the last headword in the block has incorrect data length

xiaoyifang · Jun 23, 2022 · b617323 · b617323
1 parent 4a3905c
commit b617323
Show file tree

Hide file tree

Showing 3 changed files with 31 additions and 9 deletions.
diff --git a/mdictparser.cc b/mdictparser.cc
@@ -595,10 +595,12 @@ MdictParser::HeadWordIndex MdictParser::splitHeadWordBlock( QByteArray const & b
 }
 
 bool MdictParser::readRecordBlock( MdictParser::HeadWordIndex & headWordIndex,
-                                   MdictParser::RecordHandler & recordHandler )
+                                   MdictParser::RecordHandler & recordHandler,
+                                   bool cross_block_read )
 {
   // cache the index, the headWordIndex is already sorted
   size_t idx = 0;
+  bool readNextBlock = false;
 
   for ( HeadWordIndex::const_iterator i = headWordIndex.begin(); i != headWordIndex.end(); ++i )
   {
@@ -611,21 +613,41 @@ bool MdictParser::readRecordBlock( MdictParser::HeadWordIndex & headWordIndex,
     RecordIndex const & recordIndex = recordBlockInfos_[idx];
     HeadWordIndex::const_iterator iNext = i + 1;
     qint64 recordSize;
-    if ( iNext == headWordIndex.end() )
-      recordSize = recordIndex.shadowEndPos - i->first;
+    auto current = *i;
+
+    if( iNext == headWordIndex.end() )
+    {
+      qint64 lastWordSize = recordIndex.shadowEndPos - current.first;
+
+      readNextBlock   = cross_block_read && readNextHeadWordIndex( headWordIndex );
+      if(readNextBlock)
+      {
+        recordSize = qMin(lastWordSize, headWordIndex.begin()->first - current.first);
+      }
+      else
+      {
+        recordSize = lastWordSize;
+      }
+    }
     else
-      recordSize = iNext->first - i->first;
+      recordSize = iNext->first - current.first;
 
     RecordInfo recordInfo;
     recordInfo.compressedBlockPos = recordPos_ + recordIndex.startPos;
-    recordInfo.recordOffset = i->first - recordIndex.shadowStartPos;
+    recordInfo.recordOffset          = current.first - recordIndex.shadowStartPos;
     recordInfo.decompressedBlockSize = recordIndex.decompressedSize;
     recordInfo.compressedBlockSize = recordIndex.compressedSize;
     recordInfo.recordSize = recordSize;
 
-    recordHandler.handleRecord( i->second, recordInfo );
+    recordHandler.handleRecord( current.second, recordInfo );
+
+    if( readNextBlock )
+        break;
   }
 
+  if( readNextBlock )
+    readRecordBlock( headWordIndex, recordHandler, cross_block_read );
+
   return true;
 }
 

diff --git a/mdictparser.hh b/mdictparser.hh
@@ -158,7 +158,7 @@ public:
 
   bool open( const char * filename );
   bool readNextHeadWordIndex( HeadWordIndex & headWordIndex );
-  bool readRecordBlock( HeadWordIndex & headWordIndex, RecordHandler & recordHandler );
+  bool readRecordBlock( HeadWordIndex & headWordIndex, RecordHandler & recordHandler, bool cross_block_read=false );
 
   // helpers
   static QString toUtf16( const char * fromCode, const char * from, size_t fromSize );

diff --git a/mdx.cc b/mdx.cc
@@ -1440,9 +1440,9 @@ vector< sptr< Dictionary::Class > > makeDictionaries( vector< string > const & f
       MdictParser::HeadWordIndex headWordIndex;
 
       // enumerating word and its definition
-      while ( parser.readNextHeadWordIndex( headWordIndex ) )
+      if ( parser.readNextHeadWordIndex( headWordIndex ) )
       {
-        parser.readRecordBlock( headWordIndex, articleHandler );
+        parser.readRecordBlock( headWordIndex, articleHandler, true);
       }
 
       // enumerating resources if there's any