المصحف المحمدي — Maghreb Warsh Quran Data

0% tolerance verse-level data for the Moroccan Muhammadi mushaf (المصحف المحمدي), published by the Ministry of Habous and Islamic Affairs of the Kingdom of Morocco.

Built for Maghreb users (Maroc / Algérie / Tunisie) who follow the Warsh recitation and the Madani verse-counting tradition.

• Riwaya (روایة): Warsh 'an Nafi` (ورش عن نافع)
• Numbering: Madani / Maghrebi (6,214 verses)
• Pagination: 638 pages, matching the printed Muhammadi pagination on mushaf.ma
• Eighth index: 480 ثُمْن boundaries (60 hizbs × 8 eighths), with inclusive start and end ranges

---

مقدمة بالعربية

هذه البيانات مستخرجة من المصحف المحمدي المطبوع بالمغرب، بأرقام السور والآيات والأحزاب والأثمان كما هي معتمدة رسمياً (الرواية: ورش عن نافع، الترقيم: المدني الأخير، 6214 آية).

تم التحقق من كل ثُمْن (۞) مقابل الصورة الأصلية للصفحة المطبوعة، ليوافق الموضع الدقيق في المصحف المحمدي.

• النص المعتمد: data/quran_muhammadi.json
• حدود الأثمان (480 ثُمْن): data/eighths.json
• فهارس مجمّعة: data/surahs.json, data/hizbs.json, data/juzs.json
• صور الصفحات الأصلية: raw/muhammadi-pages/ (638 PNG)

---

What's in data/

File	Entries	Purpose
quran_muhammadi.json	6,214	Full verses with text, text_simple, page_mushafma, hizb, juz, eighth_id, word_count, is_sajda (11 — Maliki tradition)
eighths.json	480	Every ۞ boundary with inclusive start / end, verses_covered, word_count, name_ar
surahs.json	114	Name (ar/en), revelation (makki/madani), verse count, first/last eighth
hizbs.json	60	Hizb → juz mapping, start/end, pages, word count
juzs.json	30	Juz → hizb range, pages, word count
audio_sources.json	19 reciters	Warsh 'an Nafi` recitations — per-surah and per-verse streaming URLs
VERSION.json	—	Dataset version + totals

See docs/SCHEMA.md for full field documentation.

---

Quick start

JavaScript / TypeScript

// Bundle or fetch from GitHub raw CDN (jsDelivr recommended):
const BASE = "https://cdn.jsdelivr.net/gh/ELAHMADI/ma-mushaf-muhammadi-data@v1/data";

const [verses, eighths, surahs] = await Promise.all([
  fetch(`${BASE}/quran_muhammadi.json`).then(r => r.json()),
  fetch(`${BASE}/eighths.json`).then(r => r.json()),
  fetch(`${BASE}/surahs.json`).then(r => r.json()),
]);

// Find the eighth containing Al-Fatiha:5
const v = verses.find(x => x.sura === 1 && x.aya === 5);
const e = eighths.find(x => x.eighth_id === v.eighth_id);
console.log(`${e.name_ar} — hizb ${e.hizb}`);

Python

import json, urllib.request
BASE = "https://cdn.jsdelivr.net/gh/ELAHMADI/ma-mushaf-muhammadi-data@v1/data"
with urllib.request.urlopen(f"{BASE}/quran_muhammadi.json") as r:
    verses = json.load(r)

React Native (bundle at build time)

# In your project root
curl -L "https://github.com/ELAHMADI/ma-mushaf-muhammadi-data/releases/download/v1.0.0/quran_muhammadi.json" \
  -o ./assets/data/quran_muhammadi.json

---

Audio (Warsh reciters)

data/audio_sources.json lists **16 Warsh 'an Nafi reciters** with per-surah streams (001.mp3…114.mp3`), prioritizing Maghreb voices:

Reciter	Country	URL pattern
عمر القزابري — Omar Al-Kazabri	🇲🇦 Morocco	server9.mp3quran.net/omar_warsh/{001..114}.mp3
ياسين الجزائري — Yassine Al-Jazairi	🇩🇿 Algeria	server11.mp3quran.net/qari/{001..114}.mp3
رشيد بلعالية — Rachid Belaalya	🇲🇦 Morocco	server6.mp3quran.net/bl3/Rewayat-Warsh-A-n-Nafi/…
عبد المجيب بنكيران — Benkirane	🇲🇦 Morocco	server16.mp3quran.net/A-Benkirane/…
…and 12 more		see audio_sources.json

All endpoints verified 2026-04-21. Per-surah Warsh audio is fully aligned with the Moroccan numbering used by this dataset.

Reciters used by the Nur al-Hifz app

Three reciters carry ثمن-level timings in data/timings_thumn/<key>/{NNN}.json. Their MP3s are mirrored on this repository's GitHub Releases, so timings always point to a stable, byte-identical file:

Reciter	Key	Mirror (GitHub Releases)	Format
العيون الكوشي — El-Ayoun El-Kouchi	kouchi	audio-kouchi-v1/{NNN}.mp3	128 kbps CBR
عبد المجيب بنكيران — Abdul-Majeed Benkirane	benkirane	audio-benkirane-v1/{NNN}.mp3	128 kbps CBR mono, re-encoded from a 320 kbps source
ياسين الجزائري — Yassine Al-Jazairi	jazairi	audio-jazairi-v1/{NNN}.mp3	CBR, mixed: MPEG-1 96 kbps mono (62), MPEG-2 96 kbps mono 22–24 kHz (49), MPEG-1 128/192 kbps stereo (3)

Coverage (surahs done, playable ثمن) is generated in data/audio_status.json on every push by .github/workflows/timings.yml.

---

How it was built

1. Ground truth: 638 page images downloaded from mushaf.ma → raw/muhammadi-pages/.
2. Text candidate A: Al-Lawh Almahfoudh APK (extracted coran.txt → raw/lawh_verses.json).
3. Text candidate B: Zizwar/mushaf-mauri (MIT) → raw/mauri_verses.json.
4. Majority-merge (scripts/majority_merge.py) with targeted rules:
   • Both agree → use Mauri (richer vocalization)
   • Bismillah bundled into aya 1 → strip (112 cases)
   • Verse-boundary disagreements are always Lawh truncation at ۞ markers → use Mauri
5. Eighth boundaries: extracted 477 ۞ markers from Mauri, added 3 implicit boundaries (Fatiha 1:1, 75:1 Al-Qiyamah, 76:1 Al-Insan) to reach the canonical 480.
6. Visual validation: each of the 480 eighths verified against its mushaf.ma page image via data/eighths_viewer.html.

Full pipeline: scripts/*.py. Full decision log: reports/divergences.md, reports/eighths_report.md.

---

Schema highlights

Verse (quran_muhammadi.json)

{
  "gid": 1,
  "sura": 1,
  "aya": 1,
  "text": "بِسْمِ ٱللَّهِ ٱلرَّحْمَٰنِ ٱلرَّحِيمِ",
  "text_simple": "بسم الله الرحمن الرحيم",
  "page_mushafma": 3,
  "page_mauri": 2,
  "hizb": 1,
  "juz": 1,
  "eighth_id": 1,
  "word_count": 4,
  "char_count": 39,
  "is_sajda": false,
  "source_decision": "..."
}

Eighth (eighths.json)

{
  "eighth_id": 460,
  "hizb": 58,
  "pos_in_hizb": 4,
  "name_ar": "النصف",
  "juz": 29,
  "is_surah_start": false,
  "start": {"sura": 74, "aya": 31, "char_offset": 0,  "partial_verse": false, "mushafma_page": 606},
  "end":   {"sura": 74, "aya": 55, "char_offset": 87, "partial_verse": false, "mushafma_page": 607, "inclusive": true},
  "verses_covered": [{"sura": 74, "aya": 31}, …],
  "verse_count": 25,
  "word_count": 180
}

Full field docs: docs/SCHEMA.md.

---

Sajda verses

is_sajda follows the Maliki / Maghreb tradition: 11 sajda verses. See docs/SAJDA.md for the full list.

---

License

• Data (JSON, images, reports): CC BY 4.0 — free to use, share, adapt, with attribution.
• Source code (scripts/*.py, HTML viewers): MIT.
• Page images (raw/muhammadi-pages/): © Ministry of Habous, Morocco — distributed here under fair-use study provisions; contact mushaf.ma for commercial licensing.

Citation

@dataset{elahmadi_muhammadi_2026,
  author    = {El-Ahmadi, Noureddine},
  title     = {ma-mushaf-muhammadi-data: Verse-level data for the Moroccan Muhammadi Quran},
  year      = 2026,
  url       = {https://github.com/ELAHMADI/ma-mushaf-muhammadi-data},
  version   = {1.0.0},
  license   = {CC-BY-4.0}
}

Contributing

Spot a divergence between the data and the printed mushaf? Open an issue with:

• Sura:verse reference
• The mushaf.ma page number
• A screenshot or description of the mismatch

Pull requests welcome for metadata improvements (translations, transliteration, tajweed rules).