opt: add xapian support

opt: add xapian fullindex support

dictionary.cc

@@ -567,13 +567,24 @@ bool needToRebuildIndex( vector< string > const & dictionaryFiles,
     if ( ts > lastModified )
       lastModified = ts;
   }
-
+#ifndef USE_XAPIAN
+  QDir d(FsEncoding::decode( indexFile.c_str() ));
+  if(d.exists()){
+    d.removeRecursively();
+  }
   QFileInfo fileInfo( FsEncoding::decode( indexFile.c_str() ) );
 
   if ( !fileInfo.exists() )
     return true;
 
   return fileInfo.lastModified().toSecsSinceEpoch() < lastModified;
+#else
+  QDir d(FsEncoding::decode( indexFile.c_str() ));
+  if(!d.exists()){
+    return true;
+  }
+  return false;
+#endif
 }
 
 QString generateRandomDictionaryId()

file.cc

@@ -114,6 +114,7 @@ void Class::open( char const * filename, char const * mode )
 
   f.setFileName( FsEncoding::decode( filename ) );
 
+  //maybe directory, the xapian use directory to store the index.
   if ( !f.open( openMode ) )
     throw exCantOpen( std::string( filename ) + ": " + f.errorString().toUtf8().data() );
 }

ftshelpers.cc

@@ -1,6 +1,9 @@
 /* This file is (c) 2014 Abs62
  * Part of GoldenDict. Licensed under GPLv3 or later, see the LICENSE file */
-
+#ifdef USE_XAPIAN
+#include "xapian.h"
+#include <stdlib.h>
+#endif
 #include "fulltextsearch.hh"
 #include "ftshelpers.hh"
 #include "wstring_qt.hh"
@@ -33,6 +36,25 @@ namespace FtsHelpers
 bool ftsIndexIsOldOrBad( string const & indexFile,
                          BtreeIndexing::BtreeDictionary * dict )
 {
+#ifdef USE_XAPIAN
+  try
+  {
+    Xapian::WritableDatabase db( dict->ftsIndexName() );
+  }
+  catch( const Xapian::Error & e )
+  {
+    qWarning() << e.get_description().c_str();
+    //the file is corrupted,remove it.
+    QFile::remove(QString::fromStdString(dict->ftsIndexName()));
+    return true;
+  }
+  catch( ... )
+  {
+    return true;
+  }
+  return false;
+#endif
+
   File::Class idx( indexFile, "rb" );
 
   FtsIdxHeader header;
@@ -321,8 +343,6 @@ void parseArticleForFts( uint32_t articleAddress, QString & articleText,
         words[ word ].push_back( articleAddress );*/
       }
     }
-
-
   }
 
   {
@@ -337,6 +357,10 @@ void parseArticleForFts( uint32_t articleAddress, QString & articleText,
 
 void makeFTSIndex( BtreeIndexing::BtreeDictionary * dict, QAtomicInt & isCancelled )
 {
+#ifdef USE_XAPIAN
+  return makeFTSIndexXapian(dict,isCancelled);
+#endif
+
   Mutex::Lock _( dict->getFtsMutex() );
 
   if( Utils::AtomicInt::loadAcquire( isCancelled ) )
@@ -466,6 +490,83 @@ void makeFTSIndex( BtreeIndexing::BtreeDictionary * dict, QAtomicInt & isCancell
   ftsIdx.writeRecords( &ftsIdxHeader, sizeof(ftsIdxHeader), 1 );
 }
 
+// use xapian to create the index
+#ifdef USE_XAPIAN
+void makeFTSIndexXapian( BtreeIndexing::BtreeDictionary * dict, QAtomicInt & isCancelled )
+{
+  Mutex::Lock _( dict->getFtsMutex() );
+
+  try {
+  if( Utils::AtomicInt::loadAcquire( isCancelled ) )
+    throw exUserAbort();
+
+  // Open the database for update, creating a new database if necessary.
+  Xapian::WritableDatabase db(dict->ftsIndexName(), Xapian::DB_CREATE_OR_OPEN);
+
+  Xapian::TermGenerator indexer;
+  Xapian::Stem stemmer("english");
+  indexer.set_stemmer(stemmer);
+  indexer.set_stemming_strategy(indexer.STEM_SOME_FULL_POS);
+
+  BtreeIndexing::IndexedWords indexedWords;
+
+  QSet< uint32_t > setOfOffsets;
+  setOfOffsets.reserve( dict->getArticleCount() );
+
+  dict->findArticleLinks( 0, &setOfOffsets, 0, &isCancelled );
+
+  if( Utils::AtomicInt::loadAcquire( isCancelled ) )
+    throw exUserAbort();
+
+  QVector< uint32_t > offsets;
+  offsets.resize( setOfOffsets.size() );
+  uint32_t * ptr = &offsets.front();
+
+  for( QSet< uint32_t >::ConstIterator it = setOfOffsets.constBegin();
+       it != setOfOffsets.constEnd(); ++it )
+  {
+    *ptr = *it;
+    ptr++;
+  }
+
+  // Free memory
+  setOfOffsets.clear();
+
+  if( Utils::AtomicInt::loadAcquire( isCancelled ) )
+    throw exUserAbort();
+
+  dict->sortArticlesOffsetsForFTS( offsets, isCancelled );
+
+  for( auto & address : offsets )
+  {
+    if( Utils::AtomicInt::loadAcquire( isCancelled ) )
+    {
+      return;
+    }
+
+    QString headword, articleStr;
+
+    dict->getArticleText( address, headword, articleStr );
+
+    Xapian::Document doc;
+
+    indexer.set_document( doc );
+    indexer.index_text( articleStr.toStdString() );
+    doc.add_boolean_term( std::to_string( address ) );
+    doc.set_data( std::to_string( address ) );
+    // Add the document to the database.
+    db.add_document( doc );
+  }
+  // Free memory
+  offsets.clear();
+
+  db.commit();
+  } catch (Xapian::Error & e) {
+    qWarning()<<QString::fromStdString(e.get_description());
+  }
+}
+#endif
+
 bool isCJKChar( ushort ch )
 {
   return Utils::isCJKChar(ch);
@@ -1056,6 +1157,10 @@ void FTSResultsRequest::fullSearch( QStringList & searchWords, QRegExp & regexp
 
 void FTSResultsRequest::run()
 {
+#ifdef USE_XAPIAN
+  return runXapian();
+#endif
+
   if ( dict.ensureInitDone().size() )
   {
     setErrorString( QString::fromUtf8( dict.ensureInitDone().c_str() ) );
@@ -1131,5 +1236,107 @@ void FTSResultsRequest::run()
   finish();
 }
 
+#ifdef USE_XAPIAN
+void FTSResultsRequest::runXapian()
+{
+  if ( dict.ensureInitDone().size() )
+  {
+    setErrorString( QString::fromUtf8( dict.ensureInitDone().c_str() ) );
+    finish();
+    return;
+  }
+
+  try
+  {
+    if( dict.haveFTSIndex() )
+    {
+      //no need to parse the search string,  use xapian directly.
+      //if the search mode is wildcard, change xapian search query flag?
+      // Open the database for searching.
+      Xapian::Database db(dict.ftsIndexName());
+
+      // Start an enquire session.
+      Xapian::Enquire enquire( db );
+
+      // Combine the rest of the command line arguments with spaces between
+      // them, so that simple queries don't have to be quoted at the shell
+      // level.
+      string query_string( searchString.toStdString() );
+
+      // Parse the query string to produce a Xapian::Query object.
+      Xapian::QueryParser qp;
+      Xapian::Stem stemmer( "english" );
+      qp.set_stemmer( stemmer );
+      qp.set_database( db );
+      qp.set_stemming_strategy( Xapian::QueryParser::STEM_SOME );
+      Xapian::QueryParser::feature_flag flag = Xapian::QueryParser::FLAG_DEFAULT;
+      if( searchMode == FTS::Wildcards )
+        flag = Xapian::QueryParser::FLAG_WILDCARD;
+      Xapian::Query query = qp.parse_query( query_string, flag );
+      qDebug() << "Parsed query is: " << query.get_description().c_str();
+
+      // Find the top 100 results for the query.
+      enquire.set_query( query );
+      Xapian::MSet matches = enquire.get_mset( 0, 100 );
+
+      // Display the results.
+      qDebug() << matches.get_matches_estimated() << " results found.\n";
+      qDebug() << "Matches 1-" << matches.size() << ":\n\n";
+      QList< uint32_t > offsetsForHeadwords;
+      for( Xapian::MSetIterator i = matches.begin(); i != matches.end(); ++i )
+      {
+        qDebug() << i.get_rank() + 1 << ": " << i.get_weight() << " docid=" << *i << " ["
+                 << i.get_document().get_data().c_str() << "]";
+        offsetsForHeadwords.append( atoi( i.get_document().get_data().c_str() ) );
+      }
+
+      if( !offsetsForHeadwords.isEmpty() )
+      {
+        QVector< QString > headwords;
+        Mutex::Lock _( dataMutex );
+        QString id = QString::fromUtf8( dict.getId().c_str() );
+        dict.getHeadwordsFromOffsets( offsetsForHeadwords, headwords, &isCancelled );
+        for( int x = 0; x < headwords.size(); x++ )
+        {
+          foundHeadwords->append( FTS::FtsHeadword( headwords.at( x ), id, QStringList(), matchCase ) );
+        }
+      }
+    }
+    else
+    {
+      QStringList indexWords, searchWords;
+      QRegExp searchRegExp;
+      if( !FtsHelpers::parseSearchString( searchString, indexWords, searchWords, searchRegExp,
+                                          searchMode, matchCase, distanceBetweenWords, hasCJK, ignoreWordsOrder ) )
+      {
+        finish();
+        return;
+      }
+      fullSearch( searchWords, searchRegExp );
+    }
+
+    if( foundHeadwords && foundHeadwords->size() > 0 )
+    {
+      Mutex::Lock _( dataMutex );
+      data.resize( sizeof( foundHeadwords ) );
+      memcpy( &data.front(), &foundHeadwords, sizeof( foundHeadwords ) );
+      foundHeadwords = 0;
+      hasAnyData = true;
+    }
+  }
+  catch (const Xapian::Error &e) {
+    qWarning() << e.get_description().c_str();
+  }
+  catch( std::exception &ex )
+  {
+    gdWarning( "FTS: Failed full-text search for \"%s\", reason: %s\n",
+               dict.getName().c_str(), ex.what() );
+    // Results not loaded -- we don't set the hasAnyData flag then
+  }
+
+  finish();
+}
+#endif
+
 } // namespace
 

ftshelpers.hh

@@ -64,7 +64,9 @@ void parseArticleForFts( uint32_t articleAddress, QString & articleText,
                          bool handleRoundBrackets = false );
 
 void makeFTSIndex( BtreeIndexing::BtreeDictionary * dict, QAtomicInt & isCancelled );
-
+#ifdef USE_XAPIAN
+void makeFTSIndexXapian( BtreeIndexing::BtreeDictionary * dict, QAtomicInt & isCancelled );
+#endif
 bool isCJKChar( ushort ch );
 
 class FTSResultsRequest : public Dictionary::DataRequest
@@ -142,7 +144,9 @@ public:
   }
 
   void run();
-
+  #ifdef USE_XAPIAN
+  void runXapian();
+  #endif
   virtual void cancel()
   {
     isCancelled.ref();

fulltextsearch.cc

@@ -225,7 +225,19 @@ FullTextSearchDialog::FullTextSearchDialog( QWidget * parent,
   ui.searchMode->addItem( tr( "Whole words" ), WholeWords );
   ui.searchMode->addItem( tr( "Plain text"), PlainText );
   ui.searchMode->addItem( tr( "Wildcards" ), Wildcards );
+#ifndef USE_XAPIAN
   ui.searchMode->addItem( tr( "RegExp" ), RegExp );
+#else
+  ui.matchCase->hide();
+  ui.articlesPerDictionary->hide();
+  ui.checkBoxArticlesPerDictionary->hide();
+  ui.checkBoxIgnoreDiacritics->hide();
+  ui.checkBoxDistanceBetweenWords->hide();
+  ui.distanceBetweenWords->hide();
+  ui.checkBoxIgnoreWordOrder->hide();
+
+  ui.searchLine->setToolTip(tr("support xapian search syntax,such as AND OR +/- etc"));
+#endif
   ui.searchMode->setCurrentIndex( cfg.preferences.fts.searchMode );
 
   ui.searchProgressBar->hide();
@@ -550,6 +562,26 @@ void FullTextSearchDialog::itemClicked( const QModelIndex & idx )
   {
     QString headword = results[ idx.row() ].headword;
     QRegExp reg;
+#ifdef USE_XAPIAN
+    auto searchText = ui.searchLine->text();
+    searchText.replace(
+      QRegularExpression( "[\\*\\?\\+\\\"]|\\bAnd\\b|\\bOR\\b", QRegularExpression::CaseInsensitiveOption ),
+      " " );
+    auto parts = searchText.split( QRegularExpression( "\\s" ), Qt::SkipEmptyParts );
+    QString firstAvailbeItem;
+    for( auto & p : parts )
+    {
+      if( p.startsWith( '-' ) )
+        continue;
+      firstAvailbeItem = p;
+      break;
+    }
+    if( !firstAvailbeItem.isEmpty() )
+    {
+      reg = QRegExp( firstAvailbeItem, Qt::CaseInsensitive, QRegExp::RegExp2 );
+      reg.setMinimal( true );
+    }
+#else
     if( !results[ idx.row() ].foundHiliteRegExps.isEmpty() )
     {
       reg = QRegExp( results[ idx.row() ].foundHiliteRegExps.join( "|"),
@@ -559,6 +591,7 @@ void FullTextSearchDialog::itemClicked( const QModelIndex & idx )
     }
     else
       reg = searchRegExp;
+#endif
     emit showTranslationFor( headword, results[ idx.row() ].dictIDs, reg, ignoreDiacritics );
   }
 }

fulltextsearch.ui

@@ -6,7 +6,7 @@
    <rect>
     <x>0</x>
     <y>0</y>
-    <width>492</width>
+    <width>562</width>
     <height>593</height>
    </rect>
   </property>
@@ -27,10 +27,33 @@
      </property>
      <layout class="QVBoxLayout" name="verticalLayout">
       <item>
-       <widget class="QLineEdit" name="searchLine"/>
+       <layout class="QHBoxLayout" name="horizontalLayout_5">
+        <item>
+         <widget class="QLineEdit" name="searchLine"/>
+        </item>
+        <item>
+         <widget class="QLabel" name="label_8">
+          <property name="text">
+           <string>Mode:</string>
+          </property>
+         </widget>
+        </item>
+        <item>
+         <widget class="QComboBox" name="searchMode"/>
+        </item>
+       </layout>
       </item>
       <item>
-       <layout class="QGridLayout" name="gridLayout_2">
+       <layout class="QGridLayout" name="fulltext_option_container">
+        <property name="sizeConstraint">
+         <enum>QLayout::SetMinAndMaxSize</enum>
+        </property>
+        <item row="1" column="1">
+         <widget class="QSpinBox" name="articlesPerDictionary"/>
+        </item>
+        <item row="0" column="1">
+         <widget class="QSpinBox" name="distanceBetweenWords"/>
+        </item>
         <item row="1" column="2">
          <widget class="QCheckBox" name="matchCase">
           <property name="text">
@@ -38,26 +61,6 @@
           </property>
          </widget>
         </item>
-        <item row="0" column="2">
-         <layout class="QHBoxLayout" name="horizontalLayout_2">
-          <item>
-           <widget class="QLabel" name="label_8">
-            <property name="text">
-             <string>Mode:</string>
-            </property>
-           </widget>
-          </item>
-          <item>
-           <widget class="QComboBox" name="searchMode"/>
-          </item>
-         </layout>
-        </item>
-        <item row="1" column="1">
-         <widget class="QSpinBox" name="articlesPerDictionary"/>
-        </item>
-        <item row="0" column="1">
-         <widget class="QSpinBox" name="distanceBetweenWords"/>
-        </item>
         <item row="0" column="0">
          <widget class="QCheckBox" name="checkBoxDistanceBetweenWords">
           <property name="text">
@@ -72,18 +75,14 @@
           </property>
          </widget>
         </item>
-       </layout>
+        <item row="2" column="0">
-      </item>
-      <item>
-       <layout class="QHBoxLayout" name="horizontalLayout_4">
-        <item>
          <widget class="QCheckBox" name="checkBoxIgnoreWordOrder">
           <property name="text">
            <string>Ignore words order</string>
           </property>
          </widget>
         </item>
-        <item>
+        <item row="2" column="1">
          <widget class="QCheckBox" name="checkBoxIgnoreDiacritics">
           <property name="text">
            <string>Ignore diacritics</string>
@@ -272,11 +271,9 @@
   </layout>
  </widget>
  <tabstops>
-  <tabstop>searchLine</tabstop>
   <tabstop>headwordsView</tabstop>
   <tabstop>checkBoxDistanceBetweenWords</tabstop>
   <tabstop>distanceBetweenWords</tabstop>
-  <tabstop>searchMode</tabstop>
   <tabstop>checkBoxArticlesPerDictionary</tabstop>
   <tabstop>articlesPerDictionary</tabstop>
   <tabstop>matchCase</tabstop>

goldendict.pro

@@ -56,6 +56,11 @@ DEFINES += QT_DISABLE_DEPRECATED_BEFORE=0x050F00
   DEFINES += MAKE_FFMPEG_PLAYER
 }
 
+CONFIG( use_xapian ) {
+  DEFINES += USE_XAPIAN
+  LIBS+= -lxapian
+}
+
 CONFIG += exceptions \
     rtti \
     stl  \