Cache built-in binary CMap files in the worker (issue 4794)

src/core/document.js

@@ -71,13 +71,15 @@ var Page = (function PageClosure() {
   var DEFAULT_USER_UNIT = 1.0;
   var LETTER_SIZE_MEDIABOX = [0, 0, 612, 792];
 
-  function Page(pdfManager, xref, pageIndex, pageDict, ref, fontCache) {
+  function Page(pdfManager, xref, pageIndex, pageDict, ref, fontCache,
+                builtInCMapCache) {
     this.pdfManager = pdfManager;
     this.pageIndex = pageIndex;
     this.pageDict = pageDict;
     this.xref = xref;
     this.ref = ref;
     this.fontCache = fontCache;
+    this.builtInCMapCache = builtInCMapCache;
     this.evaluatorOptions = pdfManager.evaluatorOptions;
     this.resourcesPromise = null;
 
@@ -248,6 +250,7 @@ var Page = (function PageClosure() {
                                                   handler, this.pageIndex,
                                                   this.idFactory,
                                                   this.fontCache,
+                                                  this.builtInCMapCache,
                                                   this.evaluatorOptions);
 
       var dataPromises = Promise.all([contentStreamPromise, resourcesPromise]);
@@ -315,6 +318,7 @@ var Page = (function PageClosure() {
                                                     handler, self.pageIndex,
                                                     self.idFactory,
                                                     self.fontCache,
+                                                    self.builtInCMapCache,
                                                     self.evaluatorOptions);
 
         return partialEvaluator.getTextContent(contentStream,
@@ -551,9 +555,10 @@ var PDFDocument = (function PDFDocumentClosure() {
       this.xref.parse(recoveryMode);
       var self = this;
       var pageFactory = {
-        createPage: function (pageIndex, dict, ref, fontCache) {
+        createPage: function (pageIndex, dict, ref, fontCache,
+                              builtInCMapCache) {
           return new Page(self.pdfManager, self.xref, pageIndex, dict, ref,
-                          fontCache);
+                          fontCache, builtInCMapCache);
         }
       };
       this.catalog = new Catalog(this.pdfManager, this.xref, pageFactory);

src/core/evaluator.js

@@ -170,18 +170,29 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
   };
 
   function PartialEvaluator(pdfManager, xref, handler, pageIndex,
-                            idFactory, fontCache, options) {
+                            idFactory, fontCache, builtInCMapCache, options) {
     this.pdfManager = pdfManager;
     this.xref = xref;
     this.handler = handler;
     this.pageIndex = pageIndex;
     this.idFactory = idFactory;
     this.fontCache = fontCache;
+    this.builtInCMapCache = builtInCMapCache;
     this.options = options || DefaultPartialEvaluatorOptions;
 
     this.fetchBuiltInCMap = function (name) {
+      var cachedCMap = builtInCMapCache[name];
+      if (cachedCMap) {
+        return Promise.resolve(cachedCMap);
+      }
       return handler.sendWithPromise('FetchBuiltInCMap', {
         name: name,
+      }).then(function (data) {
+        if (data.compressionType !== CMapCompressionType.NONE) {
+          // Given the size of uncompressed CMaps, only cache compressed ones.
+          builtInCMapCache[name] = data;
+        }
+        return data;
       });
     };
   }

src/core/obj.js

@@ -71,8 +71,8 @@ var Catalog = (function CatalogClosure() {
     this.xref = xref;
     this.catDict = xref.getCatalogObj();
     this.fontCache = new RefSetCache();
-    assert(isDict(this.catDict),
+    this.builtInCMapCache = Object.create(null);
-      'catalog object is not a dictionary');
+    assert(isDict(this.catDict), 'catalog object is not a dictionary');
 
     // TODO refactor to move getPage() to the PDFDocument.
     this.pageFactory = pageFactory;
@@ -428,6 +428,7 @@ var Catalog = (function CatalogClosure() {
           delete font.translated;
         }
         this.fontCache.clear();
+        this.builtInCMapCache = Object.create(null);
       }.bind(this));
     },
 
@@ -438,7 +439,8 @@ var Catalog = (function CatalogClosure() {
             var dict = a[0];
             var ref = a[1];
             return this.pageFactory.createPage(pageIndex, dict, ref,
-                                               this.fontCache);
+                                               this.fontCache,
+                                               this.builtInCMapCache);
           }.bind(this)
         );
       }

test/unit/document_spec.js

@@ -34,11 +34,13 @@ describe('document', function () {
       var page1 = new Page(/* pdfManager = */ { }, /* xref = */ null,
                            /* pageIndex = */ 0,
                            /* pageDict = */ null, /* ref = */ null,
-                           /* fontCache = */ null);
+                           /* fontCache = */ null,
+                           /* builtInCMapCache = */ null);
       var page2 = new Page(/* pdfManager = */ { }, /* xref = */ null,
                            /* pageIndex = */ 1,
                            /* pageDict = */ null, /* ref = */ null,
-                           /* fontCache = */ null);
+                           /* fontCache = */ null,
+                           /* builtInCMapCache = */ null);
 
       var idFactory1 = page1.idFactory, idFactory2 = page2.idFactory;