unpdf 0.6.0 → 0.7.0

package/dist/index.mjs

@@ -0,0 +1,202 @@
+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({
+    data,
+    isEvalSupported: false,
+    ...options
+  }).promise;
+  return pdf;
+}
+async function getResolvedPDFJS() {
+  if (!resolvedModule) {
+    await resolvePDFJSImports();
+  }
+  return resolvedModule;
+}
+async function resolvePDFJSImports(pdfjsResolver, { force = false } = {}) {
+  if (resolvedModule && !force) {
+    return;
+  }
+  if (pdfjsResolver) {
+    try {
+      const _import = await pdfjsResolver();
+      resolvedModule = _import.default || _import;
+      return;
+    } catch {
+      throw new Error(
+        "Resolving failed. Please check the provided configuration."
+      );
+    }
+  }
+  try {
+    resolvedModule = await import('unpdf/pdfjs');
+  } catch {
+    throw new Error(
+      "PDF.js is not available. Please add the package as a dependency."
+    );
+  }
+}
+function isPDFDocumentProxy(data) {
+  return typeof data === "object" && data !== null && "_pdfInfo" in data;
+}
+
+async function getMeta$1(data) {
+  const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data);
+  const meta = await pdf.getMetadata();
+  return {
+    info: meta?.info ?? {},
+    metadata: meta?.metadata?.getAll() ?? {}
+  };
+}
+
+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))
+  );
+  const filteredTexts = texts.filter(Boolean);
+  return {
+    totalPages: pdf.numPages,
+    text: mergePages ? filteredTexts.join("\n\n") : filteredTexts
+  };
+}
+async function getPageText(pdf, pageNumber) {
+  const page = await pdf.getPage(pageNumber);
+  const content = await page.getTextContent();
+  const items = content.items;
+  if (items.length === 0) {
+    return;
+  }
+  return items.map(({ str }) => str).join("\n");
+}
+
+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 (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 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();
+  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() {
+  let _canvas;
+  if (isNode) {
+    try {
+      _canvas = await import('canvas');
+    } catch {
+      throw new Error(
+        'The "canvas" package is required when running in Node.js.'
+      );
+    }
+  }
+  return {
+    _createCanvas(width, height) {
+      if (isBrowser) {
+        const canvas = document.createElement("canvas");
+        canvas.width = width;
+        canvas.height = height;
+        return canvas;
+      } else if (isNode) {
+        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 getMeta = async (...args) => {
+  await resolvePDFJSImports();
+  return await getMeta$1(...args);
+};
+const extractText = async (...args) => {
+  await resolvePDFJSImports();
+  return await extractText$1(...args);
+};
+const extractPDFText = extractText;
+const extractImages = async (...args) => {
+  await resolvePDFJSImports();
+  return await extractImages$1(...args);
+};
+const renderPageAsImage = async (...args) => {
+  await resolvePDFJSImports();
+  return await renderPageAsImage$1(...args);
+};
+
+export { defineUnPDFConfig, extractImages, extractPDFText, extractText, getDocumentProxy, getMeta, getResolvedPDFJS, renderPageAsImage, resolvePDFJSImports };