation.

III·J Quick Reference Card

Edition selector: [EDITION:JOURNAL] or front-matter edition: journal. The middle register between the Module III Apex Pristine paper and the Module V Textbook editions; the three never co-fire. Type: serif throughout, EB Garamond, 9.1 pt body, justified two columns at a 20 pt gutter with a hairline column rule. Sans only on section heads, box and table labels, and running furniture. Page: US Letter 612 by 792, margins 58 / 54 / 62. Accent restrained to masthead, heads, rules, box top, table caption, labels, and superscripts. Accent: crimson default, or copper, slate, navy, forest, plum, via front-matter accent. Masthead: journal wordmark, first word regular and remainder bold, with the article type in accent caps and the DOI beneath, closed by a thick accent rule. Suppressed on the first-page running header. Title block: full-measure serif title, author line with accent superscript affiliation markers, small justified affiliations and corresponding line, bold abstract bounded by hairline rules, optional keywords line. Headings: level-two accent sans head, level-three dark sans sub-head, level-four bold-italic serif run-in, inline {. lead .} run-in. Raised accent initial on the opening paragraph only. Boxes and floats: tinted Box N panel with an accent top rule; Table N accent caption bar over a ruled zebra table; Fig. N captioned figure. Column-width by default; every box flows and breaks to fill the column, a table-box also repeating its header; * spans a wide element. References and end matter: superscript numeric citations ^n^; a References heading sets a dense numbered list flowing in the columns; :::endmatter renders acknowledgements and correspondence full width below the columns. Source: front matter (edition: journal, title, author_line, journal, article_type, doi, volume, pages, date, accent), then :::affiliations, :::abstract, optional :::keywords, then Markdown body with :::box N Title, Table:, ![](){Fig. N | ...}, a References list, and :::endmatter. Build: python3 build_journal.py source.md /mnt/user-data/outputs/{slug}.pdf; base URL the source directory; inspect the render; present the PDF; keep the Markdown source as master. Locks: no LaTeX, Unicode math only; no em-dashes; ﷺ as one glyph; no emoji; warrant tier on every claim-bearing box, table, and figure; faithful map, never inflation; the raised initial is a no-float raised cap, since float inside multicol is unsupported.

---

MODULE III·S · THE SCRIPTURE EDITION

The scripture register, journal-grade. Module III·J renders the two-column journal paper; the Scripture edition is its multi-script Quranic descendant, tuned for verse-by-verse translation and tafseer. It holds the Journal's two-column grid, masthead, running heads, and tinted box discipline, and adds three things the Journal has no need of: a combined Latin-Arabic-Bengali font stack the engine resolves per glyph, the right-to-left word grid re-rendered as flowing interlinear word-cells that fit a narrow column, and a scripture title block carrying the surah head, the basmala, and the servant's byline. It is the PDF delivery for scripture. Selected by the token [EDITION:SCRIPTURE] or front-matter edition: scripture, and by the standing scripture words scripture pdf, tafsir pdf, quran pdf, mushaf pdf. It never co-fires with the other renderers.

III·S.0 · The Design Read

A scripture page at journal grade is two-column like the Journal, not single-column, and it carries the same restraint: serif throughout, an accent that touches only the furniture, a masthead band, and quiet running heads. Eight elements carry it. First, the type is serif throughout, EB Garamond for Latin, Amiri for Arabic, Noto Serif Bengali for Bengali, set as one combined stack so each glyph resolves to its own face with no per-cell targeting. Second, the page opens with a masthead: a two-tone wordmark on the left, the edition type with the surah meta and reference in accent on the right, and a thick accent rule closing the band. Third, the title block spans the full measure and is centred: the surah name in Arabic, the English title beneath, an accent-italic subtitle, the author line, and the basmala, with an optional bold surah-overview band ruled off above and below. Fourth, the body breaks into two justified columns. Fifth, each verse opens with its reference marker and the word grid, rendered not as a wide table but as a row of interlinear word-cells, each cell an Arabic word over its transliteration over its root meaning, flowing right-to-left and wrapping inside the column. Sixth, the eight components follow as a labelled ladder, and a noteworthy geometric reading or deep-dive synthesis is lifted into a tinted text box with an accent top rule, a Geometric Insight or Deep Dive panel. Seventh, a genuine summary matrix sets as a tinted-header table. Eighth, the running furniture is quiet: the wordmark and the surah reference in small accent caps at the top of every continuation page, the servant's byline and the page number at the foot. The accent is copper, the house scripture signature.

Two further structures complete the edition. The surah is framed the way a scholarly mushaf frames a chapter: a brief introduction opens it as a full-measure band ruled above and below, thematic section headers with short overviews divide the body where the surah is long enough to carry more than one theme, and a closing synthesis of the core message seals it as a full-measure tinted panel at the end. And the body sets two-column by default, the journal-grade register, or single-column on request, the Apex Pristine register for scripture; the same content and the same frame render in either measure, selected by a single front-matter switch.

III·S.1 · The Locked Contract

Serif throughout, EB Garamond for Latin at 9.4 points on a 1.42 measure, Amiri for Arabic, Noto Serif Bengali for Bengali, set as one stack so the engine falls per glyph to the right face. Page geometry is US Letter, 612 by 792 points, margins 60 by 56 by 60, a two-column body at a 20-point gutter with a hairline column rule. The body sets two-column by default; front-matter columns: 1 renders the single-column Apex Pristine register for scripture, widening the measure and enlarging the word-cells while keeping every other lock, the frame, the word grid, the boxes, the masthead, and the multi-script stack, unchanged. The accent is copper by default and touches only the masthead, the rules, the verse markers, the box tops and labels, the table caption bars, and the word-cell meaning row; the body is black on white. The universal locks hold without exception: no LaTeX and all mathematics in Unicode, no em-dashes, honorifics and the ayah marker as single glyphs, no emoji. The word grid is borderless, rendered as interlinear cells, never a ruled table.

III·S.2 · The Masthead, Title Block, and Author Line

Front matter sets the article. wordmark is the masthead's left lockup, the first word light and the rest bold, defaulting to Geometric Quran; article_type is the right-hand edition line, defaulting to Geometric Analysis; surah_meta and reference are the two lines beneath it. The centred title block carries surah_ar the Arabic surah name large in Amiri, title the English name in bold serif, subtitle in accent italic, author_line, and basmala in Amiri beneath. The author line defaults to the scripture byline, Source Text: Arabic Quran Text · Translation and Synthesis by: A Humble Servant | Fellow Witness, and is overridden in front matter when a different attribution is wanted. An optional :::abstract block sets a bold surah-overview band, and :::keywords a small key-term line. The running heads are set by running_left and running_right, defaulting to the wordmark and the reference, and the foot by footer, defaulting to the byline; all three are injected as literal strings to sidestep the engine's named-string running-header bug. The masthead and title block are suppressed from the running header on the first page; the wordmark and reference are lifted into the top margin on every page thereafter.

III·S.3 · The Word Grid as Interlinear Cells

The translation protocol emits each verse's opening as a borderless right-to-left word grid, an HTML table set dir="rtl" with three rows, Arabic, transliteration, and root meaning, written in one shared reading order. A wide table cannot live in a narrow journal column, so the engine does not render it as a table. It parses the three rows and re-emits the verse as a row of interlinear word-cells, one cell per word, each cell stacking the Arabic word over its transliteration over its root meaning, centred, flowing right-to-left inside a dir="rtl" container and wrapping to as many lines as the column needs. Read right to left, the top of the cells is the connected ayah; beneath each word sits its gloss. The Arabic sets in Amiri at cell size, the transliteration in the combined stack at reading size, the meaning in the sans accent. This is the column-fit form of the grid, and it is what makes the two-column scripture page work. The protocol is unchanged; the transform is the engine's.

III·S.3a · The Surah Frame: Introduction, Thematic Sections, and Closing Synthesis

The translation protocol frames every full surah, its Section II.3c, and the edition renders the frame as full-measure structural furniture set apart from the verse readings. Three directives carry it, each spanning both columns. The surah introduction is a :::intro block, rendered as a full-measure band ruled above and below, a brief scholarly orientation, the surah's name and meaning, its number and setting, and the arc the reading traces, set before the first verse. A thematic section is a :::theme Headline · vv. X–Y block, rendered as a full-measure section header, the headline in accent-ruled sans with its ayah range and beneath it the short overview in serif italic, placed before the verses of the block; a long surah is divided into as many sections as it carries, and a short single-theme surah carries none. The closing synthesis is a :::closing block, rendered as a full-measure tinted panel under a Closing Synthesis label, gathering the surah's core message at the end, after the last verse and after any deep-dive box. The frame directives span the full measure in both layouts, dividing the two-column body cleanly and setting off the single-column body the same way. The introduction and the closing are written for every surah; the thematic headers appear only where the surah holds more than one theme, exactly as the protocol gates them.

III·S.4 · Text Boxes, the Component Ladder, Honorifics, and Apparatus

The eight numbered components render as the protocol emits them, the bold label leading each. A noteworthy geometric reading or a deep-dive synthesis is lifted out of the prose flow into a tinted text box: the source wraps it in a box directive, :::insight Title for a Geometric Insight panel, :::deepdive Title for a Deep Dive panel, :::synthesis Title or a bare :::box Title for a general panel, each rendered as a copper-topped tint with the label in accent caps. A box spans both columns with the star form, :::insight* Title. Honorifics ﷺ and ﷻ and the prophetic AS, and the ayah-end marker ۝, are preserved as glyphs in the Arabic face. A genuine summary matrix written as a Markdown table, optionally captioned with Table: Table N | title, sets with a tinted header and a hairline body. A concordance list or the optional audit ledger sets in a copper-ruled monospace block. Anchors, the in-tradition synthesis, and the comparative deep dive read as ordinary prose unless boxed. The silent-omission discipline holds: nothing empty is printed.

III·S.5 · The Scripture Stylesheet

The complete validated stylesheet. Self-contained, compiles under WeasyPrint as written. Save as scripture.css beside the source. The combined font stack lets the engine fall per glyph to Amiri for Arabic and Noto Serif Bengali for Bengali, with EB Garamond carrying Latin.

/* ===================================================================
   SCRIPTURE  ·  two-column journal-grade multi-script Quranic edition
   EB Garamond (Latin) · Amiri (Arabic) · Noto Serif Bengali (Bengali)
   masthead + running heads · centred surah head + basmala · two-column
   body · interlinear word-cells (column-fit) · Geometric-Insight and
   Deep-Dive text boxes · tinted summary tables · copper accent
   (WeasyPrint-validated)
   =================================================================== */
:root{
  --serif:'EBGaramond','EB Garamond','Liberation Serif','Noto Serif',serif;
  --arabic:'Amiri','Scheherazade New','Noto Naskh Arabic',serif;
  --bengali:'Noto Serif Bengali','Noto Sans Bengali',serif;
  --multi:'EBGaramond','Amiri','Noto Serif Bengali','Liberation Serif',serif;
  --sans:'Archivo','Liberation Sans','Noto Sans','DejaVu Sans',sans-serif;
  --ink:#1f1b16; --soft:#4a423a; --faint:#7a6f63; --rule:#d2c4b0;
  --accent:#B87333; --deep:#8a5523; --tint:#FAF3EA; --band:#EFE0CF;
}
.acc-copper {--accent:#B87333;--deep:#8a5523;--tint:#FAF3EA;--band:#EFE0CF;}
.acc-crimson{--accent:#A0182B;--deep:#74111F;--tint:#FBEDEF;--band:#F4DEE2;}
.acc-slate  {--accent:#2C3E50;--deep:#1b2733;--tint:#ECEFF2;--band:#D8DEE4;}
.acc-navy   {--accent:#1F3A66;--deep:#142845;--tint:#EAEEF5;--band:#D3DCEA;}
.acc-forest {--accent:#1F5C3D;--deep:#143f29;--tint:#E9F2EC;--band:#D2E5D9;}
.acc-plum   {--accent:#5E2A5E;--deep:#411d41;--tint:#F2ECF2;--band:#E2D2E2;}

/* ---------------- page furniture ---------------- */
@page{
  size:612pt 792pt; margin:60pt 56pt 60pt 56pt;
  @top-left{ content:string(jtype); font-family:var(--sans); font-size:7.4pt; font-weight:700;
    color:var(--deep); text-transform:uppercase; letter-spacing:.09em; vertical-align:bottom; padding-bottom:4pt; }
  @top-right{ content:string(jname); font-family:var(--sans); font-size:7.4pt; font-weight:700;
    color:var(--deep); text-transform:uppercase; letter-spacing:.07em; vertical-align:bottom; padding-bottom:4pt; }
  @bottom-left{ content:string(jfoot); font-family:var(--serif); font-size:7.3pt; color:var(--faint);
    vertical-align:top; padding-top:5pt; }
  @bottom-right{ content:counter(page); font-family:var(--sans); font-size:8.4pt; font-weight:700;
    color:var(--deep); vertical-align:top; padding-top:5pt; }
}
@page:first{ @top-left{content:none} @top-right{content:none} }

html{ font-family:var(--multi); font-size:9.4pt; color:var(--ink); line-height:1.42; }
body{ margin:0; }
p{ margin:0 0 5.5pt; text-align:justify; hyphens:auto; orphans:2; widows:2; }
strong,b{ font-weight:700; color:var(--ink); } em,i{ font-style:italic; }
a,.xref{ color:var(--accent); text-decoration:none; }
sup{ font-size:.66em; line-height:0; vertical-align:super; } sub{ font-size:.66em; line-height:0; vertical-align:sub; }
ul,ol{ margin:3pt 0 6pt 13pt; padding:0; } li{ margin:0 0 3pt; }
ul{ list-style:none; } ul>li::before{ content:"\2022\00A0"; color:var(--accent); margin-left:-11pt; }
code{ font-family:'DejaVu Sans Mono',monospace; font-size:.9em; }

.smeta{ position:absolute; left:-9999pt; top:0; height:0; overflow:hidden; }
.jnameRaw{ string-set:jname content(); } .jtypeRaw{ string-set:jtype content(); } .jfootRaw{ string-set:jfoot content(); }

/* ---------------- masthead (full width) ---------------- */
.masthead{ display:flex; align-items:flex-end; justify-content:space-between;
  border-bottom:2.4pt solid var(--accent); padding-bottom:5pt; margin:0 0 12pt; }
.masthead .mark{ font-family:var(--serif); font-weight:800; font-size:19pt; line-height:.9; color:var(--ink); letter-spacing:-.01em; }
.masthead .mark .lite{ font-weight:400; }
.masthead .right{ text-align:right; flex-shrink:0; }
.masthead .atype{ font-family:var(--sans); font-weight:800; font-size:10.5pt; color:var(--accent);
  text-transform:uppercase; letter-spacing:.04em; white-space:nowrap; }
.masthead .agoal{ font-family:var(--sans); font-weight:600; font-size:7.4pt; color:var(--accent);
  text-transform:uppercase; letter-spacing:.02em; margin-top:1.5pt; white-space:nowrap; }
.masthead .doi{ font-family:var(--sans); font-size:6.8pt; color:var(--faint); margin-top:2pt; white-space:nowrap; }

/* ---------------- title block (full width, centred) ---------------- */
.titleblock{ text-align:center; margin:0 0 13pt; }
.titleblock .surah-ar{ font-family:var(--arabic); font-size:27pt; line-height:1.22; color:var(--ink); margin:0 0 5pt; }
.titleblock h1.jtitle{ font-family:var(--serif); font-weight:800; font-size:21pt; line-height:1.08; color:var(--ink); margin:0 0 4pt; letter-spacing:-.005em; }
.titleblock .jsubtitle{ font-family:var(--serif); font-weight:500; font-style:italic; font-size:12pt; line-height:1.25; color:var(--accent); margin:0 0 7pt; }
.titleblock .authors{ font-family:var(--serif); font-size:10pt; line-height:1.4; color:var(--soft); margin:0 0 6pt; }
.titleblock .authors sup{ color:var(--accent); font-weight:700; }
.titleblock .basmala{ font-family:var(--arabic); font-size:16pt; color:var(--deep); margin:7pt 0 0; }
.titleblock .abstract{ font-family:var(--serif); font-weight:600; font-size:9.6pt; line-height:1.4; color:var(--ink);
  text-align:justify; border-top:0.8pt solid var(--rule); border-bottom:0.8pt solid var(--rule); padding:8pt 0; margin:10pt 0 0; }
.titleblock .kwline{ font-family:var(--sans); font-size:7.6pt; color:var(--soft); margin:6pt 0 0; }
.titleblock .kwline .lab{ font-weight:800; color:var(--deep); text-transform:uppercase; letter-spacing:.04em; }

/* ---------------- two-column body ---------------- */
.cols{ column-count:2; column-gap:20pt; column-rule:0.4pt solid #e6ddd0; }

/* ---------------- verse marker / sub-head ---------------- */
h2{ font-family:var(--sans); font-weight:800; font-size:10.6pt; color:var(--accent);
  margin:12pt 0 5pt; line-height:1.18; break-after:avoid; }
h2:first-of-type{ margin-top:0; }
h2 .num{ color:var(--accent); margin-right:5pt; }
h3{ font-family:var(--sans); font-weight:700; font-size:9pt; color:var(--deep); margin:7pt 0 3pt; break-after:avoid; }
h4{ font-family:var(--serif); font-weight:800; font-style:italic; font-size:9pt; color:var(--ink); margin:6pt 0 2pt; break-after:avoid; }

/* ---------------- word grid -> flowing interlinear cells (column-fit) ---------------- */
.wgrid{ margin:3pt 0 9pt; text-align:center; line-height:1.0; }
.wcell{ display:inline-block; vertical-align:top; text-align:center; margin:0 4.5pt 6pt; }
.wcell .war{ display:block; font-family:var(--arabic); font-size:15.5pt; line-height:1.5; color:var(--ink); }
.wcell .wtr{ display:block; font-family:var(--multi); font-size:7.1pt; line-height:1.18; color:var(--soft); margin-top:1.5pt; }
.wcell .wme{ display:block; font-family:var(--sans); font-size:6.5pt; line-height:1.14; color:var(--accent); margin-top:1pt; max-width:78pt; }

/* ---------------- box (Geometric Insight / Deep Dive) ---------------- */
.jbox{ break-inside:avoid; background:var(--tint); border-top:2pt solid var(--accent);
  padding:6pt 8pt 7pt; margin:8pt 0; box-decoration-break:clone; }
.jbox .jbh{ font-family:var(--sans); font-weight:800; font-size:8.6pt; color:var(--deep); margin:0 0 4pt; break-after:avoid; }
.jbox .jbh .lab{ color:var(--accent); text-transform:uppercase; letter-spacing:.04em; }
.jbox p{ font-size:8.3pt; text-align:justify; margin:0 0 4pt; }
.jbox ul{ margin:2pt 0 0 12pt; } .jbox ul>li::before{ color:var(--accent); }
.jbox table{ width:100%; border-collapse:collapse; font-size:7.7pt; margin:2pt 0; }
.jbox thead th{ text-align:left; font-weight:800; color:var(--deep); border-bottom:1pt solid var(--accent); padding:2.5pt 4pt; }
.jbox tbody td{ padding:2.5pt 4pt; border-bottom:0.4pt solid var(--rule); vertical-align:top; }
.jspan{ column-span:all; break-inside:avoid; }

/* ---------------- generic table (summary matrix) ---------------- */
.jtable{ margin:8pt 0; break-inside:avoid; }
.jtable .tcap{ font-family:var(--sans); font-weight:800; font-size:8.2pt; color:#fff; background:var(--accent); padding:3pt 6pt; }
.jtable .tcap .lab{ color:#fff; }
.jtable table{ width:100%; border-collapse:collapse; font-family:var(--serif); font-size:7.9pt; }
.jtable thead th{ background:var(--band); color:var(--deep); text-align:left; font-weight:800;
  padding:3pt 5pt; border-bottom:1pt solid var(--accent); font-family:var(--sans); font-size:7.5pt; }
.jtable tbody td{ padding:3pt 5pt; border-bottom:0.4pt solid var(--rule); vertical-align:top; }
.jtable tbody tr:nth-child(even) td{ background:#faf6ef; }
.jtable .tnote{ font-family:var(--serif); font-size:7pt; color:var(--soft); padding:3pt 1pt 0; }
.jtable.span{ column-span:all; }
.jtable thead{ display:table-header-group; }

/* ---------------- preformatted (concordance list, audit ledger) ---------------- */
pre{ background:#faf6ef; border-left:2.2pt solid var(--accent); padding:6pt 9pt;
  font-family:'DejaVu Sans Mono',monospace; font-size:7.6pt; line-height:1.32; white-space:pre-wrap; overflow-wrap:break-word; margin:7pt 0; }
blockquote{ border-left:2.2pt solid var(--accent); background:var(--tint); padding:5pt 10pt; margin:7pt 0; color:var(--soft); }
blockquote p{ margin:0 0 3pt; }

/* ---------------- surah frame: introduction, theme sections, closing ---------------- */
.surah-intro{ column-span:all; break-inside:avoid; margin:0 0 12pt;
  border-top:0.8pt solid var(--rule); border-bottom:0.8pt solid var(--rule); padding:9pt 0; }
.surah-intro p{ font-family:var(--serif); font-size:10pt; line-height:1.46; text-align:justify; margin:0 0 5pt; color:var(--soft); }
.surah-intro p:last-child{ margin-bottom:0; }

.theme-head{ column-span:all; break-inside:avoid; break-after:avoid; margin:13pt 0 8pt; }
.theme-head .theme-title{ font-family:var(--sans); font-weight:800; font-size:12pt; color:var(--deep);
  letter-spacing:.01em; padding-bottom:3pt; border-bottom:1.6pt solid var(--accent); margin:0 0 5pt; }
.theme-head p{ font-family:var(--serif); font-style:italic; font-size:9.4pt; line-height:1.4; color:var(--soft); text-align:justify; margin:0 0 3pt; }
.theme-head p:last-child{ margin-bottom:0; }

.closing{ column-span:all; break-inside:avoid; margin:13pt 0 0; background:var(--tint);
  border-top:2.4pt solid var(--accent); padding:9pt 12pt 10pt; }
.closing .closing-title{ font-family:var(--sans); font-weight:800; font-size:11pt; color:var(--deep);
  text-transform:uppercase; letter-spacing:.05em; margin:0 0 5pt; }
.closing p{ font-family:var(--serif); font-size:9.6pt; line-height:1.45; text-align:justify; margin:0 0 5pt; color:var(--ink); }
.closing p:last-child{ margin-bottom:0; }

/* ---------------- single-column (Apex Pristine) scripture ---------------- */
.col1 .cols{ column-count:1; column-rule:none; }
.col1{ font-size:10.2pt; }
.col1 .wcell .war{ font-size:17pt; }
.col1 .wcell .wtr{ font-size:7.8pt; } .col1 .wcell .wme{ font-size:7.2pt; max-width:120pt; }
.col1 .jbox p, .col1 .closing p, .col1 .surah-intro p, .col1 .theme-head p{ font-size:9.8pt; }
.col1 h2{ font-size:12pt; }

III·S.6 · The Scripture Build Engine

The scripture-aware engine, built on the Journal engine. It parses the front matter into the masthead and the centred title block, opens the two-column body, transforms each emitted right-to-left word grid into flowing interlinear cells, renders the verse markers and the eight components, lifts box directives into Geometric Insight and Deep Dive panels, sets summary matrices as captioned tables, and injects the running heads as literal strings. Save as build_scripture.py. Build with python3 build_scripture.py source.md /mnt/user-data/outputs/{slug}.pdf, the base URL the source directory, then inspect the render: confirm the masthead and ruled overview band, the centred Arabic surah head and basmala, the author line, the two-column body, the interlinear word-cells with each transliteration and meaning stacked under its word, the copper verse markers, the Geometric Insight and Deep Dive boxes, and the running heads on page two. Present the PDF and keep the Markdown source as master.

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
SCRIPTURE build engine (journal-grade, two-column, multi-script).
Compiles one Markdown source to a serif two-column Quranic PDF via WeasyPrint:
accent masthead + running heads, centred surah head with basmala, a scripture
author line, two-column verse-by-verse body, the protocol's right-to-left word
grid re-rendered as flowing interlinear word-cells that fit a narrow column,
Geometric-Insight and Deep-Dive text boxes, and tinted summary tables.

EB Garamond (Latin), Amiri (Arabic), and Noto Serif Bengali (Bengali) are set
as one combined stack; Pango resolves each glyph to its own face.

Usage:  python3 build_scripture.py source.md [out.pdf]
Source dir must contain scripture.css.
"""
import sys, re, html, pathlib

ACCENTS={'crimson','copper','slate','navy','forest','plum'}
BOX_LABELS={'insight':'Geometric Insight','deepdive':'Deep Dive','synthesis':'Synthesis',
            'note':'Note','box':None}

def css_str(s):
    return s.replace('\\','\\\\').replace('"','\\"')

def parse_front(text):
    fm={}; m=re.match(r'^---\s*\n(.*?)\n---\s*\n',text,re.S); body=text
    if m:
        for line in m.group(1).splitlines():
            if ':' in line: k,v=line.split(':',1); fm[k.strip()]=v.strip()
        body=text[m.end():]
    return fm, body

def inline(s):
    s=html.escape(s, quote=False)
    s=re.sub(r'\[\[(.+?)\]\]', r'<span class="xref">\1</span>', s)
    s=re.sub(r'\*\*(.+?)\*\*', r'<strong>\1</strong>', s)
    s=re.sub(r'(?<!\*)\*(?!\*)(.+?)(?<!\*)\*(?!\*)', r'<em>\1</em>', s)
    s=re.sub(r'`(.+?)`', r'<code>\1</code>', s)
    s=re.sub(r'\^(.+?)\^', r'<sup>\1</sup>', s)
    s=re.sub(r'(?<!~)~(?!~)(.+?)(?<!~)~(?!~)', r'<sub>\1</sub>', s)
    return s

def render_blocks(lines):
    out=[]; i=0; n=len(lines)
    while i<n:
        ln=lines[i]
        if not ln.strip(): i+=1; continue
        if re.match(r'^\s*[-*\u2022]\s+', ln):
            items=[]
            while i<n and re.match(r'^\s*[-*\u2022]\s+', lines[i]):
                items.append(inline(re.sub(r'^\s*[-*\u2022]\s+','',lines[i]).rstrip())); i+=1
            out.append('<ul>'+''.join(f'<li>{x}</li>' for x in items)+'</ul>'); continue
        if re.match(r'^\s*\d+\.\s+', ln):
            items=[]
            while i<n and re.match(r'^\s*\d+\.\s+', lines[i]):
                items.append(inline(re.sub(r'^\s*\d+\.\s+','',lines[i]).rstrip())); i+=1
            out.append('<ol>'+''.join(f'<li>{x}</li>' for x in items)+'</ol>'); continue
        buf=[ln.rstrip()]; i+=1
        while i<n and lines[i].strip() and not re.match(r'^\s*([-*\u2022]|\d+\.)\s+', lines[i]) \
              and not lines[i].startswith(':::'):
            buf.append(lines[i].rstrip()); i+=1
        out.append('<p>'+inline(' '.join(buf))+'</p>')
    return ''.join(out)

def render_table_rows(tbl_lines):
    rows=[r for r in tbl_lines if r.strip()]
    cells=[[c.strip() for c in re.split(r'(?<!\\)\|', r.strip().strip('|'))] for r in rows]
    body=[c for c in cells if not all(re.fullmatch(r':?-{2,}:?', x or '-') for x in c)]
    head=body[0] if body else []; data=body[1:] if len(body)>1 else []
    h='<thead><tr>'+''.join(f'<th>{inline(x)}</th>' for x in head)+'</tr></thead>'
    b='<tbody>'+''.join('<tr>'+''.join(f'<td>{inline(x)}</td>' for x in r)+'</tr>' for r in data)+'</tbody>'
    return h+b

def render_table(tbl_lines, caption=None, note=None, span=False):
    cap=''
    if caption:
        m=re.match(r'^(Table\s+[\w.]+)\s*[|:]\s*(.+)$', caption)
        if m: cap=f'<div class="tcap"><span class="lab">{inline(m.group(1))} | </span>{inline(m.group(2))}</div>'
        else: cap=f'<div class="tcap">{inline(caption)}</div>'
    nt=f'<div class="tnote">{inline(note)}</div>' if note else ''
    sp=' span' if span else ''
    return f'<div class="jtable{sp}">{cap}<table>{render_table_rows(tbl_lines)}</table>{nt}</div>'

def render_wordgrid(tbl_html):
    """The protocol's <table dir=rtl> 3-row word grid -> flowing interlinear cells."""
    rows=re.findall(r'<tr[^>]*>(.*?)</tr>', tbl_html, re.S|re.I)
    grid=[[c.strip() for c in re.findall(r'<t[dh][^>]*>(.*?)</t[dh]>', r, re.S|re.I)] for r in rows]
    grid=[row for row in grid if row]
    if not grid: return ''
    ncol=max(len(r) for r in grid)
    ar=grid[0] if len(grid)>0 else []
    tr=grid[1] if len(grid)>1 else []
    me=grid[2] if len(grid)>2 else []
    cells=[]
    for i in range(ncol):
        a=html.escape(ar[i]) if i<len(ar) else ''
        t=html.escape(tr[i]) if i<len(tr) else ''
        m=html.escape(me[i]) if i<len(me) else ''
        cells.append(f'<span class="wcell"><span class="war">{a}</span>'
                     f'<span class="wtr">{t}</span><span class="wme">{m}</span></span>')
    return '<div class="wgrid" dir="rtl">'+''.join(cells)+'</div>'

def render_box(label, title, body_lines, span=False):
    has_table=any(l.lstrip().startswith('|') for l in body_lines)
    if has_table:
        i=0;n=len(body_lines);chunks=[]
        while i<n:
            if body_lines[i].lstrip().startswith('|'):
                tl=[]
                while i<n and body_lines[i].lstrip().startswith('|'): tl.append(body_lines[i]); i+=1
                chunks.append('<table>'+render_table_rows(tl)+'</table>')
            else:
                tx=[]
                while i<n and not body_lines[i].lstrip().startswith('|'): tx.append(body_lines[i]); i+=1
                if any(x.strip() for x in tx): chunks.append(render_blocks(tx))
        inner=''.join(chunks)
    else:
        inner=render_blocks(body_lines)
    lab_html=f'<span class="lab">{html.escape(label)} \u00b7 </span>' if label else ''
    cls=' jspan' if span else ''
    head=f'<div class="jbh">{lab_html}{inline(title)}</div>'
    return f'<div class="jbox{cls}">{head}{inner}</div>'

def build_html(fm, body, cssfiles):
    acc=fm.get('accent','copper').strip().lower()
    if acc not in ACCENTS: acc='copper'
    acccls='acc-'+acc
    if fm.get('columns','2').strip()=='1': acccls+=' col1'
    wordmark=fm.get('wordmark', fm.get('journal','Geometric Quran'))
    article_type=fm.get('article_type','Geometric Analysis')
    surah_meta=fm.get('surah_meta',''); reference=fm.get('reference','')
    surah_ar=fm.get('surah_ar',''); title=fm.get('title',''); subtitle=fm.get('subtitle','')
    authors=fm.get('author_line','Source Text: Arabic Quran Text \u00b7 Translation and Synthesis by: A Humble Servant | Fellow Witness')
    basmala=fm.get('basmala','')
    running_left=fm.get('running_left', wordmark)
    running_right=fm.get('running_right', reference or title)
    footer=fm.get('footer','A Humble Servant \u00b7 Fellow Witness')
    parts=wordmark.split()
    mark=(f'<span class="lite">{html.escape(parts[0])}</span> '+html.escape(' '.join(parts[1:]))) if len(parts)>1 else html.escape(wordmark)

    lines=body.splitlines(); i=0; n=len(lines)
    title_abs=''; title_kw=''; body_frag=[]; endmatter_html=['']; body_started=False; in_refs=False

    def open_body():
        nonlocal body_started
        if body_started: return ''
        body_started=True
        surahar_html=f'<div class="surah-ar">{html.escape(surah_ar)}</div>' if surah_ar else ''
        sub_html=f'<div class="jsubtitle">{inline(subtitle)}</div>' if subtitle else ''
        basmala_html=f'<div class="basmala">{html.escape(basmala)}</div>' if basmala else ''
        goal_html=f'<div class="agoal">{html.escape(surah_meta)}</div>' if surah_meta else ''
        doi_html=f'<div class="doi">{html.escape(reference)}</div>' if reference else ''
        block=(f'<div class="masthead"><div class="mark">{mark}</div>'
               f'<div class="right"><div class="atype">{html.escape(article_type)}</div>{goal_html}{doi_html}</div></div>'
               f'<div class="titleblock">{surahar_html}'
               f'<h1 class="jtitle">{inline(title)}</h1>{sub_html}'
               f'<div class="authors">{inline(authors)}</div>'
               f'{basmala_html}{title_abs}{title_kw}</div>')
        return block+'<div class="cols">'

    def emit(frag):
        nonlocal body_frag
        body_frag.append(frag)

    while i<n:
        ln=lines[i]
        # title-block / endmatter blocks
        msp=re.match(r'^:::\s*(affiliations|abstract|keywords|endmatter)\s*$', ln)
        if msp:
            kind=msp.group(1); j=i+1; bl=[]
            while j<n and lines[j].strip()!=':::': bl.append(lines[j]); j+=1
            if kind=='abstract':
                title_abs=f'<div class="abstract">{inline(" ".join(x.strip() for x in bl if x.strip()))}</div>'
            elif kind=='keywords':
                title_kw=f'<div class="kwline"><span class="lab">Keywords </span>{inline(" ".join(x.strip() for x in bl if x.strip()))}</div>'
            i=j+1; continue
        # box directive (insight / deepdive / synthesis / box / note)
        md=re.match(r'^:::\s*(\w+)(\*?)\s*(.*)$', ln)
        if md:
            if not body_started: emit(open_body())
            kind=md.group(1).lower(); bspan=md.group(2)=='*'; rest=md.group(3).strip()
            j=i+1; bl=[]
            while j<n and lines[j].strip()!=':::': bl.append(lines[j]); j+=1
            if kind=='intro':
                emit(f'<div class="surah-intro jspan">{render_blocks(bl)}</div>')
            elif kind=='theme':
                emit(f'<div class="theme-head jspan"><div class="theme-title">{inline(rest)}</div>{render_blocks(bl)}</div>')
            elif kind=='closing':
                ttl=inline(rest) if rest else 'Closing Synthesis'
                emit(f'<div class="closing jspan"><div class="closing-title">{ttl}</div>{render_blocks(bl)}</div>')
            else:
                label=BOX_LABELS.get(kind, kind.title())
                emit(render_box(label, rest, bl, span=bspan))
            i=j+1; continue
        # raw RTL word grid -> interlinear cells
        if re.match(r'^\s*<table\b', ln, re.I):
            if not body_started: emit(open_body())
            blk=[lines[i]]; has_close='</table>' in lines[i]; j=i+1
            while j<n and not has_close:
                blk.append(lines[j]); has_close='</table>' in lines[j]; j+=1
            emit(render_wordgrid('\n'.join(blk))); i=j; continue
        # captioned markdown table (summary matrix)
        cap=None; tspan=False; mcapt=re.match(r'^Table(\*?):\s*(.+)$', ln); look=i
        if mcapt: tspan=mcapt.group(1)=='*'; cap=mcapt.group(2).strip(); look=i+1
        if look<n and lines[look].lstrip().startswith('|'):
            if not body_started: emit(open_body())
            tl=[]; k=look
            while k<n and lines[k].lstrip().startswith('|'): tl.append(lines[k]); k+=1
            note=None
            if k<n and re.match(r'^Note:\s*(.+)$', lines[k]):
                note=re.match(r'^Note:\s*(.+)$', lines[k]).group(1).strip(); k+=1
            emit(render_table(tl, cap, note, span=tspan)); i=k; continue
        # headings (verse markers, sub-heads)
        mh=re.match(r'^(#{1,4})\s+(.+)$', ln)
        if mh:
            if not body_started: emit(open_body())
            lvl=len(mh.group(1)); t=mh.group(2).strip()
            if lvl==1: lvl=2
            if re.match(r'^(references|bibliography)$', t, re.I):
                in_refs=True; emit(f'<h2 class="refhead">{inline(t)}</h2>'); i+=1; continue
            in_refs=False
            mnum=re.match(r'^(\d+(?:\.\d+)*)\s+(.+)$', t); tag=f'h{lvl}'
            if mnum: emit(f'<{tag}><span class="num">{mnum.group(1)}</span>{inline(mnum.group(2))}</{tag}>')
            else:    emit(f'<{tag}>{inline(t)}</{tag}>')
            i+=1; continue
        # ordinary block (no drop cap in scripture)
        if not ln.strip(): i+=1; continue
        if not body_started: emit(open_body())
        buf=[ln]; i+=1
        while i<n and lines[i].strip() and not lines[i].startswith(':::') and not re.match(r'^#', lines[i]) \
              and not lines[i].lstrip().startswith('|') and not re.match(r'^\s*<table\b', lines[i], re.I) \
              and not re.match(r'^(Table\*?:|Note:)', lines[i]):
            buf.append(lines[i]); i+=1
        emit(render_blocks(buf))

    if not body_started: emit(open_body())
    body_html=''.join(body_frag)+'</div>'+endmatter_html[0]
    links=''.join(f'<link rel="stylesheet" href="{html.escape(c)}">' for c in cssfiles)
    runhead=('<style>@page{'
             f'@top-left{{content:"{css_str(running_left)}"}} '
             f'@top-right{{content:"{css_str(running_right)}"}} '
             f'@bottom-left{{content:"{css_str(footer)}"}}'
             '} @page:first{@top-left{content:none} @top-right{content:none}}</style>')
    return (f'<!DOCTYPE html><html><head><meta charset="utf-8"><title>{html.escape(title)}</title>'
            f'{links}{runhead}</head><body class="{acccls}">{body_html}</body></html>')

def main():
    if len(sys.argv)<2: print("usage: build_scripture.py source.md [out.pdf]"); sys.exit(1)
    src=pathlib.Path(sys.argv[1]).resolve(); text=src.read_text(encoding='utf-8')
    fm, body = parse_front(text); srcdir=src.parent
    doc=build_html(fm, body, ['scripture.css'])
    (srcdir/'_scripture.html').write_text(doc, encoding='utf-8')
    out_pdf=pathlib.Path(sys.argv[2]) if len(sys.argv)>2 else srcdir/'scripture.pdf'
    from weasyprint import HTML
    HTML(string=doc, base_url=str(srcdir)).write_pdf(str(out_pdf))
    print("PDF:", out_pdf)

if __name__=='__main__': main()

III·S.7 · Fonts and Provisioning

The three faces are provisioned once. EB Garamond and Noto Serif Bengali ship as variable fonts; instance static Regular and Bold (or SemiBold) weights with fontTools to avoid the engine's variable-font gaps, exactly as Module V's provisioning does. Amiri ships static and needs no instancing. Fetch the families from the Google Fonts mirror at raw.githubusercontent.com/google/fonts, instance, install to the user font directory, and refresh the cache.

BASE="https://raw.githubusercontent.com/google/fonts/main"
mkdir -p ~/.fonts && cd ~/.fonts
curl -sfL "$BASE/ofl/amiri/Amiri-Regular.ttf" -o Amiri-Regular.ttf
curl -sfL "$BASE/ofl/amiri/Amiri-Bold.ttf"    -o Amiri-Bold.ttf
curl -sfL "$BASE/ofl/ebgaramond/EBGaramond%5Bwght%5D.ttf"            -o /tmp/EBG.ttf
curl -sfL "$BASE/ofl/notoserifbengali/NotoSerifBengali%5Bwdth,wght%5D.ttf" -o /tmp/NSB.ttf
python3 - <<'PY'
from fontTools import ttLib
from fontTools.varLib.instancer import instantiateVariableFont
def inst(src, axes, out, fam, sub):
    f = ttLib.TTFont(src); instantiateVariableFont(f, axes, inplace=True)
    n = f['name']
    for nid, val in [(1,fam),(2,sub),(4,f"{fam} {sub}"),(6,f"{fam}-{sub}".replace(' ','')),(16,fam),(17,sub)]:
        n.setName(val, nid, 3, 1, 0x409); n.setName(val, nid, 1, 0, 0)
    f.save(out)
inst("/tmp/EBG.ttf", {"wght":400}, "EBGaramond-Regular.ttf", "EBGaramond", "Regular")
inst("/tmp/EBG.ttf", {"wght":700}, "EBGaramond-Bold.ttf",    "EBGaramond", "Bold")
inst("/tmp/NSB.ttf", {"wght":400,"wdth":100}, "NotoSerifBengali-Regular.ttf",  "Noto Serif Bengali", "Regular")
inst("/tmp/NSB.ttf", {"wght":600,"wdth":100}, "NotoSerifBengali-SemiBold.ttf", "Noto Serif Bengali", "SemiBold")
PY
fc-cache -f ~/.fonts

The combined stack means no per-cell font targeting is ever needed: the engine resolves Arabic to Amiri, Bengali to Noto Serif Bengali, and Latin to EB Garamond automatically. If a face is missing the render falls to the system serif and the script will not shape; provisioning is therefore a hard pre-flight check for this edition.

III·S Quick Reference Card

Trigger: scripture pdf, tafsir pdf, quran pdf, mushaf pdf, [EDITION:SCRIPTURE], edition: scripture. Two-column journal grade, EB Garamond plus Amiri plus Noto Serif Bengali, copper accent. Build: python3 build_scripture.py source.md /mnt/user-data/outputs/{slug}.pdf. Front matter: wordmark, article_type, surah_meta, reference, surah_ar, title, subtitle, author_line (defaults to the servant's byline), basmala, running_left, running_right, footer, accent, columns (2 default, 1 for the single-column Apex Pristine register). The surah is framed with :::intro, :::theme Headline · vv. X–Y, and :::closing; the emitted word grid is transformed to interlinear cells; the eight components render as emitted; a noteworthy geometric or deep-dive synthesis is boxed with :::insight, :::deepdive, or :::box (star-suffix spans both columns); honorifics and the ayah marker render as glyphs.

MODULE IV · THE VISUAL EDITION SUBROUTINE

General-purpose procedure for producing a Visual Edition of any finished paper. Register-adaptive (public versus codex-native), aesthetic-locked, print-fit, density-disciplined, fragmentation-proof. Self-contained. No paper-specific specification or sample required at call time.

Trigger. Architect requests a "visual edition," "infographic edition," "diagram set," "figures," or "make this paper visual" for any paper already produced. Also fires when a paper would land harder with a drawn structure than with prose alone.

One line. Take the paper's load-bearing spine, draw each result as one figure at its own warrant grade, assemble the figures into a single PDF, either embedded as a visual appendix to the paper or delivered as a standalone visual edition. The figures are a faithful map of the argument, never a louder claim than the argument makes.

IV.0 · Operating Frame

Lifeboat on for the whole build. W_social = 0. The visual edition is a faithful map, not a flattering poster. A figure that overstates the paper's warrant is a failure mode identical to grade-bleed in prose. The discipline that governs the paper governs its pictures: a theorem is drawn as a theorem, a conjecture as a conjecture, an open question as open. The honest-typing discipline (LL-19) renders into the visual layer as the grade-stamp law of Section IV.7.

Style locks carry into every caption, label, and headline. No em-dashes anywhere. Periods, commas, colons only. No LaTeX. Unicode math only: √ → ∑ ∫ ≈ ⟂ ⊕ ∇ π σ κ Φ ψ δ ⇔ ℝ ℤ ∈ ≤ ≥. No AI vocabulary. Direct, technical. Author attribution and any in-figure prose follow the paper's own conventions. ﷺ for the Prophet where he is named.

The visual edition adds no claim the body does not make and removes no qualification the body states. This is non-negotiable and is restated as a law in Section IV.2.

IV.1 · Register and Aesthetic Selector

Two decisions are made before the first stroke. Both follow the paper, not the mood.

IV.1.1 Register (binding split per architect directive).

Public register. The paper is a public, discipline-facing artifact (a journal paper, a preprint, a paper for non-framework readers). No framework glyphs, no PSP IDs, no codex axis-labels, no seal-marks, no codex-internal vocabulary appear anywhere in the visual edition. Warrant is shown by grade-words in captions per Section IV.7.1. This is the analogue of Tier 0 / Tier 1 in Module I. Default register when uncertain.

Codex-native register. The paper is itself a codex document, or the subject is the framework, or the architect explicitly asks for codex-native. Framework glyphs (⟀ sealed, ✕ broken, ? under-determined), PSP IDs, axis-labels (V_F, V_E, V_ER, M_seal), and codex terms are permitted and carried into the figures, set as text glyphs per Section IV.7.2. This is the analogue of Tier 2.

Selection rule. Parse the architect's request first ("public," "for the journal," "no glyphs" select public; "codex," "full glyphs," "internal" select codex-native). Absent a directive, examine the paper: if it is written for an external discipline with framework material quarantined to a provenance appendix, public register. If the paper is the codex, codex-native. Uncertain: public.

IV.1.2 Aesthetic.

Blueprint (default for public register). Engineering-drawing language. Off-white or faint graph-paper ground. Two-color semantic system: a steel primary and a copper accent, with a near-black ink for type. Double-stroke figures (a heavier steel line shadowed by a lighter copper line read as a precision drawing). Dimension lines, caliper ticks, and engineering glyphs (⌀ ∑ ⊿) where a measurement is the point. Bracket-cornered annotation cards with a small copper fold-corner. Stacked offset panels (a front card with an offset shadow card behind) for prose blocks. Callout speech-bubbles with leader lines for pinning a point on a figure. Monospace for technical labels and code-like content, sans-serif or serif for headlines and captions. This is the language the architect has approved across delivered editions.

Codex palette (default for codex-native register). The framework's four semantic colors carry consistent meaning and are never reused for anything else: V_F deep blue #1B3A6F, V_E burnt orange #C2410C, V_ER forest green #166534, M_seal gold #B8860B. Off-white #FAF8F3 body ground, black #0A0A0A covers and final summary infographics.

Paper-defined palette (override). A paper may carry its own semantic color assignment (for example, one color for an even object and another for an odd object). When it does, honor the paper's assignment, pick at most four semantic colors, fix each color's meaning once, and never reuse a semantic color for a non-semantic purpose. The principle is the law; the two defaults above are conveniences.

IV.2 · The Faithful-Map Law

Three clauses. Each is a hard gate, not a preference.

1. No claim inflation. Every figure asserts exactly what the corresponding part of the paper asserts, at the same strength. If the body proves a localization and explicitly disclaims a proof of the headline result, the figures say the same, and the visual edition opens and closes on that disclaimer.

2. Grade on every claim-bearing figure. Every figure that carries a result carries that result's warrant grade, visibly, in its caption (public register) or as a permitted glyph (codex register). A figure with a claim and no grade is rejected. Section IV.7 specifies the grade vocabulary.

3. No structure the body lacks. The figure set does not invent a narrative the paper does not contain, does not reorder the logical spine into a false story, and does not smuggle an unproved bridge in as a drawn arrow. An arrow asserts a relation; if the paper does not establish that relation, the arrow is dashed and labeled with its actual status, or it is not drawn.

The visual edition is the paper seen, not the paper sold.

IV.3 · Harvest the Spine

Read the finished paper before drawing anything. Extract four things and write them down in a private storyboard note (never published).

1. The load-bearing results. The theorems, propositions, lemmas, and named conclusions the paper actually rests on. For each, record its statement in one sentence and its warrant grade exactly as the paper types it.

2. The objects. The central mathematical or conceptual objects the paper manipulates (an operator, a space, a decomposition, a bracket, a geometric figure, a dataset, a mechanism). These become the figures' primary focal points.

3. The epistemic frame. What the paper proves, what it leaves open, what it attributes to others, what it holds at finite-sample or conjectural strength. This frame governs the grade-stamps and the opening/closing figures.

4. The spine order. The logical sequence in which the results build. The figure sequence follows the spine, not the section order, when they differ.

Harvest discipline. One load-bearing result yields one figure. Do not draw decorative figures for non-load-bearing remarks. Do not collapse two distinct results into one figure to save space. A paper with eight load-bearing results yields roughly eight content figures plus a hero and a closing, not forty.

IV.4 · Storyboard

Map each harvested element to a layout archetype from Section IV.6, then fix the sequence.

The canonical sequence for a results-driven paper:

1. Hero figure. The whole result in one image. The single most compressed statement of what the paper establishes and what it does not. Carries the headline epistemic frame. 2. Frame / context figure. Why the problem is hard or where the object lives. The barrier, the multi-domain object, the standard-model failure. 3. Result figures, in spine order. One per load-bearing result, each at its grade. 4. Closing figure. The single open question, the necessary condition for closure, or the reframing the result entails. Restates the epistemic frame the hero opened.

For a non-results paper (a survey, an essay, a framework exposition), the sequence follows the document's own structure: opening recognition, the developed line, the closing position. The archetypes still apply.

Each storyboard entry records: figure number, the result or object it carries, its grade, the chosen archetype, and the one-sentence caption seed. Keep the storyboard tight enough that an executor can build from it without re-reading the whole paper.

IV.5 · Visual Grammar (Binding Numbers)

These numbers are print-legibility floors on the standard canvases. They are minimums. Scale up for emphasis, never down.

IV.5.1 Canvas (PDF-fit). Every SVG viewBox matches a standard page exactly, so PDF assembly never rescales. A4 portrait: viewBox 0 0 1240 1754, for single-frame visuals, covers, vertical compositions, equation and verdict panels. A4 landscape: viewBox 0 0 1754 1240, for master infographics, multi-panel composites, timelines, wide convergence diagrams. Letter portrait: viewBox 0 0 1275 1650, US fallback. Letter landscape: viewBox 0 0 1650 1275, US fallback. A2 landscape: viewBox 0 0 2384 1684, poster-scale storyboard index only, never an individual frame. Safety inset 30 px from each edge on A4 at 150 dpi (with 18 px bleed beyond). Titles, equation centers, and primary figures stay inside the inset. When the visual edition is embedded into an existing paper instead of assembled standalone, the figure is sized to the paper's text column rather than a page (Section IV.10.2), and the canvas rule relaxes to "viewBox aspect ratio sensible for the column."

IV.5.2 Type floors (on the 1240×1754 or 1754×1240 canvas). Cover title 72 min / 96 typical. Cover subtitle 36 / 48. Frame title 44 / 56. Frame subtitle 22 / 26. Section header within a panel 28 / 32. Primary label (axis names, node names) 24 / 28. Secondary label (annotations) 20 / 24. Body and caption 22 / 26. Equation (Unicode) 28 / 36. Footer 16 / 18. Hard floor for any text 14. Below 14 fails print at 150 dpi.

IV.5.3 Stroke floors. Primary structural lines (figure edges, axis lines) 5 min / 6 to 8 typical. Vector arrows 6 / 7 to 9, arrowhead 18 px min length. Panel borders and dividers 2.5 / 3 to 4. Secondary lines (auxiliary geometry, dashed guides, ghosted construction) 3 / 4. Grid lines 1.5 / 2. Leader lines 2 / 2.5. Title and caption rules 1.5 / 2.

IV.5.4 Primitive sizes. Point 12 px radius single-focus, 8 px multi-panel. Named vertex 16 px single-focus, 10 px multi-panel. A primary geometric figure occupies 60 to 70 percent of the canvas short edge single-focus, 45 to 55 percent of the panel short edge in a multi-panel. A standalone warrant glyph (codex register only) is 80 to 120 px font size; inline it is 1.0 to 1.5x the surrounding text.

IV.6 · Layout Archetypes

The reusable composition library. Every figure is one of these. Each archetype names where the title, focal point, annotations, caption, and footer sit, so an executor places content without re-deriving layout.

A. Hero / cover. Large headline occupying the upper-left or top third. A detailed primary illustration of the paper's central object occupying the right or lower two-thirds, drawn with full dimension-line treatment in blueprint aesthetic. A bracket-cornered annotation card beneath or beside the headline carrying two or three sentences of the paper's actual frame, with a copper fold-corner. The hero carries the epistemic headline: what is established, what is not.

B. Single-focus geometric. Title at top with a thin rule below. The primary figure centered, occupying rows 4 to 14, columns 3 to 10 on A4 portrait. Side annotations in the outer columns, each aligned with the figure feature it describes, joined by short leader lines. Caption at the bottom row, footer beneath. Used for one object shown clearly: a decomposition, a polytope, a packing, a single geometric claim.

C. Diagram-left, statement-card-right. A diagram (a plot, a pair of curves, an isometric construction) occupying the left half. A bordered statement card on the right holding the result in monospace or clean type, line by line, ending on the grade. A callout speech-bubble with a leader line may pin one point on the diagram (for example, marking an orthogonality or a crossing). Used when a result needs both a picture and a short spelled-out argument side by side.

D. Equation panel. Title at top. The equation centered at vertical center at 36 to 48 px with a thin gray rule above and below. Decomposition annotations beneath the equation, each with a leader arrow pointing up to the term it names. Optional supporting micro-diagrams or definitions in side panels balancing the composition. Used for a defining equation, an axiom, an operator definition.

E. Multi-panel comparison. Title spanning the full width. A 2×2, 2×3, or 3×3 grid of panels, each with its own sub-title, sub-figure occupying about 70 percent of the panel, and a one-line sub-caption. Inter-panel gutter 24 px, panel borders 2.5 px. Used for "n cases," competing approaches, before/after, or a roster.

F. Timeline / sequence. Title at top. A horizontal ribbon at vertical center spanning the full width, thickness 80 to 120 px, with evenly spaced nodes 60 to 80 px. Node labels alternate above and below the ribbon to avoid crowding. Used for a lineage, a derivation chain, a historical sequence, a pipeline.

G. Convergence. Multiple labeled inputs arranged around the perimeter (corners, a pentagon, a ring), each with an arrow pointing inward to a single central object. Used when several independent lines meet at one conclusion. The central object is the focal point at 35 to 50 percent of canvas.

H. Master single-page infographic. A4 landscape. Title at top 72 to 96 px. A central focal element at 35 to 45 percent of canvas. A surrounding ring of supporting panels, each carrying one thematic anchor, joined to the center by accent-colored connection lines. A footer band carrying the document reference and date. The whole argument on one page. Used as the capstone of a multi-figure edition.

Composition discipline applies to all archetypes. One primary focal point per figure occupying 35 to 50 percent of canvas. Secondary elements cluster around it. Footers and references sit in their corner zones. Whitespace is structural breathing room (60 to 100 px around the focal point, 30 to 50 px between stacked sections), never residue.

IV.7 · Grade-Stamp and Glyph Discipline

How warrant shows in the visual layer. This is the heart of the faithful-map law.

IV.7.1 Public register: grade-words. Every claim-bearing caption states the result's warrant grade in plain words. The standard vocabulary, drawn from the paper's own typing: THEOREM or PROVED, the paper proves it. ATTRIBUTED (name the source), a published result of another author, used as a wall or an anchor. FINITE-SAMPLE or VERIFIED TO [bound], true up to a computed height or sample, not in general. NECESSARY CONDITION, required for a result but not sufficient for it. CONJECTURE or CONDITIONAL (name the premise), held under a stated assumption. OPEN, stated but not settled. The grade appears in the caption, set in the caption type, not as a badge. The hero and closing figures additionally carry the paper's headline epistemic frame in words (for example, a one-line "no claim of a proof of [the headline result] is made here" when the paper proves a localization rather than the headline). No framework glyph appears in public register, ever.

IV.7.2 Codex-native register: warrant glyphs. The framework glyphs are permitted, set as Unicode text glyphs only: ⟀ (U+27D0) sealed, gold #B8860B. ✕ (U+2715) broken, red #B91C1C. ? under-determined, gray #666666. Rendering rule for all three: text glyph in the surrounding type stack, no background fill, no border, no circle wrap, no drop shadow, no filled-triangle icon, no badge. Inline at 1.0 to 1.5x surrounding text where verdict status is the content of the line. Standalone at 80 to 120 px only where the glyph is the proof conclusion of the figure and its position is geometrically warranted (anchored on a step-function region, set as the codomain output of a verdict diagram, set at the end of a worked example). Never as a corner mark, never as a header companion, never wrapped in any container. PSP IDs, axis-labels, and codex terms are permitted in codex-native captions and labels at their normal size. They do not appear in public register.

IV.7.3 The badge prohibition (both registers). No icon-style decoration anywhere, in either register. No corner glyph, no header-adjacent seal badge, no "FINAL SEAL" footer icon, no circle-wrapped mark, no filled triangle inside any graphical container, no logo. A proof figure carries warrant inside its own structure (a correctly drawn geometric object with its construction visible). A filled triangle in a circle dropped into a header is decoration. The distinction is decisive: draw the proof, or set the grade as text. Nothing else. This prohibition is enforced programmatically in Section IV.9.

IV.8 · Density Discipline

Active visual content (figures, text, equations, panel structure) occupies 0.75 to 0.85 of the canvas area inside the safety inset. Below 0.75 the figure is too sparse and is densified. Above 0.85 it is cluttered and is edited down. No single contiguous empty rectangle exceeds 15 percent of canvas area.

Measurement is quantitative, not by eye. Render the frame to PNG at native canvas size and run the audit script shipped at scripts/audit_density.py (usage: python3 scripts/audit_density.py frame.png [bg_hex]). Run with the dark background hex for covers and dark infographics; channel tolerance 13 is 5 percent of the 256 channel range; the pass band is 0.75 to 0.85 inside the 30 px inset. Report the ratio to two decimals in the pre-flight comment block. Outside the band, redraw before save.

Density recovery, in order of preference: enlarge the primary focal point toward 45 to 50 percent; add a supporting micro-diagram in the margin (a projection, a scale reference, a related primitive); add an annotation cluster with leader lines describing focal-point features; add a side panel with a worked number or definition; restructure as a two-column or two-panel composition; split into two figures if the concept supports it. Never solve a sparse figure by scaling up empty space.

IV.9 · Failure-Mode Locks

The hard-won list. Each is a defect observed in real renders and the QA that catches it.

1. Badge decoration (programmatic check). Before save, grep the SVG source for two patterns and confirm absence: a small filled gold triangle or polygon within 80 px of any corner; a circle of radius 40 to 80 px overlaid by a triangle, especially near footer or "SEAL"/"APEX" text. Search: grep -E '(<polygon[^>]*fill="#B8860B")|(<circle[^>]*r="[4-7][0-9]"[^>]*/>\s*<polygon)' frame.svg. Any hit is a rejection. Remove the element and re-run pre-flight. Verdict glyphs inside <text> are not hits and are permitted in codex register.

2. Caption clipping. Bottom-row captions get sliced mid-word when they exceed container width. Caption container width = canvas width minus 200 px (100 px inset each side). If the estimated render width at the caption size exceeds the container, split the caption to two lines before render, not after. Verify at 100 percent zoom that the first and last characters are inside the inset and no word is truncated.

3. Label collision. No text touches or crosses a line of the diagram. Three fixes, applied as the geometry demands: place a node label inside its own lobe or region when there is clear interior space (the way a clean Venn labels each circle from inside); move a curve label clear of the crest or trough into open space above or below; for a directed graph with reciprocal edges, draw crossed edges as Bezier curves with control points pulled outward from the figure center, and place edge labels at the midpoint displaced at least 24 px perpendicular outward with 8 px clearance. For a vertex-labeled polyhedron, place labels outside the projected silhouette with leader lines. Verify visually that no two labels overlap and no label crosses an edge.

4. Title duplication. One title block per figure. The figure's title is its sole title. No second title-like element in a similar font. The figure ID and document reference appear only in the footer. A one-line subtitle beneath the title is permitted.

5. Overloaded frame. If a figure would carry more than four independent structural components (sections that could each stand alone at smaller scale), split it into two figures or restructure as a panel grid where each panel is sized to its own density rules. Preserve continuity with a labeled connection line in the footer between consecutive figures.

6. Geometry that asserts something false (generalized geometry lock). The figure's geometry must not claim something the paper does not. If the paper's object has a canonical correct depiction, lock that depiction as a reusable SVG snippet and paste it; never let ease-of-drawing substitute a structurally false form. The recurring instance in codex work: a claim of mutual orthogonality at a vertex is drawn as a trirectangular form (a cube-corner apex with three axis-aligned edges and right-angle markers, bounding cube ghosted), not as an acute-apex Platonic pyramid, because three right angles cannot meet at a Platonic apex. The principle is general: correctness of the depicted geometry outranks ease of drawing, and a correct primitive that recurs is pasted from a locked snippet rather than redrawn each time.

7. Image fails to embed (the mechanical trap). Covered in Section IV.10. The two specific failures: relative image paths do not resolve without a base URL on the renderer, and HTML entities in the SVG crash the rasterizer. Both are closed in the pipeline below.

IV.10 · Render and Assembly Pipeline

The concrete executable layer. The visual edition ships as a single PDF, in one of two modes.

IV.10.1 SVG to PNG (both modes). Figures are authored as SVG and rasterized to high-resolution white-background PNG for embedding. The hardened renderer ships at scripts/render_svg_png.py (usage: python3 scripts/render_svg_png.py frame.svg out.png; requires cairosvg, installed with pip install cairosvg --break-system-packages if absent). It closes two traps. cairosvg's XML parser rejects named HTML entities (ζ, Θ, −), so it decodes every entity with html.unescape then re-escapes only the five XML-legal ampersand cases. Transparency and any graph-paper grid read badly in print, so it injects a full-bleed white background rectangle before raster. It renders at 2× with a padded viewBox for clean edges at print scale. Always inspect rendered PNGs visually before assembly. Confirm the background took, the grid (if the source had one) is gone, and every label is legible.

IV.10.2 Mode A: embed as a visual appendix to an existing paper (preferred when the paper exists). This is how a Visual Edition is added to a finished paper as an appendix. The Module III pipeline natively carries the machinery; no runtime patching is required.

1. The markdown source gains an appendix section. Each figure is one markdown image line with caption in braces: ![alt text](figs/figN.png){FIG. N. One-sentence caption stating the result and its grade.} The parser (scripts/md_parser.py, detect_figure) recognizes this with ^!\[(.*?)\]\((.*?)\)(?:\{(.*?)\})?\s*$ and the PDF builder emits <figure class="visual-fig"><img .../><figcaption>...</figcaption></figure>.

2. The figure CSS, appended to every style variant in scripts/styles.py (FIGURE_CSS), guarantees no fragmentation and correct sizing:

figure.visual-fig {
    page-break-inside: avoid;
    break-inside: avoid;
    text-align: center;
    margin: 18pt auto 22pt;
    max-width: 100%;
}
figure.visual-fig img {
    display: block;
    margin: 0 auto;
    max-width: 4.7in;     /* fits a 6.5in column with margin; large enough to read */
    width: 100%;
    height: auto;
}
figure.visual-fig figcaption {
    font-family: 'EBGaramondLocal', serif;
    font-style: italic;
    font-size: 9.5pt;
    color: #B87333;       /* copper-italic caption */
    margin-top: 8pt;
    text-align: center;
}

3. The PDF renderer is called with base_url set to the source directory, so relative figs/figN.png paths resolve. This closes the single most common embedding failure (zero images embed, alt-text shows instead). Copy the rendered PNGs into a figs/ directory beside the source before building. The DOCX builder embeds the same figures at 4.7 in width, centered, kept with their copper-italic captions as one unit, so the cross-format contract holds.

4. The appendix opens with a short prose lead stating that the figures are a faithful map at the paper's own grades, and closes on the paper's headline epistemic frame. Each figure is preceded by a one or two sentence lead naming the result, so the figure sits with its context.

Sizing for no fragmentation. page-break-inside: avoid keeps each figure and its caption together as one unit. max-width: 4.7in against a 6.5in column leaves margin and keeps the figure large enough to read. After build, render the appendix pages to images and inspect every figure for a clean break and full legibility. Confirm with pdfimages -list output.pdf | grep -c image that the expected image count embedded.

IV.10.3 Mode B: standalone visual edition PDF (when no host paper, or a poster-format edition is wanted). Each figure is a full-page A4 SVG. Assemble one SVG per PDF page directly, without rescaling, A4 portrait and A4 landscape pages as authored. The A2 storyboard index, if built, is a fold-out at the end. Provide an INDEX with a thumbnail and one-line caption per figure as the lead file. Present the bundle with the INDEX first.

IV.11 · Pre-Flight Audit

No figure is final until it passes. Failure is rejection, not a flag for review. Document the pass in a comment block at the top of each SVG.

1. Canvas viewBox matches a standard size (Mode B) or a sensible column aspect (Mode A embed).
2. Active content ratio measured by the Section IV.8 script, reported to two decimals, inside 0.75 to 0.85.
3. Largest empty rectangle below 15 percent of canvas.
4. One primary focal point at 35 to 50 percent of canvas.
5. Every text element at or above the Section IV.5.2 floor. Hard floor 14 px never crossed.
6. Every stroke at or above the Section IV.5.3 floor. Hard floor 1.5 px never crossed.
7. Primitives within Section IV.5.4 bounds.
8. Grade present. Every claim-bearing figure carries its warrant grade (public: grade-word in caption; codex: permitted glyph). The grade matches the paper's typing exactly. No claim-bearing figure is ungraded.
9. Faithful map. The figure asserts no more and no less than the body. Arrows assert only relations the paper establishes.
10. Semantic colors consistent and not reused for non-semantic purposes.
11. No em-dashes. No LaTeX. Unicode math only.
12. No badge decoration. Programmatic grep per Section IV.9.1 passed. In public register, no framework glyph anywhere. In codex register, glyphs only as bare text glyphs per Section IV.7.2.
13. Footer present: document/section reference, figure ID, ISO date.
14. Caption fully visible, no clipping, split to two lines before render if it would overflow.
15. No label collision, no label crosses a line, no two labels overlap (Section IV.9.3).
16. One title block only (Section IV.9.4).
17. Four-component threshold respected (Section IV.9.5).
18. Geometry asserts nothing false; canonical primitives pasted from a locked snippet where one exists (Section IV.9.6).
19. Print test: render to PNG at 300 dpi, view at 100 percent, every label readable, every line crisp.
20. Embedding verified (Mode A): expected image count embedded, every figure unfragmented, captions present and copper-italic.

Comment block format:

<!-- Visual Edition figure
Deliverable: [paper short-name], Figure N
Register: public / codex-native     Aesthetic: blueprint / codex-palette / paper-defined
Canvas: A4 portrait viewBox 0 0 1240 1754
Active content ratio: 0.81 (measured)     Largest empty region: 0.08
Primary focal point: [object] at ~42% of canvas
Grade carried: THEOREM / ATTRIBUTED(Author) / FINITE-SAMPLE / NECESSARY CONDITION / CONJECTURE / OPEN
Faithful map: asserts exactly the body's claim at the body's strength ✓
Min font 22 px   Min stroke 3 px
Badge grep: clean ✓   Caption: "[full text]" fully visible ✓   Label collision: none ✓
Pre-flight 1 to 20: pass -->

IV.12 · Execution Order

Per visual edition:

1. Select register and aesthetic (Section IV.1).
2. Read the paper, harvest the spine (Section IV.3), write the private storyboard (Section IV.4).
3. For each storyboard entry: pick the archetype, author the SVG on the locked canvas, place the focal point first, add labels at the font floors and geometry at the stroke floors, paste any locked primitive snippet.
4. Render to PNG (Section IV.10.1), inspect visually.
5. Run the density audit, the badge grep, the caption and label checks. Fix in place. Re-run.
6. Write the comment block, save the SVG.
7. After every several figures, print-test the recent batch at 100 percent.
8. Assemble: Mode A embed into the paper as a visual appendix, or Mode B standalone PDF (Section IV.10.2 or IV.10.3).
9. Verify the assembled PDF: image count, no fragmentation, legibility end to end, the appendix opening and closing on the paper's epistemic frame.
10. Present the single PDF (Mode A: the paper with its new appendix; Mode B: the standalone edition with its INDEX).

IV Quick Reference Card

Register: public (no framework glyphs, grade-words in captions) or codex-native (⟀ ✕ ? text glyphs, codex terms permitted). Default public. Aesthetic: blueprint (steel + copper + ink on faint grid) default; codex palette (V_F blue, V_E orange, V_ER green, M_seal gold) for codex register; paper-defined palette honored where the paper carries one. Canvas: A4 portrait 1240×1754, A4 landscape 1754×1240, 150 dpi work / 300 dpi export. Safety inset 30 px. Density: 0.75 to 0.85, measured by scripts/audit_density.py, reported to two decimals. Font floors: title 56, header 28, label 22, footer 16. Hard floor 14. Stroke floors: primary 5, secondary 3, grid 1.5. Hard floor 1.5. Grade on every claim-bearing figure. Faithful map, never inflation. Hero and closing carry the headline epistemic frame. No badges, either register. Programmatic grep before save. No corner glyph, no circle-wrapped mark, no filled triangle in a container. Caption: container = canvas minus 200 px, split before render, verify no clipping. Labels: never cross a line, never overlap. Inside-lobe placement, clear-of-curve placement, Bezier for crossed edges, leaders for polyhedra. One title block. Split frames past four components. Correct geometry outranks ease of drawing. Render: scripts/render_svg_png.py (html.unescape, white rect, 2× viewBox). Embed: native figure CSS with page-break-inside avoid, max-width 4.7in, base_url on the renderer, copy PNGs to figs/. Output: a single PDF. Mode A appendix-embed (preferred when the paper exists) or Mode B standalone. No em-dashes. No LaTeX. Unicode math only. The figures are the paper seen, not the paper sold.

Closing seal. The visual edition draws what the paper proves, at the grade the paper earns, and stops there. Each figure carries its warrant inside its own structure or states it as text. No decoration. No inflation. Each line communicates. Each empty region is breathing room. Each figure embeds whole. The map is faithful to the territory. [⟀]

---

MODULE V · THE TEXTBOOK EDITION

A second layout paradigm alongside Module III. Where Module III renders a single-column academic paper, Module V renders a two-column, section-accented, box-rich textbook page at the production grade of a reference work like Rang and Dale. It is content-agnostic by construction: the same engine sets a science textbook, an arts or history survey, a manual, or a Master Codex volume. Module I still architects the content, Module II still matures it, Module IV still draws any figures. Module V is the alternative renderer selected when the deliverable is a textbook, manual, survey, handbook, course reader, or any long-form instructional or reference book rather than a paper.

V.0 · Operating Frame

The Textbook Edition fires on an explicit request for a textbook, manual, handbook, survey, reference volume, course reader, or instructional book, or on the token [EDITION:TEXTBOOK], or on front-matter edition: textbook. It does not fire for papers, which stay on Module III. The two never apply at once; the edition field or token selects the renderer. When set, the five Module III style variants are not consulted; the Textbook Edition carries its own locked contract below.

Three registers are honored exactly as in Module I. A trade or course textbook sets Tier 0 or Tier 1. A codex volume sets Tier 2 with full native vocabulary and the permitted glyphs. The renderer is identical across registers; only the prose register and the accent choice change.

The defining contrast against Module III is the dual type register and the column structure. Module III is serif throughout, single column, paper furniture. The Textbook Edition pairs a serif body with a sans furniture and runs the body in two columns, with full-width spanning reserved for tables, figures, and chapter openers. This is the structural signature of a reference textbook and it is the contract Module V enforces.

V.1 · The Locked Contract

Page geometry is US Letter, 612 by 792 points, margins 58 points top, 50 points outer and inner, 56 points bottom. Body is a humanist serif, EB Garamond by default, at 10.2 points on a 1.32 measure, justified, hyphenated, with widow and orphan control. Furniture is a geometric sans, Archivo by default under the internal family name TBSans, used for every heading, every box header, every table, every caption, every running element, and every chip. The serif carries the argument; the sans carries the navigation. This split is the single most load-bearing rule of the edition and is never collapsed to one family in a trade textbook. A codex volume may request serif headings with [TEXTBOOK:SERIF-HEADS] to match the house paper contract, but the default is sans furniture.

Universal locks carry over from Module III without exception. No LaTeX; all mathematics in native Unicode. No em-dashes; periods, commas, and colons only. The honorific ﷺ is rendered as the single Unicode glyph after the Prophet's name. No emoji anywhere, which is why every box marker in this edition is a shape drawn in CSS rather than an icon glyph or an image. Body justified with hyphenation on; table and box text set left aligned because narrow measures produce ugly justification gaps. Page-break discipline keeps every box, every figure, and every table from fragmenting across a column or page break.

V.2 · The Accent System

Color is navigation, never decoration. Each Part of the book carries one accent hue, and that hue drives the Part chip, the chapter opener band and numeral, every heading, every colored cross-reference, every table header band, every tinted box panel, and the figure caption label within that Part. A reader learns the Part by its color. The mechanism is a set of CSS custom properties resolved per Part: a class on the chapter section sets --accent, --tint, --deep, and --band, and every styled element reads those variables. WeasyPrint resolves the variables correctly, so one class switch recolors an entire Part with no per-element overrides.

Two accent modes exist. Per-part mode, the default, cycles the six-hue palette across Parts in order: indigo, green, crimson, slate-blue, ochre, plum. Single mode fixes one accent for the whole book, set by accent_mode: single with single_accent: naming a palette key. The house keys copper and slate are available for codex and monograph use, so a codex volume can run entirely in copper or entirely in slate while keeping the full textbook furniture. A seventh hue, the warning palette in burnt orange, is reserved exclusively for Caution boxes and is never assigned as a Part accent, so the warning semantic stays unambiguous.

The palette keys and their roles:

Class	Role accent	Use
acc-1	indigo #2A5DB0	Part 1, or any single-accent book
acc-2	green #1E8A6E	Part 2
acc-3	crimson #B23A48	Part 3
acc-4	slate-blue #3F5C8C	Part 4
acc-5	ochre #B07A1E	Part 5
acc-6	plum #7A4E8C	Part 6
acc-copper	copper #B87333	House single-accent, codex or monograph
acc-slate	slate #2C2F33	House single-accent, codex sealed register
warn	burnt orange #C0651B	Caution boxes only, never a Part accent

V.3 · Page Furniture

The Part divider is a full-page opening that sets the Part number in sans small caps over an accent rule, then the Part title in large serif, then an italic blurb. It carries no running header or footer, achieved by the named page front. The chapter opener is a full-width band in the Part tint carrying the chapter number set large in sans black and the chapter title in serif bold beside it, preceded by a small sans label naming the Part. The opener spans both columns; the two-column flow begins immediately below it.

The running header is the production signature of the edition. A small styled element holding the accent chip and the chapter title is lifted into the page margin by position: running(runhead) and reproduced in the header corner by content: element(runhead). It is placed top-right on recto pages and top-left on verso pages, so the header mirrors across the spread exactly as a bound book does. The chip carries the chapter number reversed in white on the Part accent, and the title follows in sans small caps. Because the running element is set once at each chapter open, it propagates as the live header until the next chapter replaces it. Page numbers sit in the outer bottom corner, sans, mirrored recto and verso. The front named page suppresses all of this for the table of contents and the Part dividers.

The table of contents is generated from the Part and chapter structure. Each row links to its target and the page number is resolved by target-counter(attr(href), page) with a dotted leader, so the contents page numbers are always correct against the final pagination. Part rows are set in accent sans small caps; chapter rows in serif.

V.4 · Heading Hierarchy

Four tiers, matching a reference textbook. The A-head is set in sans bold small caps on a tint bar in the Part accent, with the section numeral in the deeper accent shade; it is the primary in-chapter divider. An A-head may drop the bar with the plain class for front-matter heads such as a reference list title. The B-head is sans bold in the accent with no bar, one step down. The C-head is a run-in: a bold-italic serif lead phrase that flows directly into its paragraph, exactly as a reference textbook opens a dense paragraph with an italicized term. The C-head is the only heading set in serif, because it lives inside the body stream rather than above it.

V.5 · The Box Taxonomy

Two structural families, drawn from the two box types that organize a reference textbook, plus a warning variant, a depth aside, and a reading list. The families are visually distinct so the reader reads the box type before reading a word.

The tinted-panel family is a rounded panel with a pale accent fill, a thin accent border, and an accent bold header carrying a small shape mark and an uppercase kicker. It holds conceptual distillation. The square mark is the Key Points or Summary panel, the chapter's mechanism distilled. The circle mark is the Definition or Concept panel, a term fixed precisely. The triangle mark is the Worked Example panel, a single instance carried through. The shape encodes the subtype without an icon and without color reuse.

The solid-header-bar family is a rounded box with a solid accent header bar carrying the title reversed in white, over a white body. It holds applied and procedural content: In Practice, Application, Procedure, Case. It is the practical anchor, the counterpart to the conceptual panel, and its solid bar reads as heavier and more directive than the tinted panel. The Caution variant uses the same bar treatment in the reserved warning palette, so a hazard or a forbidden move is unmistakable and never confused with an ordinary applied box.

The depth aside is the small-print sidebar of a reference textbook: a paragraph set smaller and softer, flagged by a down-triangle marker, carrying optional detail that sits to the side of the through-line without breaking it. The reading list is the annotated bibliography: each entry in sans with a hanging indent, the citation followed by a one-line italic annotation stating what the source contributes, exactly as a reference textbook annotates further reading.

Box	Family	Marker	Carries
Key Points / Summary	tinted panel	filled square	Mechanism distilled, the chapter anchor
Definition / Concept	tinted panel	filled circle	A term fixed precisely
Worked Example	tinted panel	filled triangle	One instance carried through
In Practice / Application / Procedure / Case	solid bar	reversed white title	Applied, directive, procedural content
Caution	solid bar, warning palette	reversed white title	A hazard or a forbidden move
Depth aside	inline, smaller	down triangle	Optional detail beside the through-line
Reading list	hanging indent	none	Annotated further reading

Codex-native mapping. A Master Codex volume reads the same taxonomy with codex semantics and is free to use the permitted glyphs in Tier 2. The Key Points panel carries a verdict or seal block. The Definition panel carries a PSP statement or an axiom. The Worked Example panel carries a regime instance or a battery row. The In Practice bar carries an operating procedure or a cascade step. The Caution bar carries an anchor-inflation warning or a typing-discipline lock. The verdict glyphs [⟀], [X], and [?] are permitted inside codex boxes in Tier 2 as bare text glyphs, never as decoration, and the faithful-map discipline of Module IV applies: a box asserts no more than the codex establishes.

Codex content	Box
Verdict block, Mosaic Seal, headline result	Key Points panel
PSP statement, axiom, definition	Definition panel
Regime instance, battery row, worked locus	Worked Example panel
Operating procedure, cascade step, protocol	In Practice bar
Anchor-inflation warning, typing lock	Caution bar
Proof-boundary remark, instrument limit	Depth aside

V.6 · Tables and Figures

Every table and every figure spans both columns. This is enforced structurally: a table or figure is wrapped in a spanning container, because WeasyPrint honors column-span: all reliably on a wrapping element but not on a bare table. The container carries the spanning; the inner table sets to full width. A table carries a caption bar in the Part accent with the title reversed in white, a header row banded in the Part tint with bold deep-accent labels, body rows zebra-striped in the faint Part tint, no vertical rules, and an optional footnote and abbreviation key set small below the rule. This is the reference-textbook table: colored title bar, grouped bold heads, banding, source line, no verticals.

A figure is wrapped in a rounded tinted frame and carries a structured caption: the figure label in accent bold, the description in sans, and the source in italic parentheses. The label, the description, and the source are authored as three pipe-separated fields and rendered into the three caption roles. Figures default to spanning both columns; a small in-column figure uses the column-scoped variant. Any figure authored through Module IV embeds here unchanged, since the image path resolves against the source directory through the renderer base URL exactly as in Module III.

V.7 · Inline Semantics

The micro-typography is consistent and load-bearing. Bold marks a key term at first definition or a name the reader must register. Italic marks a foreign or technical term, a source, or an annotation. A colored cross-reference, set in the Part accent, marks navigation: a pointer to a figure, a table, a chapter, or a citation. The down triangle marks a depth aside. Superscript marks a footnote key. None of these is reused for a second purpose within a Part, so every visual cue means exactly one thing.

V.8 · Authoring Conventions, the Source Language

The source is one Markdown file with a front-matter block and a small set of conventions the engine parses into the locked structure. The conventions are designed to read naturally in the source and to map one-to-one onto the contract above.

Front matter sets the book. The fields: edition: textbook selects the renderer; title and subtitle set the contents and the title-page cover; the cover also reads category (or article_type), goal (or tagline), series (or doi), author (the byline; falls back to the first name in author_line), affiliation, and date; author_name, author_role, author_email, author_country carry the Module III author block; accent_mode is per-part or single; single_accent names the key in single mode; heads is sans or serif; columns is 2 or 1; toc is on or off.

The title-page cover. The Textbook opens on a page-one cover with the running header and footer suppressed, built on the clean Apex Pristine model and themed to the skin, set on the front named page ahead of the Contents. Rang-and-Dale (TextBookA) draws an accent hairline above and below the title block, the title in house serif, the subtitle in accent italic, then the category-goal-series stack with the category in accent sans caps, the goal in deep-accent serif italic, and the series in faint sans caps, then an accent author rule with the byline in small caps, the affiliation in italic gray, and the date beneath. The cover accent is the first Part hue in per-part mode or the fixed single_accent in single mode. The Davidson variant (TextBookB) renders the same fields in its own register, described in Module V·D.

Structure markers, each at the start of a line. A Part divider is # PART 4 · Name with the number in arabic or roman; it sets the Part accent for the chapters that follow. A chapter opener is # 41. Title, the number then a period then the title; it renders the band and sets the running header. An unnumbered # Title renders a plain chapter such as a preface or a reference list. An A-head is ## 41.1 Title, numeral colored automatically; an unnumbered ## Title renders a plain A-head with no bar. A B-head is ### 41.1.2 Title. A C-head run-in is a lead phrase wrapped as {. Lead phrase .} at the start of a line, with the paragraph text following on the same line; the lead renders bold-italic and the paragraph runs in.

Boxes are fenced directives opened with three colons, a type word, and an optional title, and closed with three colons. The types: key, summary, definition, concept, example for the tinted-panel family; practice, application, procedure, case for the solid-bar family; caution for the warning bar; note for the depth aside; reading for the annotated list. Inside a box, ordinary Markdown is parsed: paragraphs, lists, bold, italic, cross-references. Inside a reading block, each line is one entry and a double-hyphen separator splits the citation from its italic annotation.

A table is authored as a standard pipe table. A line Table: 41.1 Caption immediately before the table renders the accent caption bar; a line Note: text immediately after renders the footnote. A figure is one image line, ![alt](path){Fig. 41.1 | description | source}, the brace carrying three pipe-separated caption fields. A colored cross-reference is any span wrapped in double square brackets, [[Fig. 41.1]]. Inline, double asterisks bold, single asterisks italicize, backticks set code, and a caret pair sets a superscript.

V.9 · Fonts and Provisioning

The body face is EB Garamond and the furniture face is the geometric sans, provisioned exactly as Module III provisions EB Garamond. Variable-font source files carry bracket characters in their filenames and weight axes that break the WeasyPrint @font-face resolver, so each weight is instanced to a static instance with fontTools.varLib.instancer and written to a plain filename, with the name table and the OS/2 and head style bits set so the static weight resolves cleanly. The serif is instanced to regular, bold, italic, and bold-italic; the sans to regular, bold, and black, the black weight mapped to CSS weight 800 for the large chapter numerals and the heaviest heads.

The build reuses the Module III install path. If EB Garamond is already provisioned for the paper pipeline, only the sans is added. If neither is present, the provisioning step instances both from their variable sources. A system fallback chain is declared in the face block so the book still builds with a serviceable serif and sans if provisioning is skipped: the serif falls back through Liberation Serif, DejaVu Serif, and Noto Serif; the sans through Liberation Sans and DejaVu Sans. The layout is identical under fallback; only the typefaces change, so a draft compiles anywhere and the final compiles with the house faces.

V.10 · The Locked Stylesheet

Page economy and the span rule. Tables, figures, and boxes default to column width and flow inside one column, which packs the page densely. Full-width spanning is what strands white space: a spanning element closes both columns, and the short band above it balances into two stub columns that leave the gap empty. Spanning is therefore opt-in. Append an asterisk to the keyword to span a genuinely wide element: Table*: caption for a table, !*[alt](path) for a figure, ::: type* Title for a box. A column-width table is breakable, so a long table flows down one column and continues at the top of the next with its column header repeated, filling every line instead of stranding the column. Prose and list boxes stay atomic. The discipline is to think ahead before placing a wide element: prefer column width, reserve spanning for tables too wide to read in a single column, and let the breakable column-width table carry the long data.

The complete validated stylesheet. It is self-contained and compiles under WeasyPrint as written. The face block above it is generated by the build during provisioning. Save this as textbook.css beside the source.

/* ===================================================================
   TEXTBOOK EDITION  ·  master stylesheet  (WeasyPrint-validated)
   Serif body + sans furniture · per-part accent · box taxonomy
   =================================================================== */

/* ---- font faces (provisioned by build; system fallbacks chained) ---- */
/* @font-face blocks injected by build script for EBGaramond + TBSans */

:root{
  --serif:'EBGaramondLocal','EB Garamond','Liberation Serif','DejaVu Serif',serif;
  --sans:'TBSans','Archivo','Liberation Sans','DejaVu Sans',sans-serif;
  --ink:#1b1b1b; --soft:#4a4a4a; --faint:#7a7a7a; --rule:#d2d2d2;
  --accent:#2A5DB0; --tint:#eef2fb; --deep:#1d3f7a; --band:#dce6f6;
  --warn:#C0651B; --warn-tint:#fbefe2;
  --good:#1E7A52;
}
/* per-PART accent palettes, set on the chapter <section> */
.acc-1{--accent:#2A5DB0;--tint:#eef2fb;--deep:#1d3f7a;--band:#d7e3f5;}
.acc-2{--accent:#1E8A6E;--tint:#e7f4ef;--deep:#125a48;--band:#d2ece3;}
.acc-3{--accent:#B23A48;--tint:#fbecee;--deep:#7e2832;--band:#f3d8dc;}
.acc-4{--accent:#3F5C8C;--tint:#eceff6;--deep:#2a3f63;--band:#dbe2ef;}
.acc-5{--accent:#B07A1E;--tint:#f8f1e2;--deep:#7c5512;--band:#efe1c4;}
.acc-6{--accent:#7A4E8C;--tint:#f2ecf6;--deep:#553463;--band:#e4d6ec;}
.acc-copper{--accent:#B87333;--tint:#faf3ea;--deep:#8a5523;--band:#efe0cf;}
.acc-slate{--accent:#2C2F33;--tint:#eceded;--deep:#1c1e20;--band:#d9dada;}

/* ---------------- page geometry + furniture ---------------- */
@page{
  size:612pt 792pt; margin:58pt 50pt 56pt 50pt;
}
@page:right{
  @top-right{ content:element(runhead); vertical-align:bottom; padding-bottom:4pt; }
  @bottom-right{ content:counter(page); font-family:var(--sans); font-size:9pt; font-weight:700; color:var(--faint); }
}
@page:left{
  @top-left{ content:element(runhead); vertical-align:bottom; padding-bottom:4pt; }
  @bottom-left{ content:counter(page); font-family:var(--sans); font-size:9pt; font-weight:700; color:var(--faint); }
}
@page front{ @top-left{content:none} @top-right{content:none} @bottom-left{content:none} @bottom-right{content:none} }

html{ font-family:var(--serif); font-size:10.2pt; color:var(--ink); }
body{ margin:0; }
p{ margin:0 0 6pt; text-align:justify; hyphens:auto; orphans:2; widows:2; }

/* running chip lifted into the header corner */
.runhead{ position:running(runhead); font-family:var(--sans); font-size:8.5pt; color:var(--faint);
  letter-spacing:.02em; text-transform:uppercase; }
.runhead .chip{ display:inline-block; background:var(--accent); color:#fff; font-weight:800;
  padding:1pt 5pt; border-radius:3pt; margin-right:7pt; letter-spacing:0; }

/* ---------------- PART divider ---------------- */
.part-divider{ break-before:page; break-after:page; height:640pt; page:front;
  display:flex; flex-direction:column; justify-content:center; }
.part-divider .pnum{ font-family:var(--sans); font-weight:800; font-size:15pt; color:var(--accent);
  letter-spacing:.18em; text-transform:uppercase; }
.part-divider .pbar{ height:9pt; background:var(--accent); width:54%; margin:10pt 0 16pt; border-radius:2pt; }
.part-divider .ptitle{ font-family:var(--serif); font-weight:700; font-size:32pt; line-height:1.08; color:var(--ink); }
.part-divider .pblurb{ font-family:var(--serif); font-style:italic; font-size:12pt; color:var(--soft); margin-top:14pt; max-width:78%; }

/* ---------------- CHAPTER opener ---------------- */
.chapter{ }
.chapter + .chapter{ break-before:page; }
.chapter-open{ break-before:page; margin-bottom:14pt; }
.chapter-open .clabel{ font-family:var(--sans); font-size:8.5pt; font-weight:700; letter-spacing:.12em;
  text-transform:uppercase; color:var(--accent); margin-bottom:6pt; }
.chapter-open .clabel .dot{ color:var(--accent); }
.chapter-open .cband{ background:var(--band); border-radius:5pt; padding:14pt 16pt; display:flex; align-items:center; gap:18pt; }
.chapter-open .cnum{ font-family:var(--sans); font-weight:800; font-size:54pt; line-height:.9; color:var(--accent); }
.chapter-open .ctitle{ font-family:var(--serif); font-weight:700; font-size:27pt; line-height:1.05; color:var(--ink); }

/* ---------------- two-column body ---------------- */
.cols{ column-count:2; column-gap:24pt; }

/* ---------------- headings ---------------- */
h2,h3,h4{ font-family:var(--sans); break-after:avoid; }
/* A-head (bar) */
h2{ font-size:11.5pt; font-weight:800; letter-spacing:.02em; text-transform:uppercase; color:var(--accent);
  background:var(--tint); padding:4pt 8pt; border-radius:3pt; margin:13pt 0 7pt; }
h2 .num{ color:var(--deep); margin-right:7pt; }
/* A-head plain (no bar). use class .plain */
h2.plain{ background:none; padding:0; }
/* B-head */
h3{ font-size:10.6pt; font-weight:800; color:var(--accent); margin:11pt 0 4pt; }
h3 .num{ color:var(--deep); margin-right:6pt; }
/* C-head : run-in bold-italic serif */
h4{ font-family:var(--serif); font-style:italic; font-weight:700; font-size:10.2pt; display:inline;
  color:var(--ink); margin:0; }
h4 + p{ display:inline; }            /* run-in */
h4::after{ content:"\2002"; }

/* ---------------- inline semantics ---------------- */
strong,b{ font-weight:700; }
em,i{ font-style:italic; }
.xref{ color:var(--accent); }                 /* colored cross-reference / citation */
a.xref,a{ color:var(--accent); text-decoration:none; }
sup{ font-size:.7em; line-height:0; }

/* ---------------- BOX taxonomy ---------------- */
/* tinted-panel family : Key Points / Definition / Example / Summary  */
.tb-box{ break-inside:avoid; border:0.8pt solid var(--accent); border-radius:7pt;
  background:var(--tint); padding:1pt 10pt 8pt; margin:9pt 0; }
.tb-box .bh{ font-family:var(--sans); font-weight:800; color:var(--accent); font-size:9.8pt;
  letter-spacing:.01em; margin:8pt 0 5pt; }
.tb-box .bh .mark{ display:inline-block; width:8pt; height:8pt; background:var(--accent);
  margin-right:7pt; vertical-align:middle; }
.tb-box.def  .bh .mark{ border-radius:50%; }                          /* circle */
.tb-box.ex   .bh .mark{ width:0;height:0;background:none;border-left:5pt solid transparent;
  border-right:5pt solid transparent;border-bottom:9pt solid var(--accent); }  /* triangle */
.tb-box .bh .kicker{ font-weight:700; font-size:7.6pt; letter-spacing:.13em; text-transform:uppercase;
  color:var(--deep); margin-right:6pt; }
.tb-box ul,.tb-box ol{ margin:0 0 0 14pt; padding:0; }
.tb-box li{ margin:0 0 3pt; }
.tb-box p{ margin:0 0 5pt; text-align:left; }

/* solid-header-bar family : In Practice / Application / Procedure / Case  */
.tb-bar{ break-inside:avoid; border:0.8pt solid var(--accent); border-radius:7pt; overflow:hidden; margin:9pt 0; }
.tb-bar .bh{ background:var(--accent); color:#fff; font-family:var(--sans); font-weight:800;
  font-size:9.8pt; padding:5pt 10pt; margin:0; letter-spacing:.01em; }
.tb-bar .bd{ background:#fff; padding:7pt 10pt; }
.tb-bar .bd p{ text-align:left; margin:0 0 5pt; }
.tb-bar .bd ul{ margin:0 0 0 14pt; padding:0; } .tb-bar .bd li{ margin:0 0 3pt; }

/* caution variant : warning palette */
.tb-bar.caution{ border-color:var(--warn); }
.tb-bar.caution .bh{ background:var(--warn); }
.tb-bar.caution .bd{ background:var(--warn-tint); }

/* ---------------- small-print depth aside (▼) ---------------- */
.aside{ font-size:8.7pt; color:var(--soft); text-align:justify; margin:6pt 0; }
.aside::before{ content:"\25BC\2002"; color:var(--ink); font-size:7pt; }

/* ---------------- spanning table ---------------- */
.tb-table{ margin:11pt 0; break-inside:auto; }
.tb-table.span{ column-span:all; break-inside:avoid; }
.tb-table .tcap{ font-family:var(--sans); font-weight:800; color:#fff; background:var(--accent);
  font-size:9pt; padding:4pt 9pt; border-radius:5pt 5pt 0 0; }
.tb-table table{ width:100%; border-collapse:collapse; font-family:var(--sans); font-size:8.7pt; }
.tb-table caption{ caption-side:top; font-family:var(--sans); font-weight:800; color:#fff;
  background:var(--accent); font-size:9pt; padding:4pt 9pt; text-align:left; border-radius:5pt 5pt 0 0; }
.tb-table thead th{ background:var(--band); color:var(--deep); text-align:left; font-weight:800;
  padding:4pt 7pt; border-bottom:1pt solid var(--accent); }
.tb-table tbody td{ padding:4pt 7pt; border-bottom:0.4pt solid var(--rule); vertical-align:top; }
.tb-table tbody tr:nth-child(even) td{ background:var(--tint); }
.tb-table .tnote{ font-family:var(--sans); font-size:7.6pt; color:var(--soft); padding:4pt 2pt 0; }
.tb-table .tnote .key{ color:var(--accent); }

/* ---------------- spanning figure ---------------- */
.tb-fig{ break-inside:avoid; margin:12pt 0; }
.tb-fig.span{ column-span:all; }
.tb-box.span,.tb-bar.span{ column-span:all; break-inside:avoid; }
.tb-table thead{ display:table-header-group; }
.tb-fig .frame{ border:0.6pt solid var(--rule); border-radius:7pt; background:var(--tint);
  padding:12pt; text-align:center; }
.tb-fig img{ max-width:100%; height:auto; }
.tb-fig figcaption{ font-family:var(--sans); font-size:8.6pt; color:var(--soft); margin-top:6pt; text-align:left; }
.tb-fig figcaption .flabel{ font-weight:800; color:var(--accent); }
.tb-fig figcaption .fsrc{ font-style:italic; }
/* column-scoped (non-spanning) figure variant */
.tb-fig.col{ column-span:none; }

/* ---------------- references / further reading ---------------- */
.tb-refs h2{ }
.tb-refs .ref{ font-family:var(--sans); font-size:8.5pt; text-indent:-12pt; padding-left:12pt;
  margin:0 0 5pt; text-align:left; }
.tb-refs .ref .anno{ font-style:italic; color:var(--soft); }

/* ---------------- table of contents ---------------- */
.toc{ page:front; break-after:page; }
.toc h1{ font-family:var(--sans); font-size:18pt; font-weight:800; text-transform:uppercase;
  letter-spacing:.04em; color:var(--ink); margin:0 0 14pt; }
.toc .row{ font-family:var(--serif); font-size:10.5pt; display:flex; align-items:baseline; margin:0 0 5pt; }
.toc .row.part{ font-family:var(--sans); font-weight:800; text-transform:uppercase; font-size:9.5pt;
  letter-spacing:.06em; color:var(--accent); margin-top:11pt; }
.toc a{ color:var(--ink); text-decoration:none; }
.toc a::after{ content:leader('.') target-counter(attr(href), page); color:var(--faint); }

/* ---------------- title-page cover (Rang-and-Dale) ---------------- */
.cover{ page:front; break-after:page; min-height:8.7in; display:flex; flex-direction:column;
  justify-content:center; align-items:center; text-align:center; padding:0 0.5in; }
.cover .ctop{ flex:0 0 auto; width:100%; }
.cover .cbot{ flex:0 0 auto; margin-top:0.8in; width:100%; }
.cover .crule{ width:56%; height:1pt; background:var(--accent); margin:22pt auto; }
.cover .ctitle{ font-family:var(--serif); font-weight:800; font-size:30pt; line-height:1.1;
  letter-spacing:.01em; color:var(--ink); margin:0; max-width:8in; }
.cover .csub{ font-family:var(--serif); font-style:italic; font-size:15pt; line-height:1.3; color:var(--deep);
  margin:16pt auto 0; max-width:6.2in; }
.cover .cmeta{ margin:15pt 0 0; }
.cover .cmeta *{ text-decoration:none; }
.cover .ccat{ font-family:var(--sans); font-weight:800; text-transform:uppercase; letter-spacing:.13em;
  font-size:9.5pt; color:var(--accent); }
.cover .cgoal{ font-family:var(--serif); font-style:italic; font-size:11.5pt; color:var(--deep); margin-top:5pt; }
.cover .cseries{ font-family:var(--sans); font-size:8pt; text-transform:uppercase; letter-spacing:.09em;
  color:var(--faint); margin-top:7pt; }
.cover .arule{ width:28%; height:1pt; background:var(--accent); margin:0 auto 9pt; }
.cover .aname{ font-family:var(--serif); font-variant:small-caps; letter-spacing:.06em; font-size:13pt; color:var(--ink); }
.cover .ameta{ font-family:var(--serif); font-style:italic; font-size:10pt; color:var(--faint); margin-top:3pt; }
.cover .cdate{ font-family:var(--serif); font-style:italic; font-size:9.5pt; color:var(--faint); margin-top:4pt; }

V.11 · The Build Engine

The engine compiles the source language of Section V.8 into the HTML structure the stylesheet styles, then renders the PDF through WeasyPrint with the source directory as the base URL so figure and font paths resolve. It parses front matter, Part and chapter markers, the fenced directive boxes, pipe tables with optional caption and note lines, image figures with pipe-separated caption fields, the four heading tiers, the run-in C-head, colored cross-references, lists, and the inline marks. Each chapter body is wrapped in the two-column container; openers, dividers, tables, and figures span. The table of contents is assembled from the Part and chapter structure with live page targets.

Build pipeline. Step one, place the source Markdown with complete front matter in the build directory beside textbook.css, with any figures in the directory the source references. Step two, provision once: instance EB Garamond and the sans to static weights and write the face block, reusing the Module III install path; idempotent. Step three, run python3 build_textbook.py source.md /mnt/user-data/outputs/{slug}.pdf. Step four, render the result to images and inspect: confirm the title-page cover on page one, the two-column flow, the spanning tables and figures, the mirrored running header with its accent chip, the box family treatments, and the contents page numbers. Step five, present the PDF, and retain the Markdown source as the editable master. The PDF is the textbook artifact; a DOCX is not produced for this edition because the two-column box-rich layout does not survive the DOCX model, and a codex or paper needing DOCX uses Module III.

The complete validated engine. Save as build_textbook.py.

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
TEXTBOOK EDITION build engine.
Compiles one Markdown source (textbook authoring conventions) to a two-column,
section-accented, box-rich textbook PDF via WeasyPrint.

Usage:  python3 build_textbook.py source.md [out.pdf]
Source dir must contain textbook.css and a fonts/ dir (or system fallbacks apply).
"""
import sys, os, re, html, pathlib

# ---------- palette cycle for per-part accent ----------
ACC_CYCLE = ["acc-1","acc-2","acc-3","acc-4","acc-5","acc-6"]
ROMAN = {'I':1,'V':5,'X':10,'L':50,'C':100,'D':500,'M':1000}

def roman_to_int(s):
    s=s.upper(); n=0; p=0
    for ch in reversed(s):
        if ch not in ROMAN: return None
        v=ROMAN[ch]; n+=v if v>=p else -v; p=v
    return n

def slug(s):
    s=re.sub(r'<[^>]+>','',s)
    s=re.sub(r'[^A-Za-z0-9]+','-',s).strip('-').lower()
    return s or 'sec'

# ---------- front matter ----------
def parse_front(text):
    fm={}
    m=re.match(r'^---\s*\n(.*?)\n---\s*\n',text,re.S)
    body=text
    if m:
        for line in m.group(1).splitlines():
            if ':' in line:
                k,v=line.split(':',1); fm[k.strip()]=v.strip()
        body=text[m.end():]
    return fm, body

# ---------- inline ----------
def inline(s):
    s=html.escape(s, quote=False)
    s=re.sub(r'\[\[(.+?)\]\]', r'<span class="xref">\1</span>', s)      # cross-ref
    s=re.sub(r'\*\*(.+?)\*\*', r'<strong>\1</strong>', s)               # bold
    s=re.sub(r'(?<!\*)\*(?!\*)(.+?)(?<!\*)\*(?!\*)', r'<em>\1</em>', s)  # italic
    s=re.sub(r'`(.+?)`', r'<code>\1</code>', s)
    s=re.sub(r'\^(\w+)\^', r'<sup>\1</sup>', s)                          # ^sup^
    return s

# ---------- block list / paragraph rendering ----------
def render_blocks(lines):
    """Render a list of raw text lines (a directive body or main flow chunk)
    into paragraphs and lists. Returns HTML string."""
    out=[]; i=0; n=len(lines)
    while i<n:
        ln=lines[i]
        if not ln.strip():
            i+=1; continue
        # unordered list
        if re.match(r'^\s*[-*\u2022]\s+', ln):
            items=[]
            while i<n and re.match(r'^\s*[-*\u2022]\s+', lines[i]):
                items.append(inline(re.sub(r'^\s*[-*\u2022]\s+','',lines[i]).rstrip())); i+=1
            out.append('<ul>'+''.join(f'<li>{x}</li>' for x in items)+'</ul>'); continue
        # ordered list
        if re.match(r'^\s*\d+\.\s+', ln):
            items=[]
            while i<n and re.match(r'^\s*\d+\.\s+', lines[i]):
                items.append(inline(re.sub(r'^\s*\d+\.\s+','',lines[i]).rstrip())); i+=1
            out.append('<ol>'+''.join(f'<li>{x}</li>' for x in items)+'</ol>'); continue
        # run-in C-head: {. lead phrase .} optionally followed by the paragraph text
        mrun=re.match(r'^\{\.\s*(.+?)\s*\.\}\s*(.*)$', ln)
        if mrun:
            out.append(f'<h4>{inline(mrun.group(1))}.</h4>')
            buf=[mrun.group(2).rstrip()] if mrun.group(2).strip() else []
            i+=1
            while i<n and lines[i].strip() and not re.match(r'^\s*([-*\u2022]|\d+\.)\s+', lines[i]) \
                  and not lines[i].startswith(':::') and not re.match(r'^\{\.', lines[i]):
                buf.append(lines[i].rstrip()); i+=1
            if buf: out.append('<p>'+inline(' '.join(buf))+'</p>')
            continue
        # paragraph (gather until blank)
        buf=[ln.rstrip()]; i+=1
        while i<n and lines[i].strip() and not re.match(r'^\s*([-*\u2022]|\d+\.)\s+', lines[i]) \
              and not lines[i].startswith(':::') and not re.match(r'^\{\.', lines[i]):
            buf.append(lines[i].rstrip()); i+=1
        out.append('<p>'+inline(' '.join(buf))+'</p>')
    return ''.join(out)

# ---------- table ----------
def render_table(tbl_lines, caption=None, note=None, span=False):
    rows=[r for r in tbl_lines if r.strip()]
    cells=[[c.strip() for c in re.split(r'(?<!\\)\|', r.strip().strip('|'))] for r in rows]
    # drop alignment separator row
    body=[c for c in cells if not all(re.fullmatch(r':?-{2,}:?', x or '-') for x in c)]
    head=body[0] if body else []
    data=body[1:] if len(body)>1 else []
    h='<thead><tr>'+''.join(f'<th>{inline(x)}</th>' for x in head)+'</tr></thead>'
    b='<tbody>'+''.join('<tr>'+''.join(f'<td>{inline(x)}</td>' for x in r)+'</tr>' for r in data)+'</tbody>'
    cap=f'<caption>{inline(caption)}</caption>' if caption else ''
    nt=f'<div class="tnote">{inline(note)}</div>' if note else ''
    sp=' span' if span else ''
    return f'<div class="tb-table{sp}"><table>{cap}{h}{b}</table>{nt}</div>'

# ---------- figure ----------
def render_figure(alt, path, brace, span=False):
    label=src=''; bodytxt=brace or alt
    if brace and '|' in brace:
        parts=[p.strip() for p in brace.split('|')]
        label=parts[0] if len(parts)>0 else ''
        bodytxt=parts[1] if len(parts)>1 else ''
        src=parts[2] if len(parts)>2 else ''
    cap=''
    if label: cap+=f'<span class="flabel">{inline(label)}</span> '
    cap+=inline(bodytxt)
    if src: cap+=f' <span class="fsrc">({inline(src)})</span>'
    sp=' span' if span else ''
    return (f'<div class="tb-fig{sp}"><div class="frame"><img src="{html.escape(path)}" alt="{html.escape(alt)}"></div>'
            f'<figcaption>{cap}</figcaption></div>')

# ---------- directive boxes ----------
PANEL={'key':('','KEY POINTS'),'summary':('','SUMMARY'),
       'definition':('def','DEFINITION'),'concept':('def','CONCEPT'),
       'example':('ex','WORKED EXAMPLE')}
BAR={'practice':'In Practice','application':'Application','procedure':'Procedure','case':'Case'}

def render_directive(kind, title, body_lines, span=False):
    inner=render_blocks(body_lines)
    if kind in PANEL:
        cls,kick=PANEL[kind]
        head=(f'<p class="bh"><span class="mark"></span>'
              f'<span class="kicker">{html.escape(kick)}</span>{inline(title)}</p>')
        sp=' span' if span else ''
        return f'<div class="tb-box {cls}{sp}">{head}{inner}</div>'
    if kind in BAR:
        bh=title if title else BAR[kind]
        sp=' span' if span else ''
        return f'<div class="tb-bar{sp}"><p class="bh">{inline(bh)}</p><div class="bd">{inner}</div></div>'
    if kind=='caution':
        bh=title if title else 'Caution'
        return f'<div class="tb-bar caution"><p class="bh">{inline(bh)}</p><div class="bd">{inner}</div></div>'
    if kind=='note':
        return f'<p class="aside">{render_blocks(body_lines).replace("<p>","").replace("</p>"," ")}</p>'
    if kind=='reading':
        rows=[]
        for ln in body_lines:
            if not ln.strip(): continue
            if ' -- ' in ln:
                cit,anno=ln.split(' -- ',1)
                rows.append(f'<div class="ref">{inline(cit.strip())} <span class="anno">{inline(anno.strip())}</span></div>')
            else:
                rows.append(f'<div class="ref">{inline(ln.strip())}</div>')
        return ''.join(rows)
    return inner

# ---------- main document assembly ----------
def build_html(fm, body, cssfiles):
    single = fm.get('accent_mode','per-part')=='single'
    single_acc = fm.get('single_accent','acc-1')
    if not single_acc.startswith('acc-'): single_acc='acc-'+single_acc
    toc_on = fm.get('toc','on').lower() in ('on','true','yes','1')

    lines=body.splitlines()
    i=0; n=len(lines)
    parts=[]          # for TOC: (level, id, label)
    chapters=[]
    html_parts=[]     # full body html in order
    part_idx=0
    cur_acc = single_acc if single else 'acc-1'
    open_chapter=False
    chap_buf=[]       # html fragments for current chapter (goes inside .cols)

    def flush_chapter():
        nonlocal chap_buf, open_chapter
        if open_chapter:
            html_parts.append('<div class="cols">'+''.join(chap_buf)+'</div></section>')
            chap_buf=[]; open_chapter=False

    while i<n:
        ln=lines[i]
        # ----- PART -----
        mp=re.match(r'^#\s+PART\s+([IVXLCDM]+|\d+)\s*[\u00b7\u2014\-\.:]?\s*(.*)$', ln, re.I)
        if mp:
            flush_chapter()
            num=mp.group(1); name=mp.group(2).strip()
            part_idx+=1
            if not single:
                cur_acc=ACC_CYCLE[(part_idx-1)%len(ACC_CYCLE)]
            pid='part-'+slug(num+'-'+name)
            parts.append(('part',pid,f'Part {num} &#183; {name}' if name else f'Part {num}'))
            html_parts.append(
                f'<section class="part-divider {cur_acc}" id="{pid}">'
                f'<div class="pnum">Part {html.escape(num)}</div><div class="pbar"></div>'
                f'<div class="ptitle">{inline(name)}</div></section>')
            i+=1; continue
        # ----- CHAPTER (numbered) -----
        mc=re.match(r'^#\s+(\d+)\.\s+(.+)$', ln)
        if mc:
            flush_chapter()
            cnum=mc.group(1); ctitle=mc.group(2).strip()
            cid='ch-'+slug(cnum+'-'+ctitle)
            parts.append(('chapter',cid,f'{cnum} &#183; {ctitle}'))
            part_label = parts and ''
            html_parts.append(
                f'<section class="chapter {cur_acc}" id="{cid}">'
                f'<div class="chapter-open">'
                f'<div class="cband"><div class="cnum">{html.escape(cnum)}</div>'
                f'<div class="ctitle">{inline(ctitle)}</div></div></div>'
                f'<div class="runhead"><span class="chip">{html.escape(cnum)}</span> {inline(ctitle)}</div>')
            open_chapter=True; i+=1; continue
        # ----- plain chapter (unnumbered H1) -----
        m1=re.match(r'^#\s+(.+)$', ln)
        if m1:
            flush_chapter()
            t=m1.group(1).strip(); cid='ch-'+slug(t)
            parts.append(('chapter',cid,t))
            html_parts.append(
                f'<section class="chapter {cur_acc}" id="{cid}">'
                f'<div class="chapter-open"><h1 class="cplain">{inline(t)}</h1></div>'
                f'<div class="runhead"><span class="chip">&#9679;</span> {inline(t)}</div>')
            open_chapter=True; i+=1; continue
        # ----- directive fence -----
        md=re.match(r'^:::\s*(\w+)(\*?)\s*(.*)$', ln)
        if md:
            kind=md.group(1).lower(); bspan=md.group(2)=='*'; title=md.group(3).strip()
            j=i+1; bl=[]
            while j<n and not lines[j].strip()==':::':
                bl.append(lines[j]); j+=1
            frag=render_directive(kind, title, bl, span=bspan)
            (chap_buf if open_chapter else html_parts).append(frag)
            i=j+1; continue
        # ----- table with optional caption / note -----
        cap=None; tspan=False
        mcapt=re.match(r'^Table(\*?):\s*(.+)$', ln)
        look=i
        if mcapt:
            tspan=mcapt.group(1)=='*'; cap=mcapt.group(2).strip(); look=i+1
        if look<n and lines[look].lstrip().startswith('|'):
            tl=[]; k=look
            while k<n and lines[k].lstrip().startswith('|'):
                tl.append(lines[k]); k+=1
            note=None
            if k<n and re.match(r'^Note:\s*(.+)$', lines[k]):
                note=re.match(r'^Note:\s*(.+)$', lines[k]).group(1).strip(); k+=1
            frag=render_table(tl, cap, note, span=tspan)
            (chap_buf if open_chapter else html_parts).append(frag)
            i=k; continue
        # ----- figure -----
        mf=re.match(r'^!(\*?)\[(.*?)\]\((.*?)\)(?:\{(.*?)\})?\s*$', ln)
        if mf:
            frag=render_figure(mf.group(2), mf.group(3), mf.group(4), span=mf.group(1)=='*')
            (chap_buf if open_chapter else html_parts).append(frag); i+=1; continue
        # ----- headings H2/H3 -----
        mh2=re.match(r'^##\s+(.+)$', ln)
        mh3=re.match(r'^###\s+(.+)$', ln)
        if mh3:
            t=mh3.group(1).strip()
            mnum=re.match(r'^(\d+(?:\.\d+)+)\s+(.+)$', t)
            if mnum: frag=f'<h3><span class="num">{mnum.group(1)}</span>{inline(mnum.group(2))}</h3>'
            else:    frag=f'<h3>{inline(t)}</h3>'
            (chap_buf if open_chapter else html_parts).append(frag); i+=1; continue
        if mh2:
            t=mh2.group(1).strip()
            mnum=re.match(r'^(\d+(?:\.\d+)*)\s+(.+)$', t)
            if mnum: frag=f'<h2><span class="num">{mnum.group(1)}</span>{inline(mnum.group(2))}</h2>'
            else:    frag=f'<h2 class="plain">{inline(t)}</h2>'
            (chap_buf if open_chapter else html_parts).append(frag); i+=1; continue
        # ----- ordinary block (paragraph / list / run-in) -----
        buf=[ln]; i+=1
        while i<n and lines[i].strip() and not lines[i].startswith(':::') \
              and not re.match(r'^#', lines[i]) and not lines[i].lstrip().startswith('|') \
              and not re.match(r'^(Table\*?:|Note:|!\*?\[)', lines[i]):
            buf.append(lines[i]); i+=1
        frag=render_blocks(buf)
        (chap_buf if open_chapter else html_parts).append(frag)

    flush_chapter()

    # ----- TOC -----
    toc=''
    if toc_on and parts:
        rows=['<h1>Contents</h1>']
        for lvl,pid,label in parts:
            if lvl=='part':
                rows.append(f'<div class="row part">{label}</div>')
            else:
                rows.append(f'<div class="row"><a href="#{pid}">{label}</a></div>')
        toc=f'<nav class="toc {single_acc if single else ACC_CYCLE[0]}">'+''.join(rows)+'</nav>'

    # ----- title-page cover (front matter) -----
    import re as _rc
    _cv_name = fm.get('author','').strip() or _rc.sub(r'\^\d+\^','',fm.get('author_line','')).split(',')[0].strip()
    _cv_cat  = fm.get('category', fm.get('article_type',''))
    _cv_goal = fm.get('goal', fm.get('tagline',''))
    _cv_ser  = fm.get('series', fm.get('doi',''))
    _cv_sub  = fm.get('subtitle',''); _cv_date=fm.get('date',''); _cv_aff=fm.get('affiliation','')
    _cv_title= fm.get('title','')
    _cv_acc  = single_acc if single else ACC_CYCLE[0]
    cover=''
    if _cv_title:
        _cs=''
        if _cv_cat:  _cs+=f'<div class="ccat">{inline(_cv_cat)}</div>'
        if _cv_goal: _cs+=f'<div class="cgoal">{inline(_cv_goal)}</div>'
        if _cv_ser:  _cs+=f'<div class="cseries">{inline(_cv_ser)}</div>'
        cover=(f'<div class="cover {_cv_acc}"><div class="ctop"><div class="crule"></div>'
               f'<h1 class="ctitle">{inline(_cv_title)}</h1>'
               + (f'<div class="csub">{inline(_cv_sub)}</div>' if _cv_sub else '')
               + (f'<div class="cmeta">{_cs}</div>' if _cs else '')
               + '<div class="crule"></div></div><div class="cbot"><div class="arule"></div>'
               + f'<div class="aname">{inline(_cv_name)}</div>'
               + (f'<div class="ameta">{inline(_cv_aff)}</div>' if _cv_aff else '')
               + (f'<div class="cdate">{inline(_cv_date)}</div>' if _cv_date else '')
               + '</div></div>')
    links=''.join(f'<link rel="stylesheet" href="{html.escape(c)}">' for c in cssfiles)
    title=fm.get('title','')
    return (f'<!DOCTYPE html><html><head><meta charset="utf-8"><title>{html.escape(title)}</title>'
            f'{links}</head><body>{cover}{toc}{"".join(html_parts)}</body></html>')

def main():
    if len(sys.argv)<2:
        print("usage: build_textbook.py source.md [out.pdf]"); sys.exit(1)
    src=pathlib.Path(sys.argv[1]).resolve()
    text=src.read_text(encoding='utf-8')
    fm, body = parse_front(text)
    srcdir=src.parent
    css=[]
    if (srcdir/'face.css').exists(): css.append('face.css')
    css.append('textbook.css')
    doc=build_html(fm, body, css)
    out_html=srcdir/'_textbook.html'
    out_html.write_text(doc, encoding='utf-8')
    out_pdf=pathlib.Path(sys.argv[2]) if len(sys.argv)>2 else srcdir/'textbook.pdf'
    from weasyprint import HTML
    HTML(string=doc, base_url=str(srcdir)).write_pdf(str(out_pdf))
    print("PDF:", out_pdf)

if __name__=='__main__':
    main()

V.12 · Codex-Native Application

The Master Codex sets this edition at Tier 2 with full native register. It selects a single house accent, copper or slate, through accent_mode: single with single_accent: copper or single_accent: slate, so the volume reads in one sealed color rather than the rotating Part palette, though a long codex with distinct Parts may use the per-part palette instead. The permitted verdict glyphs appear as bare text glyphs inside boxes and body per the Module IV glyph discipline, never as decoration. The box taxonomy carries codex content by the mapping in Section V.5: verdicts and seals in Key Points panels, PSPs and axioms in Definition panels, regime instances and battery rows in Worked Example panels, procedures and cascade steps in In Practice bars, anchor-inflation and typing locks in Caution bars, proof-boundary remarks in depth asides. Every claim-bearing box and figure carries its warrant tier exactly as typed in the codex, and the faithful-map law holds: the page asserts no more and no less than the codex establishes. Honest typing discipline governs as everywhere: a box stating a result types it Type T, Type C, or Type S, and overclaiming a warrant grade in a box header is the same failure it is in prose.

V.13 · Pre-Flight and Gate Integration

A Textbook Edition deliverable passes the standard pre-delivery gate sequence: the silent-integration scan, the tier-threshold audit at the book's register, and the format-contract verification. The internal V-FIO Diagnostic Log is produced for a codex volume exactly as for a paper. In addition, before presentation, the edition passes a short structural checklist. Every table and figure is wrapped in a spanning container and renders full width with no fragmentation. Every Part sets its accent class and the accent carries through chip, heads, cross-references, table bands, and panels. No emoji anywhere; every box marker is a CSS shape. The running header mirrors recto and verso and the accent chip carries the live chapter number. The contents page numbers resolve against final pagination. Every figure embeds, confirmed by image count. Every claim-bearing box and figure carries its warrant tier with no inflation. Failure of any item returns the book for repair before any presentation.

V Quick Reference Card

Edition selector: [EDITION:TEXTBOOK], [EDITION:TEXTBOOK_A], or front-matter edition: textbook / edition: textbook_a; addressable in prose as TextBookA. Distinct from the Module III paper renderer and the Journal edition; they never co-fire. Auto-selected for codex, master, and internal-volume deliverables whose content is prose-dense reference material. Type register: serif body, EB Garamond default, 10.2 pt; sans furniture, Archivo as TBSans, for every head, box, table, caption, chip. Never collapsed in a trade textbook. Page: US Letter 612 by 792, margins 58 / 50 / 56 / 50 pt. Two columns, 24 pt gutter. Spanning reserved for openers, tables, figures. Accent: per-part six-hue cycle, or single mode with copper or slate for the house. Warning palette reserved for Caution only. Accent drives chip, opener, heads, cross-references, table bands, panels. Furniture: title-page cover on the front named page ahead of contents; Part divider and chapter opener band with large numeral; running chip lifted into mirrored recto and verso headers; outer page numbers; front named page suppresses furniture on the cover, contents, and dividers; contents with target-counter leaders. Title page: clean Apex-model cover in the Davidson register. A filled accent band carries the category in white caps at the top in place of a top rule; the title sets in bold sans, the subtitle in deep-accent serif italic, the goal and series follow in the meta block, and a thick 2.4 pt accent author rule sits above the bold-sans byline, with the affiliation in italic gray and the date beneath. Reads from title, subtitle, category (or article_type), goal (or tagline), series (or doi), author, affiliation, date. Headings: A-head sans bar with accent numeral, plain variant for front matter; B-head sans accent; C-head run-in bold-italic serif. Boxes: tinted panel with square, circle, triangle mark for Key Points, Definition, Worked Example; solid bar for In Practice and kin; warning bar for Caution; down-triangle depth aside; annotated reading list. Codex maps verdict, PSP, regime, procedure, warning, boundary onto the same set. Tables and figures: always wrapped in a spanning container; accent caption bar, banded head, zebra body, footnote and abbreviation key; figures in a tinted frame with label, description, source caption fields. Source language: front matter; # PART n · Name; # n. Title; # Title; ## n.n Title; ### n.n.n Title; {. run-in .}; ::: type Title ... ::: boxes; Table: and Note: around a pipe table; ![alt](path){label | desc | source}; [[cross-ref]]; bold, italic, code, superscript. Fonts: instance EB Garamond and the sans to static weights, set name and style bits, map sans black to weight 800; declare a system fallback chain so a draft compiles unprovisioned. Reuse the Module III install path. Build: python3 build_textbook.py source.md /mnt/user-data/outputs/{slug}.pdf; base URL is the source directory; inspect the render; present the PDF; keep the Markdown source as master. No DOCX for this edition. Locks: no LaTeX, Unicode math only; no em-dashes; ﷺ as one glyph; no emoji, CSS shapes only; no fragmentation; warrant tier on every claim-bearing box and figure; faithful map, never inflation.

---

MODULE V·D · THE TEXTBOOK_D (DAVIDSON) VARIANT

A second skin for the Textbook Edition. Where the default Module V renders a calm Rang-and-Dale page, serif body with tinted panels, TextBook_D renders an aggressive Davidson page: sans throughout, solid-colour banner headings, numbered icon boxes typed by colour, and a chapter thumb-tab riding the outer margin. Same content pipeline and the same authoring conventions, a different stylesheet and a Davidson-aware engine. Selected by the token [EDITION:TEXTBOOK_D] or front-matter edition: textbook_d. It transforms any subject, and any Master Codex volume, into the Davidson register.

V·D.0 · The Design Read

Davidson's Principles and Practice of Medicine sets a denser, louder page than Rang and Dale, and the difference is systematic. Five elements carry it. First, the type is sans-serif throughout: a humanist grotesque sets the body, the headings, the boxes, the tables, and the captions, with no serif anywhere, which reads as clinical and modern rather than literary. Second, headings are solid colour, not tinted: a major section opens with a filled steel banner carrying white reversed text, where Rang and Dale would use a tint bar with coloured text. Third, the signature element is the numbered icon box: every box carries a chapter-and-box number, a solid colour header bar, a small icon square at the bar's left edge, and a white reversed title, with the body frequently wrapping a full table; the box is the primary unit of organised content and there are many per chapter. Fourth, boxes are typed by colour and icon together, so the reader reads the box's purpose before its words: an information or data box in steel with an i, a practice box in green with a plus, a caution box in red with an exclamation, a special-context box in purple. Fifth, the page carries a coloured chapter thumb-tab in the outer margin, a bound-book index that lets a reader flick to a chapter by colour and number. The grid is two columns, tighter than Rang and Dale, with full-width spanning reserved for the wide table-boxes and figures. Cross-references run in the accent colour as heavy navigation. The whole is more saturated, more boxed, and more directive: the aggression is the boxing and the solid colour, captured by making the box a numbered, icon-marked, colour-typed object and the headings solid banners.

V·D.1 · The Locked Contract

Sans-serif throughout is the load-bearing rule and is never broken in this variant. The body sets in a humanist grotesque, Archivo as TBSans by default, at 9.8 points on a 1.34 measure, justified and hyphenated; every heading, box, table, caption, chip, and tab sets in the same family at its own weight. Page geometry is US Letter, 612 by 792 points, with the outer margin widened to 66 points to seat the thumb-tab and the inner margin held at 50; the outer margin mirrors, right on recto and left on verso, so the tab always rides the open edge. The grid is two columns with a 22-point gutter, denser than the default edition. The universal locks hold without exception: no LaTeX and all mathematics in Unicode, no em-dashes, the honorific ﷺ as one glyph, no emoji anywhere, every box marker a CSS-drawn square carrying a typed letter rather than an icon image, and no box, figure, or table fragmenting across a break.

V·D.2 · The Heading Ladder

Four levels below the chapter, matching Davidson. The major-section heading is a solid accent banner with white reversed bold text that may wrap two lines, authored as a level-two heading; it is the loudest mark on the page and is used sparingly for the chapter's main divisions. The sub-section heading is a pale accent bar with bold deep-accent text, authored as a level-three heading. The sub-sub heading is bold deep-accent text with a small accent square at its left and a thin rule beneath, authored as a level-four heading. The run-in is a bold-italic lead phrase that flows into its paragraph. Any heading may carry a numeral, rendered in the deeper accent shade. The banner is the element that reads as Davidson; reserve it for true section breaks and let the bar carry the numbered subsections.

V·D.3 · The Numbered Icon Box

The signature. A box carries a solid colour header bar; at the bar's left edge a solid white icon square holds a single typed letter in the box colour; the bar then carries the box number and the title in white reversed bold; the body sits below in a pale tint of the box colour and holds paragraphs, lists, or a table. The box is typed by colour and letter together, and the type set is content-agnostic so it serves any subject and any codex. The default information or data box is the accent steel and carries an i; a key or summary box is ochre and carries a star; a definition or concept box is the accent and carries an i; a worked example box carries an e; a practice, application, or procedure box is green and carries a plus; a caution, warning, or emergency box is red and carries an exclamation; a special-context or note box is purple and carries an i. A box whose body contains a wide table spans both columns automatically, which is how Davidson sets its large data boxes; a box of prose or a short list stays within its column. The box number is optional: supply it after the type word and it renders before the title, omit it and the title stands alone.

Codex-native mapping. A Master Codex volume reads the same box set with codex semantics in Tier 2 and may use the permitted glyphs. A verdict or seal block sits in a key box, gold and starred. A PSP or an axiom sits in a definition box. A regime instance or a battery row sits in a worked example box. An operating procedure or a cascade step sits in a practice box, green. An anchor-inflation warning or a typing-discipline lock sits in a caution box, red. A proof-boundary remark or an instrument limit sits in a depth aside. Every claim-bearing box carries its warrant tier exactly as typed, and the faithful-map law holds: the box asserts no more than the codex establishes.

V·D.4 · Page Furniture

The chapter opener is a solid accent band carrying the chapter number set large in white beside the title in white reversed bold; it spans both columns and the two-column flow begins below it. The chapter thumb-tab is a small solid accent square carrying the chapter number, lifted into the outer side margin by a running element and placed in the right-middle margin box on recto pages and the left-middle on verso, so it rides the open edge of the spread as a bound index. The running header carries the chapter title in deep-accent bold small caps with the page number after a bullet, mirrored to the top-outer corner. The front named page suppresses the header, the tab, and the footer for the table of contents and the part dividers. The table of contents opens with a solid accent banner reading Contents, then part rows in accent small caps and chapter rows linked, with dotted leaders and live page targets resolved against final pagination.

V·D.5 · Authoring Conventions, the Delta from Module V

The source language is the Textbook Edition's, with three deltas. First, select the variant with edition: textbook_d in front matter or the token [EDITION:TEXTBOOK_D]. Second, the heading levels map to the Davidson ladder: a level-two heading renders as a solid banner, a level-three heading as a tinted bar, a level-four heading as a square-marked rule head; author the few major sections at level two and the numbered subsections at level three. Third, a box directive may carry a number directly after the type word: ::: info 4.6 Title renders box 4.6, while ::: practice Title renders an unnumbered practice box. The box types are info, data, key, summary, definition, concept, example, practice, application, procedure, clinical, caution, warning, emergency, note, and context, each mapping to a colour and a letter. A box that contains a pipe table spans both columns; a Note: line inside the box body renders as the box footnote. Everything else is unchanged from the Textbook Edition: front matter fields, part and chapter markers, the run-in {. lead .}, pipe tables with Table: captions, image figures with pipe-separated caption fields, [[cross-references]], and the inline marks. Accent is per-part by default and single house accent, copper or slate, for codex volumes.

V·D.6 · The Davidson Stylesheet

Page economy and the span rule. Tables, figures, and boxes default to column width and flow inside one column, which packs the dense Davidson grid without stranded space. Full-width spanning is what wastes space: a spanning element closes both columns and the short band above it balances into two stub columns, leaving the gap empty. Spanning is opt-in. Append an asterisk to span a genuinely wide element: ::: type* Title for a box, Table*: caption for a standalone table, !*[alt](path) for a figure. A numbered table-box is breakable by default: its data flows down one column and continues at the top of the next with the column header repeated, the coloured box header carried once at the start, so a long audit table fills the column instead of stranding it. Prose and list boxes stay atomic. Think ahead before placing a wide element: prefer column width, reserve spanning for tables too wide to read in a column, and let the breakable column-width table-box carry the long data.

The complete validated stylesheet. Self-contained, compiles under WeasyPrint as written. Save as textbook_d.css beside the source; the face block of Module V provisions the sans, and the system fallback chain applies if it is skipped.

/* ===================================================================
   TEXTBOOK_D  ·  Davidson-style master stylesheet  (WeasyPrint-validated)
   Sans throughout · solid-banner headings · numbered icon boxes typed by
   colour · chapter thumb-tab in the outer margin · dense two-column grid
   =================================================================== */

:root{
  --sans:'TBSans','Archivo','Liberation Sans','Noto Sans','DejaVu Sans',sans-serif;
  --ink:#1c1c1c; --soft:#454545; --faint:#737373; --rule:#cdcdcd;
  /* primary accent (Davidson steel), drives chapter tab, banners, bars, xrefs, info boxes */
  --accent:#3E7CA8; --deep:#28567C; --bar:#B7C5D6; --tint:#EAF0F5; --band:#D7E2EC;
  /* semantic box palette */
  --c-info:#3E7CA8;    --t-info:#EAF1F6;
  --c-practice:#4E8A5B; --t-practice:#ECF4ee;
  --c-caution:#C0392B;  --t-caution:#FBECEA;
  --c-note:#8068B0;     --t-note:#F1ECF8;
  --c-key:#B07A1E;      --t-key:#F8F1E2;
}
/* per-PART accent (sets the dominant colour); single-mode books pick one */
.acc-1{--accent:#3E7CA8;--deep:#28567C;--bar:#B7C5D6;--tint:#EAF1F6;--band:#D7E2EC;}
.acc-2{--accent:#4E8A5B;--deep:#33603e;--bar:#BCD3C2;--tint:#ECF4EE;--band:#D5E7DA;}
.acc-3{--accent:#B23A48;--deep:#7e2832;--bar:#E0BDC2;--tint:#FBECEE;--band:#F0D4D8;}
.acc-4{--accent:#3F5C8C;--deep:#2a3f63;--bar:#BDC8DC;--tint:#ECEFF6;--band:#D6DEEF;}
.acc-5{--accent:#B07A1E;--deep:#7c5512;--bar:#DDC9A4;--tint:#F8F1E2;--band:#EFE1C4;}
.acc-6{--accent:#7A4E8C;--deep:#553463;--bar:#CDBBD8;--tint:#F2ECF6;--band:#E4D6EC;}
.acc-copper{--accent:#B87333;--deep:#8a5523;--bar:#E1C9AE;--tint:#FAF3EA;--band:#EFE0CF;}
.acc-slate{--accent:#2C2F33;--deep:#1c1e20;--bar:#C2C4C6;--tint:#ECEDED;--band:#D9DADA;}

/* ---------------- page geometry + furniture ---------------- */
@page{ size:612pt 792pt; margin:54pt 50pt 54pt 50pt; }
@page:right{
  margin-right:66pt;                       /* outer margin holds the thumb tab */
  @top-right{ content:string(dxchap) "  \2022  " counter(page);
    font-family:var(--sans); font-size:8.5pt; font-weight:800; color:var(--deep);
    text-transform:uppercase; letter-spacing:.02em; vertical-align:bottom; padding-bottom:3pt; }
  @right-middle{ content:element(thumb); }
}
@page:left{
  margin-left:66pt;
  @top-left{ content:counter(page) "  \2022  " string(dxchap);
    font-family:var(--sans); font-size:8.5pt; font-weight:800; color:var(--deep);
    text-transform:uppercase; letter-spacing:.02em; vertical-align:bottom; padding-bottom:3pt; }
  @left-middle{ content:element(thumb); }
}
@page front{ @top-left{content:none} @top-right{content:none}
  @right-middle{content:none} @left-middle{content:none} }

html{ font-family:var(--sans); font-size:9.8pt; color:var(--ink); line-height:1.34; }
body{ margin:0; }
p{ margin:0 0 6pt; text-align:justify; hyphens:auto; orphans:2; widows:2; }
ul,ol{ margin:2pt 0 6pt 15pt; padding:0; } li{ margin:0 0 3pt; }
ul{ list-style:none; } ul>li::before{ content:"\25AA\00A0"; color:var(--accent); margin-left:-12pt; }

/* chapter thumb-tab, lifted into the outer side margin */
.dx-thumb{ position:running(thumb); }
.dx-thumb .tabnum{ display:block; width:30pt; height:30pt; line-height:30pt; text-align:center;
  background:var(--accent); color:#fff; font-family:var(--sans); font-weight:800; font-size:14pt;
  border-radius:3pt; }

/* ---------------- PART divider ---------------- */
.part-divider{ break-before:page; break-after:page; height:640pt; page:front;
  display:flex; flex-direction:column; justify-content:center; }
.part-divider .pnum{ font-family:var(--sans); font-weight:800; font-size:14pt; color:var(--accent);
  letter-spacing:.16em; text-transform:uppercase; }
.part-divider .pbar{ height:10pt; background:var(--accent); width:56%; margin:10pt 0 16pt; }
.part-divider .ptitle{ font-family:var(--sans); font-weight:800; font-size:32pt; line-height:1.05; color:var(--ink); }
.part-divider .pblurb{ font-family:var(--sans); font-size:11.5pt; color:var(--soft); margin-top:14pt; max-width:80%; }

/* ---------------- CHAPTER opener ---------------- */
.chapter + .chapter{ break-before:page; }
.chapter-open{ break-before:page; margin:0 0 13pt; }
.chapter-open .cband{ background:var(--accent); padding:16pt 18pt; display:flex; align-items:center; gap:18pt; }
.chapter-open .cnum{ font-family:var(--sans); font-weight:800; font-size:52pt; line-height:.85; color:#fff; }
.chapter-open .ctitle{ font-family:var(--sans); font-weight:800; font-size:25pt; line-height:1.04; color:#fff; string-set:dxchap content(); }
.chapter-open h1.cplain{ font-family:var(--sans); font-weight:800; font-size:23pt; color:#fff;
  background:var(--accent); padding:13pt 16pt; margin:0; string-set:dxchap content(); }

/* ---------------- two-column body ---------------- */
.cols{ column-count:2; column-gap:22pt; }

/* ---------------- heading ladder ---------------- */
/* H2 = solid banner (major section), white reversed */
h2{ font-family:var(--sans); font-weight:800; font-size:12.5pt; line-height:1.12; color:#fff;
  background:var(--accent); padding:6pt 10pt; margin:13pt 0 8pt; break-after:avoid; }
h2 .num{ color:#fff; opacity:.85; margin-right:8pt; }
/* H3 = tinted bar (sub-section), bold deep text */
h3{ font-family:var(--sans); font-weight:800; font-size:10.8pt; color:var(--deep);
  background:var(--bar); padding:4pt 9pt; margin:12pt 0 6pt; break-after:avoid; }
h3 .num{ color:var(--deep); margin-right:7pt; }
/* H4 = square-marked underlined head (sub-sub) */
h4{ font-family:var(--sans); font-weight:800; font-size:10.4pt; color:var(--deep);
  border-bottom:1pt solid var(--bar); padding:0 0 3pt; margin:11pt 0 5pt; break-after:avoid; }
h4::before{ content:""; display:inline-block; width:8pt; height:8pt; background:var(--accent);
  margin-right:7pt; vertical-align:middle; }
/* run-in C-head */
.runin{ font-family:var(--sans); font-style:italic; font-weight:800; color:var(--ink); }

/* ---------------- inline semantics ---------------- */
strong,b{ font-weight:800; } em,i{ font-style:italic; }
.xref{ color:var(--accent); font-weight:600; } a,a.xref{ color:var(--accent); text-decoration:none; }
sup{ font-size:.7em; line-height:0; }

/* ---------------- NUMBERED ICON BOX (the Davidson signature) ---------------- */
.dx-box{ --bc:var(--c-info); --bt:var(--t-info); break-inside:avoid; border:0.7pt solid var(--bc);
  margin:9pt 0; }
.dx-box.practice{ --bc:var(--c-practice); --bt:var(--t-practice); }
.dx-box.caution { --bc:var(--c-caution);  --bt:var(--t-caution); }
.dx-box.note    { --bc:var(--c-note);     --bt:var(--t-note); }
.dx-box.key     { --bc:var(--c-key);      --bt:var(--t-key); }
.dx-box .dxhead{ background:var(--bc); color:#fff; font-family:var(--sans); font-weight:800;
  font-size:9.6pt; padding:0; margin:0; display:flex; align-items:stretch; }
.dx-box .dxicon{ flex:0 0 auto; width:19pt; min-height:19pt; background:#fff;
  color:var(--bc); font-weight:800; font-size:11pt; text-align:center; line-height:19pt; font-style:normal; }
.dx-box .dxttl{ padding:4pt 9pt; }
.dx-box .dxttl .dxnum{ margin-right:7pt; }
.dx-box .dxbody{ background:var(--bt); padding:7pt 10pt; }
.dx-box .dxbody p{ text-align:left; margin:0 0 5pt; }
.dx-box .dxbody ul{ margin:0 0 0 14pt; } .dx-box .dxbody ul>li::before{ color:var(--bc); }
/* table living inside a box: tighten to the box */
.dx-box .dxbody table{ width:100%; border-collapse:collapse; font-size:8.6pt; margin:2pt 0; }
.dx-box .dxbody thead th{ text-align:left; font-weight:800; color:var(--deep);
  border-bottom:1.2pt solid var(--bc); padding:3pt 6pt; }
.dx-box .dxbody tbody td{ padding:3pt 6pt; border-bottom:0.4pt solid var(--rule); vertical-align:top; }
.dx-box .dxbody .tnote{ font-size:7.6pt; color:var(--soft); font-style:italic; padding-top:4pt; }
/* a box that wraps a wide table spans both columns */
.dx-box.span{ column-span:all; break-inside:avoid; }
.dx-box.flow{ break-inside:auto; }
.dx-box.flow .dxbody{ break-inside:auto; }
.dx-box.flow .dxbody table{ break-inside:auto; }
.dx-box.flow .dxbody thead{ display:table-header-group; }

/* ---------------- small-print aside (▼) ---------------- */
.aside{ font-size:8.6pt; color:var(--soft); text-align:justify; margin:6pt 0; }
.aside::before{ content:"\25BC\2002"; color:var(--ink); font-size:7pt; }

/* ---------------- spanning standalone table ---------------- */
.tb-table{ margin:11pt 0; break-inside:auto; }
.tb-table.span{ column-span:all; break-inside:avoid; }
.tb-table table{ width:100%; border-collapse:collapse; font-family:var(--sans); font-size:8.7pt; }
.tb-table caption{ caption-side:top; font-family:var(--sans); font-weight:800; color:#fff;
  background:var(--accent); font-size:9pt; padding:4pt 9pt; text-align:left; }
.tb-table thead th{ background:var(--band); color:var(--deep); text-align:left; font-weight:800;
  padding:4pt 7pt; border-bottom:1.2pt solid var(--accent); }
.tb-table tbody td{ padding:4pt 7pt; border-bottom:0.4pt solid var(--rule); vertical-align:top; }
.tb-table tbody tr:nth-child(even) td{ background:var(--tint); }
.tb-table .tnote{ font-family:var(--sans); font-size:7.6pt; color:var(--soft); padding:4pt 2pt 0; }

/* ---------------- spanning figure ---------------- */
.tb-fig{ break-inside:avoid; margin:12pt 0; }
.tb-fig.span{ column-span:all; }
.tb-fig.col{ column-span:none; }
.tb-fig .frame{ border:0.6pt solid var(--rule); background:#fff; padding:10pt; text-align:center; }
.tb-fig img{ max-width:100%; height:auto; }
.tb-fig figcaption{ font-family:var(--sans); font-size:8.5pt; color:var(--soft); margin-top:6pt; text-align:left; }
.tb-fig figcaption .flabel{ font-weight:800; color:var(--accent); }
.tb-fig figcaption .fsrc{ font-style:italic; }

/* ---------------- references / reading ---------------- */
.tb-refs .ref{ font-family:var(--sans); font-size:8.5pt; text-indent:-12pt; padding-left:12pt;
  margin:0 0 5pt; text-align:left; }
.tb-refs .ref .anno{ font-style:italic; color:var(--soft); }

/* ---------------- table of contents ---------------- */
.toc{ page:front; break-after:page; }
.toc h1{ font-family:var(--sans); font-size:17pt; font-weight:800; text-transform:uppercase;
  letter-spacing:.04em; color:#fff; background:var(--accent); padding:7pt 11pt; margin:0 0 14pt; }
.toc .row{ font-family:var(--sans); font-size:10pt; display:flex; align-items:baseline; margin:0 0 5pt; }
.toc .row.part{ font-weight:800; text-transform:uppercase; font-size:9.5pt; letter-spacing:.05em;
  color:var(--accent); margin-top:11pt; }
.toc a{ color:var(--ink); text-decoration:none; }
.toc a::after{ content:leader('.') target-counter(attr(href), page); color:var(--faint); }

/* ---------------- title-page cover (Davidson) ---------------- */
.cover{ page:front; break-after:page; min-height:8.7in; display:flex; flex-direction:column;
  justify-content:center; align-items:center; text-align:center; padding:0 0.5in; }
.cover .ctop{ flex:0 0 auto; width:100%; }
.cover .cbot{ flex:0 0 auto; margin-top:0.85in; width:100%; }
.cover .cband{ display:inline-block; background:var(--accent); padding:6pt 16pt; margin:0 0 20pt; }
.cover .cband .ccat{ font-family:var(--sans); font-weight:800; text-transform:uppercase; letter-spacing:.14em;
  font-size:10pt; color:#fff; }
.cover .ctitle{ font-family:var(--sans); font-weight:800; font-size:31pt; line-height:1.05;
  letter-spacing:-.01em; color:var(--ink); margin:0; max-width:8in; }
.cover .csub{ font-family:var(--serif); font-style:italic; font-size:15pt; line-height:1.3; color:var(--deep);
  margin:16pt auto 0; max-width:6.2in; }
.cover .cmeta{ margin:16pt 0 0; }
.cover .cgoal{ font-family:var(--serif); font-style:italic; font-size:11.5pt; color:var(--deep); }
.cover .cseries{ font-family:var(--sans); font-size:8pt; text-transform:uppercase; letter-spacing:.09em;
  color:var(--faint); margin-top:7pt; }
.cover .arule{ width:26%; height:2.4pt; background:var(--accent); margin:0 auto 10pt; }
.cover .aname{ font-family:var(--sans); font-weight:700; letter-spacing:.04em; font-size:13pt; color:var(--ink); }
.cover .ameta{ font-family:var(--serif); font-style:italic; font-size:10pt; color:var(--faint); margin-top:3pt; }
.cover .cdate{ font-family:var(--serif); font-style:italic; font-size:9.5pt; color:var(--faint); margin-top:4pt; }

V·D.7 · The Davidson Build Engine

The Davidson-aware engine. It shares the Textbook Edition's parsing of front matter, parts, chapters, run-ins, pipe tables, figures, cross-references, and inline marks, and adds the Davidson rendering: level-two to level-four headings become banner, bar, and square heads; the box directive parses an optional number and renders the numbered icon box with its typed colour and letter; a box wrapping a table is given the spanning class; the chapter opener emits the solid band and the outer-margin thumb-tab running element; the running header binds to the chapter title. Save as build_textbook_d.py. Build with python3 build_textbook_d.py source.md /mnt/user-data/outputs/{slug}.pdf, the base URL the source directory, then inspect the render: confirm the banners, the numbered icon boxes with solid white glyph squares, the spanning table-boxes, the outer-margin thumb-tab mirrored recto and verso, and the contents page numbers. Present the PDF and keep the Markdown source as master.

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
TEXTBOOK_D build engine (Davidson style).
Compiles one Markdown source to a dense two-column, solid-banner, numbered-icon-box
PDF via WeasyPrint. Sans throughout; chapter thumb-tab in the outer margin.

Usage:  python3 build_textbook_d.py source.md [out.pdf]
Source dir must contain textbook_d.css (and optionally face.css + fonts/).
"""
import sys, os, re, html, pathlib

ACC_CYCLE = ["acc-1","acc-2","acc-3","acc-4","acc-5","acc-6"]
ROMAN = {'I':1,'V':5,'X':10,'L':50,'C':100,'D':500,'M':1000}

# box type -> (css class, icon letter)
BOXMAP = {
  'info':('','i'), 'data':('','i'), 'key':('key','\u2605'), 'summary':('key','\u2605'),
  'definition':('','i'), 'concept':('','i'), 'example':('','e'),
  'practice':('practice','+'), 'application':('practice','+'), 'procedure':('practice','+'),
  'clinical':('practice','+'),
  'caution':('caution','!'), 'warning':('caution','!'), 'emergency':('caution','!'),
  'note':('note','i'), 'context':('note','i'),
}

def roman_to_int(s):
    s=s.upper(); n=0; p=0
    for ch in reversed(s):
        if ch not in ROMAN: return None
        v=ROMAN[ch]; n+=v if v>=p else -v; p=v
    return n

def slug(s):
    s=re.sub(r'<[^>]+>','',s); s=re.sub(r'[^A-Za-z0-9]+','-',s).strip('-').lower()
    return s or 'sec'

def parse_front(text):
    fm={}; m=re.match(r'^---\s*\n(.*?)\n---\s*\n',text,re.S); body=text
    if m:
        for line in m.group(1).splitlines():
            if ':' in line: k,v=line.split(':',1); fm[k.strip()]=v.strip()
        body=text[m.end():]
    return fm, body

def inline(s):
    s=html.escape(s, quote=False)
    s=re.sub(r'\[\[(.+?)\]\]', r'<span class="xref">\1</span>', s)
    s=re.sub(r'\*\*(.+?)\*\*', r'<strong>\1</strong>', s)
    s=re.sub(r'(?<!\*)\*(?!\*)(.+?)(?<!\*)\*(?!\*)', r'<em>\1</em>', s)
    s=re.sub(r'`(.+?)`', r'<code>\1</code>', s)
    s=re.sub(r'\^(\w+)\^', r'<sup>\1</sup>', s)
    return s

def render_blocks(lines):
    out=[]; i=0; n=len(lines)
    while i<n:
        ln=lines[i]
        if not ln.strip(): i+=1; continue
        if re.match(r'^\s*[-*\u2022]\s+', ln):
            items=[]
            while i<n and re.match(r'^\s*[-*\u2022]\s+', lines[i]):
                items.append(inline(re.sub(r'^\s*[-*\u2022]\s+','',lines[i]).rstrip())); i+=1
            out.append('<ul>'+''.join(f'<li>{x}</li>' for x in items)+'</ul>'); continue
        if re.match(r'^\s*\d+\.\s+', ln):
            items=[]
            while i<n and re.match(r'^\s*\d+\.\s+', lines[i]):
                items.append(inline(re.sub(r'^\s*\d+\.\s+','',lines[i]).rstrip())); i+=1
            out.append('<ol>'+''.join(f'<li>{x}</li>' for x in items)+'</ol>'); continue
        mrun=re.match(r'^\{\.\s*(.+?)\s*\.\}\s*(.*)$', ln)
        if mrun:
            out.append(f'<span class="runin">{inline(mrun.group(1))}.</span> ')
            buf=[mrun.group(2).rstrip()] if mrun.group(2).strip() else []
            i+=1
            while i<n and lines[i].strip() and not re.match(r'^\s*([-*\u2022]|\d+\.)\s+', lines[i]) \
                  and not lines[i].startswith(':::') and not re.match(r'^\{\.', lines[i]):
                buf.append(lines[i].rstrip()); i+=1
            out[-1]='<p>'+out[-1]+inline(' '.join(buf))+'</p>'; continue
        buf=[ln.rstrip()]; i+=1
        while i<n and lines[i].strip() and not re.match(r'^\s*([-*\u2022]|\d+\.)\s+', lines[i]) \
              and not lines[i].startswith(':::') and not re.match(r'^\{\.', lines[i]):
            buf.append(lines[i].rstrip()); i+=1
        out.append('<p>'+inline(' '.join(buf))+'</p>')
    return ''.join(out)

def render_table_rows(tbl_lines, in_box=False):
    rows=[r for r in tbl_lines if r.strip()]
    cells=[[c.strip() for c in re.split(r'(?<!\\)\|', r.strip().strip('|'))] for r in rows]
    body=[c for c in cells if not all(re.fullmatch(r':?-{2,}:?', x or '-') for x in c)]
    head=body[0] if body else []; data=body[1:] if len(body)>1 else []
    h='<thead><tr>'+''.join(f'<th>{inline(x)}</th>' for x in head)+'</tr></thead>'
    b='<tbody>'+''.join('<tr>'+''.join(f'<td>{inline(x)}</td>' for x in r)+'</tr>' for r in data)+'</tbody>'
    return h+b

def render_table(tbl_lines, caption=None, note=None, span=False):
    cap=f'<caption>{inline(caption)}</caption>' if caption else ''
    nt=f'<div class="tnote">{inline(note)}</div>' if note else ''
    sp=' span' if span else ''
    return f'<div class="tb-table{sp}"><table>{cap}{render_table_rows(tbl_lines)}</table>{nt}</div>'

def render_figure(alt, path, brace, span=False):
    label=src=''; bodytxt=brace or alt
    if brace and '|' in brace:
        parts=[p.strip() for p in brace.split('|')]
        label=parts[0] if len(parts)>0 else ''
        bodytxt=parts[1] if len(parts)>1 else ''
        src=parts[2] if len(parts)>2 else ''
    cap=''
    if label: cap+=f'<span class="flabel">{inline(label)}</span> '
    cap+=inline(bodytxt)
    if src: cap+=f' <span class="fsrc">({inline(src)})</span>'
    sp=' span' if span else ''
    return (f'<div class="tb-fig{sp}"><div class="frame"><img src="{html.escape(path)}" alt="{html.escape(alt)}"></div>'
            f'<figcaption>{cap}</figcaption></div>')

def render_box(kind, number, title, body_lines, span=False):
    if kind=='note' and not title and not number:
        return f'<p class="aside">{render_blocks(body_lines).replace("<p>","").replace("</p>"," ")}</p>'
    if kind=='reading':
        rows=[]
        for ln in body_lines:
            if not ln.strip(): continue
            if ' -- ' in ln:
                cit,anno=ln.split(' -- ',1)
                rows.append(f'<div class="ref">{inline(cit.strip())} <span class="anno">{inline(anno.strip())}</span></div>')
            else: rows.append(f'<div class="ref">{inline(ln.strip())}</div>')
        return ''.join(rows)
    cls,icon = BOXMAP.get(kind, ('','i'))
    # detect an embedded table inside the box body
    span=''; inner=''
    has_table = any(l.lstrip().startswith('|') for l in body_lines)
    if has_table:
        # split body into pre-text, table, post-text
        i=0; n=len(body_lines); chunks=[]
        while i<n:
            if body_lines[i].lstrip().startswith('|'):
                tl=[]
                while i<n and body_lines[i].lstrip().startswith('|'): tl.append(body_lines[i]); i+=1
                note=None
                if i<n and re.match(r'^Note:\s*(.+)$', body_lines[i]):
                    note=re.match(r'^Note:\s*(.+)$', body_lines[i]).group(1).strip(); i+=1
                chunks.append('<table>'+render_table_rows(tl,True)+'</table>'+(f'<div class="tnote">{inline(note)}</div>' if note else ''))
            else:
                tx=[]
                while i<n and not body_lines[i].lstrip().startswith('|'): tx.append(body_lines[i]); i+=1
                if any(x.strip() for x in tx): chunks.append(render_blocks(tx))
        inner=''.join(chunks)
    else:
        inner=render_blocks(body_lines)
    span_cls=' span' if span else (' flow' if has_table else '')
    num=f'<span class="dxnum">{html.escape(number)}</span>' if number else ''
    head=(f'<div class="dxhead"><span class="dxicon">{icon}</span>'
          f'<span class="dxttl">{num}{inline(title)}</span></div>')
    return f'<div class="dx-box {cls}{span_cls}">{head}<div class="dxbody">{inner}</div></div>'

def build_html(fm, body, cssfiles):
    single = fm.get('accent_mode','per-part')=='single'
    single_acc = fm.get('single_accent','acc-1')
    if not single_acc.startswith('acc-'): single_acc='acc-'+single_acc
    toc_on = fm.get('toc','on').lower() in ('on','true','yes','1')

    lines=body.splitlines(); i=0; n=len(lines)
    parts=[]; html_parts=[]; part_idx=0
    cur_acc = single_acc if single else 'acc-1'
    open_chapter=False; chap_buf=[]

    def flush_chapter():
        nonlocal chap_buf, open_chapter
        if open_chapter:
            html_parts.append('<div class="cols">'+''.join(chap_buf)+'</div></section>')
            chap_buf=[]; open_chapter=False

    while i<n:
        ln=lines[i]
        mp=re.match(r'^#\s+PART\s+([IVXLCDM]+|\d+)\s*[\u00b7\u2014\-\.:]?\s*(.*)$', ln, re.I)
        if mp:
            flush_chapter(); num=mp.group(1); name=mp.group(2).strip(); part_idx+=1
            if not single: cur_acc=ACC_CYCLE[(part_idx-1)%len(ACC_CYCLE)]
            pid='part-'+slug(num+'-'+name)
            parts.append(('part',pid,f'Part {num} &#183; {name}' if name else f'Part {num}'))
            html_parts.append(f'<section class="part-divider {cur_acc}" id="{pid}">'
                f'<div class="pnum">Part {html.escape(num)}</div><div class="pbar"></div>'
                f'<div class="ptitle">{inline(name)}</div></section>')
            i+=1; continue
        mc=re.match(r'^#\s+(\d+)\.\s+(.+)$', ln)
        if mc:
            flush_chapter(); cnum=mc.group(1); ctitle=mc.group(2).strip()
            cid='ch-'+slug(cnum+'-'+ctitle); parts.append(('chapter',cid,f'{cnum} &#183; {ctitle}'))
            html_parts.append(f'<section class="chapter {cur_acc}" id="{cid}">'
                f'<div class="chapter-open"><div class="cband"><div class="cnum">{html.escape(cnum)}</div>'
                f'<div class="ctitle">{inline(ctitle)}</div></div></div>'
                f'<div class="dx-thumb"><span class="tabnum">{html.escape(cnum)}</span></div>')
            open_chapter=True; i+=1; continue
        m1=re.match(r'^#\s+(.+)$', ln)
        if m1:
            flush_chapter(); t=m1.group(1).strip(); cid='ch-'+slug(t)
            parts.append(('chapter',cid,t))
            html_parts.append(f'<section class="chapter {cur_acc}" id="{cid}">'
                f'<div class="chapter-open"><h1 class="cplain">{inline(t)}</h1></div>'
                f'<div class="dx-thumb"><span class="tabnum">&#9679;</span></div>')
            open_chapter=True; i+=1; continue
        # directive box: ::: type [number] Title
        md=re.match(r'^:::\s*(\w+)(\*?)\s*(.*)$', ln)
        if md:
            kind=md.group(1).lower(); bspan=md.group(2)=='*'; rest=md.group(3).strip()
            num=''; mnum=re.match(r'^([\dA-Za-z]+\.[\d]+[A-Za-z]?|\d+[A-Za-z]?)\s+(.*)$', rest)
            if mnum and re.search(r'\d', mnum.group(1)):
                num=mnum.group(1); title=mnum.group(2).strip()
            else: title=rest
            j=i+1; bl=[]
            while j<n and not lines[j].strip()==':::': bl.append(lines[j]); j+=1
            frag=render_box(kind, num, title, bl, span=bspan)
            (chap_buf if open_chapter else html_parts).append(frag); i=j+1; continue
        # table with optional caption/note
        cap=None; tspan=False; mcapt=re.match(r'^Table(\*?):\s*(.+)$', ln); look=i
        if mcapt: tspan=mcapt.group(1)=='*'; cap=mcapt.group(2).strip(); look=i+1
        if look<n and lines[look].lstrip().startswith('|'):
            tl=[]; k=look
            while k<n and lines[k].lstrip().startswith('|'): tl.append(lines[k]); k+=1
            note=None
            if k<n and re.match(r'^Note:\s*(.+)$', lines[k]):
                note=re.match(r'^Note:\s*(.+)$', lines[k]).group(1).strip(); k+=1
            frag=render_table(tl, cap, note, span=tspan)
            (chap_buf if open_chapter else html_parts).append(frag); i=k; continue
        # figure
        mf=re.match(r'^!(\*?)\[(.*?)\]\((.*?)\)(?:\{(.*?)\})?\s*$', ln)
        if mf:
            (chap_buf if open_chapter else html_parts).append(render_figure(mf.group(2),mf.group(3),mf.group(4),span=mf.group(1)=='*')); i+=1; continue
        # headings
        mh2=re.match(r'^##\s+(.+)$', ln); mh3=re.match(r'^###\s+(.+)$', ln); mh4=re.match(r'^####\s+(.+)$', ln)
        if mh4:
            t=mh4.group(1).strip(); mnum=re.match(r'^(\d+(?:\.\d+)+)\s+(.+)$', t)
            frag=f'<h4><span class="num">{mnum.group(1)}</span>{inline(mnum.group(2))}</h4>' if mnum else f'<h4>{inline(t)}</h4>'
            (chap_buf if open_chapter else html_parts).append(frag); i+=1; continue
        if mh3:
            t=mh3.group(1).strip(); mnum=re.match(r'^(\d+(?:\.\d+)+)\s+(.+)$', t)
            frag=f'<h3><span class="num">{mnum.group(1)}</span>{inline(mnum.group(2))}</h3>' if mnum else f'<h3>{inline(t)}</h3>'
            (chap_buf if open_chapter else html_parts).append(frag); i+=1; continue
        if mh2:
            t=mh2.group(1).strip(); mnum=re.match(r'^(\d+(?:\.\d+)*)\s+(.+)$', t)
            frag=f'<h2><span class="num">{mnum.group(1)}</span>{inline(mnum.group(2))}</h2>' if mnum else f'<h2>{inline(t)}</h2>'
            (chap_buf if open_chapter else html_parts).append(frag); i+=1; continue
        # ordinary block
        buf=[ln]; i+=1
        while i<n and lines[i].strip() and not lines[i].startswith(':::') and not re.match(r'^#', lines[i]) \
              and not lines[i].lstrip().startswith('|') and not re.match(r'^(Table\*?:|Note:|!\*?\[)', lines[i]):
            buf.append(lines[i]); i+=1
        (chap_buf if open_chapter else html_parts).append(render_blocks(buf))

    flush_chapter()

    toc=''
    if toc_on and parts:
        rows=['<h1>Contents</h1>']
        for lvl,pid,label in parts:
            rows.append(f'<div class="row part">{label}</div>' if lvl=='part'
                        else f'<div class="row"><a href="#{pid}">{label}</a></div>')
        toc=f'<nav class="toc {single_acc if single else ACC_CYCLE[0]}">'+''.join(rows)+'</nav>'
    # ----- title-page cover (front matter) -----
    import re as _rc
    _cv_name = fm.get('author','').strip() or _rc.sub(r'\^\d+\^','',fm.get('author_line','')).split(',')[0].strip()
    _cv_cat  = fm.get('category', fm.get('article_type',''))
    _cv_goal = fm.get('goal', fm.get('tagline',''))
    _cv_ser  = fm.get('series', fm.get('doi',''))
    _cv_sub  = fm.get('subtitle',''); _cv_date=fm.get('date',''); _cv_aff=fm.get('affiliation','')
    _cv_title= fm.get('title','')
    _cv_acc  = single_acc if single else ACC_CYCLE[0]
    cover=''
    if _cv_title:
        _cs=''
        if _cv_cat:  _cs+=f'<div class="ccat">{inline(_cv_cat)}</div>'
        if _cv_goal: _cs+=f'<div class="cgoal">{inline(_cv_goal)}</div>'
        if _cv_ser:  _cs+=f'<div class="cseries">{inline(_cv_ser)}</div>'
        cover=(f'<div class="cover {_cv_acc}"><div class="ctop"><div class="cband"><div class="ccat">{inline(_cv_cat)}</div></div>'
               f'<h1 class="ctitle">{inline(_cv_title)}</h1>'
               + (f'<div class="csub">{inline(_cv_sub)}</div>' if _cv_sub else '')
               + (f'<div class="cmeta">' + (f'<div class="cgoal">{inline(_cv_goal)}</div>' if _cv_goal else '') + (f'<div class="cseries">{inline(_cv_ser)}</div>' if _cv_ser else '') + '</div>' if (_cv_goal or _cv_ser) else '')
               + '</div><div class="cbot"><div class="arule"></div>'
               + f'<div class="aname">{inline(_cv_name)}</div>'
               + (f'<div class="ameta">{inline(_cv_aff)}</div>' if _cv_aff else '')
               + (f'<div class="cdate">{inline(_cv_date)}</div>' if _cv_date else '')
               + '</div></div>')
    links=''.join(f'<link rel="stylesheet" href="{html.escape(c)}">' for c in cssfiles)
    title=fm.get('title','')
    return (f'<!DOCTYPE html><html><head><meta charset="utf-8"><title>{html.escape(title)}</title>'
            f'{links}</head><body>{cover}{toc}{"".join(html_parts)}</body></html>')

def main():
    if len(sys.argv)<2: print("usage: build_textbook_d.py source.md [out.pdf]"); sys.exit(1)
    src=pathlib.Path(sys.argv[1]).resolve(); text=src.read_text(encoding='utf-8')
    fm, body = parse_front(text); srcdir=src.parent
    css=[]
    if (srcdir/'face.css').exists(): css.append('face.css')
    css.append('textbook_d.css')
    doc=build_html(fm, body, css)
    (srcdir/'_textbook_d.html').write_text(doc, encoding='utf-8')
    out_pdf=pathlib.Path(sys.argv[2]) if len(sys.argv)>2 else srcdir/'textbook_d.pdf'
    from weasyprint import HTML
    HTML(string=doc, base_url=str(srcdir)).write_pdf(str(out_pdf))
    print("PDF:", out_pdf)

if __name__=='__main__': main()

V·D.8 · Codex-Native Davidson

A Master Codex volume in the Davidson register sets edition: textbook_d with accent_mode: single and single_accent: copper or single_accent: slate, so the volume reads in one sealed house colour with the full Davidson furniture, the banners, the numbered icon boxes, and the thumb-tab all in copper or slate. The verdict glyphs appear as bare text glyphs inside boxes per the Module IV discipline. The box mapping of Section V·D.3 carries codex content: verdicts in gold key boxes, PSPs and axioms in definition boxes, regime instances in worked example boxes, procedures and cascade steps in green practice boxes, anchor-inflation and typing locks in red caution boxes, proof-boundary remarks in depth asides. Where a chapter must read in a distinct colour, set accent_mode: per-part and the thumb-tab, banners, and bars recolour per Part while the semantic box colours, green for practice, red for caution, hold across the whole volume. The faithful-map law and honest typing govern as everywhere: a box stating a result types it Type T, Type C, or Type S, and overclaiming a warrant grade in a box header is the same failure it is in prose.

V·D.9 · Pre-Flight and Gate Integration

A TextBook_D deliverable passes the standard pre-delivery gate sequence and, for a codex volume, produces the V-FIO Diagnostic Log exactly as the default edition. Before presentation it passes a short structural checklist. The body and every element of furniture set in the sans, with no serif anywhere. Every box carries its colour and its typed letter square, and every wide table-box spans the page. The chapter thumb-tab rides the outer margin and mirrors recto and verso, and the running header binds the live chapter title. Every banner is a true major-section break, not an over-used decoration. The contents page numbers resolve against final pagination, and every figure embeds. Every claim-bearing box and figure carries its warrant tier with no inflation. Failure of any item returns the book for repair before any presentation.

V·D Quick Reference Card

Variant selector: [EDITION:TEXTBOOK_D], [EDITION:TEXTBOOK_B], or front-matter edition: textbook_d / edition: textbook_b; addressable in prose as TextBookB. Auto-selected for codex and master deliverables whose content is strongly instructional or box- and table-heavy. A skin over the Textbook Edition; the two editions and the Module III paper renderer never co-fire. Type: sans-serif throughout, Archivo as TBSans, 9.8 pt body. Never any serif. Dense two-column grid, 22 pt gutter. Page: US Letter 612 by 792, margins 54 / 50 / 54 with the outer margin widened to 66 pt for the thumb-tab, mirrored to the open edge. Accent: per-part six-hue cycle, or single copper or slate for the house. Drives the chapter band, thumb-tab, banners, bars, square marks, cross-references, and the info and key boxes. Semantic box colours, green practice and red caution, hold across the volume. Headings: level-two solid banner, level-three tinted bar, level-four square-marked rule head, run-in bold-italic lead. Reserve the banner for true section breaks. Boxes: numbered icon box, solid colour header with a white glyph square and white reversed number and title. Types info and data and definition and concept in accent with i, key and summary in gold with a star, example with e, practice and application and procedure in green with a plus, caution and warning and emergency in red with an exclamation, note and context in purple. A box wrapping a table spans both columns. Furniture: solid-band chapter opener with large white numeral; thumb-tab in the outer margin mirrored recto and verso; deep-accent running header with the chapter title and page; front named page suppresses furniture on contents and dividers; contents banner with leaders. Source language: the Textbook Edition's, with edition: textbook_d, the level two to four heading ladder mapping to banner, bar, and square, and ::: type [number] Title boxes. Everything else identical. Build: python3 build_textbook_d.py source.md /mnt/user-data/outputs/{slug}.pdf; base URL the source directory; inspect the render; present the PDF; keep the Markdown source as master. No DOCX for this variant. Locks: no LaTeX, Unicode math only; no em-dashes; ﷺ as one glyph; no emoji, CSS glyph squares only; no fragmentation; warrant tier on every claim-bearing box and figure; faithful map, never inflation.

---

MODULE VI · THE PAPER-UPGRADE SUBROUTINE (RE-FORGE PROTOCOL)

Purpose. Reforge an existing Geometric Trisduction paper into a hardened single-voice replacement that preserves the original's spine, schema, and verdicts maximally, tightens the language, removes loose and imprecise claims, and fortifies the mathematics to its honest warrant grade. The output reads as one paper that was always this paper. No revision history. This module composes through the existing edition routing; it is not itself an edition.

Trigger. Token [UPGRADE]. Phrases: update paper, upgrade paper, re-forge paper, reforge paper, fortify paper, harden paper, modernize paper, bring the paper up to current law, replace this paper with a hardened edition, and any request pointing at an existing paper (URL or file) with intent to issue a stronger replacement of it.

VI.A · The Preservation Lock

The spine, the schema, and the verdicts are preserved by default. The burden is on a change to justify itself, never on preservation. Three objects carry forward intact unless a change earns its place against this lock.

The spine. The paper's thesis and rhetorical arc, the order in which it makes its case, the framing that makes the argument land.

The schema. The section architecture, the part division, the table and figure placement, the titling. Section order is held. A section is added only to carry genuinely new mass and removed only when it carried a claim now cut for imprecision.

The verdicts. Every seal [⟀], break [X], under-determined [?], and survivable [△] the paper issues, with its conclusion. A verdict moves only on new structural argument or new mathematical mass outside the existing span, never on reframing. This is the Revision Mandate, binding here. An author-authorized claim, a stress-tested position the architect has placed under standing permission, is never stripped. If it was over-typed it is re-typed to its honest grade and kept in full force at that grade. It is not softened toward orthodoxy and it is not removed.

VI.B · The Warrant-Rectification Pass

Walk every claim and re-type it to its honest grade. Three operations, applied symmetrically.

Pull-back. An over-typed claim is pulled to the grade it actually earns. Theorem-grade rhetoric on premise-grade content, a corroboration stated as load-bearing, an exhibit stated as a bijection, a definitional closure stated as a proof: each is pulled back. Anchor inflation in the sealing direction is the named target.

Promotion. Content that is genuinely theorem-grade but was buried, hedged, or stated below grade is surfaced and stated at grade. Under-claim is a defect symmetric to over-claim.

Fortification. Where the spine rests on weak scaffolding, a scorecard, an analogy, an appeal, an enumeration that cannot survive self-application, supply or strengthen a theorem-grade core that carries the same conclusion. The original's argument is kept and a load-bearing floor is placed under it, so the conclusion survives being turned on itself. Hardening is always correct. Watering down toward orthodoxy is never correct. The discipline is symmetric: credit genuine mass, refuse to inflate it.

VI.C · Language Tightening and the Loose-Claim Cut

Remove loose and imprecise claims. Purge any apologetic or outside-reader register; it is incompatible with the single native voice and is a corruption to be cut, not balanced. Cut hedging, redundancy, and any sentence that overstates reach. A decline for want of purchase is stated as method-silence about the instrument's reach, never as a claim that the object is empty or structureless. Every quantity and every verdict carries its warrant tier in the rebuilt text. Telegraphic density, paragraphs over bullets, no em-dashes, periods between thoughts.

VI.D · Numerical Rectification

Every structural claim that cites a number is reproduced at a fixed seed before it ships, and the worked validation travels with the claim. A number the upgrade cannot reproduce is cut or marked not reached. A numerical trace for a stage the audit did not reach is forbidden. Where the original carried a stale or wrong constant, the corrected value is integrated silently as standing law, never flagged as a correction.

VI.E · Silent Integration and Single Voice

Every fix lands as native discipline under the Module II Silent-Integration Protocol. No revised, no previously, no this version, no the earlier paper, no now corrected, no version or evolution marker anywhere in the body. The paper speaks once. Gate 1 of the gate sequence enforces this on the rebuilt artifact.

VI.F · Edition, Build, and Delivery

Compose through the existing edition routing. Default to the edition the original used; a published paper re-forges as the Professional Journal Paper edition (Module III·J) unless an explicit selector says otherwise. Secular body for a published paper, no honorifics and no ceremonial closure in the body; native apparatus, glyphs, and warrant grades permitted at the paper's tier. Build PDF and MD. Run the full pre-delivery gate sequence: Gate 1 silent-integration scan empty, the em-dash lock empty, the secular-body check empty for a published paper, every reference cited inline and every inline citation resolved, the warrant-typing pass showing neither inflation nor understatement, and the numerical reproduction confirmed. Any failure repairs the source and rebuilds before delivery. Present the PDF as primary with the MD as master, a succinct summary naming what was preserved, what was tightened, what was fortified, and what was re-typed, and no postamble.

VI.G · Execution Order

1. Ingest the original. Web-fetch the URL or read the file. Extract the spine, the schema, and the verdict ledger.
2. Lock preservation (VI.A). Hold spine, schema, verdicts, and author-authorized claims.
3. Warrant-rectification pass (VI.B). Pull back, promote, fortify.
4. Language tightening (VI.C). Cut loose and imprecise claims; purge apologetic register.
5. Numerical rectification (VI.D). Reproduce every cited number at a fixed seed.
6. Compose the replacement in single voice through the edition renderer.
7. Silent integration (VI.E). Strip every evolution marker.
8. Build PDF and MD; run the gate sequence (VI.F).
9. Deliver. Present PDF and MD; succinct summary; no postamble.
10. Produce the V-FIO diagnostic log internally, what moved, what grade changed, what was reproduced, retained in the working directory, never published.

---

PRE-DELIVERY GATE SEQUENCE

Run all applicable gates after composition and integration, before any build. Any failure returns the manuscript for repair, then re-runs the gates. Visual-edition figures additionally pass their own twenty-point pre-flight (IV.11) per figure before assembly; the host paper still passes the gates below.

Gate 1 · Silent-Integration Scan (every paper). Search the artifact for: "v1," "v2," "v3," "vN.0," "with worked counter-example," "with discriminating," "audit," "revised," "this version," "the present paper specifies," "addressing," "in response to," "earlier," "previously," "fully executes" (often defensive), "now [restricted/specified/stated]" (often defensive). Most should be removed or rewritten as native discipline. Case-by-case judgment; some are content-legitimate in context.

Gate 2 · Tier Threshold Audit (Forge-mode papers). Operates per tier. Tier 0: rejects any hit on Trisductive vocabulary, glyph, PSP ID, verdict tag, axis variable name, ceremonial closure, or L_1 / L_2 / L_3 layer reference in body text; any hit fails the audit and returns the manuscript for re-translation. Tier 1: verifies every Trisductive term in body text carries a complete inline gloss on first use, no internal codex IDs appear in body text, no glyphs appear in body text, and the count of distinct Trisductive terms does not exceed three across the full manuscript; any failure returns the manuscript for re-translation. Tier 2: verifies internal codex coherence only. The Anti-Dramatization Filter and the no-biological-phenomenology-for-structural-corrections rule apply across all three tiers.

Gate 3 · Format Contract Verification (every deliverable). Front matter complete and version-marker free. Title and subtitle satisfy the titling discipline. Headings follow the numbering conventions. Universal locks intact. Style, cover, and format selections match tokens and defaults.

---

INTERNAL V-FIO DIAGNOSTIC LOG (Retained, Never Published)

For every Forge-mode paper regardless of output tier, produce a parallel V-FIO_Diagnostic_Log containing: full PSP sweep with HEM unpacking; triaxial decomposition of every major claim; 12-Gate Cascade record (which gates fired, which were skipped under PSP coverage, which failure modes if any were caught and resolved); CDT projection on residue; verdict ledger with [⟀] / [X] / [△] / [?] tags; and Fractal Sweep results. The log is retained internally in the working directory, never placed in /mnt/user-data/outputs/ and never presented unless the architect explicitly requests it. The architect may consult it during peer review to defend any claim in the published paper from the underlying cascade audit. The log uses the full native Trisductive register regardless of the published paper's tier.

---

DEFAULT BEHAVIOR SUMMARY

Edition by trigger, an explicit selector always overriding. A bare create a pdf, make a pdf, or create a document with no other signal renders Apex Pristine, the single-column house paper. Any paper, external paper, publication paper, journal paper, or the bare word paper renders the Professional Journal Paper edition. A codex pdf, master pdf, master document, internal codex, or any book, textbook, manual, or handbook renders the Textbook edition, the skin auto-selected by content, Rang-and-Dale (TextBookA) for prose-dense reference and Davidson (TextBookB) for instructional box-heavy material, overridable by TextBookA or TextBookB. A token [EDITION:...] or front-matter edition: overrides the trigger; the four editions never co-fire.

Pure disciplinary problem, no framework-specific load-bearing claim: Tier 0, twelve-section template, no Trisductive lingo, full disciplinary translation. Problem where one or two framework constructs genuinely sharpen the argument: Tier 1, twelve-section template, minimum lingo with full inline glosses. Problem about the framework itself or codex-internal deliverable: Tier 2, twelve-section template, full native register. Uncertain: Tier 1, twelve-section template, conservative gloss discipline. Visual edition requested: Module IV, public register and blueprint aesthetic unless directed otherwise, Mode A appendix-embed when a host paper exists, single PDF output presented INDEX-first in Mode B. Any paper, an external paper, a publication paper, a journal paper, the bare word paper, or the token [EDITION:JOURNAL]: Module III·J Professional Journal Paper edition, serif two columns, an accent masthead carrying the journal name and article type, a bold ruled abstract, a raised initial, tinted Box N panels, superscript citations with a numbered reference list, and full-width end matter, the middle register between the Apex Pristine paper and the Textbook editions; present the PDF and keep the Markdown source as master. Textbook, manual, handbook, survey, or Master Codex volume requested as a book: Module V Textbook Edition, two columns, serif body with sans furniture, per-part accent or a single house accent, the box taxonomy, spanning tables and figures, mirrored running header; present the PDF and keep the Markdown source as master. The skin is auto-selected by content, the calm Rang-and-Dale (TextBookA, [EDITION:TEXTBOOK_A]) for encyclopedic and prose-dense codex material and the aggressive Davidson (TextBookB, [EDITION:TEXTBOOK_B]) for instructional, box-, and table-heavy material; Davidson sets sans throughout, solid banner headings, numbered icon boxes typed by colour, and an outer-margin chapter thumb-tab. Codex pdf, master pdf, master document, and internal codex requests route here. Update paper, upgrade paper, reforge paper, fortify paper, or the token [UPGRADE] on an existing paper: Module VI Paper-Upgrade subroutine, preserve spine, schema, and verdicts maximally, pull back over-typed claims and fortify weak scaffolding to honest grade, cut loose and imprecise claims and any apologetic register, reproduce every cited number at a fixed seed, silent integration with no revision history, compose through the edition the original used (Journal for a published paper), present PDF then MD. Every deliverable: silent integration of any fixes, Gate 1 scan, titling discipline, Apex Pristine style unless overridden, cover on over four pages, all three formats unless restricted, present PDF then DOCX then MD, no postamble.

ENFORCEMENT

Producing a document deliverable without invoking this skill is a [DR:format-drift] event and is nullified. Variant directives, if specified, are honored exactly; silent substitution is forbidden. Font fallback to anything other than the named alternates is a [DR:typographic-drift] event. Cover-page suppression on documents over four pages requires explicit [COVER:OFF]; the skill never strips a cover silently. The architect's preferred author block is the default; substitution requires explicit user instruction in the same turn. Delivering an artifact that fails Gate 1 or Gate 2 is an integration-drift event; the build scripts enforce the visual contract, the gates enforce the voice contract. A claim-bearing figure without its grade, a badge-decorated frame, or a figure asserting beyond the body is a faithful-map violation and is rejected at pre-flight, never delivered. Do not bypass any of these.

MODULE VII · THE MEDIUM / SUBSTACK / BLOG FORMAT

A delivery format, not an edition. Where the PDF and DOCX editions render a page, the Blog format emits a single Markdown file tuned to survive copy-paste into the Medium or Substack editor with nothing broken. It applies to any deliverable, a codex, a paper, or scripture, and is requested by the words Medium, Substack, blog, or post, or the token [FORMAT:BLOG]. The output is .md only; no PDF is produced. It composes through the edition routing rather than replacing it: the source is authored in the normal house Markdown, then transformed.

VII.0 · The Problem It Solves

The Medium editor has no table support at all and no mathematics; a pasted Markdown table shows its pipes and pasted LaTeX shows its source. The Substack editor is friendlier, but its table paste is unreliable and its equations are inserted through the editor rather than through paste. So a document dense with tables, the verdict apparatus, scientific notation, and the right-to-left word grid breaks on paste unless it is transformed first. The Blog format transforms it, targeting the lower common denominator so one file works on both platforms.

VII.1 · The Transform

Six moves. Tables are linearised: every pipe table becomes a labelled list, each row a bold lead followed by its header-named fields, and the right-to-left word grid becomes one inline gloss, each word bold with its transliteration and meaning in parentheses, in reading order, set as a blockquote. Raw HTML is flattened: the word grid is linearised, any other table is dropped to a marker, and stray wrappers and styled spans are stripped. Super and subscript are converted to Unicode glyphs where they exist, ^x^ and <sup> to superscript, ~x~ and <sub> to subscript, with a graceful fallback to plain text where no Unicode glyph exists for a character. Mathematics is already Unicode by house rule and passes through untouched; any LaTeX is reported for hand-fixing, since neither platform renders it from paste. Headers, bold, italic, blockquotes, bullet and numbered lists, links, and fenced code are left exactly as written, since all of these paste faithfully on both platforms. Blank runs are normalised.

VII.2 · What Survives, What Does Not

Survives on both platforms: headers, bold, italic, blockquotes, bullet and numbered lists, links, fenced code blocks, and all Unicode text, including Arabic, Bengali, and Unicode mathematics with super and subscript. Linearised to prose: every table and the word grid. Not carried: native rendered tables, LaTeX, and arbitrary HTML, none of which paste reliably. On Substack, simple tables and footnotes may in fact survive the paste; the format does not depend on it, which is exactly why the same file is also safe on Medium.

VII.3 · The Blog Builder

The complete builder. Pure text processing, no fonts and no renderer, so it runs anywhere. Save as build_blog.py.

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""MEDIUM / SUBSTACK / BLOG format builder.

Takes any source .md (codex, paper, or scripture output) and emits a single
.md that survives copy-paste into the Medium or Substack editor without
breaking. The two editors have no reliable markdown-table paste and no LaTeX,
so the transform linearises every table into labelled prose, flattens raw HTML
(including the RTL word grid) into inline gloss, converts ^x^/~x~/<sup>/<sub>
to Unicode super/subscript where the glyph exists, and leaves headers, bold,
italic, blockquotes, lists, links, and fenced code untouched, since those paste
faithfully on both platforms. All mathematics is already Unicode by house rule;
any LaTeX is reported so it can be hand-fixed.

  python3 build_blog.py source.md /mnt/user-data/outputs/{slug}.blog.md
"""
import sys, re
from pathlib import Path

SUP = {'0':'⁰','1':'¹','2':'²','3':'³','4':'⁴','5':'⁵','6':'⁶','7':'⁷','8':'⁸','9':'⁹',
       '+':'⁺','-':'⁻','=':'⁼','(':'⁽',')':'⁾','n':'ⁿ','i':'ⁱ','a':'ᵃ','b':'ᵇ','c':'ᶜ',
       'd':'ᵈ','e':'ᵉ','k':'ᵏ','m':'ᵐ','t':'ᵗ','x':'ˣ'}
SUB = {'0':'₀','1':'₁','2':'₂','3':'₃','4':'₄','5':'₅','6':'₆','7':'₇','8':'₈','9':'₉',
       '+':'₊','-':'₋','=':'₌','(':'₍',')':'₎','a':'ₐ','e':'ₑ','k':'ₖ','m':'ₘ','n':'ₙ',
       'x':'ₓ','t':'ₜ','i':'ᵢ','s':'ₛ','p':'ₚ'}

def to_unicode_run(s, table):
    if all(ch in table for ch in s):
        return ''.join(table[ch] for ch in s)
    return None

def convert_supsub(text):
    # <sup>..</sup> and ^..^  ;  <sub>..</sub> and ~..~
    def rep_sup(m):
        u = to_unicode_run(m.group(1), SUP); return u if u else m.group(1)
    def rep_sub(m):
        u = to_unicode_run(m.group(1), SUB); return u if u else m.group(1)
    text = re.sub(r'<sup>(.*?)</sup>', rep_sup, text, flags=re.S)
    text = re.sub(r'<sub>(.*?)</sub>', rep_sub, text, flags=re.S)
    text = re.sub(r'\^([^\s^]{1,8})\^', rep_sup, text)
    text = re.sub(r'(?<!~)~([^\s~]{1,8})~(?!~)', rep_sub, text)
    return text

def linearize_rtl_grid(html):
    """The protocol's <table dir="rtl"> word grid -> inline per-word gloss."""
    rows = re.findall(r'<tr>(.*?)</tr>', html, flags=re.S)
    cells = [re.findall(r'<td[^>]*>(.*?)</td>', r, flags=re.S) for r in rows]
    cells = [[c.strip() for c in row] for row in cells]
    if not cells:
        return ''
    ncol = max(len(r) for r in cells)
    ar = cells[0] if len(cells) > 0 else []
    tr = cells[1] if len(cells) > 1 else []
    me = cells[2] if len(cells) > 2 else []
    parts = []
    for i in range(ncol):
        a = ar[i] if i < len(ar) else ''
        t = tr[i] if i < len(tr) else ''
        m = me[i] if i < len(me) else ''
        gloss = ', '.join(x for x in (t, m) if x)
        parts.append(f'**{a}** ({gloss})' if gloss else f'**{a}**')
    return '> ' + '  ·  '.join(parts) + '\n'

def linearize_pipe_table(block):
    lines = [ln for ln in block.strip('\n').splitlines() if ln.strip()]
    rows = []
    for ln in lines:
        if re.match(r'^\s*\|?\s*:?-{2,}', ln.replace('|', '').strip()[:3]) or re.match(r'^[\s|:-]+$', ln):
            continue
        cells = [c.strip() for c in ln.strip().strip('|').split('|')]
        rows.append(cells)
    if len(rows) < 2:
        return block
    header, body = rows[0], rows[1:]
    out = []
    for r in body:
        first = r[0] if r else ''
        rest = []
        for j in range(1, len(r)):
            lab = header[j] if j < len(header) else ''
            val = r[j]
            if val:
                rest.append(f'{lab}: {val}' if lab else val)
        line = f'- **{first}**' + (' · ' + '; '.join(rest) if rest else '')
        out.append(line)
    return '\n'.join(out) + '\n'

def transform(md):
    notes = []
    # 1. RTL word grids -> inline gloss
    md = re.sub(r'<table[^>]*dir="rtl"[^>]*>.*?</table>',
                lambda m: linearize_rtl_grid(m.group(0)), md, flags=re.S)
    # 2. any other raw <table> ... </table> (HTML) -> drop to a note (rare)
    md = re.sub(r'<table.*?</table>',
                lambda m: '*[table flattened for blog; see source]*\n', md, flags=re.S)
    # 3. pipe tables -> labelled lists  (protect fenced code first)
    fences = []
    def stash(m):
        fences.append(m.group(0)); return f'\x00F{len(fences)-1}\x00'
    md = re.sub(r'```.*?```', stash, md, flags=re.S)
    # find blocks of >=2 consecutive lines containing a pipe, with a separator row
    out_lines, buf = [], []
    def flush():
        nonlocal buf
        if buf:
            block = '\n'.join(buf)
            if re.search(r'^\s*\|?[\s:-]*-{2,}[\s:|-]*$', block, flags=re.M) and block.count('|') >= 2:
                out_lines.append(linearize_pipe_table(block))
            else:
                out_lines.extend(buf)
            buf = []
    for ln in md.splitlines():
        if '|' in ln and ln.strip():
            buf.append(ln)
        else:
            flush(); out_lines.append(ln)
    flush()
    md = '\n'.join(out_lines)
    for i, f in enumerate(fences):
        md = md.replace(f'\x00F{i}\x00', f)
    # 4. strip leftover block HTML wrappers + inline style spans
    md = re.sub(r'</?div[^>]*>', '', md)
    md = re.sub(r'<span[^>]*>(.*?)</span>', r'\1', md, flags=re.S)
    # 5. super/subscript to Unicode
    md = convert_supsub(md)
    # 6. flag LaTeX
    if re.search(r'(?<!\\)\$[^$\n]+\$|\\\(|\\\[|\\begin\{', md):
        notes.append('LaTeX detected: convert to Unicode by hand; Medium and Substack do not render it from paste.')
    md = re.sub(r'\n{3,}', '\n\n', md)
    return md, notes

def main():
    if len(sys.argv) < 2:
        sys.exit('usage: build_blog.py source.md [out.blog.md]')
    src = Path(sys.argv[1])
    out = sys.argv[2] if len(sys.argv) > 2 else str(src.with_suffix('.blog.md'))
    md, notes = transform(src.read_text(encoding='utf-8'))
    Path(out).write_text(md, encoding='utf-8')
    print('wrote', out)
    for n in notes:
        print('NOTE:', n)

if __name__ == '__main__':
    main()

VII.4 · Authoring and Use

Write the document once in the normal house Markdown, tables and word grids and all. Run the builder to emit the blog file, read the reported notes, hand-fix any flagged LaTeX into Unicode, then copy the rendered file into the Medium or Substack editor. Keep the original Markdown as master; the blog file is a derived delivery artifact, never the source of truth.

VII Quick Reference Card

Trigger: medium, substack, blog, post, [FORMAT:BLOG]. Output: .md only, copy-paste-safe for Medium and Substack. Build: python3 build_blog.py source.md /mnt/user-data/outputs/{slug}.blog.md. Linearises every table and the right-to-left word grid, converts super and subscript to Unicode, flags LaTeX, and preserves headers, emphasis, lists, links, and code verbatim.

---

MODULE VIII · THE AUTHOR'S PROVENANCE AND METHOD DISCLOSURE SUBROUTINE

VIII.0 · Activation and Tier-Awareness

Module VIII is a mandatory subroutine, not an edition. It composes through the edition routing exactly as Module VI does and appends an Author's Provenance and Method Disclosure to every public-facing paper. It fires for the Apex Pristine mid-tier house paper, the Professional Journal fully-public paper, and the Medium/Substack/Blog format. It is skipped for the Textbook and Scripture editions, which are framework-native or scriptural and carry no external-reception burden. The disclosure exists because a translated paper severs its own provenance: no matter how faithfully the method is rendered into conventional language, some omission or incompleteness in explaining the root principle is unavoidable, and the disclosure is the fixed pointer back to the method that the conventional body cannot carry.

The disclosure has two halves. The first is a fixed generic block, identical in every paper, stating what the method is and proving its one load-bearing identity. The second is auto-generated per paper and carries the value, instantiating the method on the paper at hand. The instructions below are generic and hold for any paper; only the second half is customized.

Tier-awareness. The disclosure renders at one of three weights, selected by edition and overridable by token. FULL, the generic half plus the specific half, is the default for Apex Pristine and Blog, where the venue is the author's own and full method transparency is the point. LINE, a single pointer sentence naming the method and the master reference with no apparatus, is the default for the Journal edition, where a blind external reviewer meeting the full framework would recode the argument as the output of a private system and dismiss it on pattern-match; the paper's conclusions stand on their cited results alone, and the LINE weight protects the blind review while still disclosing provenance. OFF suppresses the block. The three override tokens are DISCLOSURE:FULL, DISCLOSURE:LINE, and DISCLOSURE:OFF, an explicit token always winning over the edition default.

The independence clause is never dropped at any weight. The disclosure always states that the paper's conclusions rest solely on the standard results cited in the body and that the method adds them no warrant. Without that clause the block would contradict the paper's own Mosaic Seal and make the result look dependent on the framework, which is the opposite of what a Trisduction paper claims. The whole block is framed as an author's method note, never as a claim about the result, and never as a step in the argument.

Placement. The disclosure renders as a demarcated appendix after the Conclusion and after any content appendices, immediately before the References section. The master reference is injected once into the References section per VIII.3. At LINE weight the appendix is a single sentence; at FULL weight it is the full block below.

VIII.1 · Part One, the Fixed Generic Half

Emitted verbatim at FULL weight. It is a constant, identical across all papers, never paraphrased and never auto-edited. The single insertion point, marked below, is where the auto-generated specific half of VIII.2 is spliced. Emit exactly:

Author's Provenance and Method Disclosure

This paper was developed under Trisduction, a verification and organizing discipline, not a source of results. Its conclusions rest solely on the standard results cited in the body. Trisduction is the method under which those results were decomposed, assembled, and audited. It adds them no warrant and claims no authorship over them.

Trisduction was used for fidelity. It forces a claim onto three independent axes so no single persuasive line carries it alone, requires every verdict to resolve to one of three states, sealed, broken, or open, each with a named failure mechanism, and attaches an explicit warrant grade, theorem, conditional, structural, or premise, to every claim so nothing is stated above its strength. Two errors it is built to catch are inflation, reading an internal lock as a proof, and circularity, reading a restatement of a claim as a derivation of it.

The root axiom, RA, is that to exist is to actuate: every physical existent carries a strictly positive energetic cost, grounded in the Heisenberg energy floor, the zero-point energy, and Landauer's bound. The formal root inside it is that to formally be is to be grounded, with provability and computation levels of access to a determinacy fixed at the ground rather than ingredients of it.

The key procedures, in brief. Orthogonal triaxial convergence: a proposition is split into three disjoint axes, formal-structural, empirical or dynamical, and registrational, verified by three separate instrument sets, agreement across the three the operational meaning of a seal. The Geometric Orthogonal Lock, GOL: the three axes are read as vectors and their independence is tested by the determinant of their correlation matrix, a determinant clear of zero a genuine three-dimensional lock, a collapse to zero one axis dissolving into the plane of the other two, a broken lock. The Convergence Dissolution Test, CDT: before the lock is read, any mass-bearing common cause the three axes might share is projected out, so an apparent convergence that traces to a shared source rather than to independent roads dissolves and carries no warrant, the lock computed on the residue that survives. The twelve-gate cascade: twelve directed failure screens, self-reference, frame-dependence, missing mechanism, and the rest, run in order, the first failure terminating the verdict with its mechanism named. The verdict issues in the three-state economy with its warrant grade attached, and where the argument stops short the open direction is named rather than filled.

:::box G1 The GOL kernel identity, proved The lock is a determinant identity, not a metaphor. Let the three axes, after normalization, be unit vectors expressed in an orthonormal basis of the subspace they span and read as pure quaternions a, b, c, with norm one each. Hamilton's product of two pure quaternions is p q = −(p·q) + p×q, the real part the negative dot product and the imaginary part the cross product, checked on the units by i j = k = i×j and i i = −1 = −(i·i). The lock scalar is the real part of the triple product, λ = Re(a b c).

Expanding, a b = −(a·b) + a×b, so Re(a b c) = −(a·b) Re(c) + Re((a×b) c). The first term is zero since c is pure, and the second is −(a×b)·c by the same product rule, giving

λ = −(a×b)·c = −det[a b c],

minus the signed volume of the parallelepiped the three axes span. Writing A for the matrix whose columns are a, b, c, the correlation matrix is R = A^T^A, its entries the pairwise dot products, so

det(R) = det(A^T^A) = det(A)² = λ².

The kernel identity λ² = det(R) is that the squared signed volume equals the Gram determinant, with the quaternion triple product the machine that computes the volume. It is confirmed at machine precision, the residual |λ² − det(R)| at 2 × 10⁻¹⁶ on the recorded battery. :::

Four consequences fix the reading, and they are why the number can be trusted to say what the lock says. The determinant is bounded, det(R) in the interval from zero to one for unit axes, by Hadamard's inequality above and positive-semidefiniteness below, so the lock has a hard ceiling at one, the fully orthogonal frame, and a hard floor at zero. The floor is the break: det(R) equals zero exactly when the three axes are linearly dependent, one axis lying in the plane of the other two, which is the geometric content of a collapsed lock. The magnitude is orientation-blind: reflecting any axis sends A to a matrix of opposite determinant, flipping the sign of λ while det(R) equals λ² is unchanged, so the number certifies the dimensionality of the lock and never its truth-sign, which is read from the ordered axes and not from the scalar. And the magnitude is frame-invariant: rotating all three axes together is conjugation q to u q ū on the imaginary quaternions, under which the real part is preserved and dot and cross products are covariant, so λ and det(R) depend on the configuration and not on the coordinate labels.

The axis count is three because the algebra forces it, not because three was chosen. The audit-composition law requires associativity, so iterated audits bracket the same way, and the absence of zero divisors, so nonzero warrants never compound to nothing. By Frobenius's theorem the only finite-dimensional associative division algebras over the reals are the reals, the complex numbers, and the quaternions, and three mutually orthogonal imaginary axes exist only in the quaternions, whose imaginary units i, j, k are the three axes. The next normed division algebra, the octonions with seven imaginary units, fails associativity, witnessed by the nonzero associator of e₁, e₂, e₄, so it cannot carry the composition law and the construction stops at three. The three axes are the imaginary part of the unique associative real division algebra that supports the audit.

[[SPECIFIC-HALF INSERTION POINT · the paragraph auto-generated per VIII.2 is spliced here, between the Frobenius paragraph and the closing pointer]]

The canonical statement of the method, its axioms, and its executable batteries lives in the reference cited below, continuously updated at the same location. This note is a pointer, not a substitute.

VIII.2 · Part Two, the Auto-Generated Specific Half

The specific half is one paragraph, auto-generated per paper from the paper's own content, and it carries the disclosure's value because it ties the method to the result the reader just finished. It opens with the words "This paper in particular." and fills five slots in fixed order.

Slot one, the three axes as instantiated: name what the formal-structural axis, the empirical-or-dynamical axis, and the registrational axis were for this specific paper. Slot two, where the lock held or collapsed, with the actual numbers: the determinant value and the rank at the decisive lock, and the determinant and rank at any decisive collapse, quoting the reproduced figures and the stated seed so a reader can re-run them. Slot three, the load-bearing gate or lemma, the single step the verdict most depends on. Slot four, the three-state verdict with its warrant tier, sealed or broken or open, and theorem, conditional, structural, or premise. Slot five, the Mosaic-Seal statement, whether a new theorem is claimed, ΔM.

Three disciplines bind the generation. The paragraph is drawn only from content already established in the body and introduces no new claim; it is a summary, not an argument. It never inflates a grade above what the body earned, the warrant-typing law of the parent discipline applying to the disclosure exactly as to the paper. And it names numbers only where the paper reproduced them, no fabricated trace, the same Fidelity Lock that governs the body.

Worked instance, a completed paper, showing the target shape. "This paper in particular. The three axes were the thermodynamic cost of storing a bit, the descent of description cost toward Kolmogorov complexity, and the independence of the memory carrier from the manifest dynamics. The lock is the rank test of Section 10, run at seed 20260705: a genuine substrate memory is an independent third axis and the determinant stands clear of zero at 0.615, rank three, while a non-energetic field carries no independent energetic dimension, collapses into the plane of the other two, and the determinant is machine zero, rank two. The load-bearing gate is the memory lemma, that cumulative memory requires an energetic substrate. The verdict is a split: the non-energetic non-local carrier is refuted at theorem grade conditional on the cost principle, the habit half of the thesis is accommodated at structural grade, and the empirical truth-value is left open, formal-alone. No new theorem is claimed; the paper reorganizes established results and applies them." This paragraph is spliced at the marked insertion point in the generic block of VIII.1.

The Hard Rule: Mandatory Codex-Integration

Before the Scribe is permitted to draft the specific paragraph (VIII.2), it must pass a Codex-Inclusion Audit. Any disclosure drafted without successfully anchoring its content in the following foundational layers is discarded as a violation of the Mosaic Seal.

The Scribe must execute these mandatory steps in order:

1. Topological Alignment Check: The Scribe must map the paper’s result back to the Achiral Ground ($Fix(\sigma) = \mathbb{R}$). It must explicitly verify that the result does not rely on a "Platonic Ghost" (e.g., an ungrounded CH-dependent assumption) unless that assumption is quarantined in the formal-only layer.

2. Category-Theoretic Logos Audit: The Scribe must identify the specific "Fertile Logos" operation (the categorical composition) used in the paper. It must define whether the result is a product of Higher-Category Laddering ($L2m$) or Grounded Composition ($L1m$).

3. RA-RAM-CH Nesting Verification: The Scribe must perform a cross-reference between the paper’s internal axioms and the Nested Root Axiom ($RA$) configuration. If the paper claims a theorem, the Scribe must verify that the Nested RA is identified as an external empirical boundary condition, not a syntactically derived internal theorem.

4. Triaxial Load-Test: The Scribe must confirm that the paper’s argument provides independent warrant rows for all three axes—Formal-Structural ($V_F$), Empirical-Dynamical ($V_E$), and Registrational ($V_{ER}$). If any axis is "hollow" (carrying no independent empirical data), the Scribe is mandated to downgrade the disclosure’s verdict from [⟀] (Sealed) to [?] (Open), regardless of the paper's claimed certainty.

Execution Enforcement

• Zero-Fabrication Policy: The Scribe is prohibited from using generic placeholders. If the paper's body does not contain the decisive determinant value or the load-bearing gate, the Scribe must refuse to generate the disclosure, signaling an Audit Failure.

• Pre-Generation Verification: The Scribe must "read" the Master Codex (specifically the Logos Fertility and Continuum/Category modules) into its active context window before drafting the disclosure. This ensures the disclosure reflects the Pre-Gödelian status of the architecture.

VIII.3 · The Master Reference, Injected into the References Section

The disclosure closes on a pointer, and the full citation is injected once as an entry of the paper's References section, Section 10 of the Plenous template, or as the sole reference when the paper carries none of its own. The DOI resolves to the current version, so the citation never goes stale as the master reference is updated at the same location. The entry, emitted verbatim:

Islam, M. TRISDUCTION: A Linguistically, Topologically, and Mathematically Sealed Verification Architecture. Triaxial Orthogonality, Twelve-Gate Closure, the Quaternionic Completion, the Root Axiom, and the Master Pre-Sealed Proposition Ledger. Zenodo, version 4, 19 June 2026. DOI 10.5281/zenodo.20757507. https://zenodo.org/records/20757507. Mirror: PhilArchive record ISLTTG, https://philpapers.org/rec/ISLTTG. Master reference, continuously updated at the same location.

The reference is injected at every weight except when the paper is Textbook or Scripture, so even a DISCLOSURE:OFF paper in a public edition still carries the citation.

VIII.4 · Rendering, the LINE Weight, and the Tier Defaults

At FULL weight the block of VIII.1 with the specific paragraph of VIII.2 renders as a demarcated appendix, the GOL identity in a boxed callout, on the same EB Garamond copper contract as the body. The box directive is :::box G1 in the PDF editions. For the Blog target the appendix renders identically except the identity box is flattened to a bold-titled paragraph, since the blog format linearizes every box; the mathematics stays Unicode throughout.

At LINE weight the entire block collapses to one sentence placed in the same appendix position: "This paper was developed under Trisduction, a verification and organizing discipline that adds the cited results no warrant; the method and its executable batteries are stated in full at the reference below." The independence clause survives at LINE weight because it is the sentence, and the master reference is still injected into References. This is the default for a blind Journal submission, where the full apparatus would invite a pattern-match dismissal rather than a reading on the merits.

At OFF weight nothing renders in the appendix, and the master reference is still injected into References unless the edition is Textbook or Scripture.

Tier defaults and tokens. Apex Pristine and Blog default to FULL. Journal defaults to LINE. Textbook and Scripture default to OFF and skip both the block and the reference injection, being framework-native. The tokens DISCLOSURE:FULL, DISCLOSURE:LINE, and DISCLOSURE:OFF override the default, an explicit token always winning, so a codex volume can carry the full disclosure on request and a house paper can suppress it.

VIII Quick Reference Card

Trigger: fires automatically on every Apex Pristine, Journal, and Blog deliverable; skipped for Textbook and Scripture. Output: a demarcated Author's Provenance and Method Disclosure appendix placed after the Conclusion and content appendices and before References, plus the master reference injected into References. Two halves: a fixed generic block (VIII.1, emitted verbatim, carrying the boxed GOL kernel-identity proof, the root axioms, and the four named procedures) and an auto-generated specific paragraph (VIII.2, five slots drawn from the paper: the three axes as instantiated, the lock or collapse with numbers and seed, the load-bearing gate, the three-state verdict with warrant tier, and ΔM). Weights: FULL for Apex and Blog, LINE for Journal as one sentence, OFF for Textbook and Scripture. Tokens: DISCLOSURE:FULL, DISCLOSURE:LINE, DISCLOSURE:OFF, explicit token wins. Mandatory at every weight: the independence clause, that the paper rests solely on its cited results and the method adds them no warrant. Build: the subroutine runs at Step 8 of the Execution Pipeline, before the format-lock build, appending the appendix to the authored Markdown and injecting the reference; the edition renderer of Module III, III·J, or VII then renders it under the active visual contract.