Add a small LRU cache for the JBIG2 symbol dictionary.

core/src/fxcodec/jbig2/JBig2_Context.cpp

 // Copyright 2014 PDFium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
  
 // Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
 
+#include <map>
+#include <list>
 #include "JBig2_Context.h"
+
+// Implement a very small least recently used (LRU) cache. It is very
+// common for a JBIG2 dictionary to span multiple page in a PDF file,
+// and we do not want to decode the same dictionary over an over
+// again. We key off of the memory location of the dictionary. The
+// list keeps track of the freshness of enties, with freshest ones are
+// at the front. Even a tiny cache size like 2 makes a dramatic
+// difference for typical JBIG2 documents.
+const int kSymbolDictCacheMaxSize = 2;
+static list<pair<FX_BYTE*, CJBig2_SymbolDict*> > symbol_dict_cache;

[Lei Zhang, 2014/12/12 00:01:20]

You need std::

[Lei Zhang, 2014/12/12 00:01:20]

I'd add a typedef for "pair<FX_BYTE*, CJBig2_SymbolDict*>" so it's less tedious
to type everywhere.

[jab, 2014/12/13 01:46:40]

Done.

[jab, 2014/12/13 01:46:41]

Done.

+
 void OutputBitmap(CJBig2_Image* pImage)
 {
     if(!pImage) {
         return;
     }
 }
 CJBig2_Context *CJBig2_Context::CreateContext(CJBig2_Module *pModule, FX_BYTE *pGlobalData, FX_DWORD dwGlobalLength,
         FX_BYTE *pData, FX_DWORD dwLength, FX_INT32 nStreamType, IFX_Pause* pPause)
 {
     return new(pModule) CJBig2_Context(pGlobalData, dwGlobalLength, pData, dwLength, nStreamType, pPause);
@@ skipping 589 matching lines @@
     FX_BYTE cSDHUFFDH, cSDHUFFDW, cSDHUFFBMSIZE, cSDHUFFAGGINST;
     CJBig2_HuffmanTable *Table_B1 = NULL, *Table_B2 = NULL, *Table_B3 = NULL, *Table_B4 = NULL, *Table_B5 = NULL;
     FX_INT32 i, nIndex, nRet;
     CJBig2_Segment *pSeg = NULL, *pLRSeg = NULL;
     FX_BOOL bUsed;
     CJBig2_Image ** SDINSYMS = NULL;
     CJBig2_SDDProc *pSymbolDictDecoder;
     JBig2ArithCtx *gbContext = NULL, *grContext = NULL;
     CJBig2_ArithDecoder *pArithDecoder;
     JBIG2_ALLOC(pSymbolDictDecoder, CJBig2_SDDProc());
+    FX_BYTE *key = pSegment->m_pData;
+    FX_BOOL cache_hit = false;
     if(m_pStream->readShortInteger(&wFlags) != 0) {
         m_pModule->JBig2_Error("symbol dictionary segment : data header too short.");
         nRet = JBIG2_ERROR_TOO_SHORT;
         goto failed;
     }
     pSymbolDictDecoder->SDHUFF = wFlags & 0x0001;
     pSymbolDictDecoder->SDREFAGG = (wFlags >> 1) & 0x0001;
     pSymbolDictDecoder->SDTEMPLATE = (wFlags >> 10) & 0x0003;
     pSymbolDictDecoder->SDRTEMPLATE = (wFlags >> 12) & 0x0003;
     cSDHUFFDH = (wFlags >> 2) & 0x0003;
@@ skipping 157 matching lines @@
             gbContext = (JBig2ArithCtx*)m_pModule->JBig2_Malloc2(sizeof(JBig2ArithCtx), dwTemp);
             JBIG2_memset(gbContext, 0, sizeof(JBig2ArithCtx)*dwTemp);
         }
         if (pSymbolDictDecoder->SDREFAGG == 1) {
             dwTemp = pSymbolDictDecoder->SDRTEMPLATE ? 1 << 10 : 1 << 13;
             grContext = (JBig2ArithCtx*)m_pModule->JBig2_Malloc2(sizeof(JBig2ArithCtx), dwTemp);
             JBIG2_memset(grContext, 0, sizeof(JBig2ArithCtx)*dwTemp);
         }
     }
     pSegment->m_nResultType = JBIG2_SYMBOL_DICT_POINTER;
-    if(pSymbolDictDecoder->SDHUFF == 0) {
-        JBIG2_ALLOC(pArithDecoder, CJBig2_ArithDecoder(m_pStream));
-        pSegment->m_Result.sd = pSymbolDictDecoder->decode_Arith(pArithDecoder, gbContext, grContext);
-        delete pArithDecoder;
-        if(pSegment->m_Result.sd == NULL) {
-            nRet = JBIG2_ERROR_FETAL;
-            goto failed;
+    for(std::list<pair<FX_BYTE*, CJBig2_SymbolDict*> >::iterator it =
+            symbol_dict_cache.begin(); it != symbol_dict_cache.end(); ++it) {
+        if (it->first == key) {
+            pSegment->m_Result.sd = it->second->DeepCopy();
+            symbol_dict_cache.erase(it);
+            symbol_dict_cache.push_front(*it);
+            cache_hit = true;
+            break;
         }
-        m_pStream->alignByte();
-        m_pStream->offset(2);
-    } else {
-        pSegment->m_Result.sd = pSymbolDictDecoder->decode_Huffman(m_pStream, gbContext, grContext, pPause);
-        if(pSegment->m_Result.sd == NULL) {
-            nRet = JBIG2_ERROR_FETAL;
-            goto failed;
+    }
+    if (!cache_hit) {
+        if(pSymbolDictDecoder->SDHUFF == 0) {
+            JBIG2_ALLOC(pArithDecoder, CJBig2_ArithDecoder(m_pStream));
+            pSegment->m_Result.sd = pSymbolDictDecoder->decode_Arith(pArithDecoder, gbContext, grContext);
+            delete pArithDecoder;
+            if(pSegment->m_Result.sd == NULL) {
+                nRet = JBIG2_ERROR_FETAL;
+                goto failed;
+            }
+            m_pStream->alignByte();
+            m_pStream->offset(2);
+        } else {
+            pSegment->m_Result.sd = pSymbolDictDecoder->decode_Huffman(m_pStream, gbContext, grContext, pPause);
+            if(pSegment->m_Result.sd == NULL) {
+                nRet = JBIG2_ERROR_FETAL;
+                goto failed;
+            }
+            m_pStream->alignByte();
         }
-        m_pStream->alignByte();
+        CJBig2_SymbolDict *value = pSegment->m_Result.sd->DeepCopy();
+        if (value) {
+            while (symbol_dict_cache.size() >= kSymbolDictCacheMaxSize) {
+                delete symbol_dict_cache.back().second;
+                symbol_dict_cache.pop_back();
+            }
+            symbol_dict_cache.push_front(
+                std::pair<FX_BYTE*, CJBig2_SymbolDict*>(key, value));
+        }
     }
     if(wFlags & 0x0200) {
         pSegment->m_Result.sd->m_bContextRetained = TRUE;
         if(pSymbolDictDecoder->SDHUFF == 0) {
             pSegment->m_Result.sd->m_gbContext = gbContext;
         }
         if(pSymbolDictDecoder->SDREFAGG == 1) {
             pSegment->m_Result.sd->m_grContext = grContext;
         }
         bUsed = TRUE;
@@ skipping 982 matching lines @@
                 SBSYMCODES[CURTEMP].code = CURCODE;
                 CURCODE = CURCODE + 1;
             }
             CURTEMP = CURTEMP + 1;
         }
         CURLEN = CURLEN + 1;
     }
     m_pModule->JBig2_Free(LENCOUNT);
     m_pModule->JBig2_Free(FIRSTCODE);
 }