Merge pull request #18047 from Snuffleupagus/issue-18042

Avoid re-parsing global images that failed decoding (issue 18042, PR 17428 follow-up)

src/core/evaluator.js

@@ -793,30 +793,42 @@ class PartialEvaluator {
     args = [objId, w, h];
     operatorList.addImageOps(OPS.paintImageXObject, args, optionalContent);
 
-    // For large (at least 500x500) or more complex images that we'll cache
-    // globally, check if the image is still cached locally on the main-thread
-    // to avoid having to re-parse the image (since that can be slow).
-    if (
-      cacheGlobally &&
-      (w * h > 250000 || dict.has("SMask") || dict.has("Mask"))
-    ) {
-      const localLength = await this.handler.sendWithPromise("commonobj", [
-        objId,
-        "CopyLocalImage",
-        { imageRef },
-      ]);
-
-      if (localLength) {
+    if (cacheGlobally) {
+      if (this.globalImageCache.hasDecodeFailed(imageRef)) {
         this.globalImageCache.setData(imageRef, {
           objId,
           fn: OPS.paintImageXObject,
           args,
           optionalContent,
-          byteSize: 0, // Temporary entry, to avoid `setData` returning early.
+          byteSize: 0, // Data is `null`, since decoding failed previously.
         });
-        this.globalImageCache.addByteSize(imageRef, localLength);
+
+        this._sendImgData(objId, /* imgData = */ null, cacheGlobally);
         return;
       }
+
+      // For large (at least 500x500) or more complex images that we'll cache
+      // globally, check if the image is still cached locally on the main-thread
+      // to avoid having to re-parse the image (since that can be slow).
+      if (w * h > 250000 || dict.has("SMask") || dict.has("Mask")) {
+        const localLength = await this.handler.sendWithPromise("commonobj", [
+          objId,
+          "CopyLocalImage",
+          { imageRef },
+        ]);
+
+        if (localLength) {
+          this.globalImageCache.setData(imageRef, {
+            objId,
+            fn: OPS.paintImageXObject,
+            args,
+            optionalContent,
+            byteSize: 0, // Temporary entry, to avoid `setData` returning early.
+          });
+          this.globalImageCache.addByteSize(imageRef, localLength);
+          return;
+        }
+      }
     }
 
     PDFImage.buildImage({
@@ -846,6 +858,9 @@ class PartialEvaluator {
       .catch(reason => {
         warn(`Unable to decode image "${objId}": "${reason}".`);
 
+        if (imageRef) {
+          this.globalImageCache.addDecodeFailed(imageRef);
+        }
         return this._sendImgData(objId, /* imgData = */ null, cacheGlobally);
       });
 

src/core/image_utils.js

@@ -19,7 +19,7 @@ import {
   unreachable,
   warn,
 } from "../shared/util.js";
-import { RefSetCache } from "./primitives.js";
+import { RefSet, RefSetCache } from "./primitives.js";
 
 class BaseLocalCache {
   constructor(options) {
@@ -178,6 +178,8 @@ class GlobalImageCache {
 
   static MAX_BYTE_SIZE = 5 * MAX_IMAGE_SIZE_TO_CACHE;
 
+  #decodeFailedSet = new RefSet();
+
   constructor() {
     if (typeof PDFJSDev === "undefined" || PDFJSDev.test("TESTING")) {
       assert(
@@ -189,7 +191,7 @@ class GlobalImageCache {
     this._imageCache = new RefSetCache();
   }
 
-  get _byteSize() {
+  get #byteSize() {
     let byteSize = 0;
     for (const imageData of this._imageCache) {
       byteSize += imageData.byteSize;
@@ -197,11 +199,11 @@ class GlobalImageCache {
     return byteSize;
   }
 
-  get _cacheLimitReached() {
+  get #cacheLimitReached() {
     if (this._imageCache.size < GlobalImageCache.MIN_IMAGES_TO_CACHE) {
       return false;
     }
-    if (this._byteSize < GlobalImageCache.MAX_BYTE_SIZE) {
+    if (this.#byteSize < GlobalImageCache.MAX_BYTE_SIZE) {
       return false;
     }
     return true;
@@ -218,12 +220,20 @@ class GlobalImageCache {
     if (pageIndexSet.size < GlobalImageCache.NUM_PAGES_THRESHOLD) {
       return false;
     }
-    if (!this._imageCache.has(ref) && this._cacheLimitReached) {
+    if (!this._imageCache.has(ref) && this.#cacheLimitReached) {
       return false;
     }
     return true;
   }
 
+  addDecodeFailed(ref) {
+    this.#decodeFailedSet.put(ref);
+  }
+
+  hasDecodeFailed(ref) {
+    return this.#decodeFailedSet.has(ref);
+  }
+
   /**
    * PLEASE NOTE: Must be called *after* the `setData` method.
    */
@@ -265,7 +275,7 @@ class GlobalImageCache {
     if (this._imageCache.has(ref)) {
       return;
     }
-    if (this._cacheLimitReached) {
+    if (this.#cacheLimitReached) {
       warn("GlobalImageCache.setData - cache limit reached.");
       return;
     }
@@ -274,6 +284,7 @@ class GlobalImageCache {
 
   clear(onlyData = false) {
     if (!onlyData) {
+      this.#decodeFailedSet.clear();
       this._refCache.clear();
     }
     this._imageCache.clear();

src/display/api.js

@@ -2783,7 +2783,7 @@ class WorkerTransport {
 
           for (const pageProxy of this.#pageCache.values()) {
             for (const [, data] of pageProxy.objs) {
-              if (data.ref !== imageRef) {
+              if (data?.ref !== imageRef) {
                 continue;
               }
               if (!data.dataLen) {