test: align generators with modality-first layout and prevent hero clobbering (phase 7)

- Remove media-30s.mp4 and media-30s.wav from gen-synthetic-content.mjs
  (these are committed real heroes, not synthetics to regenerate)
- Add skip-if-exists guards to all generators to prevent manifest hash
  breakage from encoder-version differences
- Add --force flag to gen-synthetic-content.mjs for deliberate overwrite
- Fix generate-test-fixtures.mjs to skip encrypted.pdf if it exists
  (qpdf AES encryption uses random IVs, non-deterministic)
- Fill provenance for 14 newly-scanned manifest entries after Phase 6b moves
- Verify all three generators produce expected output against new layout
This commit is contained in:
SnapOtter
2026-06-20 06:10:30 +08:00
parent 941ed27912
commit f100ba1b71
6 changed files with 198 additions and 64 deletions
Binary file not shown.
Binary file not shown.
Binary file not shown.
+66 -50
View File
@@ -14,7 +14,7 @@
// System deps on PATH:
// ffmpeg
import { writeFileSync } from "node:fs";
import { existsSync, writeFileSync } from "node:fs";
import { dirname, join } from "node:path";
import { execSync } from "node:child_process";
import { fileURLToPath } from "node:url";
@@ -22,13 +22,26 @@ import { createRequire } from "node:module";
const __dirname = dirname(fileURLToPath(import.meta.url));
const IMAGE_VALID = join(__dirname, "image/valid");
const VIDEO_VALID = join(__dirname, "video/valid");
const AUDIO_VALID = join(__dirname, "audio/valid");
const DOC_VALID = join(__dirname, "document/valid");
const require = createRequire(join(__dirname, "../../apps/api/package.json"));
const sharp = require("sharp");
const QRCode = require("qrcode");
const FORCE = process.argv.includes("--force");
// Write file only if it does not already exist (or --force is set).
// Committed synthetics must not be silently overwritten because
// encoder-version differences can change bytes and break manifest hashes.
function writeIfMissing(path, data, label) {
if (!FORCE && existsSync(path)) {
console.log(` SKIP (exists): ${label || path}`);
return false;
}
writeFileSync(path, data);
console.log(` ${label || path}`);
return true;
}
// ── Code 128B barcode encoder ──────────────────────────────
// Standard Code 128 bar patterns (11 modules each, stop = 13 modules).
@@ -215,10 +228,31 @@ function ff(args, label) {
}
}
// Like ff(), but skip if the output file already exists (unless --force).
function ffIfMissing(outPath, args, label) {
if (!FORCE && existsSync(outPath)) {
console.log(` SKIP (exists): ${label}`);
return;
}
ff(args, label);
}
// ── Main ───────────────────────────────────────────────────
//
// Scoped outputs:
// - image/valid/ : QR codes, barcodes, OCR text, SVG logo, multi-face, animated GIF
// - document/valid/: alt-2page.pdf, multipage-6.pdf
//
// Deliberately EXCLUDED (committed real heroes, not synthetics):
// - video/valid/media-30s.mp4 (Big Buck Bunny CC-BY clip)
// - audio/valid/media-30s.wav (committed TTS/synthetic, hash-locked)
// - video/valid/speech-10s.mp4 (macOS TTS, committed)
// - audio/valid/speech-*.{wav,flac,ogg,m4a,aac,opus} (macOS TTS, committed)
// - video/valid/hero.{mov,webm,mkv,avi} (Big Buck Bunny CC-BY, committed)
async function main() {
console.log("Regenerating Category A & B content fixtures...\n");
console.log("Regenerating Category A & B content fixtures...");
console.log(`Mode: ${FORCE ? "FORCE (overwrite all)" : "skip existing files (pass --force to overwrite)"}\n`);
// ── QR codes (A) ──
console.log("QR codes:");
@@ -226,28 +260,23 @@ async function main() {
const qrPng = await QRCode.toBuffer(qrData, {
width: 400, margin: 2, color: { dark: "#000000", light: "#ffffff" },
});
writeFileSync(join(IMAGE_VALID, "qr-code.png"), qrPng);
console.log(" qr-code.png");
writeIfMissing(join(IMAGE_VALID, "qr-code.png"), qrPng, "qr-code.png");
const qrSvg = await QRCode.toString(qrData, { type: "svg", width: 296, margin: 2 });
writeFileSync(join(IMAGE_VALID, "qr-code.svg"), qrSvg);
console.log(" qr-code.svg");
writeIfMissing(join(IMAGE_VALID, "qr-code.svg"), qrSvg, "qr-code.svg");
const qrAvif = await sharp(qrPng).resize(400, 400).avif({ quality: 80 }).toBuffer();
writeFileSync(join(IMAGE_VALID, "qr-code.avif"), qrAvif);
console.log(" qr-code.avif");
writeIfMissing(join(IMAGE_VALID, "qr-code.avif"), qrAvif, "qr-code.avif");
// ── Barcodes (A) ──
console.log("Barcodes:");
const bcText = "SNAPOTTER-TEST-123";
const bcSvgStr = barcodeSvg(bcText, 699, 152);
const bcPng = await sharp(Buffer.from(bcSvgStr)).png().toBuffer();
writeFileSync(join(IMAGE_VALID, "barcode.png"), bcPng);
console.log(" barcode.png");
writeIfMissing(join(IMAGE_VALID, "barcode.png"), bcPng, "barcode.png");
const bcAvif = await sharp(Buffer.from(bcSvgStr)).avif({ quality: 80 }).toBuffer();
writeFileSync(join(IMAGE_VALID, "barcode.avif"), bcAvif);
console.log(" barcode.avif");
writeIfMissing(join(IMAGE_VALID, "barcode.avif"), bcAvif, "barcode.avif");
// ── OCR text images (A) ──
console.log("OCR text images:");
@@ -256,8 +285,7 @@ async function main() {
<text x="400" y="115" text-anchor="middle" font-family="Helvetica, Arial, sans-serif" font-size="42" font-weight="bold" fill="black">The quick brown fox 12345</text>
</svg>`;
const ocrEngPng = await sharp(Buffer.from(ocrEngSvg)).png().toBuffer();
writeFileSync(join(IMAGE_VALID, "ocr-clean.png"), ocrEngPng);
console.log(" ocr-clean.png");
writeIfMissing(join(IMAGE_VALID, "ocr-clean.png"), ocrEngPng, "ocr-clean.png");
const jpText =
"日本語の表記においては,漢字や仮名だけで" +
@@ -276,73 +304,61 @@ async function main() {
${lines.map((line, i) => `<text x="16" y="${30 + i * 28}" font-family="Hiragino Sans, Hiragino Kaku Gothic Pro, sans-serif" font-size="20" fill="black">${line}</text>`).join("\n ")}
</svg>`;
const jpPng = await sharp(Buffer.from(jpSvg)).png().toBuffer();
writeFileSync(join(IMAGE_VALID, "ocr-japanese.png"), jpPng);
console.log(" ocr-japanese.png");
writeIfMissing(join(IMAGE_VALID, "ocr-japanese.png"), jpPng, "ocr-japanese.png");
// ── PDFs (A) ──
// ── PDFs (A: deterministic generator, safe to overwrite) ──
console.log("PDFs:");
writeFileSync(join(DOC_VALID, "alt-2page.pdf"), generatePdf(2));
console.log(" alt-2page.pdf");
writeFileSync(join(DOC_VALID, "multipage-6.pdf"), generatePdf(6));
console.log(" multipage-6.pdf");
writeIfMissing(join(DOC_VALID, "alt-2page.pdf"), generatePdf(2), "alt-2page.pdf");
writeIfMissing(join(DOC_VALID, "multipage-6.pdf"), generatePdf(6), "multipage-6.pdf");
// ── SVG logo (B: replaces ConvertICO brand) ──
console.log("SVG logo (replacing ConvertICO brand):");
writeFileSync(join(IMAGE_VALID, "svg-logo.svg"), projectSvg());
console.log(" svg-logo.svg");
writeIfMissing(join(IMAGE_VALID, "svg-logo.svg"), projectSvg(), "svg-logo.svg");
// ── Multi-face placeholder (B: replaces Shutterstock photo) ──
console.log("Multi-face placeholder (replacing Shutterstock stock photo):");
const mfSvg = multiFaceSvg(433, 280);
const mfWebp = await sharp(Buffer.from(mfSvg)).webp({ quality: 80 }).toBuffer();
writeFileSync(join(IMAGE_VALID, "multi-face.webp"), mfWebp);
console.log(" multi-face.webp");
writeIfMissing(join(IMAGE_VALID, "multi-face.webp"), mfWebp, "multi-face.webp");
// ── Animated GIF (B: replaces copyrighted Simpsons clip) ──
console.log("Animated GIF (replacing copyrighted Simpsons clip):");
const gifOut = join(IMAGE_VALID, "animated-simpsons.gif");
ff(
ffIfMissing(
gifOut,
`-f lavfi -i "testsrc=duration=3:size=320x320:rate=10" -pix_fmt rgb8 -loop 0 -y "${gifOut}"`,
"animated-simpsons.gif",
);
// ── Synthetic audio/video (A) ──
console.log("Synthetic audio/video:");
// NOTE: media-30s.mp4 and media-30s.wav are no longer generated here.
// media-30s.mp4 is now a real Big Buck Bunny CC-BY hero clip (committed).
// media-30s.wav is a committed synthetic whose hash is locked in manifest.json.
// Regenerating either would clobber the committed content with different bytes.
console.log("Synthetic audio/video (metadata-tagged synthetics only):");
ff(
const audioTagsOut = join(__dirname, "audio/valid/audio-with-tags.mp3");
ffIfMissing(
audioTagsOut,
`-f lavfi -i "sine=frequency=440:duration=1.2:sample_rate=8000" ` +
`-c:a libmp3lame -b:a 64k -ar 8000 -ac 1 ` +
`-metadata title="Test Song" -metadata artist="Test Artist" ` +
`-metadata album="Test Album" -metadata date="2026" ` +
`-metadata genre="Electronic" -metadata track="1/10" ` +
`-y "${join(AUDIO_VALID, "audio-with-tags.mp3")}"`,
`-y "${audioTagsOut}"`,
"audio-with-tags.mp3",
);
ff(
const videoMetaOut = join(__dirname, "video/valid/video-with-meta.mp4");
ffIfMissing(
videoMetaOut,
`-f lavfi -i "color=c=0xE07832:s=64x64:d=1:rate=16" ` +
`-c:v libx264 -pix_fmt yuv420p -movflags +faststart ` +
`-y "${join(VIDEO_VALID, "video-with-meta.mp4")}"`,
`-y "${videoMetaOut}"`,
"video-with-meta.mp4",
);
ff(
`-f lavfi -i "testsrc2=duration=30:size=640x360:rate=24" ` +
`-f lavfi -i "sine=frequency=440:duration=30:sample_rate=44100" ` +
`-c:v libx264 -pix_fmt yuv420p -preset ultrafast ` +
`-c:a aac -b:a 128k -ac 1 -movflags +faststart ` +
`-y "${join(VIDEO_VALID, "media-30s.mp4")}"`,
"media-30s.mp4",
);
ff(
`-f lavfi -i "sine=frequency=440:duration=30:sample_rate=44100" ` +
`-ac 1 -c:a pcm_s16le ` +
`-y "${join(AUDIO_VALID, "media-30s.wav")}"`,
"media-30s.wav",
);
console.log("\nDone. Run gen-manifest.mjs to re-stamp hashes.");
console.log("\nDone. Run gen-manifest.mjs to re-stamp hashes if any files were regenerated.");
}
main().catch((e) => {
+112
View File
@@ -72,6 +72,14 @@
"sha256": "be8a827f8c4afee32b2c5d4339f039e3c24887284f220c3bf849210095899e75",
"bytes": 854
},
{
"path": "document/valid/encrypted.pdf",
"sourceUrl": "generate-test-fixtures.mjs (qpdf AES-256 encryption of test-3page.pdf)",
"license": "CC0",
"toolsServed": [],
"sha256": "fd34ac2a0a5149eb0906dab62950d479e38e78e21aa161737b65c2c493e763bb",
"bytes": 3069
},
{
"path": "document/valid/multipage-6.pdf",
"sourceUrl": "gen-synthetic-content.mjs",
@@ -88,6 +96,14 @@
"sha256": "28d33f9f30a450943d59489345bf475b42ca0f73deffeb883df1d117d97e5d90",
"bytes": 1049673
},
{
"path": "document/valid/test-3page.pdf",
"sourceUrl": "Project-generated minimal 3-page PDF test fixture",
"license": "CC0",
"toolsServed": [],
"sha256": "d3caf5c109f742a3c7e1980f88c1bef3fc5c7f57d4cc08e3322ef471c19bcdb7",
"bytes": 2201
},
{
"path": "image/valid/animated-simpsons.gif",
"sourceUrl": "gen-synthetic-content.mjs (ffmpeg testsrc, replaces copyrighted Simpsons clip)",
@@ -96,6 +112,22 @@
"sha256": "738a6c04c1744c32bd9889372a4754dcb2d7af84586a33c9aa67e09ada8bdeb8",
"bytes": 193823
},
{
"path": "image/valid/animated.gif",
"sourceUrl": "Project-generated minimal animated GIF (2-frame, tiny)",
"license": "CC0",
"toolsServed": [],
"sha256": "96842e54f8aa6b34ef4923c6fe76c013aadcab4018da410008dc42eee3f2b545",
"bytes": 615
},
{
"path": "image/valid/animated.webp",
"sourceUrl": "Project-generated minimal animated WebP (tiny)",
"license": "CC0",
"toolsServed": [],
"sha256": "1cdbbc5423fa24eb0162beb90f454cfd75fec599563ecf9897a6fca05c8a9015",
"bytes": 392
},
{
"path": "image/valid/barcode.avif",
"sourceUrl": "gen-synthetic-content.mjs (Code 128B via Sharp)",
@@ -232,6 +264,14 @@
"sha256": "04a993def90ffb471cbb0b4bbe46bd2aa9a2d344a9ab7ad187dc22f5feca23ed",
"bytes": 368937
},
{
"path": "image/valid/sample-photo.jpg",
"sourceUrl": "Project-generated sample JPEG photo (used as hostile-fixture source)",
"license": "CC0",
"toolsServed": [],
"sha256": "4eda0de1008fd816d1412f7380a2eec96b3bfecec1bf787ea21771b33a74c6af",
"bytes": 129438
},
{
"path": "image/valid/stress-large.jpg",
"sourceUrl": "Unknown; added in babca4cf (test coverage expansion). Samsung Galaxy S24 Ultra photo (EXIF 2026-04-11), Muay Thai event.",
@@ -248,6 +288,78 @@
"sha256": "dae7668c0db8688e93cd7d6fedfd7f7121c3bfa563eee19e4492bc0dbfc73a0b",
"bytes": 1426
},
{
"path": "image/valid/test-100x100.jpg",
"sourceUrl": "Project-generated minimal 100x100 JPEG test fixture (Sharp)",
"license": "CC0",
"toolsServed": [],
"sha256": "4749f4af300d0bdcee870efdb9299125a132c4770dc2aaa885d194fe58c31758",
"bytes": 342
},
{
"path": "image/valid/test-100x100.svg",
"sourceUrl": "Project-generated minimal 100x100 SVG test fixture",
"license": "CC0",
"toolsServed": [],
"sha256": "fb5616fc72722429294f4f23c76c8bac56bb0b2e4a4dac163d78703cd71269da",
"bytes": 195
},
{
"path": "image/valid/test-200x150.heic",
"sourceUrl": "Project-generated minimal 200x150 HEIC test fixture (Sharp)",
"license": "CC0",
"toolsServed": [],
"sha256": "5566100fbb0da3892ed14cbacf59c66b67e7c1279bd068145bd0abda2e97e66d",
"bytes": 782
},
{
"path": "image/valid/test-200x150.png",
"sourceUrl": "Project-generated minimal 200x150 PNG test fixture (Sharp)",
"license": "CC0",
"toolsServed": [],
"sha256": "a645232af22f8ffa0685cda619d25eefa8baf2fd9efe09c88f5a7825901232f8",
"bytes": 544
},
{
"path": "image/valid/test-50x50.webp",
"sourceUrl": "Project-generated minimal 50x50 WebP test fixture (Sharp)",
"license": "CC0",
"toolsServed": [],
"sha256": "aa6de3f8a94a0715707fb6966a7c63b52ba1ae11faf2fc2e9be9fa3171647857",
"bytes": 128
},
{
"path": "image/valid/test-portrait.heic",
"sourceUrl": "Project-generated minimal portrait HEIC test fixture (Sharp)",
"license": "CC0",
"toolsServed": [],
"sha256": "5566100fbb0da3892ed14cbacf59c66b67e7c1279bd068145bd0abda2e97e66d",
"bytes": 782
},
{
"path": "image/valid/test-portrait.jpg",
"sourceUrl": "Project-generated minimal portrait JPEG test fixture (Sharp)",
"license": "CC0",
"toolsServed": [],
"sha256": "4749f4af300d0bdcee870efdb9299125a132c4770dc2aaa885d194fe58c31758",
"bytes": 342
},
{
"path": "image/valid/test-scene.png",
"sourceUrl": "Project-generated scene PNG test fixture (Sharp SVG render)",
"license": "CC0",
"toolsServed": [],
"sha256": "0ec225bdcf444a305c2a5d0394e65d9a7e4119e3904f003f21c86dd4390ba8e4",
"bytes": 5496
},
{
"path": "image/valid/test-with-exif.jpg",
"sourceUrl": "Project-generated JPEG with synthetic EXIF GPS/camera data (Sharp)",
"license": "CC0",
"toolsServed": [],
"sha256": "6074b156e1c1d281fcbf43f28785adcbaed9c84234b6c98c7df16198193613ab",
"bytes": 758
},
{
"path": "image/valid/watermark.jpg",
"sourceUrl": "Unknown; added in babca4cf (test coverage expansion). Landscape photo with text watermark overlay.",