@sweberdev/witness 0.1.0 → 0.2.0

package/dist/index.cjs

@@ -26,6 +26,7 @@ __export(index_exports, {
   createDisclosure: () => createDisclosure,
   createMarking: () => createMarking,
   detectImageFormat: () => detectImageFormat,
+  detectMediaFormat: () => detectMediaFormat,
   disclosureText: () => disclosureText,
   escapeHtml: () => escapeHtml,
   format: () => format,
@@ -36,7 +37,9 @@ __export(index_exports, {
   isAiSourceType: () => isAiSourceType,
   labelHtml: () => labelHtml,
   labelText: () => labelText,
+  markFile: () => markFile,
   markImage: () => markImage,
+  markMedia: () => markMedia,
   markingAttributes: () => markingAttributes,
   markingJsonLd: () => markingJsonLd,
   markingMetaTags: () => markingMetaTags,
@@ -44,6 +47,8 @@ __export(index_exports, {
   nextMetadata: () => nextMetadata,
   parseSourceType: () => parseSourceType,
   readImageMarking: () => readImageMarking,
+  readMarking: () => readMarking,
+  readMediaMarking: () => readMediaMarking,
   readTextWatermark: () => readTextWatermark,
   registerLocale: () => registerLocale,
   registeredLocales: () => registeredLocales,
@@ -1182,6 +1187,293 @@ function unescapeXml(value) {
   return value.replace(/&quot;/g, '"').replace(/&lt;/g, "<").replace(/&gt;/g, ">").replace(/&apos;/g, "'").replace(/&amp;/g, "&");
 }
 
+// src/media.ts
+var XMP_UUID = hex("be7acfcb97a942e89c71999491e3afac");
+var C2PA_UUID = hex("d8fec3d61b0e483c92975828877ec481");
+function detectMediaFormat(bytes) {
+  if (bytes.length >= 10 && ascii2(bytes, 0, 3) === "ID3") return "mp3";
+  if (isMpegAudioFrame(bytes, 0)) return "mp3";
+  if (bytes.length >= 12 && ascii2(bytes, 0, 4) === "RIFF" && ascii2(bytes, 8, 4) === "WAVE") {
+    return "wav";
+  }
+  if (bytes.length >= 12 && ascii2(bytes, 4, 4) === "ftyp") return "mp4";
+  return null;
+}
+function markMedia(bytes, input = {}, options = {}) {
+  const format2 = detectMediaFormat(bytes);
+  if (!format2) return { bytes, status: "unsupported", format: format2 };
+  const parsed = parse(bytes, format2);
+  if (!parsed) return { bytes, status: "unsupported", format: format2 };
+  if (options.c2pa !== "overwrite" && parsed.c2pa) {
+    return { bytes, status: "skipped-c2pa", format: format2 };
+  }
+  const marking = "sourceType" in input && "humanReviewed" in input ? input : createMarking(input);
+  const xmp = buildXmp(marking, parsed.xmp);
+  const out = format2 === "mp3" ? writeMp3(bytes, xmp) : format2 === "wav" ? writeWav(bytes, xmp) : writeMp4(bytes, xmp);
+  if (!out) return { bytes, status: "unsupported", format: format2 };
+  return { bytes: out, status: "marked", format: format2 };
+}
+function readMediaMarking(bytes) {
+  const format2 = detectMediaFormat(bytes);
+  const parsed = format2 ? parse(bytes, format2) : null;
+  if (!format2 || !parsed) {
+    return { format: format2, xmp: null, aiGenerated: false, witness: false, c2pa: false };
+  }
+  return describe(format2, parsed.xmp, parsed.c2pa);
+}
+function markFile(bytes, input = {}, options = {}) {
+  if (detectImageFormat(bytes)) return markImage(bytes, input, options);
+  return markMedia(bytes, input, options);
+}
+function readMarking(bytes) {
+  if (detectImageFormat(bytes)) return readImageMarking(bytes);
+  return readMediaMarking(bytes);
+}
+function describe(format2, xmp, c2pa) {
+  const sourceType = xmp ? parseSourceType(xmpProperty(xmp, "Iptc4xmpExt:DigitalSourceType")) : void 0;
+  const generator = xmp ? xmpProperty(xmp, "Iptc4xmpExt:AISystemUsed") ?? xmpProperty(xmp, "xmp:CreatorTool") : void 0;
+  const info = {
+    format: format2,
+    xmp,
+    aiGenerated: isAiSourceType(sourceType),
+    witness: xmp?.includes(WITNESS_NS) ?? false,
+    c2pa
+  };
+  if (sourceType) info.sourceType = sourceType;
+  if (generator) info.generator = generator;
+  return info;
+}
+function parse(bytes, format2) {
+  if (format2 === "mp3") {
+    const tag = readId3(bytes);
+    if (tag === "unsupported") return null;
+    if (!tag) return { xmp: null, c2pa: false };
+    const xmpFrame = tag.frames.find((f) => f.id === "PRIV" && isXmpPriv(bytes, f));
+    return {
+      xmp: xmpFrame ? decode(bytes.subarray(xmpFrame.dataStart + 4, xmpFrame.end)) : null,
+      c2pa: tag.frames.some(
+        (f) => f.id === "GEOB" && ascii2(bytes, f.dataStart, Math.min(64, f.end - f.dataStart)).toLowerCase().includes("c2pa")
+      )
+    };
+  }
+  if (format2 === "wav") {
+    const chunks = riffChunks(bytes);
+    if (!chunks) return null;
+    const xmp2 = chunks.find((c) => c.fourcc === "_PMX");
+    return {
+      xmp: xmp2 ? decode(bytes.subarray(xmp2.dataStart, xmp2.dataEnd)) : null,
+      c2pa: chunks.some((c) => c.fourcc === "C2PA")
+    };
+  }
+  const boxes = mp4Boxes(bytes);
+  if (!boxes) return null;
+  const xmp = boxes.find((b) => isUuidBox(bytes, b, XMP_UUID));
+  return {
+    xmp: xmp ? decode(bytes.subarray(xmp.dataStart + 16, xmp.end)) : null,
+    c2pa: boxes.some((b) => isUuidBox(bytes, b, C2PA_UUID))
+  };
+}
+function readId3(bytes) {
+  if (ascii2(bytes, 0, 3) !== "ID3") return null;
+  const version = bytes[3];
+  const flags = bytes[5] ?? 0;
+  if (version !== 3 && version !== 4) return "unsupported";
+  if (flags & 128 || flags & 16) return "unsupported";
+  const length = 10 + syncsafe(bytes, 6);
+  if (length > bytes.length) return "unsupported";
+  let offset = 10;
+  if (flags & 64) {
+    offset += version === 4 ? syncsafe(bytes, 10) : 4 + readU32BE2(bytes, 10);
+  }
+  const frames = [];
+  while (offset + 10 <= length) {
+    const id = ascii2(bytes, offset, 4);
+    if (!/^[A-Z0-9]{4}$/.test(id)) break;
+    const size = version === 4 ? syncsafe(bytes, offset + 4) : readU32BE2(bytes, offset + 4);
+    const end = offset + 10 + size;
+    if (end > length) return "unsupported";
+    frames.push({ id, start: offset, dataStart: offset + 10, end });
+    offset = end;
+  }
+  return { version, length, frames };
+}
+function isXmpPriv(bytes, frame) {
+  return ascii2(bytes, frame.dataStart, 4) === "XMP\0";
+}
+function writeMp3(bytes, xmp) {
+  const tag = readId3(bytes);
+  if (tag === "unsupported") return null;
+  const version = tag ? tag.version : 4;
+  const kept = tag ? tag.frames.filter((f) => !(f.id === "PRIV" && isXmpPriv(bytes, f))).map((f) => bytes.subarray(f.start, f.end)) : [];
+  const data = concat2(latin12("XMP\0"), utf82(xmp));
+  const frame = new Uint8Array(10 + data.length);
+  frame.set(latin12("PRIV"), 0);
+  if (version === 4) writeSyncsafe(frame, 4, data.length);
+  else writeU32BE2(frame, 4, data.length);
+  frame.set(data, 10);
+  const body = concat2(...kept, frame, new Uint8Array(64));
+  const header = new Uint8Array(10);
+  header.set(latin12("ID3"), 0);
+  header[3] = version;
+  writeSyncsafe(header, 6, body.length);
+  return concat2(header, body, bytes.subarray(tag ? tag.length : 0));
+}
+function isMpegAudioFrame(bytes, offset) {
+  const b1 = bytes[offset + 1] ?? 0;
+  return bytes[offset] === 255 && (b1 & 224) === 224 && (b1 & 24) !== 8 && (b1 & 6) !== 0;
+}
+function syncsafe(b, o) {
+  return ((b[o] ?? 0) & 127) << 21 | ((b[o + 1] ?? 0) & 127) << 14 | ((b[o + 2] ?? 0) & 127) << 7 | (b[o + 3] ?? 0) & 127;
+}
+function writeSyncsafe(b, o, v) {
+  b[o] = v >>> 21 & 127;
+  b[o + 1] = v >>> 14 & 127;
+  b[o + 2] = v >>> 7 & 127;
+  b[o + 3] = v & 127;
+}
+function riffChunks(bytes) {
+  const chunks = [];
+  const limit = Math.min(bytes.length, 8 + readU32LE2(bytes, 4));
+  let offset = 12;
+  while (offset + 8 <= limit) {
+    const size = readU32LE2(bytes, offset + 4);
+    const dataEnd = offset + 8 + size;
+    if (dataEnd > bytes.length) return null;
+    const end = Math.min(dataEnd + size % 2, bytes.length);
+    chunks.push({
+      fourcc: ascii2(bytes, offset, 4),
+      start: offset,
+      dataStart: offset + 8,
+      dataEnd,
+      end
+    });
+    offset = end;
+  }
+  return chunks;
+}
+function writeWav(bytes, xmp) {
+  const chunks = riffChunks(bytes);
+  if (!chunks) return null;
+  const data = utf82(xmp);
+  const chunk = new Uint8Array(8 + data.length + data.length % 2);
+  chunk.set(latin12("_PMX"), 0);
+  writeU32LE2(chunk, 4, data.length);
+  chunk.set(data, 8);
+  const body = concat2(
+    ...chunks.filter((c) => c.fourcc !== "_PMX").map((c) => bytes.subarray(c.start, c.end)),
+    chunk
+  );
+  const out = concat2(bytes.subarray(0, 12), body);
+  writeU32LE2(out, 4, out.length - 8);
+  return out;
+}
+function mp4Boxes(bytes) {
+  const boxes = [];
+  let offset = 0;
+  while (offset + 8 <= bytes.length) {
+    const size = readU32BE2(bytes, offset);
+    const type = ascii2(bytes, offset + 4, 4);
+    let header = 8;
+    let end;
+    if (size === 1) {
+      if (offset + 16 > bytes.length) return null;
+      const high = readU32BE2(bytes, offset + 8);
+      end = offset + high * 2 ** 32 + readU32BE2(bytes, offset + 12);
+      header = 16;
+    } else if (size === 0) {
+      end = bytes.length;
+    } else {
+      end = offset + size;
+    }
+    if (end > bytes.length || end < offset + header) return null;
+    boxes.push({ type, start: offset, dataStart: offset + header, end, open: size === 0 });
+    offset = end;
+  }
+  return boxes;
+}
+function isUuidBox(bytes, box, uuid) {
+  if (box.type !== "uuid" || box.end - box.dataStart < 16) return false;
+  for (let i = 0; i < 16; i++) if (bytes[box.dataStart + i] !== uuid[i]) return false;
+  return true;
+}
+function writeMp4(bytes, xmp) {
+  const boxes = mp4Boxes(bytes);
+  if (!boxes) return null;
+  const last = boxes[boxes.length - 1];
+  let head = bytes;
+  let length = bytes.length;
+  if (last && isUuidBox(bytes, last, XMP_UUID)) length = last.start;
+  head = bytes.slice(0, length);
+  for (const box2 of boxes) {
+    if (box2.start < length && isUuidBox(bytes, box2, XMP_UUID)) {
+      head.set(latin12("free"), box2.start + 4);
+    }
+  }
+  const open = boxes.find((b) => b.open && b.start < length);
+  if (open) {
+    const size = length - open.start;
+    if (size > 4294967295) return null;
+    writeU32BE2(head, open.start, size);
+  }
+  const data = utf82(xmp);
+  const box = new Uint8Array(8 + 16 + data.length);
+  writeU32BE2(box, 0, box.length);
+  box.set(latin12("uuid"), 4);
+  box.set(XMP_UUID, 8);
+  box.set(data, 24);
+  return concat2(head, box);
+}
+function hex(value) {
+  const out = new Uint8Array(value.length / 2);
+  for (let i = 0; i < out.length; i++) out[i] = Number.parseInt(value.slice(i * 2, i * 2 + 2), 16);
+  return out;
+}
+function decode(bytes) {
+  return new TextDecoder().decode(bytes).replace(/\0+$/, "");
+}
+function ascii2(bytes, offset, length) {
+  let out = "";
+  for (let i = offset; i < offset + length && i < bytes.length; i++) {
+    out += String.fromCharCode(bytes[i] ?? 0);
+  }
+  return out;
+}
+function latin12(text) {
+  const out = new Uint8Array(text.length);
+  for (let i = 0; i < text.length; i++) out[i] = text.charCodeAt(i) & 255;
+  return out;
+}
+function utf82(text) {
+  return new TextEncoder().encode(text);
+}
+function concat2(...parts) {
+  const out = new Uint8Array(parts.reduce((sum, part) => sum + part.length, 0));
+  let offset = 0;
+  for (const part of parts) {
+    out.set(part, offset);
+    offset += part.length;
+  }
+  return out;
+}
+function readU32BE2(b, o) {
+  return ((b[o] ?? 0) << 24 | (b[o + 1] ?? 0) << 16 | (b[o + 2] ?? 0) << 8 | (b[o + 3] ?? 0)) >>> 0;
+}
+function readU32LE2(b, o) {
+  return ((b[o] ?? 0) | (b[o + 1] ?? 0) << 8 | (b[o + 2] ?? 0) << 16 | (b[o + 3] ?? 0) << 24) >>> 0;
+}
+function writeU32BE2(b, o, v) {
+  b[o] = v >>> 24 & 255;
+  b[o + 1] = v >>> 16 & 255;
+  b[o + 2] = v >>> 8 & 255;
+  b[o + 3] = v & 255;
+}
+function writeU32LE2(b, o, v) {
+  b[o] = v & 255;
+  b[o + 1] = v >>> 8 & 255;
+  b[o + 2] = v >>> 16 & 255;
+  b[o + 3] = v >>> 24 & 255;
+}
+
 // src/types.ts
 var DISCLOSURE_KINDS = [
   "chatbot",
@@ -1312,6 +1604,7 @@ function findInRun(bytes) {
   createDisclosure,
   createMarking,
   detectImageFormat,
+  detectMediaFormat,
   disclosureText,
   escapeHtml,
   format,
@@ -1322,7 +1615,9 @@ function findInRun(bytes) {
   isAiSourceType,
   labelHtml,
   labelText,
+  markFile,
   markImage,
+  markMedia,
   markingAttributes,
   markingJsonLd,
   markingMetaTags,
@@ -1330,6 +1625,8 @@ function findInRun(bytes) {
   nextMetadata,
   parseSourceType,
   readImageMarking,
+  readMarking,
+  readMediaMarking,
   readTextWatermark,
   registerLocale,
   registeredLocales,