Don't redo encoding detection on partial index (variar#111)

xhargh · Nov 12, 2019 · 98f7583 · 98f7583
1 parent a1fb8d7
commit 98f7583
Show file tree

Hide file tree

Showing 2 changed files with 26 additions and 19 deletions.
diff --git a/src/logdata/include/data/logdataworker.h b/src/logdata/include/data/logdataworker.h
@@ -117,14 +117,14 @@ class IndexingData {
 
 struct IndexingState {
     EncodingParameters encodingParams;
-    LineOffset::UnderlyingType pos;
-    LineLength::UnderlyingType max_length;
-    LineLength::UnderlyingType additional_spaces;
-    LineOffset::UnderlyingType end;
-    LineOffset::UnderlyingType file_size;
-
-    QTextCodec* encodingGuess;
-    QTextCodec* fileTextCodec;
+    LineOffset::UnderlyingType pos = {};
+    LineLength::UnderlyingType max_length = {};
+    LineLength::UnderlyingType additional_spaces = {};
+    LineOffset::UnderlyingType end = {};
+    LineOffset::UnderlyingType file_size = {};
+
+    QTextCodec* encodingGuess = nullptr;
+    QTextCodec* fileTextCodec = nullptr;
 
     QSemaphore indexingSem;
     QSemaphore blockSem;

diff --git a/src/logdata/src/logdataworker.cpp b/src/logdata/src/logdataworker.cpp
@@ -363,21 +363,26 @@ FastLinePositionArray IndexOperation::parseDataBlock( LineOffset::UnderlyingType
 
 void IndexOperation::guessEncoding( const QByteArray& block, IndexingState& state ) const
 {
+    if ( !state.encodingGuess ) {
+        state.encodingGuess = EncodingDetector::getInstance().detectEncoding( block );
+        LOG( logINFO ) << "Encoding guess " << state.encodingGuess->name().toStdString();
+    }
+
     if ( !state.fileTextCodec ) {
         state.fileTextCodec = indexing_data_.getForcedEncoding();
+
         if ( !state.fileTextCodec ) {
-            state.fileTextCodec = EncodingDetector::getInstance().detectEncoding( block );
+            state.fileTextCodec = indexing_data_.getEncodingGuess();
+        }
+
+        if ( !state.fileTextCodec ) {
+            state.fileTextCodec = state.encodingGuess;
         }
 
         state.encodingParams = EncodingParameters( state.fileTextCodec );
         LOG( logINFO ) << "Encoding " << state.fileTextCodec->name().toStdString()
                        << ", Char width " << state.encodingParams.lineFeedWidth;
     }
-
-    if ( !state.encodingGuess ) {
-        state.encodingGuess = EncodingDetector::getInstance().detectEncoding( block );
-        LOG( logINFO ) << "Encoding guess " << state.encodingGuess->name().toStdString();
-    }
 }
 
 auto IndexOperation::setupIndexingProcess( IndexingState& indexingState )
@@ -441,13 +446,15 @@ void IndexOperation::doIndex( LineOffset initialPosition )
 
         IndexingState state;
         state.pos = initialPosition.get();
-        state.end = 0ll;
-        state.additional_spaces = 0;
-        state.max_length = 0;
-        state.encodingGuess = nullptr;
-        state.fileTextCodec = nullptr;
         state.file_size = file.size();
 
+        state.fileTextCodec = indexing_data_.getForcedEncoding();
+        if ( !state.fileTextCodec ) {
+            state.fileTextCodec = indexing_data_.getEncodingGuess();
+        }
+
+        state.encodingGuess = indexing_data_.getEncodingGuess();
+
         auto process = setupIndexingProcess( state );
         auto blockSender = std::get<0>( process );