unpdf 1.6.2 → 1.8.0

package/dist/index.cjs

@@ -1,420 +1,453 @@
-'use strict';
-
+Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
+//#region src/_internal/dommatrix-polyfill.ts
 function polyfillDOMMatrix() {
-  if (typeof globalThis.DOMMatrix !== "undefined")
-    return;
-  globalThis.DOMMatrix = class DOMMatrix {
-    a;
-    b;
-    c;
-    d;
-    e;
-    f;
-    constructor(init) {
-      if (Array.isArray(init) && init.length === 6) {
-        this.a = init[0];
-        this.b = init[1];
-        this.c = init[2];
-        this.d = init[3];
-        this.e = init[4];
-        this.f = init[5];
-      } else {
-        this.a = 1;
-        this.b = 0;
-        this.c = 0;
-        this.d = 1;
-        this.e = 0;
-        this.f = 0;
-      }
-    }
-    translateSelf(tx, ty = 0) {
-      this.e = this.a * tx + this.c * ty + this.e;
-      this.f = this.b * tx + this.d * ty + this.f;
-      return this;
-    }
-    scaleSelf(sx, sy = sx) {
-      this.a *= sx;
-      this.b *= sx;
-      this.c *= sy;
-      this.d *= sy;
-      return this;
-    }
-  };
+	if (typeof globalThis.DOMMatrix !== "undefined") return;
+	globalThis.DOMMatrix = class DOMMatrix {
+		constructor(init) {
+			if (Array.isArray(init) && init.length === 6) {
+				this.a = init[0];
+				this.b = init[1];
+				this.c = init[2];
+				this.d = init[3];
+				this.e = init[4];
+				this.f = init[5];
+			} else {
+				this.a = 1;
+				this.b = 0;
+				this.c = 0;
+				this.d = 1;
+				this.e = 0;
+				this.f = 0;
+			}
+		}
+		translateSelf(tx, ty = 0) {
+			this.e = this.a * tx + this.c * ty + this.e;
+			this.f = this.b * tx + this.d * ty + this.f;
+			return this;
+		}
+		scaleSelf(sx, sy = sx) {
+			this.a *= sx;
+			this.b *= sx;
+			this.c *= sy;
+			this.d *= sy;
+			return this;
+		}
+	};
 }
-
+//#endregion
+//#region src/_internal/env.ts
+/**
+* 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).
+*/
 function stubBrowserGlobals() {
-  polyfillDOMMatrix();
+	polyfillDOMMatrix();
 }
-
+//#endregion
+//#region src/utils.ts
 let resolvedModule;
 const isNode = globalThis.process?.release?.name === "node";
 const isBrowser = typeof window !== "undefined";
+/**
+* Returns a PDFDocumentProxy instance from a given binary 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
+*/
 async function getDocumentProxy(data, options = {}) {
-  const { getDocument } = await getResolvedPDFJS();
-  let nodeDefaults = {};
-  if (isNode) {
-    try {
-      const base = undefined("pdfjs-dist/package.json");
-      nodeDefaults = {
-        disableFontFace: true,
-        standardFontDataUrl: new URL("./standard_fonts/", base).href
-      };
-    } catch {
-    }
-  }
-  const pdf = await getDocument({
-    data,
-    isEvalSupported: false,
-    // See: https://github.com/mozilla/pdf.js/issues/4244#issuecomment-1479534301
-    useSystemFonts: true,
-    ...nodeDefaults,
-    ...options
-  }).promise;
-  return pdf;
+	const { getDocument } = await getResolvedPDFJS();
+	let nodeDefaults = {};
+	if (isNode) try {
+		const base = {}.resolve("pdfjs-dist/package.json");
+		nodeDefaults = {
+			disableFontFace: true,
+			standardFontDataUrl: new URL("./standard_fonts/", base).href,
+			cMapUrl: new URL("./cmaps/", base).href,
+			cMapPacked: true
+		};
+	} catch {}
+	return await getDocument({
+		data,
+		useSystemFonts: true,
+		...nodeDefaults,
+		...options
+	}).promise;
+}
+/**
+* Runs an operation on a `PDFDocumentProxy` resolved from either binary data
+* or an existing proxy. Documents created here are destroyed once the
+* operation settles – caller-supplied proxies keep their lifecycle with the
+* caller.
+*/
+async function withDocument(data, operation, options) {
+	const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data, options);
+	try {
+		return await operation(pdf);
+	} finally {
+		if (pdf !== data) await pdf.loadingTask.destroy();
+	}
 }
 async function getResolvedPDFJS() {
-  if (!resolvedModule) {
-    await resolvePDFJSImport();
-  }
-  return resolvedModule;
+	if (!resolvedModule) await resolvePDFJSImport();
+	return resolvedModule;
 }
 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}`);
-  }
+	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 isPDFDocumentProxy(data) {
-  return typeof data === "object" && data !== null && "_pdfInfo" in data;
+	return typeof data === "object" && data !== null && "_pdfInfo" in data;
 }
 async function interopDefault(m) {
-  const resolved = await m;
-  return resolved.default || resolved;
+	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
- */
-class BaseCanvasFactory {
-  #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;
-  }
-  // eslint-disable-next-line unused-imports/no-unused-vars
-  _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 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
- */
-class DOMCanvasFactory extends BaseCanvasFactory {
-  _document;
-  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;
-  }
-}
-class NodeCanvasFactory 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);
-  }
-}
+* 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 resolveCanvasModule(canvasImport) {
-  resolvedCanvasModule ??= await interopDefault(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.
+*
+* @remarks
+* `DOMMatrix` is always set because {@link stubBrowserGlobals} 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;
+	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
+/**
+* Extracts images from a specific page of a PDF document, including necessary metadata,
+* such as width, height, and calculated color channels.
+*
+* @example
+* const imagesData = await extractImages(pdf, pageNum)
+*
+* for (const imgData of imagesData) {
+*   const imageIndex = totalImagesProcessed + 1
+*   await sharp(imgData.data, {
+*     raw: { width: imgData.width, height: imgData.height, channels: imgData.channels }
+*   })
+*     .png()
+*     .toFile(`${imageIndex}.png`)
+* }
+*/
 async function extractImages$1(data, pageNumber) {
-  const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data);
-  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 images = [];
-  for (let i = 0; i < operatorList.fnArray.length; i++) {
-    const op = operatorList.fnArray[i];
-    if (op !== 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: data2 } = image;
-    const calculatedChannels = data2.length / (width * height);
-    if (![1, 3, 4].includes(calculatedChannels)) {
-      continue;
-    }
-    const channels = calculatedChannels;
-    images.push({
-      data: data2,
-      width,
-      height,
-      channels,
-      key: imageKey
-    });
-  }
-  return images;
+	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 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;
+			images.push({
+				data,
+				width,
+				height,
+				channels,
+				key: imageKey
+			});
+		}
+		return images;
+	});
 }
 async function renderPageAsImage$1(data, pageNumber, options = {}) {
-  const CanvasFactory = await createIsomorphicCanvasFactory(options.canvasImport);
-  const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data, { CanvasFactory });
-  const page = await pdf.getPage(pageNumber);
-  if (pageNumber < 1 || pageNumber > pdf.numPages) {
-    throw new Error(`Invalid page number. Must be between 1 and ${pdf.numPages}.`);
-  }
-  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 drawingContext = new CanvasFactory().create(viewport.width, viewport.height);
-  await page.render({
-    canvas: drawingContext.canvas,
-    canvasContext: drawingContext.context,
-    viewport
-  }).promise;
-  const dataUrl = drawingContext.canvas.toDataURL();
-  if (options.toDataURL) {
-    return dataUrl;
-  }
-  const response = await fetch(dataUrl);
-  return await response.arrayBuffer();
+	const CanvasFactory = await createIsomorphicCanvasFactory(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);
+		try {
+			await page.render({
+				canvas: drawingContext.canvas,
+				canvasContext: drawingContext.context,
+				viewport
+			}).promise;
+			const { canvas } = drawingContext;
+			if (options.toDataURL) return canvas.toDataURL();
+			if ("encode" in canvas) {
+				const buffer = await canvas.encode("png");
+				return buffer.buffer.slice(buffer.byteOffset, buffer.byteOffset + buffer.byteLength);
+			}
+			const blob = await new Promise((resolve) => {
+				canvas.toBlob(resolve);
+			});
+			if (!blob) throw new Error("Failed to encode canvas to a PNG blob.");
+			return await blob.arrayBuffer();
+		} finally {
+			canvasFactory.destroy(drawingContext);
+		}
+	}, { CanvasFactory });
 }
 async function createIsomorphicCanvasFactory(canvasImport) {
-  if (isBrowser)
-    return DOMCanvasFactory;
-  if (isNode) {
-    if (!canvasImport) {
-      throw new Error('Parameter "canvasImport" is required in Node.js environment.');
-    }
-    await resolveCanvasModule(canvasImport);
-    injectCanvasConstructors();
-    return NodeCanvasFactory;
-  }
-  throw new Error("Unsupported environment for canvas creation.");
+	if (isBrowser) return DOMCanvasFactory;
+	if (isNode) {
+		if (!canvasImport) throw new Error("Parameter \"canvasImport\" is required in Node.js environment.");
+		await resolveCanvasModule(canvasImport);
+		injectCanvasConstructors();
+		return NodeCanvasFactory;
+	}
+	throw new Error("Unsupported environment for canvas creation.");
 }
-
+//#endregion
+//#region src/links.ts
 async function extractLinks$1(data) {
-  const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data);
-  const pageLinks = await Promise.all(
-    Array.from({ length: pdf.numPages }, (_, i) => getPageLinks(pdf, i + 1))
-  );
-  return {
-    totalPages: pdf.numPages,
-    links: pageLinks.flat()
-  };
+	return await withDocument(data, async (pdf) => {
+		const pageLinks = await Promise.all(Array.from({ length: pdf.numPages }, (_, i) => getPageLinks(pdf, i + 1)));
+		return {
+			totalPages: pdf.numPages,
+			links: pageLinks.flat()
+		};
+	});
 }
 async function getPageLinks(document, pageNumber) {
-  const page = await document.getPage(pageNumber);
-  const annotations = await page.getAnnotations();
-  const links = [];
-  for (const annotation of annotations) {
-    if (annotation.subtype === "Link" && annotation.url) {
-      links.push(annotation.url);
-    }
-  }
-  return links;
+	const annotations = await (await document.getPage(pageNumber)).getAnnotations();
+	const links = [];
+	for (const annotation of annotations) if (annotation.subtype === "Link" && annotation.url) links.push(annotation.url);
+	return links;
 }
-
+//#endregion
+//#region src/meta.ts
 const XMP_DATE_PROPERTIES = [
-  "xmp:createdate",
-  "xmp:modifydate",
-  "xmp:metadatadate",
-  "xap:createdate",
-  "xap:modifydate",
-  "xap:metadatadate"
+	"xmp:createdate",
+	"xmp:modifydate",
+	"xmp:metadatadate",
+	"xap:createdate",
+	"xap:modifydate",
+	"xap:metadatadate"
 ];
 async function getMeta$1(data, options = {}) {
-  const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data);
-  const meta = await pdf.getMetadata();
-  const info = meta?.info || {};
-  if (options.parseDates) {
-    const { PDFDateString } = await getResolvedPDFJS();
-    if (info?.CreationDate) {
-      info.CreationDate = PDFDateString.toDateObject(info?.CreationDate);
-    }
-    if (info?.ModDate) {
-      info.ModDate = PDFDateString.toDateObject(info?.ModDate);
-    }
-    if (meta.metadata) {
-      meta.metadata = new Proxy(meta.metadata, {
-        get(target, prop) {
-          if (prop === "get") {
-            return (name) => {
-              const value = target.get(name);
-              if (XMP_DATE_PROPERTIES.includes(name) && value) {
-                return parseISODateString(value);
-              }
-              return value;
-            };
-          }
-          return target[prop];
-        }
-      });
-    }
-  }
-  return {
-    info,
-    metadata: meta?.metadata || {}
-  };
+	return await withDocument(data, async (pdf) => {
+		const meta = await pdf.getMetadata();
+		const info = meta?.info || {};
+		if (options.parseDates) {
+			const { PDFDateString } = await getResolvedPDFJS();
+			if (info?.CreationDate) info.CreationDate = PDFDateString.toDateObject(info?.CreationDate);
+			if (info?.ModDate) info.ModDate = PDFDateString.toDateObject(info?.ModDate);
+			if (meta.metadata) meta.metadata = new Proxy(meta.metadata, { get(target, prop) {
+				if (prop === "get") return (name) => {
+					const value = target.get(name);
+					if (XMP_DATE_PROPERTIES.includes(name) && value) return parseISODateString(value);
+					return value;
+				};
+				return target[prop];
+			} });
+		}
+		return {
+			info,
+			metadata: meta?.metadata || {}
+		};
+	});
 }
 function parseISODateString(isoDateString) {
-  if (!isoDateString)
-    return;
-  const parsedDate = Date.parse(isoDateString);
-  if (!Number.isNaN(parsedDate)) {
-    return new Date(parsedDate);
-  }
+	if (!isoDateString) return;
+	const parsedDate = Date.parse(isoDateString);
+	if (!Number.isNaN(parsedDate)) return new Date(parsedDate);
 }
-
+//#endregion
+//#region src/text.ts
 async function extractTextItems$1(data) {
-  const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data);
-  const items = await Promise.all(
-    Array.from({ length: pdf.numPages }, (_, i) => getPageTextItems(pdf, i + 1))
-  );
-  return { totalPages: pdf.numPages, items };
+	return await withDocument(data, async (pdf) => {
+		const items = await Promise.all(Array.from({ length: pdf.numPages }, (_, i) => getPageTextItems(pdf, i + 1)));
+		return {
+			totalPages: pdf.numPages,
+			items
+		};
+	});
 }
 async function getPageTextItems(document, pageNumber) {
-  const page = await document.getPage(pageNumber);
-  const content = await page.getTextContent();
-  const styles = content.styles;
-  return content.items.filter((item) => item.str != null).map((item) => {
-    const [_a, _b, c, d, e, f] = item.transform;
-    return {
-      str: item.str,
-      x: e,
-      y: f,
-      width: item.width,
-      height: item.height,
-      fontSize: Math.hypot(c, d),
-      fontFamily: styles[item.fontName]?.fontFamily ?? "",
-      dir: item.dir,
-      hasEOL: item.hasEOL
-    };
-  });
+	const content = await (await document.getPage(pageNumber)).getTextContent();
+	const styles = content.styles;
+	return content.items.filter((item) => item.str != null).map((item) => {
+		const [_a, _b, c, d, e, f] = item.transform;
+		return {
+			str: item.str,
+			x: e,
+			y: f,
+			width: item.width,
+			height: item.height,
+			fontSize: Math.hypot(c, d),
+			fontFamily: styles[item.fontName]?.fontFamily ?? "",
+			dir: item.dir,
+			hasEOL: item.hasEOL
+		};
+	});
 }
 async function extractText$1(data, options = {}) {
-  const { mergePages = false } = options;
-  const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data);
-  const texts = await Promise.all(
-    Array.from({ length: pdf.numPages }, (_, i) => getPageText(pdf, i + 1))
-  );
-  return {
-    totalPages: pdf.numPages,
-    text: mergePages ? texts.join("\n").replace(/\s+/g, " ") : texts
-  };
+	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)));
+		return {
+			totalPages: pdf.numPages,
+			text: mergePages ? normalizeMergedText(texts) : texts
+		};
+	});
 }
 async function getPageText(document, pageNumber) {
-  const page = await document.getPage(pageNumber);
-  const content = await page.getTextContent();
-  return content.items.filter((item) => item.str != null).map((item) => item.str + (item.hasEOL ? "\n" : "")).join("");
+	return (await (await document.getPage(pageNumber)).getTextContent()).items.filter((item) => item.str != null).map((item) => item.str + (item.hasEOL ? "\n" : "")).join("");
 }
-
+/**
+* Collapses whitespace without destroying the line structure: `hasEOL` and
+* page-join line breaks survive, but at most one blank line remains.
+*/
+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
+/**
+* By default, unpdf will use the latest version of PDF.js compiled for
+* serverless environments. If you want to use a different version, you can
+* provide a custom resolver function.
+*
+* @example
+* // Use the official PDF.js build (make sure to install it first)
+* import { definePDFJSModule } from 'unpdf'
+*
+* await definePDFJSModule(() => import('pdfjs-dist'))
+*/
 async function definePDFJSModule(pdfjs) {
-  await resolvePDFJSImport(pdfjs, { reload: true });
+	await resolvePDFJSImport(pdfjs, { reload: true });
 }
+/** @deprecated Use `definePDFJSModule` instead. Will be removed in v2. */
 async function configureUnPDF(options) {
-  await resolvePDFJSImport(options.pdfjs, { reload: true });
+	await resolvePDFJSImport(options.pdfjs, { reload: true });
 }
-
+//#endregion
+//#region src/index.ts
 const getMeta = async (...args) => {
-  await resolvePDFJSImport();
-  return await getMeta$1(...args);
+	await resolvePDFJSImport();
+	return await getMeta$1(...args);
 };
 const extractText = async (...args) => {
-  await resolvePDFJSImport();
-  return await extractText$1(...args);
+	await resolvePDFJSImport();
+	return await extractText$1(...args);
 };
 const extractTextItems = async (...args) => {
-  await resolvePDFJSImport();
-  return await extractTextItems$1(...args);
+	await resolvePDFJSImport();
+	return await extractTextItems$1(...args);
 };
 const extractImages = async (...args) => {
-  await resolvePDFJSImport();
-  return await extractImages$1(...args);
+	await resolvePDFJSImport();
+	return await extractImages$1(...args);
 };
 const renderPageAsImage = async (...args) => {
-  await resolvePDFJSImport();
-  return await renderPageAsImage$1(...args);
+	await resolvePDFJSImport();
+	return await renderPageAsImage$1(...args);
 };
 const extractLinks = async (...args) => {
-  await resolvePDFJSImport();
-  return await extractLinks$1(...args);
+	await resolvePDFJSImport();
+	return await extractLinks$1(...args);
 };
-
+//#endregion
 exports.configureUnPDF = configureUnPDF;
 exports.createIsomorphicCanvasFactory = createIsomorphicCanvasFactory;
 exports.definePDFJSModule = definePDFJSModule;