pdf-parser.js 3.0.7 → 4.5.1

package/dist/read.cjs

@@ -1,18 +1,27 @@
 Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
+require("./notes-annotation-author.cjs");
+const require_objects = require("./objects.cjs");
+const require_pdf_text = require("./pdf-text.cjs");
+const require_bytes_writer = require("./bytes/writer.cjs");
+const require_serialize = require("./serialize.cjs");
+const require_matrix = require("./matrix.cjs");
+const require_annotations = require("./annotations.cjs");
+const require_filters = require("./filters.cjs");
 const require_util_base64 = require("./util/base64.cjs");
+const require_attachments = require("./attachments.cjs");
 const require_bytes_crc32 = require("./bytes/crc32.cjs");
-const require_bytes_writer = require("./bytes/writer.cjs");
+const require_form = require("./form.cjs");
+const require_xmp = require("./xmp.cjs");
+const require_optional_content = require("./optional-content.cjs");
+const require_structure = require("./structure.cjs");
 require("./layout.cjs");
 const require_diagnostics = require("./diagnostics.cjs");
-const require_objects = require("./objects.cjs");
-const require_filters = require("./filters.cjs");
 const require_document = require("./document.cjs");
 const require_util_abort = require("./util/abort.cjs");
 const require_font_read = require("./font-read.cjs");
 const require_images_read = require("./images-read.cjs");
-const require_matrix = require("./matrix.cjs");
 const require_interpret = require("./interpret.cjs");
-require("./write.cjs");
+const require_navigation = require("./navigation.cjs");
 //#region src/read.ts
 const PDF_HEADER_BYTES = new TextEncoder().encode("%PDF-");
 const HEADER_SEARCH_WINDOW = 1024;
@@ -30,6 +39,7 @@ function readPdf(bytes, options) {
 	const sink = options?.sink ?? require_diagnostics.NOOP_DIAGNOSTIC_SINK;
 	const signal = options?.signal;
 	if (!hasPdfHeader(bytes)) throw new require_diagnostics.PdfParseError("pdf/no-header", "no \"%PDF-\" header found within the first bytes of the file; this does not look like a PDF at all");
+	require_util_abort.throwIfAborted(signal);
 	const doc = require_document.openPdfDocument(bytes, sink);
 	const fontResolver = require_font_read.createFontResolver({
 		resolver: doc,
@@ -37,29 +47,44 @@ function readPdf(bytes, options) {
 	});
 	const images = {};
 	const imageIdCache = /* @__PURE__ */ new Map();
-	const pages = doc.pages().map((pageDict) => {
+	const pageDicts = doc.pages();
+	const destinationRegistry = require_navigation.createDestinationRegistry(doc.catalog, doc, (obj) => doc.pageIndex(obj), sink);
+	const outline = require_navigation.readOutline(doc.catalog, destinationRegistry, doc, sink);
+	const attachments = require_attachments.readAttachments(doc.catalog, pageDicts, doc, sink);
+	const optionalContent = require_optional_content.readOptionalContent(doc.catalog, doc, sink);
+	const form = require_form.readAcroForm(doc.catalog, doc, (obj) => doc.pageIndex(obj), sink);
+	const structure = require_structure.readStructure(doc.catalog, pageDicts, doc, sink);
+	const source = readDocumentResidue(doc, sink);
+	const pageBoxRows = [];
+	const pages = pageDicts.map((pageDict, index) => {
 		require_util_abort.throwIfAborted(signal);
-		return readPage(pageDict, doc, fontResolver, images, imageIdCache, sink);
+		return readPage(index, pageDict, doc, fontResolver, images, imageIdCache, destinationRegistry, optionalContent.layerNameOf, structure.ownerOf, pageBoxRows, sink);
 	});
+	if (pageBoxRows.length > 0) source["page-boxes"] = {
+		format: "pdf",
+		xml: require_serialize.serializeObjectToText(require_objects.pdfArray(pageBoxRows))
+	};
 	return {
 		formatVersion: 1,
-		metadata: readMetadata(doc.trailer, doc),
+		metadata: readMetadata(doc.trailer, doc, doc.catalog, sink),
 		pages,
-		images
+		images,
+		...destinationRegistry.entries.length > 0 ? { destinations: [...destinationRegistry.entries] } : {},
+		...outline.length > 0 ? { outline } : {},
+		...attachments.length > 0 ? { attachments } : {},
+		...optionalContent.layers.length > 0 ? { layers: [...optionalContent.layers] } : {},
+		...form.length > 0 ? { form } : {},
+		...structure.tree.length > 0 ? { structure: [...structure.tree] } : {},
+		...Object.keys(source).length > 0 ? { source } : {}
 	};
 }
-function readMediaBox(page) {
-	const arr = require_objects.asArray(require_objects.dictGet(page, "MediaBox"));
-	if (arr === void 0) return {
-		llx: 0,
-		lly: 0,
-		urx: DEFAULT_PAGE_WIDTH_PT,
-		ury: DEFAULT_PAGE_HEIGHT_PT
-	};
+function readDeclaredPageBox(page, key) {
+	const arr = require_objects.asArray(require_objects.dictGet(page, key));
+	if (arr === void 0) return;
 	const a = require_objects.asNumber(arr[0]) ?? 0;
 	const b = require_objects.asNumber(arr[1]) ?? 0;
-	const c = require_objects.asNumber(arr[2]) ?? DEFAULT_PAGE_WIDTH_PT;
-	const d = require_objects.asNumber(arr[3]) ?? DEFAULT_PAGE_HEIGHT_PT;
+	const c = require_objects.asNumber(arr[2]) ?? 0;
+	const d = require_objects.asNumber(arr[3]) ?? 0;
 	return {
 		llx: Math.min(a, c),
 		lly: Math.min(b, d),
@@ -67,6 +92,33 @@ function readMediaBox(page) {
 		ury: Math.max(b, d)
 	};
 }
+function readMediaBox(page) {
+	return readDeclaredPageBox(page, "MediaBox") ?? {
+		llx: 0,
+		lly: 0,
+		urx: DEFAULT_PAGE_WIDTH_PT,
+		ury: DEFAULT_PAGE_HEIGHT_PT
+	};
+}
+function rotatedRectBounds(rect, matrix) {
+	const transformed = [
+		[rect.llx, rect.lly],
+		[rect.urx, rect.lly],
+		[rect.llx, rect.ury],
+		[rect.urx, rect.ury]
+	].map(([x, y]) => require_matrix.applyMatrix(matrix, {
+		x,
+		y
+	}));
+	const xs = transformed.map((point) => point.x);
+	const ys = transformed.map((point) => point.y);
+	return {
+		minX: Math.min(...xs),
+		minY: Math.min(...ys),
+		maxX: Math.max(...xs),
+		maxY: Math.max(...ys)
+	};
+}
 function normalizeRotation(rotate) {
 	if (rotate === void 0) return 0;
 	const normalized = (Math.round(rotate / 90) * 90 % 360 + 360) % 360;
@@ -135,35 +187,150 @@ function readPageContentBytes(page, resolver, sink) {
 	}
 	return /* @__PURE__ */ new Uint8Array(0);
 }
-function readPage(page, resolver, fontResolver, images, imageIdCache, sink) {
+function contentItemBounds(item) {
+	if (item.kind === "link" || item.kind === "internalLink") return;
+	const frameBounds = (xPt, yPt, widthPt, heightPt) => ({
+		minX: xPt,
+		minY: yPt,
+		maxX: xPt + widthPt,
+		maxY: yPt + heightPt
+	});
+	if (item.kind === "text" || item.kind === "image" || item.kind === "rect" || item.kind === "ellipse") {
+		if (item.kind !== "text" || item.rotationDeg === void 0 || item.rotationDeg % 180 === 0) return frameBounds(item.xPt, item.yPt, item.widthPt ?? 0, item.kind === "text" ? item.sizePt : item.heightPt);
+		const rotationDeg = item.rotationDeg;
+		const anchorX = item.xPt;
+		const anchorY = item.yPt;
+		const rotated = [
+			[0, 0],
+			[item.widthPt ?? 0, 0],
+			[0, item.sizePt],
+			[item.widthPt ?? 0, item.sizePt]
+		].map(([dx, dy]) => require_matrix.rotatePointAboutCenter({
+			x: anchorX + dx,
+			y: anchorY + dy
+		}, {
+			x: anchorX,
+			y: anchorY
+		}, rotationDeg));
+		const xs = rotated.map((point) => point.x);
+		const ys = rotated.map((point) => point.y);
+		return {
+			minX: Math.min(...xs),
+			minY: Math.min(...ys),
+			maxX: Math.max(...xs),
+			maxY: Math.max(...ys)
+		};
+	}
+	if (item.kind === "line") return {
+		minX: Math.min(item.x1Pt, item.x2Pt),
+		minY: Math.min(item.y1Pt, item.y2Pt),
+		maxX: Math.max(item.x1Pt, item.x2Pt),
+		maxY: Math.max(item.y1Pt, item.y2Pt)
+	};
+	let minX = Number.POSITIVE_INFINITY;
+	let minY = Number.POSITIVE_INFINITY;
+	let maxX = Number.NEGATIVE_INFINITY;
+	let maxY = Number.NEGATIVE_INFINITY;
+	const include = (x, y) => {
+		minX = Math.min(minX, x);
+		minY = Math.min(minY, y);
+		maxX = Math.max(maxX, x);
+		maxY = Math.max(maxY, y);
+	};
+	for (const subpath of item.subpaths) {
+		include(subpath.startXPt, subpath.startYPt);
+		for (const segment of subpath.segments) {
+			include(segment.xPt, segment.yPt);
+			if (segment.kind === "cubic") {
+				include(segment.c1xPt, segment.c1yPt);
+				include(segment.c2xPt, segment.c2yPt);
+			}
+		}
+	}
+	return {
+		minX,
+		minY,
+		maxX,
+		maxY
+	};
+}
+function itemIntersectsVisibleRegion(item, widthPt, heightPt) {
+	if (item.kind === "link" || item.kind === "internalLink") return true;
+	const bounds = contentItemBounds(item);
+	if (bounds === void 0) return true;
+	return bounds.maxX >= 0 && bounds.minX <= widthPt && bounds.maxY >= 0 && bounds.minY <= heightPt;
+}
+function pageBoxResidueEntry(pageIndex, page, mediaBox, cropBox) {
+	const declaredBoxes = [
+		["MediaBox", require_objects.dictGet(page, "MediaBox")],
+		["CropBox", require_objects.dictGet(page, "CropBox")],
+		["BleedBox", require_objects.dictGet(page, "BleedBox")],
+		["TrimBox", require_objects.dictGet(page, "TrimBox")],
+		["ArtBox", require_objects.dictGet(page, "ArtBox")]
+	];
+	const cropDeclared = require_objects.dictGet(page, "CropBox") !== void 0;
+	const productionBoxDeclared = declaredBoxes.slice(2).some(([, value]) => value !== void 0);
+	if (!(cropDeclared && (cropBox.llx !== mediaBox.llx || cropBox.lly !== mediaBox.lly || cropBox.urx !== mediaBox.urx || cropBox.ury !== mediaBox.ury)) && !productionBoxDeclared) return;
+	const entries = { Page: require_objects.pdfNum(pageIndex) };
+	for (const [key, value] of declaredBoxes) if (value !== void 0) entries[key] = value;
+	return require_objects.pdfDict(entries);
+}
+function readPage(pageIndex, page, resolver, fontResolver, images, imageIdCache, destinationRegistry, layerNameOf, structureOwnerOf, pageBoxRows, sink) {
 	const resources = resolver.resolveDict(require_objects.dictGet(page, "Resources"));
 	const mediaBox = readMediaBox(page);
+	let cropBox = readDeclaredPageBox(page, "CropBox") ?? mediaBox;
+	if (cropBox.urx - cropBox.llx <= 0 || cropBox.ury - cropBox.lly <= 0) {
+		sink({
+			code: "pdf/invalid-crop-box",
+			severity: "warning",
+			pageIndex,
+			message: "page /CropBox is degenerate (zero width or height); falling back to the /MediaBox as the visible region"
+		});
+		cropBox = mediaBox;
+	}
 	const rotationResult = pageRotationTransform(normalizeRotation(require_objects.asNumber(require_objects.dictGet(page, "Rotate"))), mediaBox.urx - mediaBox.llx, mediaBox.ury - mediaBox.lly);
-	const pageMatrix = require_matrix.multiplyMatrices(require_matrix.translationMatrix(-mediaBox.llx, -mediaBox.lly), rotationResult.matrix);
+	const visibleRect = rotatedRectBounds(cropBox, rotationResult.matrix);
+	const widthPt = visibleRect.maxX - visibleRect.minX;
+	const heightPt = visibleRect.maxY - visibleRect.minY;
+	const pageMatrix = require_matrix.multiplyMatrices(rotationResult.matrix, require_matrix.translationMatrix(-visibleRect.minX, -visibleRect.minY));
+	const boxRow = pageBoxResidueEntry(pageIndex, page, mediaBox, cropBox);
+	if (boxRow !== void 0) pageBoxRows.push(boxRow);
 	const items = [];
 	if (resources !== void 0) {
 		const contentBytes = readPageContentBytes(page, resolver, sink);
 		const extracted = require_interpret.interpretContentStream(contentBytes, resources, {
 			fontMetrics: fontResolver.metrics,
 			resolver,
-			sink
+			sink,
+			layerNameOf
 		});
 		for (const item of extracted) {
 			const converted = convertExtractedItem(item, pageMatrix, fontResolver, images, imageIdCache, resolver, sink);
-			if (converted !== void 0) items.push(converted);
+			if (converted === void 0 || !itemIntersectsVisibleRegion(converted, widthPt, heightPt)) continue;
+			const owner = item.mcid === void 0 ? void 0 : structureOwnerOf(pageIndex, item.mcid);
+			items.push(owner === void 0 ? converted : withStructure(converted, owner));
 		}
 	} else sink({
 		code: "pdf/object-missing-value",
 		severity: "warning",
 		message: "page has no /Resources dict; its content stream cannot be interpreted"
 	});
-	items.push(...readLinkAnnotations(page, pageMatrix, resolver));
+	items.push(...readLinkAnnotations(page, pageMatrix, resolver, destinationRegistry));
 	const notes = readPageNotes(page, resolver);
+	const annotations = require_annotations.readPageAnnotations(page, pageMatrix, resolver, sink);
 	return {
-		widthPt: rotationResult.widthPt,
-		heightPt: rotationResult.heightPt,
+		widthPt,
+		heightPt,
 		items,
-		...notes !== void 0 ? { notes } : {}
+		...notes !== void 0 ? { notes } : {},
+		...annotations.length > 0 ? { annotations } : {}
+	};
+}
+function withStructure(item, structure) {
+	if (item.kind === "link" || item.kind === "internalLink") return item;
+	return {
+		...item,
+		structure
 	};
 }
 function convertExtractedItem(item, pageMatrix, fontResolver, images, imageIdCache, resolver, sink) {
@@ -198,7 +365,10 @@ function convertText(item, pageMatrix, fontResolver) {
 		sizePt: sizePt > 0 ? sizePt : item.sizePt,
 		color: item.color,
 		widthPt,
-		rotationDeg: rotationDeg !== 0 ? rotationDeg : void 0
+		rotationDeg: rotationDeg !== 0 ? rotationDeg : void 0,
+		...item.layerName !== void 0 ? { layer: item.layerName } : {},
+		...item.actualText !== void 0 ? { actualText: item.actualText } : {},
+		...item.alt !== void 0 ? { alt: item.alt } : {}
 	};
 }
 function paintFields(paint) {
@@ -227,14 +397,16 @@ function convertRect(item, pageMatrix) {
 	return {
 		kind: "rect",
 		...transformBox(item, pageMatrix),
-		...paintFields(item)
+		...paintFields(item),
+		...item.layerName !== void 0 ? { layer: item.layerName } : {}
 	};
 }
 function convertEllipse(item, pageMatrix) {
 	return {
 		kind: "ellipse",
 		...transformBox(item, pageMatrix),
-		...paintFields(item)
+		...paintFields(item),
+		...item.layerName !== void 0 ? { layer: item.layerName } : {}
 	};
 }
 function convertLine(item, pageMatrix) {
@@ -253,7 +425,9 @@ function convertLine(item, pageMatrix) {
 		x2Pt: p2.x,
 		y2Pt: p2.y,
 		color: item.color,
-		widthPt: item.widthPt
+		widthPt: item.widthPt,
+		...item.style !== void 0 ? { style: item.style } : {},
+		...item.layerName !== void 0 ? { layer: item.layerName } : {}
 	};
 }
 function transformSubpath(subpath, pageMatrix) {
@@ -308,7 +482,9 @@ function convertPath(item, pageMatrix) {
 		subpaths: item.subpaths.map((subpath) => transformSubpath(subpath, pageMatrix)),
 		...item.fill !== void 0 ? { fill: item.fill } : {},
 		...item.fill !== void 0 && item.fillRule === "evenodd" ? { fillRule: "evenodd" } : {},
-		...item.stroke !== void 0 ? { stroke: item.stroke } : {}
+		...item.stroke !== void 0 ? { stroke: item.stroke } : {},
+		...item.style !== void 0 ? { style: item.style } : {},
+		...item.layerName !== void 0 ? { layer: item.layerName } : {}
 	};
 }
 function imagePlacementFrom(matrix) {
@@ -351,7 +527,8 @@ function convertImage(item, pageMatrix, images, cache, resolver, sink) {
 	return {
 		kind: "image",
 		imageId,
-		...imagePlacementFrom(require_matrix.multiplyMatrices(item.matrix, pageMatrix))
+		...imagePlacementFrom(require_matrix.multiplyMatrices(item.matrix, pageMatrix)),
+		...item.layerName !== void 0 ? { layer: item.layerName } : {}
 	};
 }
 function convertInlineImage(item, pageMatrix, images, resolver, sink) {
@@ -360,19 +537,22 @@ function convertInlineImage(item, pageMatrix, images, resolver, sink) {
 	return {
 		kind: "image",
 		imageId: registerExtractedImage(decoded.format, decoded.bytes, decoded.widthPx, decoded.heightPx, images),
-		...imagePlacementFrom(require_matrix.multiplyMatrices(item.matrix, pageMatrix))
+		...imagePlacementFrom(require_matrix.multiplyMatrices(item.matrix, pageMatrix)),
+		...item.layerName !== void 0 ? { layer: item.layerName } : {}
 	};
 }
-function readLinkAnnotations(page, pageMatrix, resolver) {
+function readLinkAnnotations(page, pageMatrix, resolver, destinationRegistry) {
 	const annotsArr = require_objects.asArray(require_objects.dictGet(page, "Annots"));
 	if (annotsArr === void 0) return [];
 	const links = [];
 	for (const annotRef of annotsArr) {
 		const annot = resolver.resolveDict(annotRef);
 		if (annot === void 0 || require_objects.asName(require_objects.dictGet(annot, "Subtype")) !== "Link") continue;
-		const uri = readLinkUri(annot, resolver);
 		const rectArr = require_objects.asArray(require_objects.dictGet(annot, "Rect"));
-		if (uri === void 0 || rectArr === void 0) continue;
+		if (rectArr === void 0) continue;
+		const uri = readLinkUri(annot, resolver);
+		const destination = uri === void 0 ? readInternalDestination(annot, resolver, destinationRegistry) : void 0;
+		if (uri === void 0 && destination === void 0) continue;
 		const x1 = require_objects.asNumber(rectArr[0]) ?? 0;
 		const y1 = require_objects.asNumber(rectArr[1]) ?? 0;
 		const x2 = require_objects.asNumber(rectArr[2]) ?? 0;
@@ -385,13 +565,24 @@ function readLinkAnnotations(page, pageMatrix, resolver) {
 			x: Math.max(x1, x2),
 			y: Math.max(y1, y2)
 		});
-		links.push({
-			kind: "link",
-			uri,
+		const contentsObj = require_objects.dictGet(annot, "Contents");
+		const title = contentsObj?.kind === "string" ? require_pdf_text.decodePdfString(contentsObj.bytes) : void 0;
+		const box = {
 			xPt: Math.min(p1.x, p2.x),
 			yPt: Math.min(p1.y, p2.y),
 			widthPt: Math.abs(p2.x - p1.x),
-			heightPt: Math.abs(p2.y - p1.y)
+			heightPt: Math.abs(p2.y - p1.y),
+			...title !== void 0 ? { title } : {}
+		};
+		if (uri !== void 0) links.push({
+			kind: "link",
+			uri,
+			...box
+		});
+		else if (destination !== void 0) links.push({
+			kind: "internalLink",
+			destination,
+			...box
 		});
 	}
 	return links;
@@ -400,7 +591,13 @@ function readLinkUri(annot, resolver) {
 	const action = resolver.resolveDict(require_objects.dictGet(annot, "A"));
 	if (action === void 0 || require_objects.asName(require_objects.dictGet(action, "S")) !== "URI") return;
 	const uriObj = require_objects.dictGet(action, "URI");
-	return uriObj?.kind === "string" ? decodePdfString(uriObj.bytes) : void 0;
+	return uriObj?.kind === "string" ? require_pdf_text.decodePdfString(uriObj.bytes) : void 0;
+}
+function readInternalDestination(annot, resolver, destinationRegistry) {
+	const dest = require_objects.dictGet(annot, "Dest");
+	if (dest !== void 0) return destinationRegistry.intern(dest);
+	const action = resolver.resolveDict(require_objects.dictGet(annot, "A"));
+	if (action !== void 0 && require_objects.asName(require_objects.dictGet(action, "S")) === "GoTo") return destinationRegistry.intern(require_objects.dictGet(action, "D"));
 }
 function readPageNotes(page, resolver) {
 	const annotsArr = require_objects.asArray(require_objects.dictGet(page, "Annots"));
@@ -409,49 +606,67 @@ function readPageNotes(page, resolver) {
 		const annot = resolver.resolveDict(annotRef);
 		if (annot === void 0 || require_objects.asName(require_objects.dictGet(annot, "Subtype")) !== "Text") continue;
 		const titleObj = require_objects.dictGet(annot, "T");
-		if ((titleObj?.kind === "string" ? decodePdfString(titleObj.bytes) : void 0) !== "documents.js:notes") continue;
+		if ((titleObj?.kind === "string" ? require_pdf_text.decodePdfString(titleObj.bytes) : void 0) !== "documents.js:notes") continue;
 		const contentsObj = require_objects.dictGet(annot, "Contents");
-		if (contentsObj?.kind === "string") return decodePdfString(contentsObj.bytes);
+		if (contentsObj?.kind === "string") return require_pdf_text.decodePdfString(contentsObj.bytes);
 	}
 }
-function decodePdfString(bytes) {
-	if (bytes.length >= 2 && bytes[0] === 254 && bytes[1] === 255) {
-		let out = "";
-		for (let i = 2; i + 1 < bytes.length; i += 2) out += String.fromCharCode((bytes[i] ?? 0) << 8 | (bytes[i + 1] ?? 0));
-		return out;
-	}
-	return Array.from(bytes, (b) => String.fromCharCode(b)).join("");
-}
-const PDF_DATE_PATTERN = /^D:(\d{4})(\d{2})?(\d{2})?(\d{2})?(\d{2})?(\d{2})?([+\-Z])?(\d{2})?'?(\d{2})?'?$/;
-function parsePdfDate(raw) {
-	if (raw === void 0) return;
-	const match = PDF_DATE_PATTERN.exec(raw);
-	if (match === null) return;
-	const [, year, month = "01", day = "01", hour = "00", minute = "00", second = "00", tzSign, tzHour = "00", tzMinute = "00"] = match;
-	return `${year}-${month}-${day}T${hour}:${minute}:${second}${tzSign === void 0 || tzSign === "Z" ? "Z" : `${tzSign}${tzHour}:${tzMinute}`}`;
-}
-function readMetadata(trailer, resolver) {
+function readMetadata(trailer, resolver, catalog, sink) {
 	const info = resolver.resolveDict(require_objects.dictGet(trailer, "Info"));
 	if (info === void 0) return {};
 	const stringField = (key) => {
 		const obj = require_objects.dictGet(info, key);
-		return obj?.kind === "string" ? decodePdfString(obj.bytes) : void 0;
+		return obj?.kind === "string" ? require_pdf_text.decodePdfString(obj.bytes) : void 0;
 	};
 	const keywords = stringField("Keywords")?.split(",").map((k) => k.trim()).filter((k) => k.length > 0);
+	const langObj = require_objects.dictGet(catalog, "Lang");
+	const xmp = xmpPacket(catalog, resolver, sink);
+	const mirrored = require_xmp.readXmpMetadata(xmp ?? "");
 	return {
-		title: stringField("Title"),
-		author: stringField("Author"),
-		subject: stringField("Subject"),
-		keywords: keywords !== void 0 && keywords.length > 0 ? keywords : void 0,
-		creator: stringField("Creator"),
-		producer: stringField("Producer"),
-		createdIso: parsePdfDate(stringField("CreationDate")),
-		modifiedIso: parsePdfDate(stringField("ModDate"))
+		title: stringField("Title") ?? mirrored.title,
+		author: stringField("Author") ?? mirrored.author,
+		subject: stringField("Subject") ?? mirrored.subject,
+		keywords: (keywords !== void 0 && keywords.length > 0 ? keywords : void 0) ?? mirrored.keywords,
+		creator: stringField("Creator") ?? mirrored.creator,
+		producer: stringField("Producer") ?? mirrored.producer,
+		createdIso: require_pdf_text.parsePdfDate(stringField("CreationDate")) ?? mirrored.createdIso,
+		modifiedIso: require_pdf_text.parsePdfDate(stringField("ModDate")) ?? mirrored.modifiedIso,
+		...langObj?.kind === "string" ? { language: require_pdf_text.decodePdfString(langObj.bytes) } : {}
+	};
+}
+function xmpPacket(catalog, resolver, sink) {
+	const metadataObj = resolver.resolve(require_objects.dictGet(catalog, "Metadata"));
+	if (metadataObj?.kind !== "stream") return;
+	const decoded = require_filters.decodeStream(metadataObj.raw, metadataObj.dict, sink);
+	return new TextDecoder().decode(decoded.bytes);
+}
+function readDocumentResidue(doc, sink) {
+	const source = {};
+	const residue = (key, obj) => {
+		if (obj !== void 0) source[key] = {
+			format: "pdf",
+			xml: require_serialize.serializeObjectToText(obj)
+		};
+	};
+	const catalog = doc.catalog;
+	const packet = xmpPacket(catalog, doc, sink);
+	if (packet !== void 0) source.xmp = {
+		format: "pdf",
+		xml: packet
 	};
+	residue("viewer-preferences", require_objects.dictGet(catalog, "ViewerPreferences"));
+	residue("page-mode", require_objects.dictGet(catalog, "PageMode"));
+	residue("page-layout", require_objects.dictGet(catalog, "PageLayout"));
+	residue("open-action", require_objects.dictGet(catalog, "OpenAction"));
+	const outputIntents = require_objects.asArray(require_objects.dictGet(catalog, "OutputIntents"));
+	if (outputIntents !== void 0) residue("output-intents", require_objects.pdfArray(outputIntents.map((intent) => doc.resolve(intent) ?? require_objects.pdfNull())));
+	residue("piece-info", require_objects.dictGet(catalog, "PieceInfo"));
+	residue("legal", require_objects.dictGet(catalog, "Legal"));
+	residue("collection", require_objects.dictGet(catalog, "Collection"));
+	residue("trailer-id", require_objects.dictGet(doc.trailer, "ID"));
+	return source;
 }
 //#endregion
-exports.decodePdfString = decodePdfString;
 exports.normalizeRotation = normalizeRotation;
 exports.pageRotationTransform = pageRotationTransform;
-exports.parsePdfDate = parsePdfDate;
 exports.readPdf = readPdf;

package/dist/read.d.cts

@@ -1,6 +1,6 @@
 import { PdfDiagnosticSink } from "./diagnostics.cjs";
-import { LayoutDocument } from "./layout.cjs";
 import { r as Matrix } from "./matrix-B7_SBKQ5.cjs";
+import { LayoutDocument } from "./layout.cjs";
 //#region src/read.d.ts
 interface ReadPdfOptions {
   readonly sink?: PdfDiagnosticSink;
@@ -15,7 +15,5 @@ interface PageRotationResult {
   readonly heightPt: number;
 }
 declare function pageRotationTransform(rotation: PageRotation, w: number, h: number): PageRotationResult;
-declare function decodePdfString(bytes: Uint8Array<ArrayBuffer>): string;
-declare function parsePdfDate(raw: string | undefined): string | undefined;
 //#endregion
-export { ReadPdfOptions, decodePdfString, normalizeRotation, pageRotationTransform, parsePdfDate, readPdf };
+export { ReadPdfOptions, normalizeRotation, pageRotationTransform, readPdf };

package/dist/read.d.ts

@@ -1,6 +1,6 @@
 import { PdfDiagnosticSink } from "./diagnostics.js";
-import { LayoutDocument } from "./layout.js";
 import { r as Matrix } from "./matrix-B7_SBKQ5.js";
+import { LayoutDocument } from "./layout.js";
 //#region src/read.d.ts
 interface ReadPdfOptions {
   readonly sink?: PdfDiagnosticSink;
@@ -15,7 +15,5 @@ interface PageRotationResult {
   readonly heightPt: number;
 }
 declare function pageRotationTransform(rotation: PageRotation, w: number, h: number): PageRotationResult;
-declare function decodePdfString(bytes: Uint8Array<ArrayBuffer>): string;
-declare function parsePdfDate(raw: string | undefined): string | undefined;
 //#endregion
-export { ReadPdfOptions, decodePdfString, normalizeRotation, pageRotationTransform, parsePdfDate, readPdf };
+export { ReadPdfOptions, normalizeRotation, pageRotationTransform, readPdf };