Zettle

0020-worldscripture-resource-class

4 min read

ADR-0020: WorldScripture resource class

Status: Accepted Date: 2026-05-21

Context

Second UC secondary work to be ingested (after BR per ADR-0016): World Scripture II — World Scripture and the Teachings of Sun Myung Moon (Chung Hwan Kwak, ed., Universal Peace Federation, 2014). 1178-page comparative anthology that builds on Andrew Wilson’s 1991 World Scripture: A Comparative Anthology of Sacred Texts by interleaving Sun Myung Moon’s teachings alongside cross-religion quotes, organized by theme.

Structurally distinct from prior classes:

  • Anthology, not monograph. Each chapter is a curated bundle of short quotes from many traditions (Bible, Bhagavad Gita, Qur’an, Dhammapada, Tao Te Ching, SMM speeches, …) clustered under sub-themes. Wilson/Kwak’s editorial framing — the choice of sub-themes, the juxtaposition of quotes across traditions, the introductory commentary — is itself the value being cited, not the quotes in isolation.
  • Two-axis citation problem. A reader of an atomic citing “Bhagavad Gita 3:35” found via WS-II has two valid sources: the original scripture (canonical, but not ingested in our vault) and WS-II (proximate, ingested). Per ADR-0003 a citation must point to a defined class; we have no Bhagavad-Gita / Qur’an / Dhammapada classes today and adding them all up front is impractical.
  • Hierarchy. 4 Parts, 22 numbered Chapters, ~200 sub-themes. Parts are top-level navigational dividers; chapters are the natural unit of thematic coherence; sub-themes are the natural citation anchor.

Decision

  • Class code: WorldScripture
  • Path: /resources/WorldScripture/
  • File granularity: one file per chapter (per ADR-0004), nested in Part{N}/ subfolder (mirrors CSG’s Book{NN}/ pattern from ADR-0018). 22 chapter files + front matter (Preface, Invocation) and back matter (Notes, List of Sources) at root.
  • Filenames: ws-{part#}-{chapter#}-{slug}.md, e.g. ws-1-3-purpose-of-human-life.md. Single-digit part/chapter numbers (parts ≤4, chapters ≤22).
  • Internal hierarchy inside each chapter file: sub-theme = H1. Marker’s heading detection picks these up from the printed book’s visual styling; stage-2 cleanup normalizes.
  • Citation form: [[WorldScripture/Part{N}/ws-{part#}-{chapter#}-{slug}#sub-theme-slug]]
    • Example: [[WorldScripture/Part2/ws-2-6-evil-sin-and-fall#human-pride|human-pride]]
  • Hybrid citation policy for quotes inside WS-II. WS-II is the citation target for the anthology’s editorial framing (sub-theme intros, chapter intros, juxtaposition). For individual quotes whose original tradition has its own ingested resource class, atomics cite the original (e.g. [[Bible/Romans#121|1:21]]). For quotes whose original class is not yet ingested — the common case at time of writing — cite WS-II and migrate the citation later when the original lands. The pre-commit hook continues to enforce grammar; placeholder dead links are not allowed.
  • Class category: UC secondary literature, same bucket as BR. Future original-tradition classes (Bhagavad-Gita, Qur’an, Dhammapada, etc.) get their own ADRs when first needed.

Alternatives considered

  • Cite originals only (atomics never point at WS-II). Rejected — most originals aren’t ingested; would force ingesting ~15 world-scripture corpora before any WS-II atomic could be written.
  • Cite WS-II only (treat originals as unrecoverable). Rejected — collapses Wilson/Kwak’s curation with the underlying quotes; loses the ability to ever distinguish “what Wilson chose to highlight” from “what the Bhagavad Gita actually says”. Hybrid preserves the option.
  • Cite both, always, in the same atomic. Rejected — doubles frontmatter clutter and creates a maintenance burden (every original-tradition ingest triggers a migration sweep across all WS-II-derived atomics anyway).
  • Flat layout (no Part{N}/ subfolders). Rejected — 22 chapters is enough that grouping aids navigation, and the Parts are semantically real divisions in Wilson’s argument (God/Creation → Sin/Salvation → Path of Life → Family/Society) rather than arbitrary slabs.
  • Sub-theme as file granularity (one file per sub-theme, ~200 files). Rejected per ADR-0004 — sub-themes are too small to be coherent resource units; the chapter is the natural argumentative beat.

Consequences

  • (+) WS-II becomes ingestible. Stage-1 extraction via scripts/ingest_pdf.py --split-level 2 after bookmarks are added per the ADR-0019 sub-pipeline.
  • (+) Hybrid citation policy is explicit and reviewable. New atomics either cite an ingested original or cite WS-II with the understanding that migration is expected when the original arrives.
  • (+) Part-subfolder layout means URL paths on the published site are /resources/WorldScripture/Part2/ws-2-6-evil-sin-and-fall/ — readable, mirrors the print structure.
  • (−) WS-II PDF has no publisher bookmarks (1178 pages, zero outline entries) and multi-column TOC with non-monotonic reading order in pymupdf text extraction. extract_toc.py’s dotted/spaced regexes don’t match its ❄-leader sub-themes or tab-newline chapter heads. Per the script-vs-AI split (script does deterministic work, AI handles judgment), bookmarks.tsv for WS-II is hand-written by AI from the TOC text dump rather than auto-extracted. Future similar UPF/IRF anthologies may justify extending extract_toc.py.
  • (−) Original-tradition resource classes are now an explicit backlog item. Each cite-via-WS-II atomic is a future migration. Acceptable: migrations are mechanical (filename rewrite + frontmatter sources: update) and can be scripted when the volume warrants.

Graph View