unpdf 1.8.1 → 1.9.1

package/dist/index.mjs

@@ -52,7 +52,7 @@ let resolvedModule;
 const isNode = globalThis.process?.release?.name === "node";
 const isBrowser = typeof window !== "undefined";
 /**
-* Returns a `PDFDocumentProxy` instance from a given binary data.
+* Creates a `PDFDocumentProxy` from binary PDF data.
 *
 * Applies the following defaults:
 * - `useSystemFonts: true`
@@ -208,18 +208,6 @@ function injectCanvasConstructors() {
 /**
 * 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) {
 	return await withDocument(data, async (pdf) => {
@@ -326,23 +314,23 @@ const XMP_DATE_PROPERTIES = [
 async function getMeta$1(data, options = {}) {
 	return await withDocument(data, async (pdf) => {
 		const meta = await pdf.getMetadata();
-		const info = meta?.info || {};
+		const info = { ...meta.info };
+		let metadata = meta.metadata;
 		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];
-			} });
+			if (info.CreationDate) info.CreationDate = PDFDateString.toDateObject(info.CreationDate);
+			if (info.ModDate) info.ModDate = PDFDateString.toDateObject(info.ModDate);
+			if (metadata) {
+				const Metadata = metadata.constructor;
+				metadata = new Metadata({
+					parsedData: new Map(Array.from(metadata, ([name, value]) => [name, XMP_DATE_PROPERTIES.includes(name) && value ? parseISODateString(value) : value])),
+					rawData: metadata.getRaw()
+				});
+			}
 		}
 		return {
 			info,
-			metadata: meta?.metadata || {}
+			metadata: metadata || {}
 		};
 	});
 }
@@ -403,15 +391,14 @@ function normalizeMergedText(texts) {
 //#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.
+* Sets the PDF.js build every other method uses – call it before any of them.
+* Without it, the serverless build is used.
 *
 * @example
-* // Use the official PDF.js build (make sure to install it first)
+* // In Node.js, the legacy build of `pdfjs-dist` (install it first)
 * import { definePDFJSModule } from 'unpdf'
 *
-* await definePDFJSModule(() => import('pdfjs-dist'))
+* await definePDFJSModule(() => import('pdfjs-dist/legacy/build/pdf.mjs'))
 */
 async function definePDFJSModule(pdfjs) {
 	await resolvePDFJSImport(pdfjs, { reload: true });