Add a small LRU cache for the JBIG2 symbol dictionary.

core/src/fxcodec/jbig2/JBig2_Context.cpp

Index: core/src/fxcodec/jbig2/JBig2_Context.cpp
diff --git a/core/src/fxcodec/jbig2/JBig2_Context.cpp b/core/src/fxcodec/jbig2/JBig2_Context.cpp
index 83ded6fbe18f15ac81bde0b7d0de95d9b25ad2be..865549a9fe452f99b9b0580c6f22b7f3fe835e2e 100644
--- a/core/src/fxcodec/jbig2/JBig2_Context.cpp
+++ b/core/src/fxcodec/jbig2/JBig2_Context.cpp
@@ -4,7 +4,22 @@
  
 // Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
 
+#include <map>
+#include <vector>
 #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
+// vector keeps track of when an entry should be deleted, to make space
+// for a new one. Even a tiny cache size like 2 makes a dramatic difference
+// for Google Books PDF files.
+const int kMaxCacheSize = 2;
+static std::map<FX_BYTE*, CJBig2_SymbolDict*> cache;
+static std::vector<FX_BYTE*> cache_keys;
+
+
 void OutputBitmap(CJBig2_Image* pImage)
 {
     if(!pImage) {
@@ -614,6 +629,7 @@ FX_INT32 CJBig2_Context::parseSymbolDict(CJBig2_Segment *pSegment, IFX_Pause* pP
     JBig2ArithCtx *gbContext = NULL, *grContext = NULL;
     CJBig2_ArithDecoder *pArithDecoder;
     JBIG2_ALLOC(pSymbolDictDecoder, CJBig2_SDDProc());
+    FX_BYTE *key = pSegment->m_pData;
     if(m_pStream->readShortInteger(&wFlags) != 0) {
         m_pModule->JBig2_Error("symbol dictionary segment : data header too short.");
         nRet = JBIG2_ERROR_TOO_SHORT;
@@ -791,7 +807,17 @@ FX_INT32 CJBig2_Context::parseSymbolDict(CJBig2_Segment *pSegment, IFX_Pause* pP
         }
     }
     pSegment->m_nResultType = JBIG2_SYMBOL_DICT_POINTER;
-    if(pSymbolDictDecoder->SDHUFF == 0) {
+    if (cache.find(key) != cache.end()) {
+        pSegment->m_Result.sd = cache[key]->DeepCopy();
+        for(std::vector<FX_BYTE*>::iterator it = cache_keys.begin();
+            it != cache_keys.end(); ++it) {
+          if (*it == key) {
+            cache_keys.erase(cache_keys.begin());

[Bo Xu, 2014/12/06 05:59:43]

Should line 815 be: "cache_keys.erase(it);" ? I guess this will affect your test
timing?

Nit: The for loop could start from cache_keys.rbegin() as this is the most
recently used one.

[jab, 2014/12/08 21:00:02]

Done.

(I made the front hold the freshest entries)

+            break;

[Bo Xu, 2014/12/07 19:22:50]

Or can we use std:list to replace the std:vector here, so getting the key takes
const time.

[jab, 2014/12/08 21:00:02]

Done.

+          }
+        }
+        cache_keys.push_back(key);
+    } else if(pSymbolDictDecoder->SDHUFF == 0) {
         JBIG2_ALLOC(pArithDecoder, CJBig2_ArithDecoder(m_pStream));
         pSegment->m_Result.sd = pSymbolDictDecoder->decode_Arith(pArithDecoder, gbContext, grContext);
         delete pArithDecoder;
@@ -809,6 +835,17 @@ FX_INT32 CJBig2_Context::parseSymbolDict(CJBig2_Segment *pSegment, IFX_Pause* pP
         }
         m_pStream->alignByte();
     }
+    if (cache.find(key) == cache.end()) {
+        CJBig2_SymbolDict *value = pSegment->m_Result.sd->DeepCopy();

[Bo Xu, 2014/12/06 05:59:43]

should |value| be deleted after use?

[jab, 2014/12/08 21:00:02]

That you. I think I need to free the memory when we remove the entry from the
cache. I am not sure how to do this. Help!

[Bo Xu, 2014/12/08 23:00:17]

How about making the cache member variable of the CJBig2_Context class and let
the destructor do the job?

[jab, 2014/12/08 23:21:04]

I made an explicit call to delete. Is that okay? I think
it is invoking the destructor and my memory checking tool
does not seem to get upset.

[Bo Xu, 2014/12/08 23:29:54]

Sure, the explicit delete is ok here. 
I am thinking moving the static variables into CJBig2_Context, would that be
sufficient to make them available for the whole document?

+        if (value) {
+            while (cache_keys.size() >= kMaxCacheSize) {
+                cache.erase(cache_keys[0]);
+                cache_keys.erase(cache_keys.begin());
+            }
+            cache.insert(std::pair<FX_BYTE*, CJBig2_SymbolDict*>(key, value));
+            cache_keys.push_back(key);
+        }
+    }
     if(wFlags & 0x0200) {
         pSegment->m_Result.sd->m_bContextRetained = TRUE;
         if(pSymbolDictDecoder->SDHUFF == 0) {