biblical-guide.com/scripts/old/validate-bsb-md.js

#!/usr/bin/env node
const fs = require('fs')
const path = require('path')

const SRC = process.env.BSB_MD_PATH || path.join('bibles', 'bible-bsb.md')

function canon() {
  const OT = [
    ['Genesis', ['Genesis'], 50],
    ['Exodus', ['Exodus'], 40],
    ['Leviticus', ['Leviticus'], 27],
    ['Numbers', ['Numbers'], 36],
    ['Deuteronomy', ['Deuteronomy'], 34],
    ['Joshua', ['Joshua'], 24],
    ['Judges', ['Judges'], 21],
    ['Ruth', ['Ruth'], 4],
    ['1 Samuel', ['1 Samuel','1  Samuel'], 31],
    ['2 Samuel', ['2 Samuel','2  Samuel'], 24],
    ['1 Kings', ['1 Kings','1  Kings'], 22],
    ['2 Kings', ['2 Kings','2  Kings'], 25],
    ['1 Chronicles', ['1 Chronicles','1  Chronicles'], 29],
    ['2 Chronicles', ['2 Chronicles','2  Chronicles'], 36],
    ['Ezra', ['Ezra'], 10],
    ['Nehemiah', ['Nehemiah'], 13],
    ['Esther', ['Esther'], 10],
    ['Job', ['Job'], 42],
    ['Psalms', ['Psalms','Psalm'], 150],
    ['Proverbs', ['Proverbs'], 31],
    ['Ecclesiastes', ['Ecclesiastes'], 12],
    ['Song of Songs', ['Song of Songs','Song of Solomon'], 8],
    ['Isaiah', ['Isaiah'], 66],
    ['Jeremiah', ['Jeremiah'], 52],
    ['Lamentations', ['Lamentations'], 5],
    ['Ezekiel', ['Ezekiel'], 48],
    ['Daniel', ['Daniel'], 12],
    ['Hosea', ['Hosea'], 14],
    ['Joel', ['Joel'], 3],
    ['Amos', ['Amos'], 9],
    ['Obadiah', ['Obadiah'], 1],
    ['Jonah', ['Jonah'], 4],
    ['Micah', ['Micah'], 7],
    ['Nahum', ['Nahum'], 3],
    ['Habakkuk', ['Habakkuk'], 3],
    ['Zephaniah', ['Zephaniah'], 3],
    ['Haggai', ['Haggai'], 2],
    ['Zechariah', ['Zechariah'], 14],
    ['Malachi', ['Malachi'], 4]
  ]
  const NT = [
    ['Matthew', ['Matthew'], 28],
    ['Mark', ['Mark'], 16],
    ['Luke', ['Luke'], 24],
    ['John', ['John'], 21],
    ['Acts', ['Acts'], 28],
    ['Romans', ['Romans'], 16],
    ['1 Corinthians', ['1 Corinthians','1  Corinthians'], 16],
    ['2 Corinthians', ['2 Corinthians','2  Corinthians'], 13],
    ['Galatians', ['Galatians'], 6],
    ['Ephesians', ['Ephesians'], 6],
    ['Philippians', ['Philippians'], 4],
    ['Colossians', ['Colossians'], 4],
    ['1 Thessalonians', ['1 Thessalonians','1  Thessalonians'], 5],
    ['2 Thessalonians', ['2 Thessalonians','2  Thessalonians'], 3],
    ['1 Timothy', ['1 Timothy','1  Timothy'], 6],
    ['2 Timothy', ['2 Timothy','2  Timothy'], 4],
    ['Titus', ['Titus'], 3],
    ['Philemon', ['Philemon'], 1],
    ['Hebrews', ['Hebrews'], 13],
    ['James', ['James'], 5],
    ['1 Peter', ['1 Peter','1  Peter'], 5],
    ['2 Peter', ['2 Peter','2  Peter'], 3],
    ['1 John', ['1 John','1  John'], 5],
    ['2 John', ['2 John','2  John'], 1],
    ['3 John', ['3 John','3  John'], 1],
    ['Jude', ['Jude'], 1],
    ['Revelation', ['Revelation'], 22]
  ]
  return [
    ...OT.map(([n,v,c]) => ({ name:n, variants:v, expectedChapters:c, testament:'OT' })),
    ...NT.map(([n,v,c]) => ({ name:n, variants:v, expectedChapters:c, testament:'NT' })),
  ]
}

function main() {
  if (!fs.existsSync(SRC)) {
    console.error('Missing source file:', SRC)
    process.exit(1)
  }
  const md = fs.readFileSync(SRC, 'utf-8')
  const books = canon()

  const report = { file: SRC, totals: { versesTagged: 0 }, books: [] }

  for (const b of books) {
    const patterns = b.variants.map(v => v.replace(/\s+/g, '\\s+'))
    const names = patterns.join('|')
    const re = new RegExp(`(?:^|[\n\r\f\s\|\(])(?:${names})\\s+(\\d+):(\\d+)`, 'gi')
    const chapters = new Set()
    let m
    let verseCount = 0
    while ((m = re.exec(md)) !== null) {
      const nums = m.slice(1).filter(Boolean)
      const ch = parseInt(nums[0] || '0', 10)
      const vs = parseInt(nums[1] || '0', 10)
      if (Number.isFinite(ch) && ch > 0) chapters.add(ch)
      if (Number.isFinite(vs) && vs > 0) verseCount++
    }

    // Heuristic: some one-chapter books may lack inline verse references; accept header presence
    const oneChapterBooks = new Set(['Obadiah','Philemon','2 John','3 John','Jude'])
    if (chapters.size === 0 && oneChapterBooks.has(b.name)) {
      const headerRe = new RegExp(`[\f\n\r]\s*${b.variants.map(v=>v.replace(/\s+/g,'\\s+')).join('|')}\s*[\n\r]`, 'i')
      if (headerRe.test(md)) {
        chapters.add(1)
      }
    }
    report.totals.versesTagged += verseCount
    report.books.push({
      name: b.name,
      testament: b.testament,
      expectedChapters: b.expectedChapters,
      detectedChapters: Array.from(chapters).sort((a,b)=>a-b),
      detectedCount: chapters.size,
      coverage: b.expectedChapters > 0 ? +(100 * chapters.size / b.expectedChapters).toFixed(2) : null,
      verseMarkers: verseCount
    })
  }

  const missingBooks = report.books.filter(x => x.detectedCount === 0).map(x=>x.name)
  const partialBooks = report.books.filter(x => x.detectedCount > 0 && x.detectedCount < x.expectedChapters).map(x=>({name:x.name, det:x.detectedCount, exp:x.expectedChapters}))

  console.log('Validation summary for', SRC)
  console.log('Total verse markers found:', report.totals.versesTagged)
  console.log('Books missing markers:', missingBooks.length ? missingBooks.join(', ') : 'None')
  console.log('Books partially detected (chapters):', partialBooks.length ? JSON.stringify(partialBooks.slice(0,10)) : 'None')

  const outDir = path.join('data','en_bible','BSB_VALIDATION')
  fs.mkdirSync(outDir, { recursive: true })
  fs.writeFileSync(path.join(outDir,'report.json'), JSON.stringify(report, null, 2), 'utf-8')
  console.log('Wrote detailed report to', path.join(outDir,'report.json'))
}

main()