Merge pull request #20831 from calixteman/internal_viewer

Add a new internal viewer to explore the structure of PDF files.

Author: timvandermeij

gulpfile.mjs

@@ -65,6 +65,7 @@ const IMAGE_DECODERS_LEGACY_DIR = BUILD_DIR + "image_decoders-legacy/";
 const DEFAULT_PREFERENCES_DIR = BUILD_DIR + "default_preferences/";
 const MINIFIED_DIR = BUILD_DIR + "minified/";
 const MINIFIED_LEGACY_DIR = BUILD_DIR + "minified-legacy/";
+const INTERNAL_VIEWER_DIR = BUILD_DIR + "internal-viewer/";
 const JSDOC_BUILD_DIR = BUILD_DIR + "jsdoc/";
 const GH_PAGES_DIR = BUILD_DIR + "gh-pages/";
 const DIST_DIR = BUILD_DIR + "dist/";
@@ -2368,6 +2369,52 @@ gulp.task("check_l10n", function (done) {
   });
 });
 
+function createInternalViewerBundle(defines) {
+  const viewerFileConfig = createWebpackConfig(defines, {
+    filename: "pdf_internal_viewer.mjs",
+    library: {
+      type: "module",
+    },
+  });
+  return gulp
+    .src("./web/pdf_internal_viewer.js", { encoding: false })
+    .pipe(webpack2Stream(viewerFileConfig));
+}
+
+function buildInternalViewer(defines, dir) {
+  fs.rmSync(dir, { recursive: true, force: true });
+
+  return ordered([
+    createMainBundle(defines).pipe(gulp.dest(dir + "build")),
+    createWorkerBundle(defines).pipe(gulp.dest(dir + "build")),
+    createInternalViewerBundle(defines).pipe(gulp.dest(dir + "web")),
+    preprocessHTML("web/pdf_internal_viewer.html", defines).pipe(
+      gulp.dest(dir + "web")
+    ),
+    preprocessCSS("web/pdf_internal_viewer.css", defines)
+      .pipe(
+        postcss([
+          postcssDirPseudoClass(),
+          discardCommentsCSS(),
+          postcssNesting(),
+          postcssLightDarkFunction({ preserve: true }),
+          autoprefixer(AUTOPREFIXER_CONFIG),
+        ])
+      )
+      .pipe(gulp.dest(dir + "web")),
+    createWasmBundle().pipe(gulp.dest(dir + "web/wasm")),
+  ]);
+}
+
+gulp.task(
+  "internal-viewer",
+  gulp.series(createBuildNumber, function createInternalViewer() {
+    console.log("\n### Creating internal viewer");
+    const defines = { ...DEFINES, GENERIC: true };
+    return buildInternalViewer(defines, INTERNAL_VIEWER_DIR);
+  })
+);
+
 function ghPagesPrepare() {
   console.log("\n### Creating web site");
 
@@ -2391,6 +2438,13 @@ function ghPagesPrepare() {
     gulp
       .src(JSDOC_BUILD_DIR + "**/*", { base: JSDOC_BUILD_DIR, encoding: false })
       .pipe(gulp.dest(GH_PAGES_DIR + "api/draft/")),
+    gulp
+      .src(INTERNAL_VIEWER_DIR + "**/*", {
+        base: INTERNAL_VIEWER_DIR,
+        encoding: false,
+        removeBOM: false,
+      })
+      .pipe(gulp.dest(GH_PAGES_DIR + "internal-viewer/")),
   ]);
 }
 
@@ -2442,6 +2496,7 @@ gulp.task(
   gulp.series(
     "generic",
     "generic-legacy",
+    "internal-viewer",
     "jsdoc",
     ghPagesPrepare,
     "metalsmith"

src/core/document.js

@@ -22,6 +22,7 @@ import {
   isArrayEqual,
   makeArr,
   objectSize,
+  OPS,
   PageActionEventType,
   RenderingIntentFlag,
   shadow,
@@ -37,6 +38,17 @@ import {
   PopupAnnotation,
   WidgetAnnotation,
 } from "./annotation.js";
+import {
+  Cmd,
+  Dict,
+  EOF,
+  isName,
+  isRefsEqual,
+  Name,
+  Ref,
+  RefSet,
+  RefSetCache,
+} from "./primitives.js";
 import {
   collectActions,
   getInheritableProperty,
@@ -51,27 +63,21 @@ import {
   XRefEntryException,
   XRefParseException,
 } from "./core_utils.js";
-import {
-  Dict,
-  isName,
-  isRefsEqual,
-  Name,
-  Ref,
-  RefSet,
-  RefSetCache,
-} from "./primitives.js";
+import { EvaluatorPreprocessor, PartialEvaluator } from "./evaluator.js";
 import { getXfaFontDict, getXfaFontName } from "./xfa_fonts.js";
+import { Lexer, Linearization, Parser } from "./parser.js";
 import { NullStream, Stream } from "./stream.js";
 import { BaseStream } from "./base_stream.js";
 import { calculateMD5 } from "./calculate_md5.js";
 import { Catalog } from "./catalog.js";
 import { clearGlobalCaches } from "./cleanup_helper.js";
 import { DatasetReader } from "./dataset_reader.js";
 import { Intersector } from "./intersector.js";
-import { Linearization } from "./parser.js";
+import { LocalColorSpaceCache } from "./image_utils.js";
 import { ObjectLoader } from "./object_loader.js";
 import { OperatorList } from "./operator_list.js";
-import { PartialEvaluator } from "./evaluator.js";
+import { PDFFunctionFactory } from "./function.js";
+import { PDFImage } from "./image.js";
 import { StreamsSequenceStream } from "./decode_stream.js";
 import { StructTreePage } from "./struct_tree.js";
 import { XFAFactory } from "./xfa/factory.js";
@@ -2030,6 +2036,219 @@ class PDFDocument {
       AnnotationFactory.createGlobals(this.pdfManager)
     );
   }
+
+  async toJSObject(value, firstCall = true) {
+    if (typeof PDFJSDev !== "undefined" && PDFJSDev.test("MOZCENTRAL")) {
+      throw new Error("Not implemented: toJSObject");
+    }
+
+    if (value === null && firstCall) {
+      return this.toJSObject(this.xref.trailer, false);
+    }
+    if (value instanceof Dict) {
+      const obj = Object.create(null);
+      const isPage = isName(value.get("Type"), "Page");
+      for (const [key, val] of value.getRawEntries()) {
+        obj[key] =
+          isPage && key === "Contents"
+            ? _getContentTokens(val, this.xref)
+            : await this.toJSObject(val, false);
+      }
+      return obj;
+    }
+    if (Array.isArray(value)) {
+      return Promise.all(value.map(v => this.toJSObject(v, false)));
+    }
+    if (value instanceof Ref) {
+      if (firstCall) {
+        return this.toJSObject(this.xref.fetch(value), false);
+      }
+      const result = Object.create(null);
+      result.num = value.num;
+      result.gen = value.gen;
+      return result;
+    }
+    if (value instanceof BaseStream) {
+      const { dict } = value;
+      const obj = Object.create(null);
+      obj.dict = await this.toJSObject(dict, false);
+
+      if (
+        isName(dict.get("Type"), "XObject") &&
+        isName(dict.get("Subtype"), "Image")
+      ) {
+        try {
+          const pdfFunctionFactory = new PDFFunctionFactory({
+            xref: this.xref,
+            isEvalSupported: this.pdfManager.evaluatorOptions.isEvalSupported,
+          });
+          const imageObj = await PDFImage.buildImage({
+            xref: this.xref,
+            res: Dict.empty,
+            image: value,
+            pdfFunctionFactory,
+            globalColorSpaceCache: this.catalog.globalColorSpaceCache,
+            localColorSpaceCache: new LocalColorSpaceCache(),
+          });
+          const imgData = await imageObj.createImageData(
+            /* forceRGBA = */ true,
+            /* isOffscreenCanvasSupported = */ false
+          );
+          obj.imageData = {
+            width: imgData.width,
+            height: imgData.height,
+            kind: imgData.kind,
+            data: imgData.data,
+          };
+          return obj;
+        } catch {
+          // Fall through to regular byte stream if image decoding fails.
+        }
+      }
+
+      if (isName(dict.get("Subtype"), "Form")) {
+        obj.bytes = value.getString();
+        value.reset();
+        const { instructions, cmdNames } = _groupIntoInstructions(
+          _tokenizeStream(value, this.xref)
+        );
+        obj.contentStream = true;
+        obj.instructions = instructions;
+        obj.cmdNames = cmdNames;
+        return obj;
+      }
+
+      obj.bytes = value.getString();
+      return obj;
+    }
+    return value;
+  }
+}
+
+function _tokenizeStream(stream, xref) {
+  const tokens = [];
+  const parser = new Parser({
+    lexer: new Lexer(stream),
+    xref,
+    allowStreams: false,
+  });
+  while (true) {
+    let obj;
+    try {
+      obj = parser.getObj();
+    } catch {
+      break;
+    }
+    if (obj === EOF) {
+      break;
+    }
+    const token = _tokenToJSObject(obj);
+    if (token !== null) {
+      tokens.push(token);
+    }
+  }
+  return tokens;
+}
+
+function _getContentTokens(contentsVal, xref) {
+  const refs = Array.isArray(contentsVal) ? contentsVal : [contentsVal];
+  const rawContents = [];
+  const tokens = [];
+  for (const rawRef of refs) {
+    if (rawRef instanceof Ref) {
+      rawContents.push({ num: rawRef.num, gen: rawRef.gen });
+    }
+    const stream = xref.fetchIfRef(rawRef);
+    if (!(stream instanceof BaseStream)) {
+      continue;
+    }
+    tokens.push(..._tokenizeStream(stream, xref));
+  }
+  const { instructions, cmdNames } = _groupIntoInstructions(tokens);
+  return { contentStream: true, instructions, cmdNames, rawContents };
+}
+
+// Lazily-built reverse map: OPS numeric id → property name string.
+let _opsIdToName = null;
+
+function _getOpsIdToName() {
+  if (!_opsIdToName) {
+    _opsIdToName = Object.create(null);
+    for (const [name, id] of Object.entries(OPS)) {
+      _opsIdToName[id] = name;
+    }
+  }
+  return _opsIdToName;
+}
+
+function _groupIntoInstructions(tokens) {
+  const { opMap } = EvaluatorPreprocessor;
+  const opsIdToName = _getOpsIdToName();
+  const instructions = [];
+  const cmdNames = Object.create(null);
+  const argBuffer = [];
+  for (const token of tokens) {
+    if (token.type !== "cmd") {
+      argBuffer.push(token);
+      continue;
+    }
+    const op = opMap[token.value];
+    if (op && !(token.value in cmdNames)) {
+      cmdNames[token.value] = opsIdToName[op.id];
+    }
+    let args;
+    if (!op || op.variableArgs) {
+      // Unknown command or variable args: consume all pending args.
+      args = argBuffer.splice(0);
+    } else {
+      // Fixed args: consume exactly numArgs, orphan the rest.
+      const orphanCount = Math.max(0, argBuffer.length - op.numArgs);
+      for (let i = 0; i < orphanCount; i++) {
+        instructions.push({ cmd: null, args: [argBuffer.shift()] });
+      }
+      args = argBuffer.splice(0);
+    }
+    instructions.push({ cmd: token.value, args });
+  }
+  for (const t of argBuffer) {
+    instructions.push({ cmd: null, args: [t] });
+  }
+  return { instructions, cmdNames };
+}
+
+function _tokenToJSObject(obj) {
+  if (obj instanceof Cmd) {
+    return { type: "cmd", value: obj.cmd };
+  }
+  if (obj instanceof Name) {
+    return { type: "name", value: obj.name };
+  }
+  if (obj instanceof Ref) {
+    return { type: "ref", num: obj.num, gen: obj.gen };
+  }
+  if (Array.isArray(obj)) {
+    return { type: "array", value: obj.map(_tokenToJSObject) };
+  }
+  if (obj instanceof Dict) {
+    const result = Object.create(null);
+    for (const [key, val] of obj.getRawEntries()) {
+      result[key] = _tokenToJSObject(val);
+    }
+    return { type: "dict", value: result };
+  }
+  if (typeof obj === "number") {
+    return { type: "number", value: obj };
+  }
+  if (typeof obj === "string") {
+    return { type: "string", value: obj };
+  }
+  if (typeof obj === "boolean") {
+    return { type: "boolean", value: obj };
+  }
+  if (obj === null) {
+    return { type: "null" };
+  }
+  return null;
 }
 
 export { Page, PDFDocument };

src/core/worker.js

@@ -952,6 +952,22 @@ class WorkerMessageHandler {
       return pdfManager.fontFallback(data.id, handler);
     });
 
+    handler.on("GetRawData", async function ({ ref, page }) {
+      if (typeof PDFJSDev !== "undefined" && PDFJSDev.test("MOZCENTRAL")) {
+        throw new Error("Not implemented: GetRawData");
+      }
+      let value = null;
+      if (page >= 1) {
+        value = (await pdfManager.ensureCatalog("getPageDict", [page - 1]))[1];
+      } else if (ref) {
+        value =
+          typeof ref === "string"
+            ? Ref.fromString(ref)
+            : Ref.get(ref.num, ref.gen);
+      }
+      return pdfManager.ensureDoc("toJSObject", [value]);
+    });
+
     handler.on("Cleanup", function (data) {
       return pdfManager.cleanup(/* manuallyTriggered = */ true);
     });