Merge pull request #18167 from calixteman/decompress_image

Decompress when it's possible images in using DecompressionStream

src/core/base_stream.js

@@ -44,6 +44,27 @@ class BaseStream {
     unreachable("Abstract method `getBytes` called");
   }
 
+  /**
+   * NOTE: This method can only be used to get image-data that is guaranteed
+   *       to be fully loaded, since otherwise intermittent errors may occur;
+   *       note the `ObjectLoader` class.
+   */
+  async getImageData(length, ignoreColorSpace) {
+    return this.getBytes(length, ignoreColorSpace);
+  }
+
+  async asyncGetBytes() {
+    unreachable("Abstract method `asyncGetBytes` called");
+  }
+
+  get isAsync() {
+    return false;
+  }
+
+  get canAsyncDecodeImageFromBuffer() {
+    return false;
+  }
+
   peekByte() {
     const peekedByte = this.getByte();
     if (peekedByte !== -1) {

src/core/decode_stream.js

@@ -99,6 +99,14 @@ class DecodeStream extends BaseStream {
     return this.buffer.subarray(pos, end);
   }
 
+  async getImageData(length, ignoreColorSpace = false) {
+    if (!this.canAsyncDecodeImageFromBuffer) {
+      return this.getBytes(length, ignoreColorSpace);
+    }
+    const data = await this.stream.asyncGetBytes();
+    return this.decodeImage(data, ignoreColorSpace);
+  }
+
   reset() {
     this.pos = 0;
   }

src/core/flate_stream.js

@@ -21,6 +21,7 @@
 
 import { FormatError, info } from "../shared/util.js";
 import { DecodeStream } from "./decode_stream.js";
+import { Stream } from "./stream.js";
 
 const codeLenCodeMap = new Int32Array([
   16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15,
@@ -148,6 +149,57 @@ class FlateStream extends DecodeStream {
     this.codeBuf = 0;
   }
 
+  async getImageData(length, _ignoreColorSpace) {
+    const data = await this.asyncGetBytes();
+    return data?.subarray(0, length) || this.getBytes(length);
+  }
+
+  async asyncGetBytes() {
+    this.str.reset();
+    const bytes = this.str.getBytes();
+
+    try {
+      const { readable, writable } = new DecompressionStream("deflate");
+      const writer = writable.getWriter();
+      writer.write(bytes);
+      writer.close();
+
+      const chunks = [];
+      let totalLength = 0;
+
+      for await (const chunk of readable) {
+        chunks.push(chunk);
+        totalLength += chunk.byteLength;
+      }
+      const data = new Uint8Array(totalLength);
+      let offset = 0;
+      for (const chunk of chunks) {
+        data.set(chunk, offset);
+        offset += chunk.byteLength;
+      }
+
+      return data;
+    } catch {
+      // DecompressionStream failed (for example because there are some extra
+      // bytes after the end of the compressed data), so we fallback to our
+      // decoder.
+      // We already get the bytes from the underlying stream, so we just reuse
+      // them to avoid get them again.
+      this.str = new Stream(
+        bytes,
+        2 /* = header size (see ctor) */,
+        bytes.length,
+        this.str.dict
+      );
+      this.reset();
+      return null;
+    }
+  }
+
+  get isAsync() {
+    return true;
+  }
+
   getBits(bits) {
     const str = this.str;
     let codeSize = this.codeSize;

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

src/core/jbig2_stream.js

@@ -44,9 +44,14 @@ class Jbig2Stream extends DecodeStream {
   }
 
   readBlock() {
+    this.decodeImage();
+  }
+
+  decodeImage(bytes) {
     if (this.eof) {
-      return;
+      return this.buffer;
     }
+    bytes ||= this.bytes;
     const jbig2Image = new Jbig2Image();
 
     const chunks = [];
@@ -57,7 +62,7 @@ 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 +73,12 @@ class Jbig2Stream extends DecodeStream {
     this.buffer = data;
     this.bufferLength = dataLength;
     this.eof = true;
+
+    return this.buffer;
+  }
+
+  get canAsyncDecodeImageFromBuffer() {
+    return this.stream.isAsync;
   }
 }
 

src/core/jpeg_stream.js

@@ -24,16 +24,6 @@ import { shadow } from "../shared/util.js";
  */
 class JpegStream extends DecodeStream {
   constructor(stream, maybeLength, params) {
-    // Some images may contain 'junk' before the SOI (start-of-image) marker.
-    // Note: this seems to mainly affect inline images.
-    let ch;
-    while ((ch = stream.getByte()) !== -1) {
-      // Find the first byte of the SOI marker (0xFFD8).
-      if (ch === 0xff) {
-        stream.skip(-1); // Reset the stream position to the SOI.
-        break;
-      }
-    }
     super(maybeLength);
 
     this.stream = stream;
@@ -53,8 +43,24 @@ class JpegStream extends DecodeStream {
   }
 
   readBlock() {
+    this.decodeImage();
+  }
+
+  decodeImage(bytes) {
     if (this.eof) {
-      return;
+      return this.buffer;
+    }
+    bytes ||= this.bytes;
+
+    // Some images may contain 'junk' before the SOI (start-of-image) marker.
+    // Note: this seems to mainly affect inline images.
+    for (let i = 0, ii = bytes.length - 1; i < ii; i++) {
+      if (bytes[i] === 0xff && bytes[i + 1] === 0xd8) {
+        if (i > 0) {
+          bytes = bytes.subarray(i);
+        }
+        break;
+      }
     }
     const jpegOptions = {
       decodeTransform: undefined,
@@ -89,7 +95,7 @@ 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 +106,12 @@ class JpegStream extends DecodeStream {
     this.buffer = data;
     this.bufferLength = data.length;
     this.eof = true;
+
+    return this.buffer;
+  }
+
+  get canAsyncDecodeImageFromBuffer() {
+    return this.stream.isAsync;
   }
 }
 

src/core/jpx_stream.js

@@ -42,13 +42,23 @@ class JpxStream extends DecodeStream {
   }
 
   readBlock(ignoreColorSpace) {
-    if (this.eof) {
-      return;
-    }
+    this.decodeImage(null, ignoreColorSpace);
+  }
 
-    this.buffer = JpxImage.decode(this.bytes, ignoreColorSpace);
+  decodeImage(bytes, ignoreColorSpace) {
+    if (this.eof) {
+      return this.buffer;
+    }
+    bytes ||= this.bytes;
+    this.buffer = JpxImage.decode(bytes, ignoreColorSpace);
     this.bufferLength = this.buffer.length;
     this.eof = true;
+
+    return this.buffer;
+  }
+
+  get canAsyncDecodeImageFromBuffer() {
+    return this.stream.isAsync;
   }
 }