unpdf 1.9.1 → 2.0.0

package/dist/index.mjs

@@ -1,4 +1,8 @@
-//#region src/_internal/dommatrix-polyfill.ts
+// src/_internal/dommatrix-polyfill.ts
+/**
+* Stubs `DOMMatrix`, which PDF.js reads while its modules evaluate, so it must
+* run before any PDF.js module is imported.
+*/
 function polyfillDOMMatrix() {
 	if (typeof globalThis.DOMMatrix !== "undefined") return;
 	globalThis.DOMMatrix = class DOMMatrix {
@@ -33,44 +37,32 @@ function polyfillDOMMatrix() {
 		}
 	};
 }
-//#endregion
-//#region src/_internal/env.ts
+// src/utils.ts
+let pdfjsLoader;
+let resolvedModule;
 /**
-* Stubs browser globals that PDF.js expects at the module level
-* in non-browser environments.
-*
-* @remarks
-* Must be called before importing any PDF.js module, since PDF.js
-* accesses these globals at parse time (top-level constants).
+* Whether PDF.js takes its Node.js code paths, by PDF.js' own rule: a
+* `process` object outside of NW.js and Electron renderers. It decides between
+* PDF.js' Node.js and DOM factories for canvases, fonts, and CMaps.
 */
-function stubBrowserGlobals() {
-	polyfillDOMMatrix();
-}
-//#endregion
-//#region src/utils.ts
-let resolvedModule;
-const isNode = globalThis.process?.release?.name === "node";
-const isBrowser = typeof window !== "undefined";
+const isNode = isNodeProcess(globalThis.process);
 /**
-* Creates a `PDFDocumentProxy` from binary PDF data.
-*
-* Applies the following defaults:
-* - `useSystemFonts: true`
-*
-* In Node.js environments, additionally applies:
-* - `disableFontFace: true`
-* - `standardFontDataUrl` resolved from the local `pdfjs-dist` package
-* - `cMapUrl` and `cMapPacked` resolved from the local `pdfjs-dist` package
+* Creates a `PDFDocumentProxy` from binary PDF data, with `useSystemFonts:
+* true`. In Node.js with `pdfjs-dist` installed, it uses that package's
+* standard fonts instead of system fonts, and its packed CMaps.
 */
 async function getDocumentProxy(data, options = {}) {
 	const { getDocument } = await getResolvedPDFJS();
 	let nodeDefaults = {};
 	if (isNode) try {
-		const base = import.meta.resolve("pdfjs-dist/package.json");
+		const nodeProcess = globalThis.process;
+		const { dirname, join } = nodeProcess.getBuiltinModule("node:path");
+		const { fileURLToPath } = nodeProcess.getBuiltinModule("node:url");
+		const pdfjsDir = dirname(fileURLToPath(import.meta.resolve("pdfjs-dist/package.json")));
 		nodeDefaults = {
-			disableFontFace: true,
-			standardFontDataUrl: new URL("./standard_fonts/", base).href,
-			cMapUrl: new URL("./cmaps/", base).href,
+			useSystemFonts: false,
+			standardFontDataUrl: `${join(pdfjsDir, "standard_fonts")}/`,
+			cMapUrl: `${join(pdfjsDir, "cmaps")}/`,
 			cMapPacked: true
 		};
 	} catch {}
@@ -95,24 +87,27 @@ async function withDocument(data, operation, options) {
 		if (pdf !== data) await pdf.loadingTask.destroy();
 	}
 }
-async function getResolvedPDFJS() {
-	if (!resolvedModule) await resolvePDFJSImport();
-	return resolvedModule;
+/**
+* Sets the PDF.js build every other method uses. The build loads on first use,
+* so call this before any other method. Without it, the serverless build is
+* used.
+*
+* @example
+* // In Node.js, the legacy build of `pdfjs-dist` (install it first)
+* import { definePDFJSModule } from 'unpdf'
+*
+* definePDFJSModule(() => import('pdfjs-dist/legacy/build/pdf.mjs'))
+*/
+function definePDFJSModule(pdfjs) {
+	pdfjsLoader = pdfjs;
+	resolvedModule = void 0;
 }
-async function resolvePDFJSImport(pdfjsResolver, { reload = false } = {}) {
-	if (resolvedModule && !reload) return;
-	stubBrowserGlobals();
-	if (pdfjsResolver) try {
-		resolvedModule = await interopDefault(pdfjsResolver());
-		return;
-	} catch (error) {
-		throw new Error(`PDF.js could not be resolved: ${error}`);
-	}
-	try {
-		resolvedModule = await import("unpdf/pdfjs");
-	} catch (error) {
-		throw new Error(`Serverless PDF.js bundle could not be resolved: ${error}`);
-	}
+function getResolvedPDFJS() {
+	resolvedModule ??= loadPDFJS(pdfjsLoader).catch((error) => {
+		resolvedModule = void 0;
+		throw error;
+	});
+	return resolvedModule;
 }
 function isPDFDocumentProxy(data) {
 	return typeof data === "object" && data !== null && "_pdfInfo" in data;
@@ -121,177 +116,437 @@ async function interopDefault(m) {
 	const resolved = await m;
 	return resolved.default || resolved;
 }
-//#endregion
-//#region src/_internal/canvas.ts
-let resolvedCanvasModule;
-/**
-* Derived from the PDF.js project by the Mozilla Foundation.
-* @see https://github.com/mozilla/pdf.js/blob/b8de9a372f9bbf7e33adb362eeae5ef1919dba73/src/display/canvas_factory.js#L18
-* @license Apache-2.0
-*/
-var BaseCanvasFactory = class {
-	#enableHWA = false;
-	constructor({ enableHWA = false } = {}) {
-		this.#enableHWA = enableHWA;
-	}
-	create(width, height) {
-		const canvas = this._createCanvas(width, height);
-		return {
-			canvas,
-			context: canvas.getContext("2d", { willReadFrequently: !this.#enableHWA })
-		};
-	}
-	reset({ canvas }, width, height) {
-		if (!canvas) throw new Error("Canvas is not specified");
-		canvas.width = width;
-		canvas.height = height;
-	}
-	destroy(context) {
-		if (!context.canvas) throw new Error("Canvas is not specified");
-		context.canvas.width = 0;
-		context.canvas.height = 0;
-		context.canvas = void 0;
-		context.context = void 0;
-	}
-	_createCanvas(width, height) {
-		throw new Error("Not implemented");
-	}
-};
-/**
-* Derived from the PDF.js project by the Mozilla Foundation.
-* @see https://github.com/mozilla/pdf.js/blob/b8de9a372f9bbf7e33adb362eeae5ef1919dba73/src/display/canvas_factory.js#L18
-* @license Apache-2.0
-*/
-var DOMCanvasFactory = class extends BaseCanvasFactory {
-	constructor({ ownerDocument = globalThis.document, enableHWA = false } = {}) {
-		super({ enableHWA });
-		this._document = ownerDocument;
-	}
-	_createCanvas(width, height) {
-		const canvas = this._document.createElement("canvas");
-		canvas.width = width;
-		canvas.height = height;
-		return canvas;
-	}
-};
-var NodeCanvasFactory = class extends BaseCanvasFactory {
-	constructor({ enableHWA = false } = {}) {
-		super({ enableHWA });
-	}
-	_createCanvas(width, height) {
-		if (!resolvedCanvasModule) throw new Error("@napi-rs/canvas module is not resolved");
-		return resolvedCanvasModule.createCanvas(width, height);
+async function loadPDFJS(loader) {
+	polyfillDOMMatrix();
+	try {
+		return await interopDefault(loader ? loader() : import("unpdf/pdfjs"));
+	} catch (error) {
+		throw new Error(`PDF.js could not be resolved: ${error}`, { cause: error });
 	}
-};
+}
+function isNodeProcess(nodeProcess) {
+	return typeof nodeProcess === "object" && `${nodeProcess}` === "[object process]" && !nodeProcess.versions.nw && !(nodeProcess.versions.electron && nodeProcess.type && nodeProcess.type !== "browser");
+}
+// src/_internal/canvas.ts
+let resolvedCanvasModule;
 async function resolveCanvasModule(canvasImport) {
 	resolvedCanvasModule ??= await interopDefault(canvasImport());
 	globalThis[Symbol.for("unpdf.canvasModule")] = resolvedCanvasModule;
 }
 /**
-* Injects `DOMMatrix`, `ImageData` and `Path2D` constructors from the
-* `@napi-rs/canvas` package into the global scope. This is necessary for
-* Node.js environments that do not support these constructors natively.
+* Injects the `DOMMatrix` and `Path2D` constructors of `@napi-rs/canvas` into
+* the global scope, which the PDF.js display layer instantiates while
+* rendering. Only the legacy `pdfjs-dist` build polyfills them itself.
 *
 * @remarks
-* `DOMMatrix` is always set because {@link stubBrowserGlobals} may have
+* `DOMMatrix` is always set because {@link polyfillDOMMatrix} may have
 * installed a minimal stub to allow PDF.js to import – the real
 * `@napi-rs/canvas` implementation must take precedence for rendering.
 */
 function injectCanvasConstructors() {
 	if (!resolvedCanvasModule) return;
 	globalThis.DOMMatrix = resolvedCanvasModule.DOMMatrix;
-	if (typeof globalThis.ImageData === "undefined") globalThis.ImageData = resolvedCanvasModule.ImageData;
 	if (typeof globalThis.Path2D === "undefined") globalThis.Path2D = resolvedCanvasModule.Path2D;
 }
-//#endregion
-//#region src/image.ts
+// src/_internal/operator-list.ts
+const IDENTITY_MATRIX = [
+	1,
+	0,
+	0,
+	1,
+	0,
+	0
+];
+/**
+* Walks an operator list in painting order, with the graphics state each
+* operation draws with and whether it is visible. Skips the content of soft
+* masks, which only modulates transparency.
+*/
+function* walkOperatorList(operatorList, ops, optionalContentConfig) {
+	const stateStack = [];
+	let state = {
+		transform: IDENTITY_MATRIX,
+		textRenderingMode: 0
+	};
+	const softMaskGroupFlags = [];
+	const contentVisibilityFlags = [];
+	for (let i = 0; i < operatorList.fnArray.length; i++) {
+		const fn = operatorList.fnArray[i];
+		const args = operatorList.argsArray[i];
+		if (fn === ops.save) stateStack.push(state);
+		else if (fn === ops.restore || fn === ops.paintFormXObjectEnd || fn === ops.endAnnotation) state = stateStack.pop() ?? state;
+		else if (fn === ops.transform) state = {
+			...state,
+			transform: multiplyMatrices(state.transform, args)
+		};
+		else if (fn === ops.paintFormXObjectBegin) {
+			stateStack.push(state);
+			const [formMatrix] = args;
+			if (formMatrix) state = {
+				...state,
+				transform: multiplyMatrices(state.transform, formMatrix)
+			};
+		} else if (fn === ops.beginAnnotation) {
+			stateStack.push(state);
+			const [, , annotationTransform, appearanceMatrix] = args;
+			state = {
+				transform: multiplyMatrices(Array.from(annotationTransform), appearanceMatrix),
+				textRenderingMode: 0
+			};
+		} else if (fn === ops.setTextRenderingMode) state = {
+			...state,
+			textRenderingMode: args[0]
+		};
+		else if (fn === ops.beginGroup) softMaskGroupFlags.push(Boolean(args[0]?.smask));
+		else if (fn === ops.endGroup) softMaskGroupFlags.pop();
+		else if (fn === ops.beginMarkedContent) contentVisibilityFlags.push(true);
+		else if (fn === ops.beginMarkedContentProps) {
+			const [tag, properties] = args;
+			contentVisibilityFlags.push(tag !== "OC" || optionalContentConfig.isVisible(properties));
+		} else if (fn === ops.endMarkedContent) contentVisibilityFlags.pop();
+		if (softMaskGroupFlags.includes(true)) continue;
+		yield {
+			fn,
+			args,
+			...state,
+			isVisible: !contentVisibilityFlags.includes(false)
+		};
+	}
+}
 /**
-* Extracts images from a specific page of a PDF document, including necessary metadata,
-* such as width, height, and calculated color channels.
+* Maps the unit square, which PDF images are drawn into, through a transform
+* and returns the axis-aligned box around the result.
 */
-async function extractImages$1(data, pageNumber) {
+function getUnitSquareBounds(transform) {
+	const [a, b, c, d, e, f] = transform;
+	const xs = [
+		e,
+		a + e,
+		c + e,
+		a + c + e
+	];
+	const ys = [
+		f,
+		b + f,
+		d + f,
+		b + d + f
+	];
+	const minX = Math.min(...xs);
+	const minY = Math.min(...ys);
+	return {
+		x: minX,
+		y: minY,
+		width: Math.max(...xs) - minX,
+		height: Math.max(...ys) - minY
+	};
+}
+/** Applies `inner` first, then `outer`, like `cm` concatenating onto the current matrix. */
+function multiplyMatrices(outer, inner) {
+	const [a, b, c, d, e, f] = outer;
+	const [a2, b2, c2, d2, e2, f2] = Array.from(inner);
+	return [
+		a * a2 + c * b2,
+		b * a2 + d * b2,
+		a * c2 + c * d2,
+		b * c2 + d * d2,
+		a * e2 + c * f2 + e,
+		b * e2 + d * f2 + f
+	];
+}
+// src/_internal/png.ts
+const PNG_SIGNATURE = [
+	137,
+	80,
+	78,
+	71,
+	13,
+	10,
+	26,
+	10
+];
+let crcTable;
+/**
+* Encodes RGBA pixels as an unfiltered 8-bit RGBA PNG, compressed through
+* `CompressionStream`, which Node.js, Cloudflare Workers, and browsers all provide.
+*/
+async function encodePNG(rgba, width, height) {
+	const rowLength = width * 4;
+	const scanlines = new Uint8Array((rowLength + 1) * height);
+	for (let y = 0; y < height; y++) scanlines.set(rgba.subarray(y * rowLength, (y + 1) * rowLength), y * (rowLength + 1) + 1);
+	const header = /* @__PURE__ */ new Uint8Array(13);
+	const headerView = new DataView(header.buffer);
+	headerView.setUint32(0, width);
+	headerView.setUint32(4, height);
+	header[8] = 8;
+	header[9] = 6;
+	const chunks = [
+		createChunk("IHDR", header),
+		createChunk("IDAT", await deflate(scanlines)),
+		createChunk("IEND", /* @__PURE__ */ new Uint8Array(0))
+	];
+	const png = new Uint8Array(PNG_SIGNATURE.length + chunks.reduce((sum, chunk) => sum + chunk.length, 0));
+	png.set(PNG_SIGNATURE);
+	let offset = PNG_SIGNATURE.length;
+	for (const chunk of chunks) {
+		png.set(chunk, offset);
+		offset += chunk.length;
+	}
+	return png;
+}
+async function deflate(data) {
+	const stream = new CompressionStream("deflate");
+	const writer = stream.writable.getWriter();
+	const [compressed] = await Promise.all([
+		new Response(stream.readable).arrayBuffer(),
+		writer.write(data),
+		writer.close()
+	]);
+	return new Uint8Array(compressed);
+}
+/** Lays out a chunk as length, type, data, and the CRC of type and data. */
+function createChunk(type, data) {
+	const chunk = new Uint8Array(12 + data.length);
+	const view = new DataView(chunk.buffer);
+	view.setUint32(0, data.length);
+	for (let i = 0; i < 4; i++) chunk[4 + i] = type.charCodeAt(i);
+	chunk.set(data, 8);
+	view.setUint32(8 + data.length, crc32(chunk.subarray(4, 8 + data.length)));
+	return chunk;
+}
+function crc32(bytes) {
+	crcTable ??= createCRCTable();
+	let crc = 4294967295;
+	for (const byte of bytes) crc = crcTable[(crc ^ byte) & 255] ^ crc >>> 8;
+	return (crc ^ 4294967295) >>> 0;
+}
+function createCRCTable() {
+	const table = /* @__PURE__ */ new Uint32Array(256);
+	for (let n = 0; n < 256; n++) {
+		let crc = n;
+		for (let k = 0; k < 8; k++) crc = crc & 1 ? 3988292384 ^ crc >>> 1 : crc >>> 1;
+		table[n] = crc;
+	}
+	return table;
+}
+// src/image.ts
+/**
+* Extracts the images painted on a page – image XObjects, inline images, and
+* image masks – as RGBA pixels. Image masks come out opaque black where they
+* paint and transparent elsewhere, since their color is the fill color at the
+* time of painting. A 1×1 image mask that paints its pixel, which PDF.js
+* turns into a solid fill, is skipped.
+*
+* @remarks
+* A caller-supplied document proxy created with `isOffscreenCanvasSupported`
+* or `isImageDecoderSupported` (both on by default outside of Node.js) hands
+* some images out as bitmaps instead of pixels, which are skipped.
+*/
+async function extractImages(data, pageNumber, options = {}) {
 	return await withDocument(data, async (pdf) => {
 		if (pageNumber < 1 || pageNumber > pdf.numPages) throw new Error(`Invalid page number. Must be between 1 and ${pdf.numPages}.`);
 		const page = await pdf.getPage(pageNumber);
-		const operatorList = await page.getOperatorList();
-		const { OPS } = await getResolvedPDFJS();
+		const [operatorList, optionalContentConfig] = await Promise.all([page.getOperatorList(), pdf.getOptionalContentConfig()]);
+		const { ImageKind, OPS } = await getResolvedPDFJS();
 		const images = [];
-		for (let i = 0; i < operatorList.fnArray.length; i++) {
-			if (operatorList.fnArray[i] !== OPS.paintImageXObject) continue;
-			const imageKey = operatorList.argsArray[i][0];
-			const image = await new Promise((resolve) => (imageKey.startsWith("g_") ? page.commonObjs : page.objs).get(imageKey, resolve));
-			if (!image || !image.data || !image.width || !image.height) continue;
-			const { width, height, data } = image;
-			const calculatedChannels = data.length / (width * height);
-			if (![
-				1,
-				3,
-				4
-			].includes(calculatedChannels)) continue;
-			const channels = calculatedChannels;
+		for (const { fn, args, transform, isVisible } of walkOperatorList(operatorList, OPS, optionalContentConfig)) {
+			if (!isVisible) continue;
+			let key = null;
+			let image;
+			if (fn === OPS.paintImageXObject) {
+				key = args[0];
+				image = await getPageObject(page, key);
+			} else if (fn === OPS.paintInlineImageXObject) image = args[0];
+			else if (fn === OPS.paintImageMaskXObject) {
+				key = args[0].data;
+				image = await getPageObject(page, key);
+			} else continue;
+			if (!image) continue;
+			const pixels = toRGBA(image, fn === OPS.paintImageMaskXObject, ImageKind);
+			if (!pixels) continue;
 			images.push({
-				data,
-				width,
-				height,
-				channels,
-				key: imageKey
+				data: options.format === "png" ? await encodePNG(pixels, image.width, image.height) : pixels,
+				width: image.width,
+				height: image.height,
+				key,
+				bounds: getUnitSquareBounds(transform)
 			});
 		}
 		return images;
+	}, {
+		isOffscreenCanvasSupported: false,
+		isImageDecoderSupported: false
 	});
 }
-async function renderPageAsImage$1(data, pageNumber, options = {}) {
-	const CanvasFactory = await createIsomorphicCanvasFactory(options.canvasImport);
+async function renderPageAsImage(data, pageNumber, options = {}) {
+	await prepareCanvasEnvironment(options.canvasImport);
 	return await withDocument(data, async (pdf) => {
 		if (pageNumber < 1 || pageNumber > pdf.numPages) throw new Error(`Invalid page number. Must be between 1 and ${pdf.numPages}.`);
 		const page = await pdf.getPage(pageNumber);
 		const defaultViewport = page.getViewport({ scale: 1 });
-		let scale = options.scale || 1;
-		if (options.width) scale = options.width / defaultViewport.width;
-		else if (options.height) scale = options.height / defaultViewport.height;
-		const viewport = page.getViewport({ scale: Math.max(0, scale) });
-		const canvasFactory = new CanvasFactory();
-		const drawingContext = canvasFactory.create(viewport.width, viewport.height);
+		let scale = options.scale ?? 1;
+		if (options.width !== void 0) scale = options.width / defaultViewport.width;
+		else if (options.height !== void 0) scale = options.height / defaultViewport.height;
+		const viewport = page.getViewport({ scale });
+		const canvasFactory = pdf.canvasFactory;
+		const canvasAndContext = canvasFactory.create(viewport.width, viewport.height);
 		try {
+			const { canvas } = canvasAndContext;
 			await page.render({
-				canvas: drawingContext.canvas,
-				canvasContext: drawingContext.context,
+				canvas,
 				viewport
 			}).promise;
-			const { canvas } = drawingContext;
-			if (options.toDataURL) return canvas.toDataURL();
+			const { format = "png", quality } = options;
+			const mimeType = `image/${format}`;
+			if (options.toDataURL) {
+				const dataURL = canvas.toDataURL(mimeType, quality);
+				assertEncodedType(dataURL.slice(5, dataURL.indexOf(";")), mimeType);
+				return dataURL;
+			}
 			if ("encode" in canvas) {
-				const buffer = await canvas.encode("png");
-				return buffer.buffer.slice(buffer.byteOffset, buffer.byteOffset + buffer.byteLength);
+				const buffer = format === "png" ? await canvas.encode(format) : await canvas.encode(format, quality === void 0 ? void 0 : quality * 100);
+				return new Uint8Array(buffer);
 			}
 			const blob = await new Promise((resolve) => {
-				canvas.toBlob(resolve);
+				canvas.toBlob(resolve, mimeType, quality);
 			});
-			if (!blob) throw new Error("Failed to encode canvas to a PNG blob.");
-			return await blob.arrayBuffer();
+			if (!blob) throw new Error(`Failed to encode canvas as ${mimeType}.`);
+			assertEncodedType(blob.type, mimeType);
+			return new Uint8Array(await blob.arrayBuffer());
 		} finally {
-			canvasFactory.destroy(drawingContext);
+			canvasFactory.destroy(canvasAndContext);
+		}
+	});
+}
+function getPageObject(page, objectId) {
+	const objects = objectId.startsWith("g_") ? page.commonObjs : page.objs;
+	return new Promise((resolve) => objects.get(objectId, resolve));
+}
+/**
+* Converts PDF.js image data into a fresh RGBA buffer, so the result never
+* aliases a buffer PDF.js keeps cached.
+*/
+function toRGBA(image, isMask, imageKind) {
+	const { data, width, height, kind } = image;
+	if (!data) return;
+	const rgba = new Uint8Array(width * height * 4);
+	if (kind === imageKind.RGBA_32BPP) {
+		rgba.set(data.subarray(0, rgba.length));
+		return rgba;
+	}
+	if (kind === imageKind.RGB_24BPP) {
+		for (let source = 0, target = 0; target < rgba.length; source += 3, target += 4) {
+			rgba[target] = data[source];
+			rgba[target + 1] = data[source + 1];
+			rgba[target + 2] = data[source + 2];
+			rgba[target + 3] = 255;
+		}
+		return rgba;
+	}
+	if (!isMask && kind !== imageKind.GRAYSCALE_1BPP) return;
+	const bytesPerRow = Math.ceil(width / 8);
+	for (let y = 0; y < height; y++) for (let x = 0; x < width; x++) {
+		const bit = data[y * bytesPerRow + (x >> 3)] >> 7 - (x & 7) & 1;
+		const target = (y * width + x) * 4;
+		if (isMask) rgba[target + 3] = bit === 0 ? 255 : 0;
+		else {
+			rgba.fill(bit === 0 ? 0 : 255, target, target + 3);
+			rgba[target + 3] = 255;
 		}
-	}, { CanvasFactory });
+	}
+	return rgba;
 }
-async function createIsomorphicCanvasFactory(canvasImport) {
-	if (isBrowser) return DOMCanvasFactory;
+async function prepareCanvasEnvironment(canvasImport) {
 	if (isNode) {
 		if (!canvasImport) throw new Error("Parameter \"canvasImport\" is required in Node.js environment.");
 		await resolveCanvasModule(canvasImport);
 		injectCanvasConstructors();
-		return NodeCanvasFactory;
+		return;
+	}
+	if (typeof document === "undefined") throw new Error("Unsupported environment for canvas creation.");
+}
+function assertEncodedType(encodedType, mimeType) {
+	if (encodedType !== mimeType) throw new Error(`This environment can't encode ${mimeType}, only ${encodedType}.`);
+}
+// src/inspect.ts
+const WHITESPACE_PATTERN = /\s/g;
+/**
+* Measures the text and images of every page, such as to tell scanned pages
+* from born-digital ones. Where that line lies is left to the caller.
+*/
+async function inspectPages(data) {
+	return await withDocument(data, async (pdf) => {
+		const { OPS } = await getResolvedPDFJS();
+		const optionalContentConfig = await pdf.getOptionalContentConfig();
+		const shouldReleasePages = !isPDFDocumentProxy(data);
+		const pages = [];
+		for (let pageNumber = 1; pageNumber <= pdf.numPages; pageNumber++) pages.push(await inspectPage(pdf, pageNumber, OPS, optionalContentConfig, shouldReleasePages));
+		return {
+			totalPages: pdf.numPages,
+			pages
+		};
+	});
+}
+async function inspectPage(document, pageNumber, ops, optionalContentConfig, shouldReleasePage) {
+	const page = await document.getPage(pageNumber);
+	const [textContent, operatorList] = await Promise.all([page.getTextContent(), page.getOperatorList()]);
+	const textLength = textContent.items.reduce((sum, item) => sum + ("str" in item ? countCharacters(item.str) : 0), 0);
+	let invisibleTextLength = 0;
+	const imageBounds = [];
+	for (const { fn, args, transform, textRenderingMode, isVisible } of walkOperatorList(operatorList, ops, optionalContentConfig)) if (fn === ops.showText && (!isVisible || textRenderingMode === 3 || textRenderingMode === 7)) {
+		for (const glyph of args[0]) if (typeof glyph === "object" && glyph?.unicode) invisibleTextLength += countCharacters(glyph.unicode);
+	} else if (isVisible && (fn === ops.paintImageXObject || fn === ops.paintInlineImageXObject || fn === ops.paintImageMaskXObject)) imageBounds.push(getUnitSquareBounds(transform));
+	if (shouldReleasePage) page.cleanup();
+	return {
+		textLength,
+		invisibleTextLength,
+		imageCount: imageBounds.length,
+		imageCoverage: getCoverage(imageBounds, page.view)
+	};
+}
+function countCharacters(text) {
+	return Array.from(text.replace(WHITESPACE_PATTERN, "")).length;
+}
+/**
+* Computes the fraction of a view box the union of boxes covers, exactly: cuts
+* the view box into vertical slabs at every box edge and merges the vertical
+* extents of the boxes spanning each slab.
+*/
+function getCoverage(boxes, view) {
+	const [viewX0, viewY0, viewX1, viewY1] = view;
+	const viewArea = (viewX1 - viewX0) * (viewY1 - viewY0);
+	if (viewArea <= 0) return 0;
+	const clippedBoxes = boxes.map((box) => ({
+		x0: Math.max(box.x, viewX0),
+		x1: Math.min(box.x + box.width, viewX1),
+		y0: Math.max(box.y, viewY0),
+		y1: Math.min(box.y + box.height, viewY1)
+	})).filter((box) => box.x0 < box.x1 && box.y0 < box.y1);
+	const edges = [...new Set(clippedBoxes.flatMap((box) => [box.x0, box.x1]))].sort((a, b) => a - b);
+	let area = 0;
+	for (let i = 0; i < edges.length - 1; i++) {
+		const left = edges[i];
+		const right = edges[i + 1];
+		const spans = clippedBoxes.filter((box) => box.x0 <= left && box.x1 >= right).sort((a, b) => a.y0 - b.y0);
+		let coveredHeight = 0;
+		let coveredTop = -Infinity;
+		for (const { y0, y1 } of spans) {
+			const start = Math.max(y0, coveredTop);
+			if (y1 > start) {
+				coveredHeight += y1 - start;
+				coveredTop = y1;
+			}
+		}
+		area += coveredHeight * (right - left);
 	}
-	throw new Error("Unsupported environment for canvas creation.");
+	return area / viewArea;
 }
-//#endregion
-//#region src/links.ts
-async function extractLinks$1(data) {
+// src/links.ts
+async function extractLinks(data) {
 	return await withDocument(data, async (pdf) => {
-		const pageLinks = await Promise.all(Array.from({ length: pdf.numPages }, (_, i) => getPageLinks(pdf, i + 1)));
+		const links = await Promise.all(Array.from({ length: pdf.numPages }, (_, i) => getPageLinks(pdf, i + 1)));
 		return {
 			totalPages: pdf.numPages,
-			links: pageLinks.flat()
+			links
 		};
 	});
 }
@@ -301,8 +556,7 @@ async function getPageLinks(document, pageNumber) {
 	for (const annotation of annotations) if (annotation.subtype === "Link" && annotation.url) links.push(annotation.url);
 	return links;
 }
-//#endregion
-//#region src/meta.ts
+// src/meta.ts
 const XMP_DATE_PROPERTIES = [
 	"xmp:createdate",
 	"xmp:modifydate",
@@ -311,7 +565,7 @@ const XMP_DATE_PROPERTIES = [
 	"xap:modifydate",
 	"xap:metadatadate"
 ];
-async function getMeta$1(data, options = {}) {
+async function getMeta(data, options = {}) {
 	return await withDocument(data, async (pdf) => {
 		const meta = await pdf.getMetadata();
 		const info = { ...meta.info };
@@ -330,18 +584,16 @@ async function getMeta$1(data, options = {}) {
 		}
 		return {
 			info,
-			metadata: metadata || {}
+			metadata: metadata ?? null
 		};
 	});
 }
 function parseISODateString(isoDateString) {
-	if (!isoDateString) return;
 	const parsedDate = Date.parse(isoDateString);
-	if (!Number.isNaN(parsedDate)) return new Date(parsedDate);
+	return Number.isNaN(parsedDate) ? null : new Date(parsedDate);
 }
-//#endregion
-//#region src/text.ts
-async function extractTextItems$1(data) {
+// src/text.ts
+async function extractTextItems(data) {
 	return await withDocument(data, async (pdf) => {
 		const items = await Promise.all(Array.from({ length: pdf.numPages }, (_, i) => getPageTextItems(pdf, i + 1)));
 		return {
@@ -368,7 +620,7 @@ async function getPageTextItems(document, pageNumber) {
 		};
 	});
 }
-async function extractText$1(data, options = {}) {
+async function extractText(data, options = {}) {
 	const { mergePages = false } = options;
 	return await withDocument(data, async (pdf) => {
 		const texts = await Promise.all(Array.from({ length: pdf.numPages }, (_, i) => getPageText(pdf, i + 1)));
@@ -388,50 +640,4 @@ async function getPageText(document, pageNumber) {
 function normalizeMergedText(texts) {
 	return texts.join("\n").replace(/[^\S\n]+/g, " ").replace(/ ?\n ?/g, "\n").replace(/\n{3,}/g, "\n\n");
 }
-//#endregion
-//#region src/config.ts
-/**
-* Sets the PDF.js build every other method uses – call it before any of them.
-* Without it, the serverless build is used.
-*
-* @example
-* // In Node.js, the legacy build of `pdfjs-dist` (install it first)
-* import { definePDFJSModule } from 'unpdf'
-*
-* await definePDFJSModule(() => import('pdfjs-dist/legacy/build/pdf.mjs'))
-*/
-async function definePDFJSModule(pdfjs) {
-	await resolvePDFJSImport(pdfjs, { reload: true });
-}
-/** @deprecated Use `definePDFJSModule` instead. Will be removed in v2. */
-async function configureUnPDF(options) {
-	await resolvePDFJSImport(options.pdfjs, { reload: true });
-}
-//#endregion
-//#region src/index.ts
-const getMeta = async (...args) => {
-	await resolvePDFJSImport();
-	return await getMeta$1(...args);
-};
-const extractText = async (...args) => {
-	await resolvePDFJSImport();
-	return await extractText$1(...args);
-};
-const extractTextItems = async (...args) => {
-	await resolvePDFJSImport();
-	return await extractTextItems$1(...args);
-};
-const extractImages = async (...args) => {
-	await resolvePDFJSImport();
-	return await extractImages$1(...args);
-};
-const renderPageAsImage = async (...args) => {
-	await resolvePDFJSImport();
-	return await renderPageAsImage$1(...args);
-};
-const extractLinks = async (...args) => {
-	await resolvePDFJSImport();
-	return await extractLinks$1(...args);
-};
-//#endregion
-export { configureUnPDF, createIsomorphicCanvasFactory, definePDFJSModule, extractImages, extractLinks, extractText, extractTextItems, getDocumentProxy, getMeta, getResolvedPDFJS, renderPageAsImage, resolvePDFJSImport };
+export { definePDFJSModule, extractImages, extractLinks, extractText, extractTextItems, getDocumentProxy, getMeta, getResolvedPDFJS, inspectPages, renderPageAsImage };