[api-minor] Add a jpx decoder based on OpenJPEG 2.5.2

The decoder is compiled in WASM:
https://github.com/mozilla/pdf.js.openjpeg

Fixes #17289, #17061, #16485, #13051, #6365, #4648, #12213.

.eslintignore

@@ -6,6 +6,7 @@ external/bcmaps/
 external/builder/fixtures/
 external/builder/fixtures_esprima/
 external/quickjs/
+external/openjpeg/
 test/tmp/
 test/pdfs/
 web/locale/

external/openjpeg/README.md

@@ -0,0 +1,12 @@
+## Build
+
+In order to generate the file `openjpeg.js`:
+* git clone https://github.com/mozilla/pdf.js.openjpeg/
+* the build requires to have a [Docker](https://www.docker.com/) setup and then:
+  * `node build.js -C` to build the Docker image
+  * `node build.js -co /pdf.js/external/openjpeg/` to compile the decoder
+
+## Licensing
+
+[OpenJPEG](https://www.openjpeg.org/) is under [BSD 2-clause "Simplified" License](https://github.com/uclouvain/openjpeg/blob/master/LICENSE)
+and [pdf.js.openjpeg](https://github.com/mozilla/pdf.js.openjpeg/) is released under [Apache 2](https://github.com/mozilla/pdf.js.openjpeg/blob/main/LICENSE) license so `openjpeg.js` is released under [Apache 2](https://github.com/mozilla/pdf.js.openjpeg/blob/main/LICENSE) license too.

gulpfile.mjs

@@ -1603,6 +1603,7 @@ function buildLib(defines, dir) {
     ),
     gulp.src(["web/*.js", "!web/{pdfjs,viewer}.js"], { base: "." }),
     gulp.src("test/unit/*.js", { base: "." }),
+    gulp.src("external/openjpeg/*.js", { base: "openjpeg/" }),
   ]);
 
   return buildLibHelper(bundleDefines, inputStream, dir);

src/core/cleanup_helper.js

@@ -16,11 +16,16 @@
 import { clearPatternCaches } from "./pattern.js";
 import { clearPrimitiveCaches } from "./primitives.js";
 import { clearUnicodeCaches } from "./unicode.js";
+import { JpxImage } from "./jpx.js";
 
 function clearGlobalCaches() {
   clearPatternCaches();
   clearPrimitiveCaches();
   clearUnicodeCaches();
+
+  // Remove the global `JpxImage` instance, since it may hold a reference to
+  // the WebAssembly module.
+  JpxImage.cleanup();
 }
 
 export { clearGlobalCaches };

src/core/decode_stream.js

@@ -73,7 +73,7 @@ class DecodeStream extends BaseStream {
     return this.buffer[this.pos++];
   }
 
-  getBytes(length) {
+  getBytes(length, ignoreColorSpace = false) {
     const pos = this.pos;
     let end;
 
@@ -82,7 +82,7 @@ class DecodeStream extends BaseStream {
       end = pos + length;
 
       while (!this.eof && this.bufferLength < end) {
-        this.readBlock();
+        this.readBlock(ignoreColorSpace);
       }
       const bufEnd = this.bufferLength;
       if (end > bufEnd) {
@@ -90,7 +90,7 @@ class DecodeStream extends BaseStream {
       }
     } else {
       while (!this.eof) {
-        this.readBlock();
+        this.readBlock(ignoreColorSpace);
       }
       end = this.bufferLength;
     }

src/core/image.js

@@ -104,6 +104,7 @@ class PDFImage {
     localColorSpaceCache,
   }) {
     this.image = image;
+    let jpxDecode = false;
     const dict = image.dict;
 
     const filter = dict.get("F", "Filter");
@@ -118,14 +119,14 @@ class PDFImage {
     }
     switch (filterName) {
       case "JPXDecode":
-        const jpxImage = new JpxImage();
-        jpxImage.parseImageProperties(image.stream);
+        ({
+          width: image.width,
+          height: image.height,
+          componentsCount: image.numComps,
+          bitsPerComponent: image.bitsPerComponent,
+        } = JpxImage.parseImageProperties(image.stream));
         image.stream.reset();
-
-        image.width = jpxImage.width;
-        image.height = jpxImage.height;
-        image.bitsPerComponent = jpxImage.bitsPerComponent;
-        image.numComps = jpxImage.componentsCount;
+        jpxDecode = true;
         break;
       case "JBIG2Decode":
         image.bitsPerComponent = 1;
@@ -197,6 +198,7 @@ class PDFImage {
             );
         }
       }
+
       this.colorSpace = ColorSpace.parse({
         cs: colorSpace,
         xref,
@@ -205,6 +207,10 @@ class PDFImage {
         localColorSpaceCache,
       });
       this.numComps = this.colorSpace.numComps;
+
+      // If the jpx image has a color space then it musn't be used in order to
+      // be able to use the color space that comes from the pdf.
+      this.ignoreColorSpace = jpxDecode && this.colorSpace.name === "Indexed";
     }
 
     this.decode = dict.getArray("D", "Decode");
@@ -984,7 +990,7 @@ class PDFImage {
     this.image.drawHeight = drawHeight || this.height;
     this.image.forceRGBA = !!forceRGBA;
     this.image.forceRGB = !!forceRGB;
-    const imageBytes = this.image.getBytes(length);
+    const imageBytes = this.image.getBytes(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/jpx_stream.js

@@ -41,44 +41,12 @@ class JpxStream extends DecodeStream {
     // directly insert all of its data into `this.buffer`.
   }
 
-  readBlock() {
+  readBlock(ignoreColorSpace) {
     if (this.eof) {
       return;
     }
-    const jpxImage = new JpxImage();
-    jpxImage.parse(this.bytes);
 
-    const width = jpxImage.width;
-    const height = jpxImage.height;
-    const componentsCount = jpxImage.componentsCount;
-    const tileCount = jpxImage.tiles.length;
-    if (tileCount === 1) {
-      this.buffer = jpxImage.tiles[0].items;
-    } else {
-      const data = new Uint8ClampedArray(width * height * componentsCount);
-
-      for (let k = 0; k < tileCount; k++) {
-        const tileComponents = jpxImage.tiles[k];
-        const tileWidth = tileComponents.width;
-        const tileHeight = tileComponents.height;
-        const tileLeft = tileComponents.left;
-        const tileTop = tileComponents.top;
-
-        const src = tileComponents.items;
-        let srcPosition = 0;
-        let dataPosition = (width * tileTop + tileLeft) * componentsCount;
-        const imgRowSize = width * componentsCount;
-        const tileRowSize = tileWidth * componentsCount;
-
-        for (let j = 0; j < tileHeight; j++) {
-          const rowBytes = src.subarray(srcPosition, srcPosition + tileRowSize);
-          data.set(rowBytes, dataPosition);
-          srcPosition += tileRowSize;
-          dataPosition += imgRowSize;
-        }
-      }
-      this.buffer = data;
-    }
+    this.buffer = JpxImage.decode(this.bytes, ignoreColorSpace);
     this.bufferLength = this.buffer.length;
     this.eof = true;
   }

test/pdfs/.gitignore

@@ -642,3 +642,4 @@
 !issue17871_top_right.pdf
 !bug1889122.pdf
 !issue17929.pdf
+!issue12213.pdf

test/pdfs/bug1659412.pdf.link

@@ -0,0 +1,2 @@
+https://bugzilla.mozilla.org/attachment.cgi?id=9254200
+

test/pdfs/issue12213.1.pdf.link

@@ -0,0 +1,2 @@
+https://github.com/mozilla/pdf.js/files/6600376/JPXimageprobem.pdf
+

test/pdfs/issue13051.pdf.link

@@ -0,0 +1,2 @@
+https://github.com/mozilla/pdf.js/files/6083044/njp.32101064480005_page_013.2.pdf
+

test/pdfs/issue16485.pdf.link

@@ -0,0 +1,2 @@
+https://web.archive.org/web/20230617061919/https://cdn.dealereprocess.org/cdn/servicemanuals/nissan/2023-ariya.pdf
+

test/pdfs/issue17061.pdf.link

@@ -0,0 +1,2 @@
+https://web.archive.org/web/20240229104747/https://btohq.sgp1.cdn.digitaloceanspaces.com/bto/nov-2019-bto/plantation-village.pdf
+

test/pdfs/issue17289.pdf.link

@@ -0,0 +1,2 @@
+https://github.com/mozilla/pdf.js/files/13391396/example.pdf
+

test/pdfs/issue4648.pdf.link

@@ -0,0 +1,2 @@
+https://web.archive.org/web/20240415131452/http://www.braidense.it/rd/03595.pdf
+

test/pdfs/issue6365.pdf.link

@@ -0,0 +1 @@
+https://web.archive.org/web/20210403114737/http://waltercosand.com/CosandScores/Composers%20A-D/Chopin,%20Frederic/Find_by_Category/Preludes/Preludes-Op28-edDebussy.pdf

test/test_manifest.json

@@ -9840,5 +9840,84 @@
     "md5": "2f5686f77f73fb99037f5c2c7e48ce53",
     "rounds": 1,
     "type": "eq"
+  },
+  {
+    "id": "issue17289",
+    "file": "pdfs/issue17289.pdf",
+    "md5": "acf5bb9e4309ba1c4b3568e7c4a7dd73",
+    "rounds": 1,
+    "link": true,
+    "type": "eq"
+  },
+  {
+    "id": "issue17061",
+    "file": "pdfs/issue17061.pdf",
+    "md5": "21a1b887b7e657b01e4205b22d1f4378",
+    "rounds": 1,
+    "link": true,
+    "firstPage": 1,
+    "lastPage": 5,
+    "type": "eq"
+  },
+  {
+    "id": "issue16485",
+    "file": "pdfs/issue16485.pdf",
+    "md5": "8a2c43161e0d89469fc7ed345c7bd3b1",
+    "rounds": 1,
+    "link": true,
+    "firstPage": 266,
+    "lastPage": 266,
+    "type": "eq"
+  },
+  {
+    "id": "issue13051",
+    "file": "pdfs/issue13051.pdf",
+    "md5": "ead6598c79641a0d266a224160e764d8",
+    "rounds": 1,
+    "link": true,
+    "type": "eq"
+  },
+  {
+    "id": "issue6365",
+    "file": "pdfs/issue6365.pdf",
+    "md5": "fd567f3b4f7c896b5b8071ea762af069",
+    "rounds": 1,
+    "link": true,
+    "firstPage": 2,
+    "lastPage": 5,
+    "type": "eq"
+  },
+  {
+    "id": "issue4648",
+    "file": "pdfs/issue4648.pdf",
+    "md5": "9906cb55f190b8a2d340ea34093ed763",
+    "rounds": 1,
+    "link": true,
+    "firstPage": 1,
+    "lastPage": 4,
+    "type": "eq"
+  },
+  {
+    "id": "bug1659412",
+    "file": "pdfs/bug1659412.pdf",
+    "md5": "0a3220e322d7b899e7604cea947d424c",
+    "rounds": 1,
+    "link": true,
+    "type": "eq"
+  },
+  {
+    "id": "issue12213",
+    "file": "pdfs/issue12213.pdf",
+    "md5": "84df1ce3e3d96e793a2e339aaba138e9",
+    "rounds": 1,
+    "type": "eq"
+  },
+  {
+    "id": "issue12213.1",
+    "file": "pdfs/issue12213.1.pdf",
+    "md5": "8ff4af8d7fd391b7f6601406e490d865",
+    "rounds": 1,
+    "link": true,
+    "type": "eq"
   }
 ]