Divine Discourses — qurandiscourses

A Qur'an study platform directed at one goal: direct, personal, evidence-based engagement with the Qur'an, using a coherence-based reading method — read each surah as one discourse, not a string of isolated quotations. Maintained in memory of Dr. Irfan Ahmad Khan (1931–2018), whose teaching the method follows; see About for the fuller story.

It exists so that anyone — an independent student, a teacher, a skeptical reader — can test the coherence of a discourse for themselves instead of taking assertions on trust. What the project needs is careful work, not money: corrections (via issues — every one is logged publicly), help transcribing Khan's remaining published outlines into exercises, scholarly review of the method pages against their sources, and — for the 84 surahs Khan never published an outline for — your own structural readings; see CONTRIBUTING.md.

Live site: https://divinediscourses.org

What makes it different

1. Research claims expose their evidence status. A compact ●/○/~ marker opens the cited source, while canonical records separately state whether the source was checked, a computation was reproduced, the result is method-dependent, and which limitations remain. The machine-readable ledger is data/claims.json; its integrity gate is node scripts/check-claims.mjs. The reader-facing method is documented on Validation.
2. The data ships with the site. Word-by-word morphology for all 77,429 tokens (Leeds Quranic Arabic Corpus v0.4, GPL — see NOTICE.md) is bundled as JSON, so root counts, theme passages, and exercises are recomputable by anyone, offline.
3. No commentary is generated. The site orders evidence (roots, frequencies, structure, cross-references) and points to published scholarship; the reader does the engagement. Where curation is unavoidable — theme groupings and titles, short working glosses — it is labeled as editorial (Nuanced), never presented as computed data.

Architecture

Static site, no framework, no build step for pages, no external runtime dependencies (fonts self-hosted; the maintainer's own cookieless first-party analytics, reached through this site's /_ah/ paths, is the only analytics and counts usage; the only other network calls are the optional alquran.cloud verse/translation API and recitation audio).

*.html               33 pages, vanilla HTML + inline page logic
assets/              shared JS (app, nav, ask, cite-badge, glossary,
                     chart), CSS (3 palettes x light/dark), fonts
js/                  viz.js, the shared zero-dependency SVG chart
                     helpers used by the analytics pages
data/                bundled datasets: morphology/ (per-surah tokens),
                     roots-summary, root-analytics/, cooccurrence/,
                     association/ (PMI/LLR/Dice pair statistics),
                     network/ (precomputed graph layouts), centrality/,
                     coverage/, exports/ (public CSV/JSON downloads),
                     gloss/, rhyme/,
                     themes, surah profiles, sources registry
scripts/             deterministic, zero-dependency Node generators
                     that produce data/ artifacts from the morphology
                     (one Python helper, build-roots-index.py)
netlify.toml         hosting config: security headers (CSP etc.) and
                     the .html -> clean-URL redirects

Working on the site

node scripts/serve.mjs             # serve locally, open localhost:8000
                                   # (resolves /read the way Netlify does;
                                   #  python3 -m http.server cannot)
node scripts/build-themes.mjs      # regenerate data/themes.json
node scripts/build-cooccurrence.mjs
node scripts/build-root-analytics.mjs
node scripts/compute-association-stats.mjs   # data/association/
node scripts/compute-network-layout.mjs      # data/network/ (needs association/)
node scripts/compute-centrality.mjs          # data/centrality/ (needs association/)
node scripts/compute-coverage.mjs            # data/coverage/
node scripts/build-exports.mjs               # data/exports/ (needs association/)

Every generator is deterministic: run it twice and git diff must be empty. The full pipeline, in dependency order, is documented in docs/maintainer-guide.md.

Read docs/maintainer-guide.md before making content changes — it covers the editorial rules (what may be claimed, how badges are assigned, Chicago citation format), every build script, and the step-by-step recipes for common tasks (adding a source, an exercise, a theme). A ready-to-record walkthrough script for a training video is in docs/video-walkthrough-script.md. The prioritized cross-disciplinary roadmap for accessibility, scholarship, governance, privacy, learning, reach, and operational resilience is in the public-engagement and excellence review.

Licensing

• Site code (HTML/CSS/JS/scripts): MIT
• data/morphology/ and derivatives (including data/association/, data/network/, data/centrality/, data/coverage/, data/exports/): GPL, per the Leeds corpus — see NOTICE.md
• data/quran-text/: Tanzil Uthmani verse text, a verbatim copy under Tanzil's terms (CC BY 3.0, no changes allowed, notice retained)
• Translation editions: fetched at runtime, never bundled; each keeps its own copyright — see NOTICE.md
• data/gloss/ and data/khan-interpretations.json: quoted from Khan (2011), © Association for Qur'anic Understanding — see NOTICE.md
• assets/fonts/: SIL Open Font License 1.1, license texts bundled beside the binaries
• Full source list with citations: Sources

Corrections, citations, and source contributions are welcome via issues and pull requests.