Merge pull request #19893 from calixteman/issue19517

Downscale jpeg2000 images, if needed, while decoding them

src/core/image.js

@@ -112,11 +112,22 @@ class PDFImage {
           bitsPerComponent: image.bitsPerComponent,
         } = JpxImage.parseImageProperties(image.stream));
         image.stream.reset();
+        const reducePower = ImageResizer.getReducePowerForJPX(
+          image.width,
+          image.height,
+          image.numComps
+        );
         this.jpxDecoderOptions = {
           numComponents: 0,
           isIndexedColormap: false,
           smaskInData: dict.has("SMaskInData"),
+          reducePower,
         };
+        if (reducePower) {
+          const factor = 2 ** reducePower;
+          image.width = Math.ceil(image.width / factor);
+          image.height = Math.ceil(image.height / factor);
+        }
         break;
       case "JBIG2Decode":
         image.bitsPerComponent = 1;

src/core/image_resizer.js

@@ -96,6 +96,30 @@ class ImageResizer {
     return area > maxArea;
   }
 
+  static getReducePowerForJPX(width, height, componentsCount) {
+    const area = width * height;
+    // The maximum memory we've in the wasm runtime is 2GB.
+    // Each component is 4 bytes and we can't allocate all the memory just for
+    // the buffers so we limit the size to 1GB / (componentsCount * 4).
+    // We could use more than 2GB by setting MAXIMUM_MEMORY but it would take
+    // too much time to decode a big image.
+    const maxJPXArea = 2 ** 30 / (componentsCount * 4);
+    if (!this.needsToBeResized(width, height)) {
+      if (area > maxJPXArea) {
+        // The image is too large, we need to rescale it.
+        return Math.ceil(Math.log2(area / maxJPXArea));
+      }
+      return 0;
+    }
+    const { MAX_DIM, MAX_AREA } = this;
+    const minFactor = Math.max(
+      width / MAX_DIM,
+      height / MAX_DIM,
+      Math.sqrt(area / Math.min(maxJPXArea, MAX_AREA))
+    );
+    return Math.ceil(Math.log2(minFactor));
+  }
+
   static get MAX_DIM() {
     return shadow(
       this,

src/core/jpx.js

@@ -92,7 +92,12 @@ class JpxImage {
 
   static async decode(
     bytes,
-    { numComponents = 4, isIndexedColormap = false, smaskInData = false } = {}
+    {
+      numComponents = 4,
+      isIndexedColormap = false,
+      smaskInData = false,
+      reducePower = 0,
+    } = {}
   ) {
     if (!this.#modulePromise) {
       const { promise, resolve } = Promise.withResolvers();
@@ -119,13 +124,14 @@ class JpxImage {
     try {
       const size = bytes.length;
       ptr = module._malloc(size);
-      module.HEAPU8.set(bytes, ptr);
+      module.writeArrayToMemory(bytes, ptr);
       const ret = module._jp2_decode(
         ptr,
         size,
         numComponents > 0 ? numComponents : 0,
         !!isIndexedColormap,
-        !!smaskInData
+        !!smaskInData,
+        reducePower
       );
       if (ret) {
         const { errorMessages } = module;

test/pdfs/.gitignore

@@ -724,3 +724,4 @@
 !issue16742.pdf
 !chrome-text-selection-markedContent.pdf
 !bug1963407.pdf
+!issue19517.pdf

test/test_manifest.json

@@ -12107,5 +12107,12 @@
         "value": "Hello World"
       }
     }
+  },
+  {
+    "id": "issue19517",
+    "file": "pdfs/issue19517.pdf",
+    "md5": "2abe67c8b34522feb6b85d252dde9d3e",
+    "rounds": 1,
+    "type": "eq"
   }
 ]