unpdf 0.6.1 → 0.7.1

package/dist/index.mjs

@@ -1,4 +1,6 @@
 let resolvedModule;
+const isNode = globalThis.process?.release?.name === "node";
+const isBrowser = typeof window !== "undefined";
 async function getDocumentProxy(data, options = {}) {
   const { getDocument } = await getResolvedPDFJS();
   const pdf = await getDocument({
@@ -8,12 +10,6 @@ async function getDocumentProxy(data, options = {}) {
   }).promise;
   return pdf;
 }
-async function defineUnPDFConfig(options) {
-  const { pdfjs } = { ...options };
-  if (pdfjs) {
-    await resolvePDFJSImports(pdfjs, { force: true });
-  }
-}
 async function getResolvedPDFJS() {
   if (!resolvedModule) {
     await resolvePDFJSImports();
@@ -47,7 +43,7 @@ function isPDFDocumentProxy(data) {
   return typeof data === "object" && data !== null && "_pdfInfo" in data;
 }
 
-async function getPDFMeta$1(data) {
+async function getMeta$1(data) {
   const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data);
   const meta = await pdf.getMetadata();
   return {
@@ -56,7 +52,7 @@ async function getPDFMeta$1(data) {
   };
 }
 
-async function extractPDFText$1(data, options = {}) {
+async function extractText$1(data, options = {}) {
   const { mergePages = false } = { ...options };
   const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data);
   const texts = await Promise.all(
@@ -78,35 +74,123 @@ async function getPageText(pdf, pageNumber) {
   return items.map(({ str }) => str).join("\n");
 }
 
-async function getImagesFromPage$1(data, pageNumber) {
+async function extractImages$1(data, pageNumber) {
   const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data);
   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 operatorList = await page.getOperatorList();
   const { OPS } = await getResolvedPDFJS();
   const images = [];
-  for (const op of operatorList.fnArray) {
+  for (let i = 0; i < operatorList.fnArray.length; i++) {
+    const op = operatorList.fnArray[i];
     if (op !== OPS.paintImageXObject) {
       continue;
     }
-    const image = await page.objs.get(operatorList.argsArray[op][0]);
-    if (image.data) {
-      images.push(image.data.buffer);
-    }
+    const imageKey = operatorList.argsArray[i][0];
+    const image = await page.objs.get(imageKey);
+    images.push(image.data);
   }
   return images;
 }
+async function renderPageAsImage$1(data, pageNumber, options = {}) {
+  const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data);
+  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}.`
+    );
+  }
+  let outputScale = options.scale || 1;
+  let viewport = page.getViewport({ scale: outputScale });
+  if (options.width) {
+    outputScale = options.width / viewport.width;
+  } else if (options.height) {
+    outputScale = options.height / viewport.height;
+  }
+  if (outputScale !== 1 && outputScale > 0) {
+    viewport = page.getViewport({ scale: outputScale });
+  }
+  const canvasFactory = await createIsomorphicCanvasFactory(options.canvas);
+  const ctx = canvasFactory.create(viewport.width, viewport.height);
+  await page.render({
+    canvasContext: ctx.context,
+    viewport
+  }).promise;
+  const dataUrl = isBrowser ? ctx.canvas.toDataURL() : ctx.canvas.toDataURL();
+  const response = await fetch(dataUrl);
+  return await response.arrayBuffer();
+}
+async function createIsomorphicCanvasFactory(canvas) {
+  const _canvas = await canvas?.();
+  return {
+    _createCanvas(width, height) {
+      if (isBrowser) {
+        const canvas2 = document.createElement("canvas");
+        canvas2.width = width;
+        canvas2.height = height;
+        return canvas2;
+      } else if (isNode) {
+        if (!_canvas) {
+          throw new Error('Failed to resolve "canvas" package.');
+        }
+        return _canvas.createCanvas(width, height);
+      }
+      throw new Error("Unsupported environment for canvas creation.");
+    },
+    create(width, height) {
+      const _canvas2 = this._createCanvas(width, height);
+      const context = _canvas2.getContext(
+        "2d"
+      );
+      return {
+        canvas: _canvas2,
+        context
+      };
+    },
+    reset(ctx, width, height) {
+      if (ctx.canvas) {
+        ctx.canvas.width = width;
+        ctx.canvas.height = height;
+      }
+    },
+    destroy(ctx) {
+      if (ctx.canvas) {
+        ctx.canvas.width = 0;
+        ctx.canvas.height = 0;
+      }
+      ctx.canvas = void 0;
+      ctx.context = void 0;
+    }
+  };
+}
+
+async function defineUnPDFConfig(options) {
+  const { pdfjs } = { ...options };
+  if (pdfjs) {
+    await resolvePDFJSImports(pdfjs, { force: true });
+  }
+}
 
-const getPDFMeta = async (...args) => {
+const getMeta = async (...args) => {
+  await resolvePDFJSImports();
+  return await getMeta$1(...args);
+};
+const extractText = async (...args) => {
   await resolvePDFJSImports();
-  return await getPDFMeta$1(...args);
+  return await extractText$1(...args);
 };
-const extractPDFText = async (...args) => {
+const extractPDFText = extractText;
+const extractImages = async (...args) => {
   await resolvePDFJSImports();
-  return await extractPDFText$1(...args);
+  return await extractImages$1(...args);
 };
-const getImagesFromPage = async (...args) => {
+const renderPageAsImage = async (...args) => {
   await resolvePDFJSImports();
-  return await getImagesFromPage$1(...args);
+  return await renderPageAsImage$1(...args);
 };
 
-export { defineUnPDFConfig, extractPDFText, getDocumentProxy, getImagesFromPage, getPDFMeta, getResolvedPDFJS, resolvePDFJSImports };
+export { defineUnPDFConfig, extractImages, extractPDFText, extractText, getDocumentProxy, getMeta, getResolvedPDFJS, renderPageAsImage, resolvePDFJSImports };