[api-minor] Use the Fetch API, when supported, to load PDF documents in Node.js environments

Given that modern Node.js versions now implement support for a fair number of "browser" APIs, we can utilize the standard Fetch API to load PDF documents that are specified via http/https URLs.

Please find compatibility information at:
 - https://developer.mozilla.org/en-US/docs/Web/API/Fetch_API#browser_compatibility
 - https://nodejs.org/dist/latest-v18.x/docs/api/globals.html#fetch
 - https://developer.mozilla.org/en-US/docs/Web/API/Response#browser_compatibility
 - https://nodejs.org/dist/latest-v18.x/docs/api/globals.html#response

src/display/api.js

@@ -419,7 +419,16 @@ function getDocument(src) {
               PDFJSDev.test("GENERIC") &&
               isNodeJS
             ) {
-              return new PDFNodeStream(params);
+              const isFetchSupported = function () {
+                return (
+                  typeof fetch !== "undefined" &&
+                  typeof Response !== "undefined" &&
+                  "body" in Response.prototype
+                );
+              };
+              return isFetchSupported() && isValidFetchUrl(params.url)
+                ? new PDFFetchStream(params)
+                : new PDFNodeStream(params);
             }
             return isValidFetchUrl(params.url)
               ? new PDFFetchStream(params)
@@ -762,6 +771,9 @@ class PDFDocumentProxy {
 
     if (typeof PDFJSDev === "undefined" || PDFJSDev.test("TESTING")) {
       // For testing purposes.
+      Object.defineProperty(this, "getNetworkStreamName", {
+        value: () => this._transport.getNetworkStreamName(),
+      });
       Object.defineProperty(this, "getXFADatasets", {
         value: () => this._transport.getXFADatasets(),
       });
@@ -2344,6 +2356,9 @@ class WorkerTransport {
 
     if (typeof PDFJSDev === "undefined" || PDFJSDev.test("TESTING")) {
       // For testing purposes.
+      Object.defineProperty(this, "getNetworkStreamName", {
+        value: () => networkStream?.constructor?.name || null,
+      });
       Object.defineProperty(this, "getXFADatasets", {
         value: () =>
           this.messageHandler.sendWithPromise("GetXFADatasets", null),

test/unit/api_spec.js

@@ -32,6 +32,7 @@ import {
 import {
   buildGetDocumentParams,
   CMAP_URL,
+  createTemporaryNodeServer,
   DefaultFileReaderFactory,
   TEST_PDFS_PATH,
 } from "./test_utils.js";
@@ -67,13 +68,27 @@ describe("api", function () {
     buildGetDocumentParams(tracemonkeyFileName);
 
   let CanvasFactory;
+  let tempServer = null;
 
   beforeAll(function () {
     CanvasFactory = new DefaultCanvasFactory();
+
+    if (isNodeJS) {
+      tempServer = createTemporaryNodeServer();
+    }
   });
 
   afterAll(function () {
     CanvasFactory = null;
+
+    if (isNodeJS) {
+      // Close the server from accepting new connections after all test
+      // finishes.
+      const { server } = tempServer;
+      server.close();
+
+      tempServer = null;
+    }
   });
 
   function waitSome(callback) {
@@ -119,13 +134,10 @@ describe("api", function () {
     });
 
     it("creates pdf doc from URL-object", async function () {
-      if (isNodeJS) {
+      const urlObj = isNodeJS
-        pending("window.location is not supported in Node.js.");
+        ? new URL(`http://127.0.0.1:${tempServer.port}/${basicApiFileName}`)
-      }
+        : new URL(TEST_PDFS_PATH + basicApiFileName, window.location);
-      const urlObj = new URL(
+
-        TEST_PDFS_PATH + basicApiFileName,
-        window.location
-      );
       const loadingTask = getDocument(urlObj);
       expect(loadingTask instanceof PDFDocumentLoadingTask).toEqual(true);
       const pdfDocument = await loadingTask.promise;
@@ -134,6 +146,9 @@ describe("api", function () {
       expect(pdfDocument instanceof PDFDocumentProxy).toEqual(true);
       expect(pdfDocument.numPages).toEqual(3);
 
+      // Ensure that the Fetch API was used to load the PDF document.
+      expect(pdfDocument.getNetworkStreamName()).toEqual("PDFFetchStream");
+
       await loadingTask.destroy();
     });
 

test/unit/clitests.json

@@ -21,6 +21,7 @@
     "encodings_spec.js",
     "evaluator_spec.js",
     "event_utils_spec.js",
+    "fetch_stream_spec.js",
     "font_substitutions_spec.js",
     "function_spec.js",
     "message_handler_spec.js",

test/unit/fetch_stream_spec.js

@@ -13,16 +13,40 @@
  * limitations under the License.
  */
 
-import { AbortException } from "../../src/shared/util.js";
+import { AbortException, isNodeJS } from "../../src/shared/util.js";
+import { createTemporaryNodeServer } from "./test_utils.js";
 import { PDFFetchStream } from "../../src/display/fetch_stream.js";
 
 describe("fetch_stream", function () {
-  const pdfUrl = new URL("../pdfs/tracemonkey.pdf", window.location).href;
+  let tempServer = null;
+
+  function getPdfUrl() {
+    return isNodeJS
+      ? `http://127.0.0.1:${tempServer.port}/tracemonkey.pdf`
+      : new URL("../pdfs/tracemonkey.pdf", window.location).href;
+  }
   const pdfLength = 1016315;
 
+  beforeAll(function () {
+    if (isNodeJS) {
+      tempServer = createTemporaryNodeServer();
+    }
+  });
+
+  afterAll(function () {
+    if (isNodeJS) {
+      // Close the server from accepting new connections after all test
+      // finishes.
+      const { server } = tempServer;
+      server.close();
+
+      tempServer = null;
+    }
+  });
+
   it("read with streaming", async function () {
     const stream = new PDFFetchStream({
-      url: pdfUrl,
+      url: getPdfUrl(),
       disableStream: false,
       disableRange: true,
     });
@@ -57,7 +81,7 @@ describe("fetch_stream", function () {
   it("read ranges with streaming", async function () {
     const rangeSize = 32768;
     const stream = new PDFFetchStream({
-      url: pdfUrl,
+      url: getPdfUrl(),
       rangeChunkSize: rangeSize,
       disableStream: false,
       disableRange: false,

test/unit/node_stream_spec.js

@@ -14,6 +14,7 @@
  */
 
 import { AbortException, isNodeJS } from "../../src/shared/util.js";
+import { createTemporaryNodeServer } from "./test_utils.js";
 import { PDFNodeStream } from "../../src/display/node_stream.js";
 
 // Ensure that these tests only run in Node.js environments.
@@ -25,12 +26,10 @@ if (!isNodeJS) {
 
 const path = await __non_webpack_import__("path");
 const url = await __non_webpack_import__("url");
-const http = await __non_webpack_import__("http");
-const fs = await __non_webpack_import__("fs");
 
 describe("node_stream", function () {
-  let server = null;
+  let tempServer = null;
-  let port = null;
+
   const pdf = url.parse(
     encodeURI(
       "file://" + path.join(process.cwd(), "./test/pdfs/tracemonkey.pdf")
@@ -39,50 +38,20 @@ describe("node_stream", function () {
   const pdfLength = 1016315;
 
   beforeAll(function () {
-    // Create http server to serve pdf data for tests.
+    tempServer = createTemporaryNodeServer();
-    server = http
-      .createServer((request, response) => {
-        const filePath = process.cwd() + "/test/pdfs" + request.url;
-        fs.lstat(filePath, (error, stat) => {
-          if (error) {
-            response.writeHead(404);
-            response.end(`File ${request.url} not found!`);
-            return;
-          }
-          if (!request.headers.range) {
-            const contentLength = stat.size;
-            const stream = fs.createReadStream(filePath);
-            response.writeHead(200, {
-              "Content-Type": "application/pdf",
-              "Content-Length": contentLength,
-              "Accept-Ranges": "bytes",
-            });
-            stream.pipe(response);
-          } else {
-            const [start, end] = request.headers.range
-              .split("=")[1]
-              .split("-")
-              .map(x => Number(x));
-            const stream = fs.createReadStream(filePath, { start, end });
-            response.writeHead(206, {
-              "Content-Type": "application/pdf",
-            });
-            stream.pipe(response);
-          }
-        });
-      })
-      .listen(0); /* Listen on a random free port */
-    port = server.address().port;
   });
 
   afterAll(function () {
     // Close the server from accepting new connections after all test finishes.
+    const { server } = tempServer;
     server.close();
+
+    tempServer = null;
   });
 
   it("read both http(s) and filesystem pdf files", async function () {
     const stream1 = new PDFNodeStream({
-      url: `http://127.0.0.1:${port}/tracemonkey.pdf`,
+      url: `http://127.0.0.1:${tempServer.port}/tracemonkey.pdf`,
       rangeChunkSize: 65536,
       disableStream: true,
       disableRange: true,
@@ -144,7 +113,7 @@ describe("node_stream", function () {
   it("read custom ranges for both http(s) and filesystem urls", async function () {
     const rangeSize = 32768;
     const stream1 = new PDFNodeStream({
-      url: `http://127.0.0.1:${port}/tracemonkey.pdf`,
+      url: `http://127.0.0.1:${tempServer.port}/tracemonkey.pdf`,
       length: pdfLength,
       rangeChunkSize: rangeSize,
       disableStream: true,

test/unit/test_utils.js

@@ -13,15 +13,16 @@
  * limitations under the License.
  */
 
+import { assert, isNodeJS } from "../../src/shared/util.js";
 import { NullStream, StringStream } from "../../src/core/stream.js";
 import { Page, PDFDocument } from "../../src/core/document.js";
-import { isNodeJS } from "../../src/shared/util.js";
 import { Ref } from "../../src/core/primitives.js";
 
-let fs;
+let fs, http;
 if (isNodeJS) {
   // Native packages.
   fs = await __non_webpack_import__("fs");
+  http = await __non_webpack_import__("http");
 }
 
 const TEST_PDFS_PATH = isNodeJS ? "./test/pdfs/" : "../pdfs/";
@@ -144,10 +145,54 @@ function createIdFactory(pageIndex) {
   return page._localIdFactory;
 }
 
+function createTemporaryNodeServer() {
+  assert(isNodeJS, "Should only be used in Node.js environments.");
+
+  // Create http server to serve pdf data for tests.
+  const server = http
+    .createServer((request, response) => {
+      const filePath = process.cwd() + "/test/pdfs" + request.url;
+      fs.lstat(filePath, (error, stat) => {
+        if (error) {
+          response.writeHead(404);
+          response.end(`File ${request.url} not found!`);
+          return;
+        }
+        if (!request.headers.range) {
+          const contentLength = stat.size;
+          const stream = fs.createReadStream(filePath);
+          response.writeHead(200, {
+            "Content-Type": "application/pdf",
+            "Content-Length": contentLength,
+            "Accept-Ranges": "bytes",
+          });
+          stream.pipe(response);
+        } else {
+          const [start, end] = request.headers.range
+            .split("=")[1]
+            .split("-")
+            .map(x => Number(x));
+          const stream = fs.createReadStream(filePath, { start, end });
+          response.writeHead(206, {
+            "Content-Type": "application/pdf",
+          });
+          stream.pipe(response);
+        }
+      });
+    })
+    .listen(0); /* Listen on a random free port */
+
+  return {
+    server,
+    port: server.address().port,
+  };
+}
+
 export {
   buildGetDocumentParams,
   CMAP_URL,
   createIdFactory,
+  createTemporaryNodeServer,
   DefaultFileReaderFactory,
   STANDARD_FONT_DATA_URL,
   TEST_PDFS_PATH,