pdf-parser.js 5.2.30 → 5.3.0

package/dist/write.d.cts

@@ -1,5 +1,5 @@
 import { LayoutDocument } from "./layout.cjs";
-import { r as EmbeddedFaceSubstitution } from "./embedded-font-DgVBedw6.cjs";
+import { r as EmbeddedFaceSubstitution } from "./embedded-font-CqePfvrl.cjs";
 import { WinAnsiSubstitution } from "./winansi.cjs";
 import { PdfEncryptionOptions } from "./encrypt-write.cjs";
 import { FontRegistry } from "./font-registry.cjs";

package/dist/write.d.ts

@@ -1,5 +1,5 @@
 import { LayoutDocument } from "./layout.js";
-import { r as EmbeddedFaceSubstitution } from "./embedded-font-ByD7ZiQS.js";
+import { r as EmbeddedFaceSubstitution } from "./embedded-font-CQZbTTDM.js";
 import { WinAnsiSubstitution } from "./winansi.js";
 import { PdfEncryptionOptions } from "./encrypt-write.js";
 import { FontRegistry } from "./font-registry.js";

package/dist/write.js

@@ -1,26 +1,19 @@
-import { n as widthOfCode, t as STANDARD_METRICS } from "./afm-widths-B_IGbGky.js";
-import { NOTES_ANNOTATION_AUTHOR } from "./notes-annotation-author.js";
-import { pdfArray, pdfBool, pdfDict, pdfHexString, pdfLiteralString, pdfName, pdfNull, pdfNum, pdfRef, pdfStream } from "./objects.js";
-import { ByteWriter, concatBytes } from "./bytes/writer.js";
+import { n as widthOfCode, t as STANDARD_METRICS } from "./afm-widths-p3UfaXme.js";
+import { pdfArray, pdfDict, pdfHexString, pdfLiteralString, pdfName, pdfNull, pdfNum, pdfRef, pdfStream } from "./objects.js";
+import { ByteWriter } from "./bytes/writer.js";
 import { writeObject } from "./serialize.js";
-import { ByteReader } from "./bytes/reader.js";
 import { deflate } from "./bytes/flate.js";
-import { parseValue } from "./parse.js";
-import { throwIfAborted } from "./util/abort.js";
 import { randomBytes } from "./crypto/random.js";
 import { createStandardEncryptor, encryptIndirectObject } from "./encrypt-write.js";
-import { readJpegInfo as readJpegInfo$1 } from "./image/jpeg-info.js";
-import { encodeCcittFax } from "./image/ccitt-encode.js";
-import { decodePng } from "./image/png-decode.js";
-import { collectEmbeddedGlyphs } from "./embedded-font.js";
-import { writeContentStream } from "./content-write.js";
-import { buildEmbeddedFontObjects } from "./embedded-font-write.js";
+import { prepareImage } from "./write-images.js";
+import { buildInfoDict } from "./write-strings.js";
+import { restoreResidueRow, xrefEntry } from "./write-annotations.js";
+import { emitFormObjects } from "./write-form.js";
+import { emitFontObjects } from "./write-font-objects.js";
 import { resolveFaceWithRegistry } from "./font-registry.js";
-import { collectUsedGlyphs, writeFormulaContentStream } from "./math-content-write.js";
-import { loadMathFont } from "./math-font.js";
-import { buildMathFontObjects } from "./math-font-write.js";
 import { createFontMeasurer } from "./measure.js";
-import { subsetSfnt } from "./sfnt-subset.js";
+import { emitPageObjects } from "./write-page-emit.js";
+import { emitOutlineObjects } from "./write-outline.js";
 import { base64ToBytes } from "byte-codec";
 //#region src/write.ts
 const MATH_FONT_RESOURCE_NAME = "MF";
@@ -35,43 +28,6 @@ const FLAG_NONSYMBOLIC = 32;
 const FLAG_ITALIC = 64;
 const FLAG_FORCE_BOLD = 262144;
 const FILE_ID_BYTES = 16;
-const BITS_PER_BYTE = 8;
-const BYTE_SHIFT = Math.log2(BITS_PER_BYTE);
-const BYTE_MASK = 255;
-const PACKED_BIT_MSB_MASK = 128;
-const BIT_INDEX_MASK = 7;
-function textToPdfString(text) {
-	const bytes = new Uint8Array(2 + text.length * 2);
-	bytes[0] = 254;
-	bytes[1] = 255;
-	let offset = 2;
-	for (const unit of text.split("")) {
-		const code = unit.charCodeAt(0);
-		bytes[offset] = code >> BITS_PER_BYTE & BYTE_MASK;
-		bytes[offset + 1] = code & BYTE_MASK;
-		offset += 2;
-	}
-	return pdfHexString(bytes);
-}
-function pad2(n) {
-	return n.toString().padStart(2, "0");
-}
-function formatPdfDate(iso) {
-	const date = new Date(iso);
-	return `D:${date.getUTCFullYear()}${pad2(date.getUTCMonth() + 1)}${pad2(date.getUTCDate())}${pad2(date.getUTCHours())}${pad2(date.getUTCMinutes())}${pad2(date.getUTCSeconds())}Z`;
-}
-function buildInfoDict(doc) {
-	const entries = /* @__PURE__ */ new Map();
-	entries.set("Producer", textToPdfString("documents.js"));
-	if (doc.metadata.title !== void 0) entries.set("Title", textToPdfString(doc.metadata.title));
-	if (doc.metadata.author !== void 0) entries.set("Author", textToPdfString(doc.metadata.author));
-	if (doc.metadata.subject !== void 0) entries.set("Subject", textToPdfString(doc.metadata.subject));
-	if (doc.metadata.keywords !== void 0) entries.set("Keywords", textToPdfString(doc.metadata.keywords.join(", ")));
-	if (doc.metadata.creator !== void 0) entries.set("Creator", textToPdfString(doc.metadata.creator));
-	if (doc.metadata.createdIso !== void 0) entries.set("CreationDate", textToPdfString(formatPdfDate(doc.metadata.createdIso)));
-	if (doc.metadata.modifiedIso !== void 0) entries.set("ModDate", textToPdfString(formatPdfDate(doc.metadata.modifiedIso)));
-	return pdfDict(entries);
-}
 function computeFontFlags(standardName, metrics) {
 	let flags = FLAG_NONSYMBOLIC;
 	if (standardName.startsWith("Courier")) flags |= FLAG_FIXED_PITCH;
@@ -112,245 +68,6 @@ function buildFontObjects(standardName, descriptorRef) {
 function codePointsOf(texts) {
 	return new Set(texts.flatMap((text) => [...text].map((character) => character.codePointAt(0))));
 }
-const JPEG_COMPONENTS_CMYK = 4;
-function prepareJpegImage(bytes) {
-	const info = readJpegInfo$1(bytes);
-	const colorSpace = info.components === 1 ? "DeviceGray" : info.components === JPEG_COMPONENTS_CMYK ? "DeviceCMYK" : "DeviceRGB";
-	const entries = /* @__PURE__ */ new Map([
-		["Type", pdfName("XObject")],
-		["Subtype", pdfName("Image")],
-		["Width", pdfNum(info.width)],
-		["Height", pdfNum(info.height)],
-		["ColorSpace", pdfName(colorSpace)],
-		["BitsPerComponent", pdfNum(info.precision)],
-		["Filter", pdfName("DCTDecode")]
-	]);
-	if (info.components === JPEG_COMPONENTS_CMYK && (info.adobeTransform === 2 || info.adobeTransform === void 0)) entries.set("Decode", pdfArray([
-		1,
-		0,
-		1,
-		0,
-		1,
-		0,
-		1,
-		0
-	].map((n) => pdfNum(n))));
-	return {
-		dict: pdfDict(entries),
-		raw: bytes
-	};
-}
-function pngImageDict(width, height, colorSpace, compress) {
-	const entries = /* @__PURE__ */ new Map([
-		["Type", pdfName("XObject")],
-		["Subtype", pdfName("Image")],
-		["Width", pdfNum(width)],
-		["Height", pdfNum(height)],
-		["ColorSpace", pdfName(colorSpace)],
-		["BitsPerComponent", pdfNum(BITS_PER_BYTE)]
-	]);
-	if (compress) entries.set("Filter", pdfName("FlateDecode"));
-	return pdfDict(entries);
-}
-function packBilevel(raw) {
-	const rowBytes = raw.width + BIT_INDEX_MASK >> BYTE_SHIFT;
-	const packed = new Uint8Array(rowBytes * raw.height);
-	const view = new DataView(packed.buffer);
-	for (let index = 0; index < raw.data.length; index++) {
-		const sample = raw.data[index];
-		if (sample !== 0 && sample !== BYTE_MASK) return;
-		if (sample === BYTE_MASK) {
-			const x = index % raw.width;
-			const byteIndex = Math.floor(index / raw.width) * rowBytes + (x >> BYTE_SHIFT);
-			view.setUint8(byteIndex, view.getUint8(byteIndex) | PACKED_BIT_MSB_MASK >> (x & BIT_INDEX_MASK));
-		}
-	}
-	return packed;
-}
-function preparePngImage(bytes, compress) {
-	const raw = decodePng(bytes);
-	const colorSpace = raw.channels === 1 ? "DeviceGray" : "DeviceRGB";
-	if (compress && raw.channels === 1 && raw.alpha === void 0) {
-		const bilevel = packBilevel(raw);
-		if (bilevel !== void 0) {
-			const flate = deflate(raw.data);
-			const g4 = encodeCcittFax(bilevel, {
-				columns: raw.width,
-				rows: raw.height,
-				maxBytes: flate.length
-			});
-			if (g4 !== void 0) return {
-				dict: pdfDict(/* @__PURE__ */ new Map([
-					["Type", pdfName("XObject")],
-					["Subtype", pdfName("Image")],
-					["Width", pdfNum(raw.width)],
-					["Height", pdfNum(raw.height)],
-					["ColorSpace", pdfName("DeviceGray")],
-					["BitsPerComponent", pdfNum(1)],
-					["Filter", pdfName("CCITTFaxDecode")],
-					["DecodeParms", pdfDict(/* @__PURE__ */ new Map([
-						["K", pdfNum(-1)],
-						["Columns", pdfNum(raw.width)],
-						["Rows", pdfNum(raw.height)],
-						["BlackIs1", pdfBool(false)]
-					]))]
-				])),
-				raw: g4,
-				alpha: void 0
-			};
-			return {
-				dict: pngImageDict(raw.width, raw.height, colorSpace, true),
-				raw: flate,
-				alpha: void 0
-			};
-		}
-	}
-	return {
-		dict: pngImageDict(raw.width, raw.height, colorSpace, compress),
-		raw: compress ? deflate(raw.data) : raw.data,
-		alpha: raw.alpha === void 0 ? void 0 : {
-			dict: pngImageDict(raw.width, raw.height, "DeviceGray", compress),
-			raw: compress ? deflate(raw.alpha) : raw.alpha
-		}
-	};
-}
-function preparePassthroughImage(asset, original, compress) {
-	const png = preparePngImage(base64ToBytes(asset.base64), compress);
-	const entries = /* @__PURE__ */ new Map([
-		["Type", pdfName("XObject")],
-		["Subtype", pdfName("Image")],
-		["Width", pdfNum(asset.widthPx)],
-		["Height", pdfNum(asset.heightPx)],
-		["Filter", pdfName(original.filter === "jbig2" ? "JBIG2Decode" : "JPXDecode")]
-	]);
-	if (original.filter === "jbig2") {
-		entries.set("ColorSpace", pdfName("DeviceGray"));
-		entries.set("BitsPerComponent", pdfNum(1));
-	}
-	return {
-		dict: pdfDict(entries),
-		raw: base64ToBytes(original.base64),
-		alpha: png.alpha
-	};
-}
-function prepareImage(asset, compress) {
-	if (asset.original !== void 0) return preparePassthroughImage(asset, asset.original, compress);
-	const bytes = base64ToBytes(asset.base64);
-	return asset.format === "jpeg" ? prepareJpegImage(bytes) : preparePngImage(bytes, compress);
-}
-function buildLinkAnnotDict(link) {
-	return pdfDict({
-		Type: pdfName("Annot"),
-		Subtype: pdfName("Link"),
-		Rect: pdfArray([
-			link.xPt,
-			link.yPt,
-			link.xPt + link.widthPt,
-			link.yPt + link.heightPt
-		].map((n) => pdfNum(n))),
-		Border: pdfArray([
-			0,
-			0,
-			0
-		].map((n) => pdfNum(n))),
-		A: pdfDict({
-			Type: pdfName("Action"),
-			S: pdfName("URI"),
-			URI: pdfHexString(new TextEncoder().encode(link.uri))
-		})
-	});
-}
-function destinationViewArray(target) {
-	const n = (value) => value === void 0 ? pdfNull() : pdfNum(value);
-	if (target.kind === "xyz") return [
-		pdfName("XYZ"),
-		n(target.leftPt),
-		n(target.topPt),
-		n(target.zoom)
-	];
-	if (target.kind === "fitH") return [pdfName("FitH"), n(target.topPt)];
-	if (target.kind === "fitV") return [pdfName("FitV"), n(target.leftPt)];
-	if (target.kind === "fitR") return [
-		pdfName("FitR"),
-		n(target.leftPt),
-		n(target.bottomPt),
-		n(target.rightPt),
-		n(target.topPt)
-	];
-	if (target.kind === "fitBH") return [pdfName("FitBH"), n(target.topPt)];
-	if (target.kind === "fitBV") return [pdfName("FitBV"), n(target.leftPt)];
-	return [pdfName(target.kind === "fitB" ? "FitB" : "Fit")];
-}
-function resolveDestinationArray(doc, pageAllocs, name, what) {
-	const destination = doc.destinations?.find((d) => d.name === name);
-	if (destination === void 0) throw new Error(`${what} names destination "${name}", which the document's destinations table does not carry — this is a caller-invariant violation`);
-	const targetPage = pageAllocs[destination.pageIndex];
-	if (targetPage === void 0) throw new Error(`destination "${destination.name}" names page index ${destination.pageIndex}, which is beyond the document's own pages — this is a caller-invariant violation`);
-	return [pdfRef(targetPage.pageNum, 0), ...destinationViewArray(destination.target)];
-}
-function buildInternalLinkAnnotDict(link, doc, pageAllocs) {
-	return pdfDict({
-		Type: pdfName("Annot"),
-		Subtype: pdfName("Link"),
-		Rect: pdfArray([
-			link.xPt,
-			link.yPt,
-			link.xPt + link.widthPt,
-			link.yPt + link.heightPt
-		].map((v) => pdfNum(v))),
-		Border: pdfArray([
-			0,
-			0,
-			0
-		].map((v) => pdfNum(v))),
-		Dest: pdfArray(resolveDestinationArray(doc, pageAllocs, link.destination, "internal link"))
-	});
-}
-function isLinkItem(item) {
-	return item.kind === "link";
-}
-function isInternalLinkItem(item) {
-	return item.kind === "internalLink";
-}
-const NOTES_ANNOTATION_HIDDEN_FLAG = 2;
-function buildNotesAnnotDict(notes) {
-	return pdfDict({
-		Type: pdfName("Annot"),
-		Subtype: pdfName("Text"),
-		Rect: pdfArray([
-			0,
-			0,
-			0,
-			0
-		].map((n) => pdfNum(n))),
-		Contents: textToPdfString(notes),
-		T: textToPdfString(NOTES_ANNOTATION_AUTHOR),
-		F: pdfNum(NOTES_ANNOTATION_HIDDEN_FLAG)
-	});
-}
-const XREF_OFFSET_DIGITS = 10;
-const XREF_GENERATION_DIGITS = 5;
-function xrefEntry(offset, generation, inUse) {
-	return `${offset.toString().padStart(XREF_OFFSET_DIGITS, "0")} ${generation.toString().padStart(XREF_GENERATION_DIGITS, "0")} ${inUse ? "n" : "f"} \n`;
-}
-function parseResidueRow(residue) {
-	const reader = new ByteReader(new TextEncoder().encode(residue.xml));
-	return parseValue(reader, () => void 0);
-}
-function objectContainsReference(obj) {
-	if (obj.kind === "ref") return true;
-	if (obj.kind === "array") return obj.items.some(objectContainsReference);
-	if (obj.kind === "dict") return [...obj.entries.values()].some(objectContainsReference);
-	if (obj.kind === "stream") return [...obj.dict.entries.values()].some(objectContainsReference);
-	return false;
-}
-function restoreResidueRow(source, key) {
-	const row = source?.[key];
-	if (row === void 0) return;
-	const parsed = parseResidueRow(row);
-	if (parsed === void 0 || objectContainsReference(parsed)) return;
-	return parsed;
-}
 function writePdf(doc, options = {}) {
 	const compress = options.compress ?? true;
 	const registry = options.fonts;
@@ -555,43 +272,13 @@ function writePdf(doc, options = {}) {
 			value: pdfDict({ EmbeddedFiles: pdfDict({ Names: pdfArray(names) }) })
 		});
 	}
-	if (outlineRootNum !== void 0) {
-		let itemCursor = 0;
-		const emitItems = (items, parentNum) => {
-			const siblingNums = [];
-			let prevNum;
-			for (const item of items) {
-				const ownNum = outlineItemNums[itemCursor];
-				itemCursor += 1;
-				siblingNums.push(ownNum);
-				const childNums = emitItems(item.children, ownNum);
-				const entries = [["Title", pdfLiteralString(new TextEncoder().encode(item.title))], ["Parent", pdfRef(parentNum, 0)]];
-				if (item.destination !== void 0) entries.push(["Dest", pdfArray(resolveDestinationArray(doc, pageAllocs, item.destination, `outline item "${item.title}"`))]);
-				if (prevNum !== void 0) entries.push(["Prev", pdfRef(prevNum, 0)]);
-				if (siblingNums.length < items.length) entries.push(["Next", pdfRef(outlineItemNums[itemCursor], 0)]);
-				if (childNums.length > 0) {
-					entries.push(["First", pdfRef(childNums[0], 0)]);
-					entries.push(["Last", pdfRef(childNums[childNums.length - 1], 0)]);
-					entries.push(["Count", pdfNum(childNums.length)]);
-				}
-				objects.push({
-					num: ownNum,
-					value: pdfDict(Object.fromEntries(entries))
-				});
-				prevNum = ownNum;
-			}
-			return siblingNums;
-		};
-		const topLevelNums = emitItems(doc.outline ?? [], outlineRootNum);
-		objects.push({
-			num: outlineRootNum,
-			value: pdfDict({
-				Type: pdfName("Outlines"),
-				First: pdfRef(topLevelNums[0], 0),
-				Last: pdfRef(topLevelNums[topLevelNums.length - 1], 0)
-			})
-		});
-	}
+	emitOutlineObjects({
+		objects,
+		outlineItemNums,
+		outlineRootNum,
+		doc,
+		pageAllocs
+	});
 	for (const layer of doc.layers ?? []) {
 		const num = layerNumByName.get(layer.name);
 		objects.push({
@@ -603,85 +290,13 @@ function writePdf(doc, options = {}) {
 		});
 	}
 	const widgetAnnotsByPage = /* @__PURE__ */ new Map();
-	const noteWidgetAnnot = (widget, ref) => {
-		const existing = widgetAnnotsByPage.get(widget.pageIndex);
-		if (existing === void 0) widgetAnnotsByPage.set(widget.pageIndex, [ref]);
-		else existing.push(ref);
-	};
-	const widgetRectArray = (widget) => pdfArray([
-		widget.xPt,
-		widget.yPt,
-		widget.xPt + widget.widthPt,
-		widget.yPt + widget.heightPt
-	].map((n) => pdfNum(n)));
-	const widgetDict = (widget) => pdfDict({
-		Subtype: pdfName("Widget"),
-		Rect: widgetRectArray(widget),
-		P: pdfRef(pageAllocs[widget.pageIndex].pageNum, 0)
-	});
-	const FIELD_TYPE_PDF_NAME = {
-		text: "Tx",
-		checkbox: "Btn",
-		radio: "Btn",
-		button: "Btn",
-		listbox: "Ch",
-		combobox: "Ch",
-		signature: "Sig"
-	};
-	const emitFormFieldObjects = (fields, parentName) => {
-		for (const field of fields) {
-			const ownName = parentName === void 0 ? field.name : field.name.startsWith(`${parentName}.`) ? field.name.slice(parentName.length + 1) : field.name;
-			const entries = [];
-			if (ownName.length > 0) entries.push(["T", pdfLiteralString(new TextEncoder().encode(ownName))]);
-			if (field.alias !== void 0) entries.push(["TU", pdfLiteralString(new TextEncoder().encode(field.alias))]);
-			if (field.fieldType === "group") entries.push(["Kids", pdfArray(field.children.map((child) => pdfRef(formNumOf(child), 0)))]);
-			else {
-				entries.push(["FT", pdfName(FIELD_TYPE_PDF_NAME[field.fieldType])]);
-				const FLAG_READ_ONLY = 1;
-				const FLAG_PUSHBUTTON = 4;
-				const FLAG_RADIO = 32768;
-				const FLAG_COMBO = 131072;
-				let flags = 0;
-				if (field.readOnly === true) flags |= FLAG_READ_ONLY;
-				if (field.fieldType === "button") flags |= FLAG_PUSHBUTTON;
-				if (field.fieldType === "radio") flags |= FLAG_RADIO;
-				if (field.fieldType === "combobox") flags |= FLAG_COMBO;
-				if (flags !== 0) entries.push(["Ff", pdfNum(flags)]);
-				if (field.fieldType === "text" || field.fieldType === "listbox" || field.fieldType === "combobox") {
-					if (field.value !== void 0) entries.push(["V", pdfLiteralString(new TextEncoder().encode(field.value))]);
-				} else if (field.fieldType === "checkbox" || field.fieldType === "radio") entries.push(["V", pdfName(field.value ?? (field.checked === true ? "Yes" : "Off"))]);
-				if (field.options !== void 0) entries.push(["Opt", pdfArray(field.options.map((option) => pdfLiteralString(new TextEncoder().encode(option))))]);
-				const firstWidget = field.widgets[0];
-				if (field.widgets.length === 1 && firstWidget !== void 0) {
-					entries.push(["Subtype", pdfName("Widget")]);
-					entries.push(["Rect", widgetRectArray(firstWidget)]);
-					entries.push(["P", pdfRef(pageAllocs[firstWidget.pageIndex].pageNum, 0)]);
-					noteWidgetAnnot(firstWidget, pdfRef(formNumOf(field), 0));
-				} else if (formExtraWidgetNums.has(field)) {
-					const extraNums = formExtraWidgetNums.get(field) ?? [];
-					entries.push(["Kids", pdfArray(extraNums.map((num) => pdfRef(num, 0)))]);
-					for (const [index, num] of extraNums.entries()) {
-						const widget = field.widgets[index];
-						if (widget !== void 0) noteWidgetAnnot(widget, pdfRef(num, 0));
-					}
-				}
-			}
-			objects.push({
-				num: formNumOf(field),
-				value: pdfDict(Object.fromEntries(entries))
-			});
-			const extraNums = formExtraWidgetNums.get(field) ?? [];
-			for (const [index, num] of extraNums.entries()) {
-				const widget = field.widgets[index];
-				if (widget !== void 0) objects.push({
-					num,
-					value: widgetDict(widget)
-				});
-			}
-			emitFormFieldObjects(field.children, field.name);
-		}
-	};
-	emitFormFieldObjects(doc.form ?? [], void 0);
+	emitFormObjects({
+		objects,
+		widgetAnnotsByPage,
+		pageAllocs,
+		formNumOf,
+		formExtraWidgetNums
+	}, doc.form ?? []);
 	if (structRootNum !== void 0) {
 		const emitStructureElement = (element, parentNum) => {
 			const entries = [
@@ -747,67 +362,14 @@ function writePdf(doc, options = {}) {
 			value: pdfStream(alloc.prepared.dict, alloc.prepared.raw)
 		});
 	}
-	const mathFont = mathFontAlloc === void 0 ? void 0 : loadMathFont().font;
-	const usedGlyphs = mathFontAlloc === void 0 || mathFont === void 0 ? void 0 : collectUsedGlyphs(formulas, mathFont);
-	if (mathFontAlloc !== void 0 && mathFont !== void 0 && usedGlyphs !== void 0) {
-		const built = buildMathFontObjects(mathFont, usedGlyphs, {
-			cidFontRef: pdfRef(mathFontAlloc.cidFontNum, 0),
-			descriptorRef: pdfRef(mathFontAlloc.descriptorNum, 0),
-			fontFileRef: pdfRef(mathFontAlloc.fontFileNum, 0),
-			toUnicodeRef: pdfRef(mathFontAlloc.toUnicodeNum, 0)
-		}, compress);
-		objects.push({
-			num: mathFontAlloc.type0Num,
-			value: built.type0
-		});
-		objects.push({
-			num: mathFontAlloc.cidFontNum,
-			value: built.cidFont
-		});
-		objects.push({
-			num: mathFontAlloc.descriptorNum,
-			value: built.descriptor
-		});
-		objects.push({
-			num: mathFontAlloc.fontFileNum,
-			value: built.fontFile
-		});
-		objects.push({
-			num: mathFontAlloc.toUnicodeNum,
-			value: built.toUnicode
-		});
-	}
-	for (const [face, alloc] of embeddedAllocs) {
-		const faceUsedGlyphs = collectEmbeddedGlyphs(alloc.texts, face);
-		const subset = subsetSfnt(face.font, [...alloc.codePoints], [...faceUsedGlyphs.keys()]);
-		if (subset === void 0) throw new Error(`font "${face.postScriptName}" resolved to an embeddable face, but its glyph outlines could not be subsetted — only TrueType-outline ('glyf') fonts can be embedded, so supply a TrueType face for this family or drop it from the registry`);
-		const built = buildEmbeddedFontObjects(face, subset, faceUsedGlyphs, {
-			cidFontRef: pdfRef(alloc.cidFontNum, 0),
-			descriptorRef: pdfRef(alloc.descriptorNum, 0),
-			fontFileRef: pdfRef(alloc.fontFileNum, 0),
-			toUnicodeRef: pdfRef(alloc.toUnicodeNum, 0)
-		}, compress);
-		objects.push({
-			num: alloc.type0Num,
-			value: built.type0
-		});
-		objects.push({
-			num: alloc.cidFontNum,
-			value: built.cidFont
-		});
-		objects.push({
-			num: alloc.descriptorNum,
-			value: built.descriptor
-		});
-		objects.push({
-			num: alloc.fontFileNum,
-			value: built.fontFile
-		});
-		objects.push({
-			num: alloc.toUnicodeNum,
-			value: built.toUnicode
-		});
-	}
+	const mathFontResult = emitFontObjects({
+		objects,
+		formulas,
+		mathFontAlloc,
+		embeddedAllocs,
+		compress
+	});
+	const mathFont = mathFontResult === void 0 ? void 0 : mathFontResult.font;
 	const resourceEntries = /* @__PURE__ */ new Map();
 	if (fontAllocs.size > 0 || embeddedAllocs.size > 0 || mathFontAlloc !== void 0) {
 		const fontEntries = new Map([...fontAllocs.values()].map((alloc) => [alloc.resourceName, pdfRef(alloc.fontNum, 0)]));
@@ -817,86 +379,24 @@ function writePdf(doc, options = {}) {
 	}
 	if (imageAllocs.size > 0) resourceEntries.set("XObject", pdfDict(new Map([...imageAllocs.values()].map((alloc) => [alloc.resourceName, pdfRef(alloc.imageNum, 0)]))));
 	const resourcesDict = pdfDict(resourceEntries);
-	const markedStructureByPage = /* @__PURE__ */ new Map();
-	const formulasByPage = /* @__PURE__ */ new Map();
-	for (const formula of formulas) {
-		const forPage = formulasByPage.get(formula.pageIndex);
-		if (forPage === void 0) formulasByPage.set(formula.pageIndex, [formula]);
-		else forPage.push(formula);
-	}
-	const context = {
+	const markedStructureByPage = emitPageObjects({
+		doc,
+		options,
+		objects,
+		compress,
 		measurer,
-		resolveFont: (font) => {
-			const resolved = resolveFaceWithRegistry(registry, font);
-			if (resolved.kind === "embedded") {
-				const alloc = embeddedAllocs.get(resolved.face);
-				if (alloc === void 0) throw new Error(`embedded font "${resolved.face.postScriptName}" was not pre-allocated — this is a writePdf internal invariant violation`);
-				return {
-					kind: "embedded",
-					resourceName: alloc.resourceName,
-					face: resolved.face
-				};
-			}
-			const alloc = fontAllocs.get(resolved.standardName);
-			if (alloc === void 0) throw new Error(`font "${resolved.standardName}" was not pre-allocated — this is a writePdf internal invariant violation`);
-			return {
-				kind: "standard",
-				resourceName: alloc.resourceName,
-				standardName: resolved.standardName
-			};
-		},
-		resolveImage: (imageId) => {
-			const alloc = imageAllocs.get(imageId);
-			if (alloc === void 0) throw new Error(`image "${imageId}" was not pre-allocated — this is a writePdf internal invariant violation`);
-			return { resourceName: alloc.resourceName };
-		}
-	};
-	doc.pages.forEach((page, pageIndex) => {
-		throwIfAborted(options.signal);
-		const { pageNum, contentsNum } = pageAllocs[pageIndex];
-		let pageMcid = 0;
-		const pageContext = {
-			...context,
-			nextMcid: () => pageMcid++,
-			layerObjectNumberOf: (name) => layerNumByName.get(name)
-		};
-		const { bytes: contentBytes, substitutions, missingGlyphs, markedStructure } = writeContentStream(page.items, pageContext);
-		if (markedStructure.length > 0) markedStructureByPage.set(pageIndex, [...markedStructure]);
-		for (const substitution of substitutions) options.onSubstitution?.(substitution, { pageIndex });
-		for (const missing of missingGlyphs) options.onMissingGlyph?.(missing, { pageIndex });
-		const pageFormulas = formulasByPage.get(pageIndex);
-		const formulaBytes = pageFormulas === void 0 || mathFontAlloc === void 0 || mathFont === void 0 ? void 0 : writeFormulaContentStream(pageFormulas, {
-			font: mathFont,
-			resourceName: mathFontAlloc.resourceName
-		});
-		const combinedContentBytes = formulaBytes === void 0 ? contentBytes : concatBytes([contentBytes, formulaBytes]);
-		const finalContentBytes = compress ? deflate(combinedContentBytes) : combinedContentBytes;
-		const contentsDict = pdfDict(compress ? { Filter: pdfName("FlateDecode") } : {});
-		objects.push({
-			num: contentsNum,
-			value: pdfStream(contentsDict, finalContentBytes)
-		});
-		const annots = [...page.items.filter(isLinkItem).map((link) => buildLinkAnnotDict(link)), ...page.items.filter(isInternalLinkItem).map((link) => buildInternalLinkAnnotDict(link, doc, pageAllocs))];
-		if (page.notes !== void 0 && page.notes.length > 0) annots.push(buildNotesAnnotDict(page.notes));
-		annots.push(...widgetAnnotsByPage.get(pageIndex) ?? []);
-		const pageEntries = /* @__PURE__ */ new Map([
-			["Type", pdfName("Page")],
-			["Parent", pdfRef(pagesNum, 0)],
-			["MediaBox", pdfArray([
-				0,
-				0,
-				page.widthPt,
-				page.heightPt
-			].map((n) => pdfNum(n)))],
-			["Resources", resourcesDict],
-			["Contents", pdfRef(contentsNum, 0)]
-		]);
-		if (annots.length > 0) pageEntries.set("Annots", pdfArray(annots));
-		if (markedStructureByPage.has(pageIndex)) pageEntries.set("StructParents", pdfNum(pageIndex));
-		objects.push({
-			num: pageNum,
-			value: pdfDict(pageEntries)
-		});
+		registry,
+		fontAllocs,
+		embeddedAllocs,
+		imageAllocs,
+		layerNumByName,
+		formulas,
+		mathFontAlloc,
+		mathFont,
+		pageAllocs,
+		pagesNum,
+		resourcesDict,
+		widgetAnnotsByPage
 	});
 	if (structRootNum !== void 0 && structParentTreeNum !== void 0) {
 		const nums = [];

package/dist/xref.d.cts

@@ -1,4 +1,4 @@
-import { PdfDiagnosticSink } from "./diagnostics.cjs";
+import { i as PdfDiagnosticSink } from "./diagnostics-C9FxfZ6H.cjs";
 import { n as PdfDict } from "./objects-e2RS3-yT.cjs";
 //#region src/xref.d.ts
 interface XrefOffsetEntry {

package/dist/xref.d.ts

@@ -1,4 +1,4 @@
-import { PdfDiagnosticSink } from "./diagnostics.js";
+import { i as PdfDiagnosticSink } from "./diagnostics-C9FxfZ6H.js";
 import { n as PdfDict } from "./objects-e2RS3-yT.js";
 //#region src/xref.d.ts
 interface XrefOffsetEntry {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "pdf-parser.js",
-  "version": "5.2.30",
+  "version": "5.3.0",
   "description": "Hand-written, dependency-minimal PDF codec: parses arbitrary real-world PDFs and generates new ones, built on its own codec-owned LayoutDocument item model and Zod 4 codecs.",
   "type": "module",
   "repository": {