@segnavia/render 0.12.0 → 0.13.0

package/dist/book-pdf.js

@@ -1,12 +1,13 @@
 import { marginDocument } from "./book-margins.js";
 import { RenderRefusedError } from "./errors.js";
 import { namedDestinations, rewriteBookLinks } from "./book-destinations.js";
+import { loadPdf } from "./pdf-load.js";
 import { Book, formatNumber, resolveDirection, sheetMm } from "@segnavia/format";
-import { PDFDocument } from "pdf-lib";
 import { bookDocument, bookPrintPart } from "@segnavia/html";
+import { PDFDocument } from "pdf-lib";
 //#region src/book-pdf.ts
 async function inspect(output) {
-	const document = await PDFDocument.load(output.bytes);
+	const document = await loadPdf(output.bytes);
 	return {
 		bytes: output.bytes,
 		overflow: output.overflow,
@@ -51,7 +52,7 @@ function verifyStarts(label, sections, expected, actual) {
 }
 async function appendBlank(part) {
 	if (!part.bytes) return part;
-	const document = await PDFDocument.load(part.bytes);
+	const document = await loadPdf(part.bytes);
 	const last = document.getPage(document.getPageCount() - 1);
 	document.addPage([last.getWidth(), last.getHeight()]);
 	return {
@@ -193,10 +194,10 @@ async function join(front, body, marginBytes) {
 	const output = await PDFDocument.create();
 	for (const part of [front, body]) {
 		if (!part.bytes) continue;
-		const source = await PDFDocument.load(part.bytes);
+		const source = await loadPdf(part.bytes);
 		for (const page of await output.copyPages(source, source.getPageIndices())) output.addPage(page);
 	}
-	const layer = await PDFDocument.load(marginBytes);
+	const layer = await loadPdf(marginBytes);
 	if (layer.getPageCount() !== output.getPageCount()) throw new RenderRefusedError(`The margin layer has ${layer.getPageCount()} pages, but the book has ${output.getPageCount()}.`);
 	for (const [index, page] of layer.getPages().entries()) {
 		const [embedded] = await output.embedPdf(layer, [index]);

package/dist/deck-pdf.js

@@ -1,11 +1,11 @@
 import { RenderRefusedError } from "./errors.js";
+import { loadPdf } from "./pdf-load.js";
 import { containFixedPageOverflow } from "./overflow-containment.js";
 import { pdfPixelBox } from "./pdf-geometry.js";
 import { composeImposedPdf, drawClippedPage, embedOnePage } from "./pdf-compose.js";
 import { Imposition, fitPage, impose, pagePlacementLayout } from "@segnavia/format";
 import { newNonce } from "@segnavia/frame/bridge";
 import { htmlSlideDocument, notesLayoutGeometry, paginateNotes, printableDeck } from "@segnavia/frame/print-document";
-import { PDFDocument } from "pdf-lib";
 import { fitImage, imageLayerStyles } from "@segnavia/frame/image-fit";
 //#region src/deck-pdf.ts
 function nameSlideError(error, id) {
@@ -118,7 +118,7 @@ async function regularDeckPdf(deck, opts, services) {
 			nameSlideError(error, id);
 		}
 	}))]);
-	const output = await PDFDocument.load(printable);
+	const output = await loadPdf(printable);
 	const pages = output.getPages();
 	const slideBox = notesGeometry?.slide ?? {
 		x: 0,

package/dist/effects/adapter.js

@@ -1,4 +1,5 @@
 import { RenderRefusedError } from "../errors.js";
+import { loadPdf } from "../pdf-load.js";
 import { IDENTITY_MATRIX, multiplyAffine } from "../press/affine.js";
 import { visitContentOperations } from "../press/content.js";
 import { decodeContent, dictionaryOf, nameOf, resolvedColorSpace } from "../press/pdf-objects.js";
@@ -8,7 +9,7 @@ import { softMaskAction } from "./policy.js";
 import { rule1EffectImage } from "./rule-1.js";
 import { EMPTY_EFFECT_GRAPHICS_STATE, advanceEffectGraphicsState, effectFormStructure, effectGraphicsStateOpacity, effectGraphicsStateStructure, effectOpacity } from "./structure.js";
 import { buildRuleCImage, ruleCPlan } from "./rule-c.js";
-import { PDFArray, PDFDict, PDFDocument, PDFName, PDFNumber, PDFRawStream, PDFRef } from "pdf-lib";
+import { PDFArray, PDFDict, PDFName, PDFNumber, PDFRawStream, PDFRef } from "pdf-lib";
 import { deflateSync } from "node:zlib";
 //#region src/effects/adapter.ts
 const key = (name) => PDFName.of(name);
@@ -321,7 +322,7 @@ function rewriteEffectsDocument(document) {
 }
 async function rewriteEffectsPdf(bytes) {
 	try {
-		const document = await PDFDocument.load(bytes, { updateMetadata: false });
+		const document = await loadPdf(bytes, { updateMetadata: false });
 		const report = rewriteEffectsDocument(document);
 		return {
 			bytes: report.rule1 === 0 && report.ruleC === 0 ? bytes : await document.save(),

package/dist/index.js

@@ -1,4 +1,5 @@
 import { RenderRefusedError, RendererUnavailableError } from "./errors.js";
+import { loadPdf } from "./pdf-load.js";
 import { renderBrowserBookPdf } from "./book-render.js";
 import { browserJpegDecoder } from "./browser-jpeg.js";
 import { containFixedPageOverflow } from "./overflow-containment.js";
@@ -12,7 +13,6 @@ import { rewritePressPdf } from "./press/adapter.js";
 import { Page, printProperties } from "@segnavia/format";
 import { frameDocument, newNonce } from "@segnavia/frame/bridge";
 import { printableDeck } from "@segnavia/frame/print-document";
-import { PDFDocument } from "pdf-lib";
 import { chromium } from "playwright-core";
 //#region src/index.ts
 const RENDER_BOUND_MS = 6e4;
@@ -47,7 +47,7 @@ async function withinBound(ms, work) {
 	}
 }
 async function sheets(pdf) {
-	return (await PDFDocument.load(pdf)).getPageCount();
+	return (await loadPdf(pdf)).getPageCount();
 }
 async function enforce(pdf, limits) {
 	if (limits?.maxBytes !== void 0 && pdf.byteLength > limits.maxBytes) throw new RenderRefusedError(`The PDF is ${pdf.byteLength} bytes; the limit is ${limits.maxBytes}.`);

package/dist/page-paper.js

@@ -1,8 +1,9 @@
 import { RenderRefusedError } from "./errors.js";
+import { loadPdf } from "./pdf-load.js";
 import { pdfBox, pdfPoint, pt } from "./pdf-geometry.js";
 import { printWithOverflow } from "./overflow.js";
 import { PX_PER_MM, printBoxes, printPadding, printProperties, sheetMm } from "@segnavia/format";
-import { PDFDocument, PDFName, rgb } from "pdf-lib";
+import { PDFName, rgb } from "pdf-lib";
 //#region src/page-paper.ts
 const usesPressLayout = (print) => print.bleed > 0 || print.marks;
 function printCss(print) {
@@ -33,18 +34,16 @@ function assertPrintOptions(print, preferCssPageSize) {
 }
 async function addPressBoxes(pdf, print) {
 	const boxes = printBoxes(print);
-	const document = await PDFDocument.load(pdf);
+	const document = await loadPdf(pdf);
 	const source = document.getPage(0);
 	const content = await document.embedPage(source);
 	const page = document.addPage([pt(boxes.media.width), pt(boxes.media.height)]);
 	const bleedBox = pdfBox(boxes.bleedBox, boxes.media.height);
 	const trimBox = pdfBox(boxes.trim, boxes.media.height);
-	page.drawPage(content, {
-		x: bleedBox.x,
-		y: bleedBox.y,
+	page.drawPage(content, pressContentPlacement({
 		width: source.getWidth(),
 		height: source.getHeight()
-	});
+	}, bleedBox));
 	document.removePage(0);
 	page.node.delete(PDFName.of("CropBox"));
 	page.setBleedBox(bleedBox.x, bleedBox.y, bleedBox.width, bleedBox.height);
@@ -57,6 +56,14 @@ async function addPressBoxes(pdf, print) {
 	});
 	return document.save();
 }
+function pressContentPlacement(source, bleedBox) {
+	return {
+		x: bleedBox.x,
+		y: bleedBox.y + bleedBox.height - source.height,
+		width: source.width,
+		height: source.height
+	};
+}
 async function printPage(tab, print, preferCssPageSize, inputIndex) {
 	if (!usesPressLayout(print)) {
 		const mm = sheetMm(print);
@@ -88,7 +95,7 @@ async function printPage(tab, print, preferCssPageSize, inputIndex) {
 			left: "0mm"
 		}
 	}, inputIndex);
-	const papers = (await PDFDocument.load(printed.bytes)).getPageCount();
+	const papers = (await loadPdf(printed.bytes)).getPageCount();
 	if (papers > 1) throw new RenderRefusedError(`A page with bleed or marks is one printed piece, but the content needs ${papers} papers. Shorten the content, or remove the bleed and marks.`);
 	return {
 		bytes: await addPressBoxes(printed.bytes, print),
@@ -96,4 +103,4 @@ async function printPage(tab, print, preferCssPageSize, inputIndex) {
 	};
 }
 //#endregion
-export { assertPrintOptions, printCss, printPage, printViewport };
+export { assertPrintOptions, pressContentPlacement, printCss, printPage, printViewport };

package/dist/pages-pdf.js

@@ -1,4 +1,5 @@
 import { RenderRefusedError } from "./errors.js";
+import { loadPdf } from "./pdf-load.js";
 import { groupPieces, pieceGeometry } from "./pages-imposition.js";
 import { composeImposedPdf } from "./pdf-compose.js";
 import { Imposition, Page, impose, printPadding, printProperties } from "@segnavia/format";
@@ -7,7 +8,7 @@ import { PDFDocument } from "pdf-lib";
 async function joinPdfs(pdfs) {
 	const joined = await PDFDocument.create();
 	for (const bytes of pdfs) {
-		const source = await PDFDocument.load(bytes);
+		const source = await loadPdf(bytes);
 		const pages = await joined.copyPages(source, source.getPageIndices());
 		for (const page of pages) joined.addPage(page);
 	}

package/dist/pdf-compose.js

@@ -1,9 +1,11 @@
+import { loadPdf } from "./pdf-load.js";
 import { placementPdfGeometry } from "./pages-imposition.js";
 import { MM_PER_PT } from "@segnavia/format";
 import { PDFDocument, clip, endPath, popGraphicsState, pushGraphicsState, rectangle } from "pdf-lib";
 //#region src/pdf-compose.ts
 async function embedOnePage(output, bytes, name) {
-	const [embedded] = await output.embedPdf(bytes, [0]);
+	const source = await loadPdf(bytes);
+	const [embedded] = await output.embedPdf(source, [0]);
 	if (!embedded) throw new Error(`${name} has no PDF page.`);
 	return embedded;
 }

package/dist/pdf-load.js

@@ -0,0 +1,120 @@
+import { RenderRefusedError } from "./errors.js";
+import { PDFDocument } from "pdf-lib";
+import { Buffer } from "node:buffer";
+import { inflateSync } from "node:zlib";
+//#region src/pdf-load.ts
+const MAX_PDF_STREAM_BYTES = 16777216;
+const whitespace = "[\\x00\\x09\\x0a\\x0c\\x0d\\x20]";
+function nextObject(bytes, from) {
+	let marker = bytes.indexOf(" obj", from, "latin1");
+	while (marker !== -1) {
+		const match = bytes.toString("latin1", Math.max(0, marker - 48), marker).match(/(\d+)\s+(\d+)\s*$/);
+		if (match?.[1]) return {
+			body: marker + 4,
+			number: Number(match[1])
+		};
+		marker = bytes.indexOf(" obj", marker + 4, "latin1");
+	}
+}
+function streamStart(bytes, from, end) {
+	let marker = bytes.indexOf("stream", from, "latin1");
+	while (marker !== -1 && marker < end) {
+		const before = bytes[marker - 1];
+		const after = bytes[marker + 6];
+		const delimited = before !== void 0 && (before === 32 || before === 10 || before === 13 || before === 9);
+		if (delimited && after === 10) return marker + 7;
+		if (delimited && after === 13 && bytes[marker + 7] === 10) return marker + 8;
+		marker = bytes.indexOf("stream", marker + 6, "latin1");
+	}
+}
+function referencedValue(bytes, object, generation) {
+	const marker = Buffer.from(`${object} ${generation} obj`, "latin1");
+	let start = bytes.indexOf(marker);
+	while (start !== -1) {
+		const before = bytes[start - 1];
+		if (start === 0 || before === 10 || before === 13 || before === 32) {
+			const body = start + marker.length;
+			const end = bytes.indexOf("endobj", body, "latin1");
+			if (end !== -1) return bytes.toString("latin1", body, end).trim();
+		}
+		start = bytes.indexOf(marker, start + marker.length);
+	}
+}
+function dictionaryValue(dictionary, key, bytes) {
+	const direct = dictionary.match(new RegExp(`/${key}${whitespace}+(\\d+)(?:${whitespace}+(\\d+)${whitespace}+R)?`));
+	if (direct?.[1] && direct[2]) return referencedValue(bytes, Number(direct[1]), Number(direct[2]));
+	return direct?.[1];
+}
+function streamType(dictionary, bytes) {
+	const direct = dictionary.match(new RegExp(`/Type${whitespace}*/(ObjStm|XRef)\\b`))?.[1];
+	if (direct === "ObjStm" || direct === "XRef") return direct;
+	const reference = dictionary.match(new RegExp(`/Type${whitespace}+(\\d+)${whitespace}+(\\d+)${whitespace}+R`));
+	if (!reference?.[1] || !reference[2]) return void 0;
+	const value = referencedValue(bytes, Number(reference[1]), Number(reference[2]));
+	return value === "/ObjStm" ? "ObjStm" : value === "/XRef" ? "XRef" : void 0;
+}
+function filters(dictionary) {
+	const value = dictionary.match(new RegExp(`/Filter${whitespace}+(\\[[^\\]]*\\]|/[A-Za-z0-9]+)`))?.[1];
+	if (!value) return [];
+	const names = [...value.matchAll(/\/([A-Za-z0-9]+)/g)].map((match) => match[1]);
+	return names.length > 0 ? names : void 0;
+}
+function refuseUnknown(number, fact) {
+	throw new RenderRefusedError(`PDF stream object ${number}'s inflated size could not be established before loading because its ${fact} is unsupported. Use a valid PDF produced by this renderer with a direct length and no filter or one FlateDecode filter to clear this refusal.`);
+}
+function streamObjects(pdf) {
+	const bytes = Buffer.from(pdf.buffer, pdf.byteOffset, pdf.byteLength);
+	const streams = [];
+	let cursor = 0;
+	while (cursor < bytes.length) {
+		const object = nextObject(bytes, cursor);
+		if (!object) break;
+		const endObject = bytes.indexOf("endobj", object.body, "latin1");
+		if (endObject === -1) break;
+		const start = streamStart(bytes, object.body, endObject);
+		if (start === void 0) {
+			cursor = endObject + 6;
+			continue;
+		}
+		const dictionary = bytes.toString("latin1", object.body, start - 6);
+		const length = dictionaryValue(dictionary, "Length", bytes)?.match(/^\s*(\d+)\s*$/)?.[1];
+		if (streamType(dictionary, bytes)) {
+			if (!length) refuseUnknown(object.number, "declared length");
+			streams.push({
+				dictionary,
+				number: object.number,
+				start
+			});
+		}
+		cursor = length ? Math.min(bytes.length, start + Number(length)) : endObject + 6;
+	}
+	return streams;
+}
+function refusal(number, size, ceiling) {
+	return new RenderRefusedError(`PDF stream object ${number} inflated past the ceiling after ${size} bytes; the ceiling is ${ceiling} bytes. Reduce the page content that Chromium packs into one object or cross-reference stream to clear this refusal.`);
+}
+function assertPdfStreamsBounded(pdf, ceiling = MAX_PDF_STREAM_BYTES) {
+	const bytes = Buffer.from(pdf.buffer, pdf.byteOffset, pdf.byteLength);
+	for (const stream of streamObjects(pdf)) {
+		const length = Number(dictionaryValue(stream.dictionary, "Length", bytes));
+		if (!Number.isSafeInteger(length) || length < 0 || stream.start + length > bytes.length) refuseUnknown(stream.number, "declared length");
+		const encodings = filters(stream.dictionary);
+		if (!encodings || encodings.length > 1 || encodings.length === 1 && encodings[0] !== "FlateDecode") refuseUnknown(stream.number, `decode filter ${encodings?.join(", ") || "value"}`);
+		if (encodings.length === 0) {
+			if (length > ceiling) throw refusal(stream.number, length, ceiling);
+			continue;
+		}
+		try {
+			inflateSync(bytes.subarray(stream.start, stream.start + length), { maxOutputLength: ceiling });
+		} catch (error) {
+			if (error instanceof Error && "code" in error && error.code === "ERR_BUFFER_TOO_LARGE") throw refusal(stream.number, ceiling, ceiling);
+			refuseUnknown(stream.number, "FlateDecode data");
+		}
+	}
+}
+async function loadPdf(pdf, options) {
+	assertPdfStreamsBounded(pdf);
+	return PDFDocument.load(pdf, options);
+}
+//#endregion
+export { MAX_PDF_STREAM_BYTES, assertPdfStreamsBounded, loadPdf };

package/dist/press/adapter.js

@@ -1,4 +1,5 @@
 import { RenderRefusedError } from "../errors.js";
+import { loadPdf } from "../pdf-load.js";
 import { cachedPressColor, destinationColor, newlySnapsToNeutral, pdfMaskLuminosity, pressDestination } from "./color-policy.js";
 import { groupAction, shadingAction } from "./policy.js";
 import { rewriteContent } from "./content.js";
@@ -10,7 +11,7 @@ import { readProfile } from "../color/profile.js";
 import { convertSuppliedImage, convertedImageCount, createImageState } from "./image.js";
 import { createPaintedBoundsTracker, recordPatternPaintedBounds } from "./pattern-bounds.js";
 import { inspectPressPdf } from "./report.js";
-import { PDFDict, PDFDocument, PDFHexString, PDFName, PDFRef, PDFString } from "pdf-lib";
+import { PDFDict, PDFHexString, PDFName, PDFRef, PDFString } from "pdf-lib";
 import { deflateSync } from "node:zlib";
 //#region src/press/adapter.ts
 const key = (name) => PDFName.of(name);
@@ -319,7 +320,7 @@ async function rewritePressPdf(bytes, profileBytes, tolerance = 3, decodeJpeg) {
 	try {
 		const profile = readProfile(profileBytes);
 		if (profile.colorSpace !== "CMYK" && profile.colorSpace !== "GRAY") refuse(`press profile color space ${profile.colorSpace}`, "CMYK or GRAY");
-		const document = await PDFDocument.load(bytes, { updateMetadata: false });
+		const document = await loadPdf(bytes, { updateMetadata: false });
 		const counts = rewritePressDocument(document, profile, tolerance);
 		let convertedImages = 0;
 		const destination = pressDestination(profile);

package/dist/press/report.js

@@ -1,9 +1,10 @@
 import { RenderRefusedError } from "../errors.js";
+import { loadPdf } from "../pdf-load.js";
 import { groupAction, shadingAction } from "./policy.js";
 import { inspectContent } from "./content.js";
 import { decodeContent, dictionaryOf, nameOf, numberOf, rawStreamOf, resolvedColorSpace } from "./pdf-objects.js";
 import { walkPressGraph } from "./visitor.js";
-import { PDFArray, PDFDict, PDFDocument, PDFName, PDFRef } from "pdf-lib";
+import { PDFArray, PDFDict, PDFName, PDFRef } from "pdf-lib";
 //#region src/press/report.ts
 const key = (name) => PDFName.of(name);
 const refuse = (found, supported) => {
@@ -64,7 +65,7 @@ function outputDestination(document) {
 	return profile && numberOf(profile.dict, "N") === 1 ? "gray" : "cmyk";
 }
 async function inspectPressPdf(bytes, requested) {
-	const document = await PDFDocument.load(bytes, { updateMetadata: false });
+	const document = await loadPdf(bytes, { updateMetadata: false });
 	const destination = requested ?? outputDestination(document);
 	const state = {
 		budget: { decodedBytes: 0 },

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@segnavia/render",
-  "version": "0.12.0",
+  "version": "0.13.0",
   "type": "module",
   "license": "MIT",
   "repository": {
@@ -26,9 +26,9 @@
     }
   },
   "dependencies": {
-    "@segnavia/format": "0.12.0",
-    "@segnavia/frame": "0.12.0",
-    "@segnavia/html": "0.12.0",
+    "@segnavia/format": "0.13.0",
+    "@segnavia/frame": "0.13.0",
+    "@segnavia/html": "0.13.0",
     "playwright-core": "1.55.1",
     "pdf-lib": "^1.17.1"
   }