Decompress FlatDecode images in using DecompressStream stuff

mozilla · May 24, 2024 · 4878ce4 · 4878ce4
1 parent 18a7bd6
commit 4878ce4
Show file tree

Hide file tree

Showing 8 changed files with 125 additions and 20 deletions.
diff --git a/src/core/base_stream.js b/src/core/base_stream.js
@@ -44,6 +44,18 @@ class BaseStream {
     unreachable("Abstract method `getBytes` called");
   }
 
+  async getImageData(_ignoreColorSpace) {
+    unreachable("Abstract method `getImageData` called");
+  }
+
+  get isAsync() {
+    return false;
+  }
+
+  get canDecodeImageFromBuffer() {
+    return false;
+  }
+
   peekByte() {
     const peekedByte = this.getByte();
     if (peekedByte !== -1) {

diff --git a/src/core/decode_stream.js b/src/core/decode_stream.js
@@ -99,6 +99,24 @@ class DecodeStream extends BaseStream {
     return this.buffer.subarray(pos, end);
   }
 
+  async getImageData(ignoreColorSpace = false) {
+    if (!this.canDecodeImageFromBuffer) {
+      return this.getBytes(0, ignoreColorSpace);
+    }
+    if (!this.stream.isAsync) {
+      return this.decodeImage(null, ignoreColorSpace);
+    }
+    let data;
+    try {
+      data = await this.stream.asyncGetBytes();
+    } catch {
+      this.reset();
+      return this.decodeImage(null, ignoreColorSpace);
+    }
+
+    return this.decodeImage(data, ignoreColorSpace);
+  }
+
   reset() {
     this.pos = 0;
   }

diff --git a/src/core/evaluator.js b/src/core/evaluator.js
@@ -3405,7 +3405,7 @@ class PartialEvaluator {
               });
             }
             break;
-          case OPS.beginMarkedContentProps:
+          /*case OPS.beginMarkedContentProps:
             flushTextContentItem();
             if (includeMarkedContent) {
               markedContentData.level++;
@@ -3422,7 +3422,7 @@ class PartialEvaluator {
                 tag: args[0] instanceof Name ? args[0].name : null,
               });
             }
-            break;
+            break;*/
           case OPS.endMarkedContent:
             flushTextContentItem();
             if (includeMarkedContent) {

diff --git a/src/core/flate_stream.js b/src/core/flate_stream.js
@@ -148,6 +148,38 @@ class FlateStream extends DecodeStream {
     this.codeBuf = 0;
   }
 
+  get isAsync() {
+    return true;
+  }
+
+  async asyncGetBytes() {
+    const ds = new DecompressionStream("deflate");
+    const writer = ds.writable.getWriter();
+    this.stream.str.reset();
+    const bytes = this.stream.str.getBytes();
+    writer.write(bytes);
+    writer.close();
+
+    const chunks = [];
+    let totalLength = 0;
+    const reader = ds.readable.getReader();
+    while (true) {
+      const { value, done } = await reader.read();
+      if (done) {
+        break;
+      }
+      chunks.push(value);
+      totalLength += value.byteLength;
+    }
+    reader.releaseLock();
+    const data = new Uint8Array(totalLength);
+    let offset = 0;
+    for (const chunk of chunks) {
+      data.set(new Uint8Array(chunk), offset);
+      offset += chunk.byteLength;
+    }
+  }
+
   getBits(bits) {
     const str = this.str;
     let codeSize = this.codeSize;

diff --git a/src/core/image.js b/src/core/image.js
@@ -565,7 +565,7 @@ class PDFImage {
     return output;
   }
 
-  fillOpacity(rgbaBuf, width, height, actualHeight, image) {
+  async fillOpacity(rgbaBuf, width, height, actualHeight, image) {
     if (typeof PDFJSDev === "undefined" || PDFJSDev.test("TESTING")) {
       assert(
         rgbaBuf instanceof Uint8ClampedArray,
@@ -580,7 +580,7 @@ class PDFImage {
       sw = smask.width;
       sh = smask.height;
       alphaBuf = new Uint8ClampedArray(sw * sh);
-      smask.fillGrayBuffer(alphaBuf);
+      await smask.fillGrayBuffer(alphaBuf);
       if (sw !== width || sh !== height) {
         alphaBuf = resizeImageMask(alphaBuf, smask.bpc, sw, sh, width, height);
       }
@@ -590,7 +590,7 @@ class PDFImage {
         sh = mask.height;
         alphaBuf = new Uint8ClampedArray(sw * sh);
         mask.numComps = 1;
-        mask.fillGrayBuffer(alphaBuf);
+        await mask.fillGrayBuffer(alphaBuf);
 
         // Need to invert values in rgbaBuf
         for (i = 0, ii = sw * sh; i < ii; ++i) {
@@ -716,7 +716,7 @@ class PDFImage {
         drawWidth === originalWidth &&
         drawHeight === originalHeight
       ) {
-        const data = this.getImageBytes(originalHeight * rowBytes, {});
+        const data = await this.getImageBytes(originalHeight * rowBytes, {});
         if (isOffscreenCanvasSupported) {
           if (mustBeResized) {
             return ImageResizer.createImage(
@@ -774,7 +774,7 @@ class PDFImage {
           }
 
           if (isHandled) {
-            const rgba = this.getImageBytes(imageLength, {
+            const rgba = await this.getImageBytes(imageLength, {
               drawWidth,
               drawHeight,
               forceRGBA: true,
@@ -794,7 +794,7 @@ class PDFImage {
             case "DeviceRGB":
             case "DeviceCMYK":
               imgData.kind = ImageKind.RGB_24BPP;
-              imgData.data = this.getImageBytes(imageLength, {
+              imgData.data = await this.getImageBytes(imageLength, {
                 drawWidth,
                 drawHeight,
                 forceRGB: true,
@@ -809,7 +809,7 @@ class PDFImage {
       }
     }
 
-    const imgArray = this.getImageBytes(originalHeight * rowBytes, {
+    const imgArray = await this.getImageBytes(originalHeight * rowBytes, {
       internal: true,
     });
     // imgArray can be incomplete (e.g. after CCITT fax encoding).
@@ -852,7 +852,7 @@ class PDFImage {
       maybeUndoPreblend = true;
 
       // Color key masking (opacity) must be performed before decoding.
-      this.fillOpacity(data, drawWidth, drawHeight, actualHeight, comps);
+      await this.fillOpacity(data, drawWidth, drawHeight, actualHeight, comps);
     }
 
     if (this.needsDecode) {
@@ -893,7 +893,7 @@ class PDFImage {
     return imgData;
   }
 
-  fillGrayBuffer(buffer) {
+  async fillGrayBuffer(buffer) {
     if (typeof PDFJSDev === "undefined" || PDFJSDev.test("TESTING")) {
       assert(
         buffer instanceof Uint8ClampedArray,
@@ -913,7 +913,9 @@ class PDFImage {
 
     // rows start at byte boundary
     const rowBytes = (width * numComps * bpc + 7) >> 3;
-    const imgArray = this.getImageBytes(height * rowBytes, { internal: true });
+    const imgArray = await this.getImageBytes(height * rowBytes, {
+      internal: true,
+    });
 
     const comps = this.getComponents(imgArray);
     let i, length;
@@ -975,7 +977,7 @@ class PDFImage {
     };
   }
 
-  getImageBytes(
+  async getImageBytes(
     length,
     {
       drawWidth,
@@ -990,7 +992,10 @@ class PDFImage {
     this.image.drawHeight = drawHeight || this.height;
     this.image.forceRGBA = !!forceRGBA;
     this.image.forceRGB = !!forceRGB;
-    const imageBytes = this.image.getBytes(length, this.ignoreColorSpace);
+    const imageBytes = await this.image.getImageData(
+      length,
+      this.ignoreColorSpace
+    );
 
     // If imageBytes came from a DecodeStream, we're safe to transfer it
     // (and thus detach its underlying buffer) because it will constitute

diff --git a/src/core/jbig2_stream.js b/src/core/jbig2_stream.js
@@ -44,8 +44,15 @@ class Jbig2Stream extends DecodeStream {
   }
 
   readBlock() {
+    return this.decodeImage();
+  }
+
+  decodeImage(bytes) {
     if (this.eof) {
-      return;
+      return this.buffer;
+    }
+    if (!bytes) {
+      bytes = this.bytes;
     }
     const jbig2Image = new Jbig2Image();
 
@@ -57,7 +64,7 @@ class Jbig2Stream extends DecodeStream {
         chunks.push({ data: globals, start: 0, end: globals.length });
       }
     }
-    chunks.push({ data: this.bytes, start: 0, end: this.bytes.length });
+    chunks.push({ data: bytes, start: 0, end: bytes.length });
     const data = jbig2Image.parseChunks(chunks);
     const dataLength = data.length;
 
@@ -68,6 +75,12 @@ class Jbig2Stream extends DecodeStream {
     this.buffer = data;
     this.bufferLength = dataLength;
     this.eof = true;
+
+    return this.buffer;
+  }
+
+  get canDecodeImageFromBuffer() {
+    return true;
   }
 }
 

diff --git a/src/core/jpeg_stream.js b/src/core/jpeg_stream.js
@@ -53,8 +53,15 @@ class JpegStream extends DecodeStream {
   }
 
   readBlock() {
+    return this.decodeImage();
+  }
+
+  decodeImage(bytes) {
     if (this.eof) {
-      return;
+      return this.buffer;
+    }
+    if (!bytes) {
+      bytes = this.bytes;
     }
     const jpegOptions = {
       decodeTransform: undefined,
@@ -89,7 +96,7 @@ class JpegStream extends DecodeStream {
     }
     const jpegImage = new JpegImage(jpegOptions);
 
-    jpegImage.parse(this.bytes);
+    jpegImage.parse(bytes);
     const data = jpegImage.getData({
       width: this.drawWidth,
       height: this.drawHeight,
@@ -100,6 +107,12 @@ class JpegStream extends DecodeStream {
     this.buffer = data;
     this.bufferLength = data.length;
     this.eof = true;
+
+    return this.buffer;
+  }
+
+  get canDecodeImageFromBuffer() {
+    return true;
   }
 }
 

diff --git a/src/core/jpx_stream.js b/src/core/jpx_stream.js
@@ -42,13 +42,25 @@ class JpxStream extends DecodeStream {
   }
 
   readBlock(ignoreColorSpace) {
+    return this.decodeImage(null, ignoreColorSpace);
+  }
+
+  decodeImage(bytes, ignoreColorSpace) {
     if (this.eof) {
-      return;
+      return this.buffer;
+    }
+    if (!bytes) {
+      bytes = this.bytes;
     }
-
     this.buffer = JpxImage.decode(this.bytes, ignoreColorSpace);
     this.bufferLength = this.buffer.length;
     this.eof = true;
+
+    return this.buffer;
+  }
+
+  get canDecodeImageFromBuffer() {
+    return true;
   }
 }