biblical-guide.com/scripts/old/parse-bsb-md-samples.ts

#!/usr/bin/env tsx
import fs from 'fs'
import path from 'path'

/*
  Quick sample extractor from bibles/bible-bsb.md to our OT/NT JSON format.
  - Looks for Genesis 3:20–24 markers and builds a small sample JSON.
  - Output directory: data/en_bible/BSB_SAMPLES
  - Intended for demo/import testing without hitting API limits.
*/

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

function ensureDir(p: string) { fs.mkdirSync(p, { recursive: true }) }
function writeJson(file: string, obj: any) { ensureDir(path.dirname(file)); fs.writeFileSync(file, JSON.stringify(obj, null, 2), 'utf-8') }

function extractGenesis3Samples(md: string): { chapterNum: number; verses: { verseNum: number; text: string }[] } {
  // Find all markers like "Genesis 3:20" and capture their file offsets
  const regex = /Genesis\s+3:(\d+)/g
  const indices: { verse: number; index: number }[] = []
  for (const m of md.matchAll(regex) as any) {
    const verse = parseInt(m[1], 10)
    indices.push({ verse, index: m.index })
  }

  // We'll only keep verses 20..24 as a small sample
  const keep = new Set([20, 21, 22, 23, 24])
  const kept = indices.filter(x => keep.has(x.verse)).sort((a,b) => a.verse - b.verse)

  const verses: { verseNum: number; text: string }[] = []
  for (let i = 0; i < kept.length; i++) {
    const cur = kept[i]
    const next = kept[i+1]
    const start = cur.index!
    const end = next ? next.index! : Math.min(md.length, start + 2000) // cap window
    let chunk = md.slice(start, end)
    // Remove the marker itself and nearby page headers/footers and footnote junk
    chunk = chunk.replace(/Genesis\s+3:\d+.*\n?/,'')
    chunk = chunk.replace(/\f\d+\s*\|\s*Genesis\s*3:\d+.*\n?/g,'')
    chunk = chunk.replace(/[\u000c\r]+/g,'\n') // form feed cleanup
    chunk = chunk.replace(/\s+/g,' ').trim()
    // Try to cut off before the next verse number embedded as an isolated number
    const stop = chunk.search(/\s(?:2[1-9]|3\d|\d{1,2})\s/) // heuristic
    const clean = (stop > 40 ? chunk.slice(0, stop) : chunk).trim()
    if (clean.length > 0) verses.push({ verseNum: cur.verse, text: clean })
  }

  // Fallback if nothing captured
  if (verses.length === 0) {
    verses.push({ verseNum: 20, text: 'And Adam named his wife Eve, because she would be the mother of all the living.' })
  }

  return { chapterNum: 3, verses }
}

function main() {
  if (!fs.existsSync(SRC)) {
    console.error('Missing source file:', SRC)
    process.exit(1)
  }
  const md = fs.readFileSync(SRC, 'utf-8')
  const gen3 = extractGenesis3Samples(md)

  const ot = {
    testament: 'Old Testament',
    books: [
      {
        name: 'Genesis',
        chapters: [gen3]
      }
    ]
  }
  // Minimal NT placeholder for structure completeness
  const nt = { testament: 'New Testament', books: [] as any[] }

  writeJson(path.join(OUT, 'old_testament.json'), ot)
  writeJson(path.join(OUT, 'new_testament.json'), nt)
  console.log('Wrote samples to', OUT)
}

main()