pdf-codec 1.1.5 → 1.2.0

package/dist/index.cjs

@@ -329,7 +329,6 @@ function unfilterScanlines(data, height, bytesPerRow, bpp) {
 	for (let y = 0; y < height; y++) {
 		const filterByte = data[y * stride];
 		if (filterByte === void 0 || !isPngFilterType(filterByte)) throw new Error(`unknown PNG filter type: ${String(filterByte)}`);
-		const filterType = filterByte;
 		const rowStart = y * stride + 1;
 		const outRowStart = y * bytesPerRow;
 		const prevOutRowStart = y > 0 ? outRowStart - bytesPerRow : void 0;
@@ -338,7 +337,7 @@ function unfilterScanlines(data, height, bytesPerRow, bpp) {
 			const a = x >= bpp ? out[outRowStart + x - bpp] : 0;
 			const b = prevOutRowStart === void 0 ? 0 : out[prevOutRowStart + x];
 			const c = x >= bpp && prevOutRowStart !== void 0 ? out[prevOutRowStart + x - bpp] : 0;
-			out[outRowStart + x] = raw + predictorValue(filterType, a, b, c) & 255;
+			out[outRowStart + x] = raw + predictorValue(filterByte, a, b, c) & 255;
 		}
 	}
 	return out;
@@ -1244,8 +1243,8 @@ function readXrefStreamSection(dict, raw, sink) {
 	const entries = /* @__PURE__ */ new Map();
 	let rowOffset = 0;
 	for (const [start, count] of ranges) for (let i = 0; i < count; i++) {
-		const base = rowOffset;
 		rowOffset += rowLength;
+		const base = rowOffset - rowLength;
 		if (rowLength === 0 || base + rowLength > decoded.bytes.length) {
 			sink({
 				code: "pdf/xref-entry-invalid",
@@ -1302,17 +1301,17 @@ function findAllKeywordPositions(bytes, needle) {
 function isDigitByte(byte) {
 	return byte !== void 0 && byte >= 48 && byte <= 57;
 }
+function scanBackWhile(bytes, end, predicate) {
+	let i = end;
+	while (i > 0 && predicate(bytes[i - 1])) i--;
+	return i;
+}
 function scanObjectHeaderBackward(bytes, objKeywordStart) {
-	let i = objKeywordStart;
-	while (i > 0 && isAsciiWhitespace(bytes[i - 1])) i--;
-	const genEnd = i;
-	while (i > 0 && isDigitByte(bytes[i - 1])) i--;
-	const genStart = i;
+	const genEnd = scanBackWhile(bytes, objKeywordStart, isAsciiWhitespace);
+	const genStart = scanBackWhile(bytes, genEnd, isDigitByte);
 	if (genStart === genEnd) return;
-	while (i > 0 && isAsciiWhitespace(bytes[i - 1])) i--;
-	const numEnd = i;
-	while (i > 0 && isDigitByte(bytes[i - 1])) i--;
-	const numStart = i;
+	const numEnd = scanBackWhile(bytes, genStart, isAsciiWhitespace);
+	const numStart = scanBackWhile(bytes, numEnd, isDigitByte);
 	if (numStart === numEnd) return;
 	const decoder = new TextDecoder("latin1");
 	return {
@@ -1486,9 +1485,11 @@ function openPdfDocument(bytes, sink) {
 	function resolveDict(obj) {
 		return asDict(resolve(obj));
 	}
-	const resolvedRoot = resolveDict(dictGet(xref.trailer, "Root"));
-	if (resolvedRoot === void 0 || !isName(dictGet(resolvedRoot, "Type"), "Catalog")) throw new PdfParseError("pdf/no-root", "no resolvable /Root catalog was found, even after cross-reference recovery");
-	const catalog = resolvedRoot;
+	function requireCatalog(dict) {
+		if (dict === void 0 || !isName(dictGet(dict, "Type"), "Catalog")) throw new PdfParseError("pdf/no-root", "no resolvable /Root catalog was found, even after cross-reference recovery");
+		return dict;
+	}
+	const catalog = requireCatalog(resolveDict(dictGet(xref.trailer, "Root")));
 	function pages() {
 		const pagesRoot = resolveDict(dictGet(catalog, "Pages"));
 		if (pagesRoot === void 0) return [];
@@ -5264,20 +5265,19 @@ function readPdf(bytes, options) {
 	const signal = options?.signal;
 	if (!hasPdfHeader(bytes)) throw new PdfParseError("pdf/no-header", "no \"%PDF-\" header found within the first bytes of the file; this does not look like a PDF at all");
 	const doc = openPdfDocument(bytes, sink);
-	const resolver = doc;
 	const fontResolver = createFontResolver({
-		resolver,
+		resolver: doc,
 		sink
 	});
 	const images = {};
 	const imageIdCache = /* @__PURE__ */ new Map();
 	const pages = doc.pages().map((pageDict) => {
 		throwIfAborted(signal);
-		return readPage(pageDict, resolver, fontResolver, images, imageIdCache, sink);
+		return readPage(pageDict, doc, fontResolver, images, imageIdCache, sink);
 	});
 	return {
 		formatVersion: document_schema_js.LAYOUT_FORMAT_VERSION,
-		metadata: readMetadata(doc.trailer, resolver),
+		metadata: readMetadata(doc.trailer, doc),
 		pages,
 		images
 	};

package/dist/index.js

@@ -328,7 +328,6 @@ function unfilterScanlines(data, height, bytesPerRow, bpp) {
 	for (let y = 0; y < height; y++) {
 		const filterByte = data[y * stride];
 		if (filterByte === void 0 || !isPngFilterType(filterByte)) throw new Error(`unknown PNG filter type: ${String(filterByte)}`);
-		const filterType = filterByte;
 		const rowStart = y * stride + 1;
 		const outRowStart = y * bytesPerRow;
 		const prevOutRowStart = y > 0 ? outRowStart - bytesPerRow : void 0;
@@ -337,7 +336,7 @@ function unfilterScanlines(data, height, bytesPerRow, bpp) {
 			const a = x >= bpp ? out[outRowStart + x - bpp] : 0;
 			const b = prevOutRowStart === void 0 ? 0 : out[prevOutRowStart + x];
 			const c = x >= bpp && prevOutRowStart !== void 0 ? out[prevOutRowStart + x - bpp] : 0;
-			out[outRowStart + x] = raw + predictorValue(filterType, a, b, c) & 255;
+			out[outRowStart + x] = raw + predictorValue(filterByte, a, b, c) & 255;
 		}
 	}
 	return out;
@@ -1243,8 +1242,8 @@ function readXrefStreamSection(dict, raw, sink) {
 	const entries = /* @__PURE__ */ new Map();
 	let rowOffset = 0;
 	for (const [start, count] of ranges) for (let i = 0; i < count; i++) {
-		const base = rowOffset;
 		rowOffset += rowLength;
+		const base = rowOffset - rowLength;
 		if (rowLength === 0 || base + rowLength > decoded.bytes.length) {
 			sink({
 				code: "pdf/xref-entry-invalid",
@@ -1301,17 +1300,17 @@ function findAllKeywordPositions(bytes, needle) {
 function isDigitByte(byte) {
 	return byte !== void 0 && byte >= 48 && byte <= 57;
 }
+function scanBackWhile(bytes, end, predicate) {
+	let i = end;
+	while (i > 0 && predicate(bytes[i - 1])) i--;
+	return i;
+}
 function scanObjectHeaderBackward(bytes, objKeywordStart) {
-	let i = objKeywordStart;
-	while (i > 0 && isAsciiWhitespace(bytes[i - 1])) i--;
-	const genEnd = i;
-	while (i > 0 && isDigitByte(bytes[i - 1])) i--;
-	const genStart = i;
+	const genEnd = scanBackWhile(bytes, objKeywordStart, isAsciiWhitespace);
+	const genStart = scanBackWhile(bytes, genEnd, isDigitByte);
 	if (genStart === genEnd) return;
-	while (i > 0 && isAsciiWhitespace(bytes[i - 1])) i--;
-	const numEnd = i;
-	while (i > 0 && isDigitByte(bytes[i - 1])) i--;
-	const numStart = i;
+	const numEnd = scanBackWhile(bytes, genStart, isAsciiWhitespace);
+	const numStart = scanBackWhile(bytes, numEnd, isDigitByte);
 	if (numStart === numEnd) return;
 	const decoder = new TextDecoder("latin1");
 	return {
@@ -1485,9 +1484,11 @@ function openPdfDocument(bytes, sink) {
 	function resolveDict(obj) {
 		return asDict(resolve(obj));
 	}
-	const resolvedRoot = resolveDict(dictGet(xref.trailer, "Root"));
-	if (resolvedRoot === void 0 || !isName(dictGet(resolvedRoot, "Type"), "Catalog")) throw new PdfParseError("pdf/no-root", "no resolvable /Root catalog was found, even after cross-reference recovery");
-	const catalog = resolvedRoot;
+	function requireCatalog(dict) {
+		if (dict === void 0 || !isName(dictGet(dict, "Type"), "Catalog")) throw new PdfParseError("pdf/no-root", "no resolvable /Root catalog was found, even after cross-reference recovery");
+		return dict;
+	}
+	const catalog = requireCatalog(resolveDict(dictGet(xref.trailer, "Root")));
 	function pages() {
 		const pagesRoot = resolveDict(dictGet(catalog, "Pages"));
 		if (pagesRoot === void 0) return [];
@@ -5263,20 +5264,19 @@ function readPdf(bytes, options) {
 	const signal = options?.signal;
 	if (!hasPdfHeader(bytes)) throw new PdfParseError("pdf/no-header", "no \"%PDF-\" header found within the first bytes of the file; this does not look like a PDF at all");
 	const doc = openPdfDocument(bytes, sink);
-	const resolver = doc;
 	const fontResolver = createFontResolver({
-		resolver,
+		resolver: doc,
 		sink
 	});
 	const images = {};
 	const imageIdCache = /* @__PURE__ */ new Map();
 	const pages = doc.pages().map((pageDict) => {
 		throwIfAborted(signal);
-		return readPage(pageDict, resolver, fontResolver, images, imageIdCache, sink);
+		return readPage(pageDict, doc, fontResolver, images, imageIdCache, sink);
 	});
 	return {
 		formatVersion: LAYOUT_FORMAT_VERSION,
-		metadata: readMetadata(doc.trailer, resolver),
+		metadata: readMetadata(doc.trailer, doc),
 		pages,
 		images
 	};

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "pdf-codec",
-  "version": "1.1.5",
+  "version": "1.2.0",
   "description": "Hand-written, dependency-minimal PDF codec: parses arbitrary real-world PDFs and generates new ones, built on document-schema.js's LayoutDocument pivot and Zod 4 codecs.",
   "type": "module",
   "repository": {