Batch rendering rejects JSON files saved with a byte order mark #6604

Describe the bug

render --batch rejects otherwise valid batch data saved as UTF-8 with a byte order mark (BOM), or as BOM-marked UTF-16LE. This blocks batch files written by Windows editors and Windows PowerShell 5.1. The CLI already has a decoder for these encodings in jsonFlag.ts, but the batch reader bypasses it.

Minimal reproduction

From a repository checkout after the normal bun install and bun run build, create a one-second project and the two batch files:

node --input-type=module <<'JS'
import { mkdirSync, writeFileSync, copyFileSync } from 'node:fs';
mkdirSync('bom-batch');
writeFileSync('bom-batch/index.html', `<!doctype html>
<html lang="en" data-composition-variables='[{"id":"title","type":"string","label":"Title","default":"Default"}]'>
<head><meta charset="UTF-8"><title>Batch encoding reproduction</title>
<script src="gsap.min.js"></script>
<style>body{margin:0;background:#111;color:#fff;font-family:sans-serif}#root{width:100%;height:100%;display:grid;place-items:center}h1{font-size:36px}</style></head>
<body><div id="root" data-composition-id="main" data-width="320" data-height="180" data-duration="1">
<h1 class="clip" data-start="0" data-duration="1" data-track-index="0"><span data-var-text="title">Default</span></h1>
</div><script>window.__timelines["main"]=gsap.timeline({paused:true});</script></body></html>
`);
copyFileSync('packages/player/node_modules/gsap/dist/gsap.min.js', 'bom-batch/gsap.min.js');
for (const encoding of ['utf8', 'utf16le']) {
  writeFileSync(`bom-batch/rows-${encoding}.json`, Buffer.from('\uFEFF[{"title":"Café"}]', encoding));
}
JS

Steps to reproduce

cd bom-batch
node --import ../node_modules/tsx/dist/loader.mjs ../packages/cli/src/cli.ts render \
  --batch rows-utf8.json --output 'renders/{index}.mp4' \
  --fps 1 --workers 1 --json --quiet --strict-variables

Repeat with --batch rows-utf16le.json. The same problem occurs through the built packages/cli/bin/hyperframes.mjs entry point.

Expected behavior

Decode the file's encoding marker, parse the row and render one video containing Café, as with the same JSON in unmarked UTF-8.

Actual behavior

Both commands exit 1 before rendering with Invalid JSON in --batch. UTF-8 reports Unexpected token at U+FEFF; UTF-16LE reports Unexpected token at U+FFFD after the bytes are decoded as UTF-8. No batch output is produced.

Environment

• Baseline: 7e92aec502c6a56488807c3cf9edb8371c854abd (CLI 0.8.209).
• Linux arm64, Node 24.22.1, Bun 1.4.4, FFmpeg/FFprobe 7.0.2, system Chromium 156.0.8078.47.
• Byte-exact UTF-8 BOM and UTF-16LE BOM fixtures were generated on Linux. Native Windows shell execution was not tested.

Additional context

The matching unmarked UTF-8 control renders successfully. Reusing the existing decoder makes both marked inputs render successfully too: each result reports total: 1, completed: 1, failed: 0, durationMs: 1000, with title: "Café". FFprobe confirms H.264, 320×180 and a 1-second duration; decoded frames were inspected.

Related work checked: #5725 covers project configuration, #6580 covers HTML decoding, and #6591 covers undefined injected readers for --variables-file. None changes this batch-file reader.