thiagobodruk/bible

What it is

Bible: XML and JSON is a data repository containing 141 Bible versions across 85 languages (including English, Portuguese, Chinese, Arabic, French, German, Japanese, and many others). The text is organized verse by verse in two formats — JSON and XML. The stated goal is to let people build Bible apps, verse search tools, and NLP/AI datasets without scraping or cleaning text themselves.

How the data was created

The files were generated using a Python-based crawler. The README warns this means files may contain minor encoding or syntax issues.

Data formats

XML — UTF-8 encoded, structured like:

<bible>
  <b id="abbrev" n="Book Name">
    <c n="1">
      <v n="1">Verse text</v>
    </c>
  </b>
</bible>

JSON — UTF-8 encoded, an array of book objects, where each book has:

• abbrev: a stable identifier (same across all languages)
• name: the book name in that version's language (e.g., "Genesis" in English, "創世記" in Japanese)
• chapters: an array of arrays — each inner array is a chapter, and each element is a verse string. Chapter/verse numbers come from array indices.

Example JSON shape:

[
  {
    "abbrev": "en_kjv",
    "name": "Genesis",
    "chapters": [
      ["Verse 1", "Verse 2"],
      ["Verse 1", "Verse 2"]
    ]
  }
]

Coverage details

• Files follow the pattern json/<code>.json and xml/<code>.xml (e.g., json/pt_nvi.json).
• json/index.json lists the full catalog in machine-readable form.
• json/book_names.json lists the book names for every version.
• Some versions are partial: e.g., several are New Testament only (27 books), some have only Psalms + New Testament (28 books), and a few are missing books like Esther and Daniel.
• The README documents known quirks: some editions have Malachi with 3 chapters instead of 4, one Turkish version lost its dotless "ı" diacritic, and two Oromo book names may be mis-transliterated.

Licensing

Code and data are under the MIT License. The Bible versions themselves remain the property of their respective owners — all rights reserved to them.

How to contribute

The README invites help with: reviewing code, enhancing structures, and adding new versions. Donations are accepted via PayPal.

Bottom line

It's a free, permissively licensed, ready-to-use collection of structured Bible text aimed at app developers and AI/LLM workflows (embeddings, RAG pipelines, fine-tuning, cross-lingual alignment), with the caveat that the data came from web scraping and may contain small imperfections.