Mostly working

* ranges passed through
* updates possible

examples/forge-functionality/api_examples.js

@@ -11,7 +11,7 @@ const STANDARD_FONT_DATA_URL =
 
 // Loading file from file system into typed array.
 const pdfPath =
-  process.argv[2] || "../../web/compressed.tracemonkey-pldi-09.pdf";
+  process.argv[2] || "C:\\Users\\kj131\\pdf-forge\\test_pdfs\\ISO_32000-2_2020(en).pdf";
 const data = new Uint8Array(fs.readFileSync(pdfPath));
 
 // Load the PDF file.
@@ -21,12 +21,46 @@ const loadingTask = getDocument({
   cMapPacked: CMAP_PACKED,
   standardFontDataUrl: STANDARD_FONT_DATA_URL,
 });
+test(loadingTask);
+async function test(loading) {
   try {
-  const pdfDocument = await loadingTask.promise;
+    const pdfDocument = await loading.promise;
     console.log("# PDF document loaded.");
-  const page = await pdfDocument.getPage(1);
+    const page = await pdfDocument.getPage(4);
+    printOpList(page);
+    console.time("contents");
+    const contents = await page.getContents();
+    console.timeEnd("contents");
+    console.time("oplist");
     const opList = await page.getOperatorList();
-  console.log(opList);
+    console.timeEnd("oplist");
+    // console.log(opList);
+    let newContents = "";
+    for (let i = 0; i < 5; i++) {
+      const range = opList.rangeArray[i];
+      if (range) {
+        newContents += contents.slice(range[0], range[1]);
+        newContents += "\n";
+      }
+    }
+    console.log(newContents);
+    await page.updateContents(newContents);
+    await printOpList(page);
   } catch (e) {
     console.error(e);
   }
+}
+
+async function printOpList(page) {
+  const contents = await page.getContents();
+  const opList = await page.getOperatorList();
+  // console.log(opList);
+  const ops = [];
+  for (let i = 0; i < opList.rangeArray.length; i++) {
+    const range = opList.rangeArray[i];
+    if (range) {
+      ops.push(contents.slice(range[0], range[1]));
+    }
+  }
+  console.log(ops.slice(0, 100));
+}

examples/forge-functionality/fetch_primitives.js

@@ -45,7 +45,7 @@ function bytesToString(bytes) {
 }
 
 async function parse(doc) {
-  const path = "/Page2/Contents/2";
+  const path = "/Page2/Contents/1";
   let [stream] = await getPrimitive(path, doc);
   const lexer = new Lexer(stream);
   const parser = new Parser({ lexer, xref: doc.xref, trackRanges: true });
@@ -63,12 +63,12 @@ async function parse(doc) {
   const bytes = stream.getBytes();
   const classes = new Set();
   for (const o of objs) {
-    console.log(o[0].constructor.name);
+    // console.log(o[0].constructor.name);
     classes.add(o[0].constructor.name);
     const lexemmeBytes = bytes.slice(o[1], o[2]);
-    console.log(bytesToString(lexemmeBytes));
+    // console.log(bytesToString(lexemmeBytes));
   }
-  console.log("unique classes", classes);
+  // console.log("unique classes", classes);
   [stream] = await getPrimitive(path, doc);
   const preprocessor = new EvaluatorPreprocessor(stream, doc.xref);
   const operation = {};
@@ -78,10 +78,10 @@ async function parse(doc) {
     const fn = operation.fn;
     const range = operation.range;
     const op = bytesToString(bytes.slice(range[0], range[1]));
-    console.log(args, fn);
-    console.log(`${range[0]}------------------------------------`);
-    console.log(op);
-    console.log(`${range[1]}------------------------------------`);
+    // console.log(args, fn);
+    // console.log(`----------------- ${range} -------------------`);
+    console.log(`${fn}: ${op}`);
+    // console.log(`---------------------------------------------`);
   }
   // console.time("xref");
   // let table = await retrieveXref(doc);

src/core/document.js

@@ -65,7 +65,7 @@ import { Catalog } from "./catalog.js";
 import { clearGlobalCaches } from "./cleanup_helper.js";
 import { DatasetReader } from "./dataset_reader.js";
 import { Linearization } from "./parser.js";
-import { NullStream } from "./stream.js";
+import { NullStream, StringStream } from "./stream.js";
 import { ObjectLoader } from "./object_loader.js";
 import { OperatorList } from "./operator_list.js";
 import { PartialEvaluator } from "./evaluator.js";
@@ -107,6 +107,8 @@ class Page {
     this.resourcesPromise = null;
     this.xfaFactory = xfaFactory;
 
+    this.updatedContents = null;
+
     const idCounters = {
       obj: 0,
     };
@@ -246,10 +248,19 @@ class Page {
     throw reason;
   }
 
+  setContents(newContents) {
+    this.updatedContents = newContents;
+  }
+
   /**
    * @returns {Promise<BaseStream>}
    */
   getContentStream() {
+    if (this.updatedContents !== null) {
+      return new Promise(resolve => {
+        resolve(new StringStream(this.updatedContents));
+      });
+    }
     return this.pdfManager.ensure(this, "content").then(content => {
       if (content instanceof BaseStream) {
         return content;
@@ -426,8 +437,11 @@ class Page {
     cacheKey,
     annotationStorage = null,
     modifiedIds = null,
+    contentOverride = null,
   }) {
-    const contentStreamPromise = this.getContentStream();
+    const contentStreamPromise = contentOverride
+      ? new StringStream(contentOverride)
+      : this.getContentStream();
     const resourcesPromise = this.loadResources([
       "ColorSpace",
       "ExtGState",

src/core/evaluator.js

@@ -802,6 +802,7 @@ class PartialEvaluator {
     operatorList.addImageOps(
       OPS.paintImageXObject,
       args,
+      range,
       optionalContent,
       hasMask
     );
@@ -1445,15 +1446,20 @@ class PartialEvaluator {
           minMax = [Infinity, Infinity, -Infinity, -Infinity];
           break;
       }
-      operatorList.addOp(OPS.constructPath, [[fn], args, minMax], range);
+      operatorList.addOp(
+        OPS.constructPath,
+        [[fn], args, minMax, [range[0], range[1]]],
+        range
+      );
 
       if (parsingText) {
-        operatorList.addOp(OPS.restore, null, range);
+        operatorList.addOp(OPS.restore, null);
       }
     } else {
       const opArgs = operatorList.argsArray[lastIndex];
       opArgs[0].push(fn);
       opArgs[1].push(...args);
+      opArgs[3].push(...range);
       const minMax = opArgs[2];
 
       const opRange = operatorList.rangeArray[lastIndex];
@@ -5215,7 +5221,7 @@ class EvaluatorPreprocessor {
 
         operation.fn = fn;
         operation.args = args;
-        const end = this.parser.getPosition();
+        const end = this.parser.getEnd();
         operation.range = [start, end];
         return true;
       }

src/core/parser.js

@@ -59,12 +59,7 @@ function getInlineImageCacheKey(bytes) {
 }
 
 class Parser {
-  constructor({
-    lexer,
-    xref,
-    allowStreams = false,
-    recoveryMode = false,
-  }) {
+  constructor({ lexer, xref, allowStreams = false, recoveryMode = false }) {
     this.lexer = lexer;
     this.xref = xref;
     this.allowStreams = allowStreams;
@@ -75,8 +70,8 @@ class Parser {
   }
 
   refill() {
-    const [buf1, start1, end1] = this.lexer.getObj(true);
-    const [buf2, start2, end2] = this.lexer.getObj(true);
+    const [buf1, start1, end1] = this.lexer.getObjWithRange();
+    const [buf2, start2, end2] = this.lexer.getObjWithRange();
     this.buf1 = buf1;
     this.range1 = [start1, end1];
     this.buf2 = buf2;
@@ -87,12 +82,14 @@ class Parser {
     if (this.buf2 instanceof Cmd && this.buf2.cmd === "ID") {
       this.buf1 = this.buf2;
       this.buf2 = null;
+      this.lastEnd = this.range1[1];
       this.range1 = this.range2;
       this.range2 = null;
     } else {
       this.buf1 = this.buf2;
+      this.lastEnd = this.range1[1];
       this.range1 = this.range2;
-      const [buf2, start2, end2] = this.lexer.getObj(true);
+      const [buf2, start2, end2] = this.lexer.getObjWithRange();
       this.buf2 = buf2;
       this.range2 = [start2, end2];
     }
@@ -116,6 +113,10 @@ class Parser {
     return this.range1 ? this.range1[0] : 0;
   }
 
+  getEnd() {
+    return this.lastEnd ?? 0;
+  }
+
   getObjWithRange(cipherTransform = null) {
     const start = this.range1[0];
     const obj = this.getObj(cipherTransform);
@@ -1229,14 +1230,18 @@ class Lexer {
   }
 
   getObjWithRange() {
-    // at the start of getObj() the stream has stepped beyond currentChar by one
-    const start = this.stream.pos - 1;
+    const ch = this._skipWhitespaceAndComments();
+    if (ch === EOF) {
+      return [ch, -1, -1];
+    }
+    // currentChar is always at pos - 1
+    const start = Math.max(this.stream.pos - 1, 0);
     const obj = this.getObj();
-    const end = this.stream.pos;
+    const end = this.stream.pos - 1;
     return [obj, start, end];
   }
 
-  getObj(withRange = false) {
+  _skipWhitespaceAndComments() {
     // Skip whitespace and comments.
     let comment = false;
     let ch = this.currentChar;
@@ -1255,16 +1260,14 @@ class Lexer {
       }
       ch = this.nextChar();
     }
-    const start = this.stream.pos - 1;
-    const obj = this.extracted(ch);
-    const end = this.stream.pos - 1;
-    if (withRange) {
-      return [start, obj, end];
-    }
-    return obj;
+    return ch;
   }
 
-  extracted(ch) {
+  getObj() {
+    let ch = this._skipWhitespaceAndComments();
+    if (ch === EOF) {
+      return ch;
+    }
     // Start reading a token.
     switch (ch | 0) {
       case 0x30: // '0'

src/core/worker.js

@@ -748,6 +748,31 @@ class WorkerMessageHandler {
       }
     );
 
+    handler.on("StreamContents", function(data, sink) {
+      const pageIndex = data.pageIndex;
+      pdfManager.getPage(pageIndex).then(function(page) {
+        page.getContentStream().then(stream => {
+          let byte;
+          let string = "";
+          while ((byte = stream.getByte()) !== -1) {
+            string += String.fromCharCode(byte);
+          }
+          sink.enqueue(string, string.length);
+          sink.close();
+        });
+      });
+    });
+
+    handler.on("UpdateContents", function (data) {
+      return new Promise(resolve => {
+        const pageIndex = data.pageIndex;
+        pdfManager.getPage(pageIndex).then(function (page) {
+          page.setContents(data.value);
+          resolve();
+        });
+      });
+    });
+
     handler.on("GetOperatorList", function (data, sink) {
       const pageIndex = data.pageIndex;
       pdfManager.getPage(pageIndex).then(function (page) {
@@ -767,6 +792,7 @@ class WorkerMessageHandler {
             cacheKey: data.cacheKey,
             annotationStorage: data.annotationStorage,
             modifiedIds: data.modifiedIds,
+            contentOverride: data.contentOverride,
           })
           .then(
             function (operatorListInfo) {

src/display/api.js

@@ -35,6 +35,7 @@ import {
   AbortException,
   AnnotationMode,
   assert,
+  djb2Hash,
   FeatureTest,
   getVerbosityLevel,
   info,
@@ -1660,6 +1661,37 @@ class PDFPageProxy {
     return renderTask;
   }
 
+  updateContents(newContents) {
+    if (!newContents) {
+      throw new Error("Contents may not be null or undefined");
+    }
+    this._intentStates.clear();
+    return this._transport.updateContents(newContents, this._pageIndex);
+  }
+
+  getContents() {
+    const readableStream = this._transport.streamContents(this._pageIndex);
+
+    return new Promise(function (resolve, reject) {
+      function pump() {
+        reader.read().then(function ({ value, done }) {
+          if (done) {
+            resolve(textContent.text);
+            return;
+          }
+          textContent.text += value;
+          pump();
+        }, reject);
+      }
+
+      const reader = readableStream.getReader();
+      const textContent = {
+        text: "",
+      };
+      pump();
+    });
+  }
+
   /**
    * @param {GetOperatorListParameters} params - Page getOperatorList
    *   parameters.
@@ -1671,6 +1703,7 @@ class PDFPageProxy {
     annotationMode = AnnotationMode.ENABLE,
     printAnnotationStorage = null,
     isEditing = false,
+    contentOverride = null,
   } = {}) {
     if (typeof PDFJSDev !== "undefined" && !PDFJSDev.test("GENERIC")) {
       throw new Error("Not implemented: getOperatorList");
@@ -1688,7 +1721,8 @@ class PDFPageProxy {
       annotationMode,
       printAnnotationStorage,
       isEditing,
-      /* isOpList = */ true
+      /* isOpList = */ true,
+      contentOverride
     );
     let intentState = this._intentStates.get(intentArgs.cacheKey);
     if (!intentState) {
@@ -1910,6 +1944,7 @@ class PDFPageProxy {
     cacheKey,
     annotationStorageSerializable,
     modifiedIds,
+    contentOverride,
   }) {
     if (typeof PDFJSDev === "undefined" || PDFJSDev.test("TESTING")) {
       assert(
@@ -1927,6 +1962,7 @@ class PDFPageProxy {
         cacheKey,
         annotationStorage: map,
         modifiedIds,
+        contentOverride,
       },
       transfer
     );
@@ -2528,7 +2564,8 @@ class WorkerTransport {
     annotationMode = AnnotationMode.ENABLE,
     printAnnotationStorage = null,
     isEditing = false,
-    isOpList = false
+    isOpList = false,
+    contentOverride = null
   ) {
     let renderingIntent = RenderingIntentFlag.DISPLAY; // Default value.
     let annotationStorageSerializable = SerializableEmpty;
@@ -2586,11 +2623,16 @@ class WorkerTransport {
       modifiedIdsHash,
     ];
 
+    if (contentOverride) {
+      cacheKeyBuf.push(djb2Hash(contentOverride));
+    }
+
     return {
       renderingIntent,
       cacheKey: cacheKeyBuf.join("_"),
       annotationStorageSerializable,
       modifiedIds,
+      contentOverride,
     };
   }
 
@@ -2974,6 +3016,23 @@ class WorkerTransport {
     return this.messageHandler.sendWithPromise("GetXRefEntries", null);
   }
 
+  streamContents(pageIndex) {
+    return this.messageHandler.sendWithStream("StreamContents", {
+      pageIndex,
+    });
+  }
+
+  /**
+   * @returns {Promise<void>} A promise that is resolved once the contents
+   * are updated.
+   */
+  updateContents(newContents, pageIndex) {
+    return this.messageHandler.sendWithPromise("UpdateContents", {
+      value: newContents,
+      pageIndex,
+    });
+  }
+
   saveDocument() {
     if (this.annotationStorage.size <= 0) {
       warn(

src/shared/util.js

@@ -578,6 +578,16 @@ function objectFromMap(map) {
   return obj;
 }
 
+// fast and easy hash
+function djb2Hash(str) {
+  let hash = 5381;
+  for (let i = 0; i < str.length; i++) {
+    hash = (hash << 5) + hash + str.charCodeAt(i);
+    hash &= hash; // Convert to 32-bit integer
+  }
+  return hash >>> 0; // Convert to unsigned
+}
+
 // Checks the endianness of the platform.
 function isLittleEndian() {
   const buffer8 = new Uint8Array(4);
@@ -1136,6 +1146,7 @@ export {
   BASELINE_FACTOR,
   bytesToString,
   createValidAbsoluteUrl,
+  djb2Hash,
   DocumentActionEventType,
   FeatureTest,
   FONT_IDENTITY_MATRIX,