Language Data for the Bible Passage Reference Parser

The code in this repository is mostly AI-generated, though the source data is not.

This repository contains the YAML to Javascript build pipeline for Bible Passage Reference Parser language data.

Structure

• data/_defaults.yaml: shared defaults for variables/options.
• data/*.yaml: per-language data (ISO 639-3 codes like eng, zho).
• translation_systems/*.yaml: versification systems and translation aliases.
• src/: TypeScript source for building language output.
• bin/compile.sh: bundles TypeScript CLI tools in src/ to bin/*.js.
• bin/build_spec.js: builds localized-book Jasmine specs from book_names/all/.
• bin/build_all_langs.js: builds all languages in parallel and runs specs for each language.
• lang/: generated output files (optional; regenerate as needed).
• book_names/all/: generated book-name lists used to build tests.
• book_names/preferred/: preferred display names (default + optional translation overrides).
• test/: generated localized-book specs.

Quick start

npm install
bin/compile.sh
node bin/build_lang.js eng

Build a cross-language parser:

node bin/build_lang.js --cross eng spa --out eng_spa
node bin/build_spec.js eng_spa

Load a language module programmatically:

import load_language_code from "bible-passage-reference-parser-languages";
const lang = await load_language_code("eng");
// Reserved Windows code example:
const con = await load_language_code("con");

Generate specs:

node bin/build_spec.js        # all languages
node bin/build_spec.js eng    # single language

Build all languages:

Cross-language options:
- `--cross`: enable cross-language build mode.
- `--out <code>`: output language code (must not be 3 characters).
- `--merge-mode append|smart`: book merge mode (default `append`).
node bin/build_all_langs.js          # parallel build + specs + tests
node bin/build_all_langs.js -j 4     # set worker count
node bin/build_all_langs.js --test-only   # skip lang rebuild, build specs + run tests

Output format

The build outputs three JS classes:

• bcv_regexps
• bcv_translations
• bcv_grammar_options_default

The output matches the expected format in Bible-Passage-Reference-Parser.

Creating a new language YAML

1. Pick an existing language as a starting point (for example data/eng.yaml) and copy it to a new ISO 639-3 code, like data/isl.yaml.
2. Update the language file contents:
   • variables: text tokens and patterns used by the parser (titles, next, ff, etc.).
   • options: language-specific parsing options; any missing options fall back to data/_defaults.yaml.
   • books: list of Bible book names/abbreviations and regex-related data for the language.
   • ordinals and translations are optional; include them if needed for the language.
3. Build and verify:

bin/compile.sh
node bin/build_lang.js isl

Notes for new languages:

• The first language file passed to build_lang sets the primary variables and options; additional languages are only used to merge books.
• data/_defaults.yaml provides required defaults for variables and options, so you only need to override what differs.

YAML structure (data/*.yaml)

variables

Used to build the core grammar and separator patterns. Values can be:

• simple strings: - "cap."
• objects for fine control:
  • text: string value
  • regexp: raw regex (no escaping)
  • regexp_after: appended raw regex after the text/regexp
  • normalize: none to skip combining-character normalization for that item

Example:

variables:
  and:
    - text: a
      regexp_after: (?!\p{L})
    - vedi
  to:
    - "-"
    - "a"

options

Common options (see data/_defaults.yaml for full list):

• normalize: combining_characters (default) or none
• trailing_dots_in_variables: optional or as_is
• expand_characters: array of { character, expand: [ ... ] } to allow alternates anywhere that character appears in book names or variables. Example:

  expand_characters:
    - character: "'"
      expand: ["'", "’"]

• replace_characters_with: { regexp, replacement } (default converts spaces to \s*)
• before_book_allowed_characters, after_book_allowed_characters: regex character classes used to enforce valid boundaries before/after a matched book name.
• before_every_book, after_every_book: regex patterns inserted immediately before/after every book match. Use these to add language-specific required prefixes/suffixes around all books (rare). These are applied in addition to the before/after allowed character boundaries.
• join_before, join_after: default join strings used when expanding before/after book patterns (for example the default space between an ordinal and the book name). Override to control whether the joiner is a space, empty string, punctuation, etc.

books

Each entry declares OSIS code(s) and the localized texts. Forms:

• osis: "Gen" or osis: ["Jonah","Job"]
• osis objects with before, after, join for numbered books:

  - osis:
      - osis: 1Sam
        before: *first
      - osis: 2Sam
        before: *second
    texts:
      - Samuel

• texts can be strings or objects with text and optional normalize: none, which prevents diacritics and spacing from changing.

ordinals

Defines ordinal suffixes and optional Psalm handling:

ordinals:
  - after: ["st"]
    numbers: [1, 21, 31]
  - after: ["nd"]
    numbers: [2, 22, 32]
  - between:
      regexp: \s*
    texts: ["Psalm"]

book_names/all

node bin/build_lang.js <lang> writes book_names/all/<lang>.yaml, which is a normalized list of book texts used by bin/build_spec.js. Names output collapses whitespace to single spaces and normalizes to NFC (combining-character variants are unified), but does not add extra variants.

book_names/preferred

book_names/preferred/<lang>.yaml documents preferred book names for display and UI. Each file has:

• default: preferred names by OSIS code.
• translations (optional): translation-specific overrides (currently only in eng.yaml).

Example (book_names/preferred/eng.yaml):

default:
  Gen:
    long: Genesis
    short: Gen
    shorter: Ge
  Ps:
    long: Psalms
    long_single: Psalm
    short: Ps
    shorter: Ps
translations:
  niv:
    Ps:
      short_plural: Pss
    1Sam:
      shorter: 1Sa

Keys used in preferred names can include:

• long: full name
• long_single: singular form (e.g., Psalm vs Psalms)
• short: common short form
• shorter: shortest form
• short_plural: translation-specific plural short form

tests

Optional per-language Jasmine tests can be added to data/<lang>.yaml (see pol.yaml for an example):

tests:
  - text: "Rdz 1:1"
    osis: "Gen.1.1"
  - it: "should handle odd spacing"
    text: "Rdz   1:1"
    osis: "Gen.1.1"

bin/build_spec.js will emit these into test/<lang>.spec.js. Entries with it get their own it(<label>) block; entries without it are grouped under it("should handle custom tests").

Notes

• lang/ can be regenerated at any time; it is not a source of truth.
• Language codes are ISO 639-3 (e.g., eng, zho).
• Windows-reserved 3-letter basenames (con, prn, aux, nul) are stored on disk with a trailing underscore (for example, logical code con maps to con_.yaml / con_.js / con_.spec.js).

ISO 639-2 to ISO 639-3 mapping

Mappings used by tooling and for compatibility with the older abbreviations used in Bible Passage Reference Parser:

ISO 639-2	ISO 639-3	English name
ar	ara	Arabic
bg	bul	Bulgarian
cs	ces	Czech
cy	cym	Welsh
da	dan	Danish
de	deu	German
el	grc	Greek
en	eng	English
es	spa	Spanish
fa	fas	Persian
fi	fin	Finnish
fr	fra	French
he	heb	Hebrew
hi	hin	Hindi
hr	hrv	Croatian
ht	hat	Haitian Creole
hu	hun	Hungarian
id	ind	Indonesian
is	isl	Icelandic
it	ita	Italian
ja	jpn	Japanese
jv	jav	Javanese
kn	kan	Kannada
ko	kor	Korean
la	lat	Latin
lg	lug	Ganda
mk	mkd	Macedonian
mr	mar	Marathi
ne	nep	Nepali
nl	nld	Dutch
no	nor	Norwegian
ny	nya	Nyanja
or	ori	Odia
pa	pan	Punjabi
pl	pol	Polish
pt	por	Portuguese
ro	ron	Romanian
ru	rus	Russian
sk	slk	Slovak
sl	slv	Slovenian
so	som	Somali
sq	sqi	Albanian
sr	srp	Serbian
sv	swe	Swedish
sw	swa	Swahili
ta	tam	Tamil
te	tel	Telugu
th	tha	Thai
tl	tgl	Tagalog
tr	tur	Turkish
uk	ukr	Ukrainian
ur	urd	Urdu
vi	vie	Vietnamese
yo	yor	Yoruba
zh	zho	Chinese

Todo

1. Move spec building process into the main build_lang script so that it's a one-step process.
2. Add more translation-specific versification.
3. Improve English translation representation.

Changelog

February 7, 2026. Add a loader function to handle file renaming for the "con" language so that this repo works on Windows.

January 31, 2026. Rework folder naming and include preferred book names from source data. Add an additional 2,100 languages from YouVersion data.

January 29, 2026. First release.