// A NUL byte in a source file is invisible in an editor and changes what git thinks the file IS:
// git classifies any file containing one as binary, and from then on a diff of it carries no body.
// That is not cosmetic. A forge that ships pull requests as patches — tangled builds every round
// with `git format-patch` — then emits `Binary files a/x and b/x differ` for a `.ts` file, and the
// round cannot be applied by the other side. `packages/ui/src/lib/store.ts` reached `main` with
// three of them, written as literal bytes where `\u0000` was meant, and the first sign of it was a
// pull request reporting a conflict against a branch that merged as a fast-forward.
//
// So: every tracked file that is not a known binary asset must be free of NUL bytes. Escapes are
// unaffected — `"\u0000"` is six ASCII characters and produces the identical string at runtime.

import { spawnSync } from 'node:child_process'
import { readFileSync } from 'node:fs'

/** Extensions whose CONTENT is legitimately binary. Everything else must be text. */
const BINARY = new Set([
  'png', 'jpg', 'jpeg', 'gif', 'webp', 'avif', 'ico', 'icns', 'bmp',
  'woff', 'woff2', 'ttf', 'otf', 'eot',
  'wasm', 'gz', 'zip', 'tar', 'br', 'zst', 'pdf', 'mp4', 'webm', 'mp3', 'wav', 'ogg',
])

const listed = spawnSync('git', ['ls-files', '-z'], { encoding: 'buffer', maxBuffer: 1 << 28 })
if (listed.status !== 0) {
  console.error(`check-no-nul: git ls-files failed: ${listed.stderr?.toString() ?? ''}`)
  process.exit(2)
}

const paths = listed.stdout
  .toString('utf8')
  .split('\0')
  .filter((path) => path.length > 0)

const offenders = []
for (const path of paths) {
  const dot = path.lastIndexOf('.')
  const extension = dot === -1 ? '' : path.slice(dot + 1).toLowerCase()
  if (BINARY.has(extension)) continue

  let bytes
  try {
    bytes = readFileSync(path)
  } catch {
    continue // a submodule, a broken symlink, or a path removed since ls-files ran
  }

  const at = bytes.indexOf(0)
  if (at === -1) continue

  const line = bytes.subarray(0, at).toString('utf8').split('\n').length
  const total = bytes.reduce((count, byte) => (byte === 0 ? count + 1 : count), 0)
  offenders.push({ path, line, total })
}

if (offenders.length > 0) {
  console.error('check-no-nul: NUL bytes in files git will therefore treat as BINARY:\n')
  for (const { path, line, total } of offenders) {
    console.error(`  ${path}:${line}  (${total} NUL byte${total === 1 ? '' : 's'})`)
  }
  console.error('\nWrite the escape instead of the raw byte — `\\u0000` in a JS/TS string or template')
  console.error('literal produces the identical value and keeps the file text.')
  console.error('If the file really is binary, add its extension to BINARY in scripts/check-no-nul.mjs.')
  process.exit(1)
}

console.log(`check-no-nul: ${paths.length} tracked files, no NUL bytes outside binary assets`)