write phrase index logger in progress

4 files changed, 45 insertions, 6 deletions

diff --git a/src/pinyin.cpp b/src/pinyin.cpp
index 0233400..8d08555 100644
--- a/src/pinyin.cpp
+++ b/src/pinyin.cpp
@@ -37,7 +37,7 @@ bool pinyin_set_pinyin_keys(pinyin_context_t * context,
 
 
 /* the returned sentence should be freed by g_free(). */
-bool pinyin_get_guessed_setence(pinyin_context_t * context,
+bool pinyin_get_guessed_sentence(pinyin_context_t * context,
                                 char ** sentence);
 
 bool pinyin_parse_one(pinyin_context_t * context,
diff --git a/src/storage/phrase_index.cpp b/src/storage/phrase_index.cpp
index 33da334..59b166f 100644
--- a/src/storage/phrase_index.cpp
+++ b/src/storage/phrase_index.cpp
@@ -213,6 +213,32 @@ bool FacadePhraseIndex::unload(guint8 phrase_index){
     return true;
 }
 
+bool FacadePhraseIndex::diff(guint8 phrase_index, MemoryChunk * oldchunk,
+                             MemoryChunk * newlog){
+    SubPhraseIndex * & sub_phrases = m_sub_phrase_indices[phrase_index];
+    if ( !sub_phrases )
+        return false;
+
+    SubPhraseIndex old_sub_phrases;
+    old_sub_phrases.load(oldchunk, 0, oldchunk->size());
+    PhraseIndexLogger logger;
+
+    bool retval = sub_phrases->diff(&old_sub_phrases, &logger);
+    logger.store(newlog);
+    return retval;
+}
+
+bool FacadePhraseIndex::merge(guint8 phrase_index, MemoryChunk * log){
+    SubPhraseIndex * & sub_phrases = m_sub_phrase_indices[phrase_index];
+    if ( !sub_phrases )
+        return false;
+
+    PhraseIndexLogger logger;
+    logger.load(log);
+
+    return sub_phrases->merge(&logger);
+}
+
 bool SubPhraseIndex::load(MemoryChunk * chunk, 
 			  table_offset_t offset, table_offset_t end){
     //save the memory chunk
diff --git a/src/storage/phrase_index.h b/src/storage/phrase_index.h
index 2dcad26..f2648e8 100644
--- a/src/storage/phrase_index.h
+++ b/src/storage/phrase_index.h
@@ -168,6 +168,10 @@ public:
     bool store(MemoryChunk * new_chunk, 
 	       table_offset_t offset, table_offset_t & end);
 
+    /* switch to logger format to reduce user storage */
+    bool diff(SubPhraseIndex * oldone, PhraseIndexLogger * logger);
+    bool merge(PhraseIndexLogger * logger);
+
     /* get token range in this sub phrase */
     int get_range(/* out */ PhraseIndexRange & range);
     
@@ -183,6 +187,7 @@ public:
      * from m_total_freq
      */
     int remove_phrase_item(phrase_token_t token, /* out */ PhraseItem * & item);
+
 };
 
 class FacadePhraseIndex{
@@ -211,6 +216,12 @@ public:
     bool store(guint8 phrase_index, MemoryChunk * new_chunk);
     bool unload(guint8 phrase_index);
 
+    /* load/store logger format.
+       the ownership of oldchunk and log is transfered to here. */
+    bool diff(guint8 phrase_index, MemoryChunk * oldchunk,
+              MemoryChunk * newlog);
+    bool merge(guint8 phrase_index, MemoryChunk * log);
+
     /* compat all SubPhraseIndex m_phrase_content memory usage.*/
     bool compat();
 
@@ -262,6 +273,7 @@ public:
 	m_total_freq -= item->get_unigram_frequency();
 	return result;
     }
+
 };
  
 };
diff --git a/src/storage/phrase_index_logger.h b/src/storage/phrase_index_logger.h
index 7641524..c961112 100644
--- a/src/storage/phrase_index_logger.h
+++ b/src/storage/phrase_index_logger.h
@@ -59,7 +59,7 @@ protected:
     }
 public:
     PhraseIndexLogger():m_offset(0){
-        m_chunk = NULL;
+        m_chunk = new MemoryChunk;
     }
 
     ~PhraseIndexLogger(){
@@ -102,7 +102,7 @@ public:
             size_t len = 0;
             m_chunk->get_content(offset, &len, sizeof(size_t));
             offset += sizeof(size_t);
-            newone->set_content(0, m_chunk->begin() + offset, len);
+            newone->set_content(0, ((char *)m_chunk->begin()) + offset, len);
             offset += len;
             break;
         }
@@ -112,7 +112,7 @@ public:
             size_t len = 0;
             m_chunk->get_content(offset, &len, sizeof(size_t));
             offset += sizeof(size_t);
-            oldone->set_content(0, m_chunk->begin() + offset, len);
+            oldone->set_content(0, ((char *)m_chunk->begin()) + offset, len);
             offset += len;
             break;
         }
@@ -124,9 +124,10 @@ public:
             offset += sizeof(size_t);
             m_chunk->get_content(offset, &newlen, sizeof(size_t));
             offset += sizeof(size_t);
-            oldone->set_content(0, m_chunk->begin() + offset, oldlen);
+            oldone->set_content(0, ((char *)m_chunk->begin()) + offset,
+                                oldlen);
             offset += oldlen;
-            newone->set_content(0, m_chunk->begin() + offset, newlen);
+            newone->set_content(0, ((char *)m_chunk->begin()) + offset, newlen);
             offset += newlen;
             break;
         }