Merge pull request #12034 from Snuffleupagus/Function-local-cache-3

Add local caching of `Function`s, by reference, in the `PDFFunctionFactory` (issue 2541)

src/core/colorspace.js

@@ -535,7 +535,7 @@ class ColorSpace {
           const name = xref.fetchIfRef(cs[1]);
           numComps = Array.isArray(name) ? name.length : 1;
           alt = this.parseToIR(cs[2], xref, resources, pdfFunctionFactory);
-          const tintFn = pdfFunctionFactory.create(xref.fetchIfRef(cs[3]));
+          const tintFn = pdfFunctionFactory.create(cs[3]);
           return ["AlternateCS", numComps, alt, tintFn];
         case "Lab":
           params = xref.fetchIfRef(cs[1]);

src/core/document.js

@@ -53,7 +53,6 @@ import { calculateMD5 } from "./crypto.js";
 import { Linearization } from "./parser.js";
 import { OperatorList } from "./operator_list.js";
 import { PartialEvaluator } from "./evaluator.js";
-import { PDFFunctionFactory } from "./function.js";
 
 const DEFAULT_USER_UNIT = 1.0;
 const LETTER_SIZE_MEDIABOX = [0, 0, 612, 792];
@@ -75,7 +74,6 @@ class Page {
     fontCache,
     builtInCMapCache,
     globalImageCache,
-    pdfFunctionFactory,
   }) {
     this.pdfManager = pdfManager;
     this.pageIndex = pageIndex;
@@ -85,7 +83,6 @@ class Page {
     this.fontCache = fontCache;
     this.builtInCMapCache = builtInCMapCache;
     this.globalImageCache = globalImageCache;
-    this.pdfFunctionFactory = pdfFunctionFactory;
     this.evaluatorOptions = pdfManager.evaluatorOptions;
     this.resourcesPromise = null;
 
@@ -265,7 +262,6 @@ class Page {
       builtInCMapCache: this.builtInCMapCache,
       globalImageCache: this.globalImageCache,
       options: this.evaluatorOptions,
-      pdfFunctionFactory: this.pdfFunctionFactory,
     });
 
     const dataPromises = Promise.all([contentStreamPromise, resourcesPromise]);
@@ -359,7 +355,6 @@ class Page {
         builtInCMapCache: this.builtInCMapCache,
         globalImageCache: this.globalImageCache,
         options: this.evaluatorOptions,
-        pdfFunctionFactory: this.pdfFunctionFactory,
       });
 
       return partialEvaluator.getTextContent({
@@ -508,11 +503,6 @@ class PDFDocument {
     this.pdfManager = pdfManager;
     this.stream = stream;
     this.xref = new XRef(stream, pdfManager);
-
-    this.pdfFunctionFactory = new PDFFunctionFactory({
-      xref: this.xref,
-      isEvalSupported: pdfManager.evaluatorOptions.isEvalSupported,
-    });
     this._pagePromises = [];
   }
 
@@ -821,7 +811,6 @@ class PDFDocument {
         fontCache: catalog.fontCache,
         builtInCMapCache: catalog.builtInCMapCache,
         globalImageCache: catalog.globalImageCache,
-        pdfFunctionFactory: this.pdfFunctionFactory,
       });
     }));
   }

src/core/evaluator.js

@@ -26,6 +26,7 @@ import {
   isNum,
   isString,
   OPS,
+  shadow,
   stringToPDFString,
   TextRenderingMode,
   UNSUPPORTED_FEATURES,
@@ -72,6 +73,7 @@ import {
   getSymbolsFonts,
 } from "./standard_fonts.js";
 import { getTilingPatternIR, Pattern } from "./pattern.js";
+import { isPDFFunction, PDFFunctionFactory } from "./function.js";
 import { Lexer, Parser } from "./parser.js";
 import { LocalColorSpaceCache, LocalImageCache } from "./image_utils.js";
 import { bidi } from "./bidi.js";
@@ -79,7 +81,6 @@ import { ColorSpace } from "./colorspace.js";
 import { DecodeStream } from "./stream.js";
 import { getGlyphsUnicode } from "./glyphlist.js";
 import { getMetrics } from "./metrics.js";
-import { isPDFFunction } from "./function.js";
 import { MurmurHash3_64 } from "./murmurhash3.js";
 import { OperatorList } from "./operator_list.js";
 import { PDFImage } from "./image.js";
@@ -103,7 +104,6 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
     builtInCMapCache,
     globalImageCache,
     options = null,
-    pdfFunctionFactory,
   }) {
     this.xref = xref;
     this.handler = handler;
@@ -113,7 +113,6 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
     this.builtInCMapCache = builtInCMapCache;
     this.globalImageCache = globalImageCache;
     this.options = options || DefaultPartialEvaluatorOptions;
-    this.pdfFunctionFactory = pdfFunctionFactory;
     this.parsingType3Font = false;
 
     this._fetchBuiltInCMapBound = this.fetchBuiltInCMap.bind(this);
@@ -207,6 +206,18 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
     SHADING_PATTERN = 2;
 
   PartialEvaluator.prototype = {
+    /**
+     * Since Functions are only cached (locally) by reference, we can share one
+     * `PDFFunctionFactory` instance within this `PartialEvaluator` instance.
+     */
+    get _pdfFunctionFactory() {
+      const pdfFunctionFactory = new PDFFunctionFactory({
+        xref: this.xref,
+        isEvalSupported: this.options.isEvalSupported,
+      });
+      return shadow(this, "_pdfFunctionFactory", pdfFunctionFactory);
+    },
+
     clone(newOptions = DefaultPartialEvaluatorOptions) {
       var newEvaluator = Object.create(this);
       newEvaluator.options = newOptions;
@@ -552,7 +563,7 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
           res: resources,
           image,
           isInline,
-          pdfFunctionFactory: this.pdfFunctionFactory,
+          pdfFunctionFactory: this._pdfFunctionFactory,
           localColorSpaceCache,
         });
         // We force the use of RGBA_32BPP images here, because we can't handle
@@ -589,7 +600,7 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
         res: resources,
         image,
         isInline,
-        pdfFunctionFactory: this.pdfFunctionFactory,
+        pdfFunctionFactory: this._pdfFunctionFactory,
         localColorSpaceCache,
       })
         .then(imageObj => {
@@ -651,7 +662,7 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
       // we will build a map of integer values in range 0..255 to be fast.
       var transferObj = smask.get("TR");
       if (isPDFFunction(transferObj)) {
-        const transferFn = this.pdfFunctionFactory.create(transferObj);
+        const transferFn = this._pdfFunctionFactory.create(transferObj);
         var transferMap = new Uint8Array(256);
         var tmp = new Float32Array(1);
         for (var i = 0; i < 256; i++) {
@@ -1145,7 +1156,7 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
         cs,
         xref: this.xref,
         resources,
-        pdfFunctionFactory: this.pdfFunctionFactory,
+        pdfFunctionFactory: this._pdfFunctionFactory,
         localColorSpaceCache,
       }).catch(reason => {
         if (reason instanceof AbortException) {
@@ -1202,7 +1213,7 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
             this.xref,
             resources,
             this.handler,
-            this.pdfFunctionFactory,
+            this._pdfFunctionFactory,
             localColorSpaceCache
           );
           operatorList.addOp(fn, pattern.getIR());
@@ -1641,7 +1652,7 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
                 xref,
                 resources,
                 self.handler,
-                self.pdfFunctionFactory,
+                self._pdfFunctionFactory,
                 localColorSpaceCache
               );
               var patternIR = shadingFill.getIR();

src/core/function.js

@@ -13,6 +13,7 @@
  * limitations under the License.
  */
 
+import { Dict, isDict, isStream, Ref } from "./primitives.js";
 import {
   FormatError,
   info,
@@ -20,29 +21,94 @@ import {
   IsEvalSupportedCached,
   unreachable,
 } from "../shared/util.js";
-import { isDict, isStream } from "./primitives.js";
 import { PostScriptLexer, PostScriptParser } from "./ps_parser.js";
+import { LocalFunctionCache } from "./image_utils.js";
 
 class PDFFunctionFactory {
   constructor({ xref, isEvalSupported = true }) {
     this.xref = xref;
     this.isEvalSupported = isEvalSupported !== false;
+    this._localFunctionCache = null; // Initialized lazily.
   }
 
   create(fn) {
-    return PDFFunction.parse({
+    const cachedFunction = this.getCached(fn);
+    if (cachedFunction) {
+      return cachedFunction;
+    }
+    const parsedFunction = PDFFunction.parse({
       xref: this.xref,
       isEvalSupported: this.isEvalSupported,
-      fn,
+      fn: fn instanceof Ref ? this.xref.fetch(fn) : fn,
     });
+
+    // Attempt to cache the parsed Function, by reference.
+    this._cache(fn, parsedFunction);
+
+    return parsedFunction;
   }
 
   createFromArray(fnObj) {
-    return PDFFunction.parseArray({
+    const cachedFunction = this.getCached(fnObj);
+    if (cachedFunction) {
+      return cachedFunction;
+    }
+    const parsedFunction = PDFFunction.parseArray({
       xref: this.xref,
       isEvalSupported: this.isEvalSupported,
-      fnObj,
+      fnObj: fnObj instanceof Ref ? this.xref.fetch(fnObj) : fnObj,
     });
+
+    // Attempt to cache the parsed Function, by reference.
+    this._cache(fnObj, parsedFunction);
+
+    return parsedFunction;
+  }
+
+  getCached(cacheKey) {
+    let fnRef;
+    if (cacheKey instanceof Ref) {
+      fnRef = cacheKey;
+    } else if (cacheKey instanceof Dict) {
+      fnRef = cacheKey.objId;
+    } else if (isStream(cacheKey)) {
+      fnRef = cacheKey.dict && cacheKey.dict.objId;
+    }
+    if (fnRef) {
+      if (!this._localFunctionCache) {
+        this._localFunctionCache = new LocalFunctionCache();
+      }
+      const localFunction = this._localFunctionCache.getByRef(fnRef);
+      if (localFunction) {
+        return localFunction;
+      }
+    }
+    return null;
+  }
+
+  /**
+   * @private
+   */
+  _cache(cacheKey, parsedFunction) {
+    if (!parsedFunction) {
+      throw new Error(
+        'PDFFunctionFactory._cache - expected "parsedFunction" argument.'
+      );
+    }
+    let fnRef;
+    if (cacheKey instanceof Ref) {
+      fnRef = cacheKey;
+    } else if (cacheKey instanceof Dict) {
+      fnRef = cacheKey.objId;
+    } else if (isStream(cacheKey)) {
+      fnRef = cacheKey.dict && cacheKey.dict.objId;
+    }
+    if (fnRef) {
+      if (!this._localFunctionCache) {
+        this._localFunctionCache = new LocalFunctionCache();
+      }
+      this._localFunctionCache.set(/* name = */ null, fnRef, parsedFunction);
+    }
   }
 }
 

src/core/image_utils.js

@@ -91,6 +91,22 @@ class LocalColorSpaceCache extends BaseLocalCache {
   }
 }
 
+class LocalFunctionCache extends BaseLocalCache {
+  getByName(name) {
+    unreachable("Should not call `getByName` method.");
+  }
+
+  set(name = null, ref, data) {
+    if (!ref) {
+      throw new Error('LocalFunctionCache.set - expected "ref" argument.');
+    }
+    if (this._imageCache.has(ref)) {
+      return;
+    }
+    this._imageCache.put(ref, data);
+  }
+}
+
 class GlobalImageCache {
   static get NUM_PAGES_THRESHOLD() {
     return shadow(this, "NUM_PAGES_THRESHOLD", 2);
@@ -184,4 +200,9 @@ class GlobalImageCache {
   }
 }
 
-export { LocalImageCache, LocalColorSpaceCache, GlobalImageCache };
+export {
+  LocalImageCache,
+  LocalColorSpaceCache,
+  LocalFunctionCache,
+  GlobalImageCache,
+};

src/core/pattern.js

@@ -178,7 +178,7 @@ Shadings.RadialAxial = (function RadialAxialClosure() {
     this.extendStart = extendStart;
     this.extendEnd = extendEnd;
 
-    var fnObj = dict.get("Function");
+    var fnObj = dict.getRaw("Function");
     var fn = pdfFunctionFactory.createFromArray(fnObj);
 
     // 10 samples seems good enough for now, but probably won't work
@@ -878,7 +878,7 @@ Shadings.Mesh = (function MeshClosure() {
       ? cs.getRgb(dict.get("Background"), 0)
       : null;
 
-    var fnObj = dict.get("Function");
+    var fnObj = dict.getRaw("Function");
     var fn = fnObj ? pdfFunctionFactory.createFromArray(fnObj) : null;
 
     this.coords = [];