Fix stream parsing bug

bcmsbond · Jul 6, 2019 · 764fb4b · 764fb4b
1 parent 18878dc
commit 764fb4b
Show file tree

Hide file tree

Showing 7 changed files with 38 additions and 8 deletions.
diff --git a/src/core/errors.ts b/src/core/errors.ts
@@ -51,7 +51,7 @@ export class MissingCatalogError extends Error {
 
 /***** Parser Errors ******/
 
-interface Position {
+export interface Position {
   line: number;
   column: number;
   offset: number;

diff --git a/src/core/parser/PDFObjectParser.ts b/src/core/parser/PDFObjectParser.ts
@@ -1,6 +1,7 @@
 import {
   PDFObjectParsingError,
   PDFStreamParsingError,
+  Position,
   UnbalancedParenthesisError,
 } from 'src/core/errors';
 import PDFArray from 'src/core/objects/PDFArray';
@@ -221,9 +222,27 @@ class PDFObjectParser extends BaseParser {
     }
 
     const start = this.bytes.offset();
+    let end: number;
 
-    // TODO: Try to use dict's `Length` entry, but use this as backup...
+    const Length = dict.get(PDFName.of('Length'));
+    if (Length instanceof PDFNumber) {
+      end = start + Length.value();
+      this.bytes.moveTo(end);
+      this.skipWhitespaceAndComments();
+      if (!this.matchKeyword(Keywords.endstream)) {
+        this.bytes.moveTo(start);
+        end = this.findEndOfStreamFallback(startPos);
+      }
+    } else {
+      end = this.findEndOfStreamFallback(startPos);
+    }
+
+    const contents = this.bytes.slice(start, end);
 
+    return PDFRawStream.of(dict, contents);
+  }
+
+  protected findEndOfStreamFallback(startPos: Position) {
     // Move to end of stream, while handling nested streams
     let nestingLvl = 1;
     let end = this.bytes.offset();
@@ -249,9 +268,7 @@ class PDFObjectParser extends BaseParser {
 
     if (nestingLvl !== 0) throw new PDFStreamParsingError(startPos);
 
-    const contents = this.bytes.slice(start, end);
-
-    return PDFRawStream.of(dict, contents);
+    return end;
   }
 }
 

diff --git a/src/core/parser/PDFParser.ts b/src/core/parser/PDFParser.ts
@@ -165,8 +165,14 @@ class PDFParser extends PDFObjectParser {
   private tryToParseInvalidIndirectObject() {
     const startPos = this.bytes.position();
 
+    console.warn(
+      `Trying to parse invalid object: ${JSON.stringify(startPos)})`,
+    );
+
     const ref = this.parseIndirectObjectHeader();
 
+    console.warn(`Invalid object ref: ${ref}`);
+
     this.skipWhitespaceAndComments();
     const start = this.bytes.offset();
 

diff --git a/src/core/structures/PDFObjectStream.ts b/src/core/structures/PDFObjectStream.ts
@@ -16,7 +16,7 @@ class PDFObjectStream extends PDFFlateStream {
     encode = true,
   ) => new PDFObjectStream(context, objects, encode);
 
-  readonly objects: IndirectObject[];
+  private readonly objects: IndirectObject[];
   private readonly offsets: Array<[number, number]>;
   private readonly offsetsString: string;
 
@@ -36,6 +36,10 @@ class PDFObjectStream extends PDFFlateStream {
     this.dict.set(PDFName.of('First'), PDFNumber.of(this.offsetsString.length));
   }
 
+  getObjectsCount(): number {
+    return this.objects.length;
+  }
+
   clone(context?: PDFContext): PDFObjectStream {
     return PDFObjectStream.withContextAndObjects(
       context || this.dict.context,

diff --git a/src/core/writers/PDFStreamWriter.ts b/src/core/writers/PDFStreamWriter.ts
@@ -1,5 +1,6 @@
 import PDFHeader from 'src/core/document/PDFHeader';
 import PDFTrailer from 'src/core/document/PDFTrailer';
+import PDFInvalidObject from 'src/core/objects/PDFInvalidObject';
 import PDFName from 'src/core/objects/PDFName';
 import PDFNumber from 'src/core/objects/PDFNumber';
 import PDFObject from 'src/core/objects/PDFObject';
@@ -64,6 +65,7 @@ class PDFStreamWriter extends PDFWriter {
       const shouldNotCompress =
         ref === this.context.trailerInfo.Encrypt ||
         object instanceof PDFStream ||
+        object instanceof PDFInvalidObject ||
         ref.generationNumber !== 0;
 
       if (shouldNotCompress) {

diff --git a/src/core/writers/PDFWriter.ts b/src/core/writers/PDFWriter.ts
@@ -78,7 +78,8 @@ class PDFWriter {
       buffer[offset++] = CharCodes.Newline;
       buffer[offset++] = CharCodes.Newline;
 
-      const n = object instanceof PDFObjectStream ? object.objects.length : 1;
+      const n =
+        object instanceof PDFObjectStream ? object.getObjectsCount() : 1;
       if (this.shouldWaitForTick(n)) await waitForTick();
     }
 

diff --git a/tests/core/parser/PDFParser.spec.ts b/tests/core/parser/PDFParser.spec.ts
@@ -190,7 +190,7 @@ describe(`PDFParser`, () => {
 
     expect(context.header).toBeInstanceOf(PDFHeader);
     expect(context.header.toString()).toEqual('%PDF-1.6\n%');
-    expect(context.enumerateIndirectObjects().length).toBe(208);
+    expect(context.enumerateIndirectObjects().length).toBe(17);
   });
 
   it(`can fix incorrect values for /Root`, async () => {