Merge pull request #143 from iaom/0219-dev

Optimized Chinese phonetic alphabet index.

libsearch/index/construct-document.cpp

@@ -50,14 +50,14 @@ void ConstructDocumentForPath::run()
     //多音字版
     //现加入首字母
     QStringList pinyin_text_list = FileUtils::findMultiToneWords(QString(m_list.at(0)).replace(".",""));
-    if(!pinyin_text_list.isEmpty())
+//    if(!pinyin_text_list.isEmpty())
-    {
+//    {
-        for (QString& i : pinyin_text_list){
+//        for (QString& i : pinyin_text_list){
-            i.replace("", " ");
+//            i.replace("", " ");
-            i = i.simplified();
+//            i = i.simplified();
-        }
+//        }
-        doc.setIndexText(pinyin_text_list);
+//        doc.setIndexText(pinyin_text_list);
-    }
+//    }
 
     QString uniqueterm = QString::fromStdString(FileUtils::makeDocUterm(sourcePath));
     QString upTerm = QString::fromStdString(FileUtils::makeDocUterm(sourcePath.section("/",0,-2,QString::SectionIncludeLeadingSep)));
@@ -78,11 +78,22 @@ void ConstructDocumentForPath::run()
     int postingCount = 0;
     while(postingCount < index_text.size())
     {
-        QVector<size_t> p;
+//        QVector<size_t> p;
-        p.append(postingCount);
+//        p.append(postingCount);
-        doc.addPosting(QUrl::toPercentEncoding(index_text.at(postingCount)).toStdString(),p);
+        doc.addPosting(QUrl::toPercentEncoding(index_text.at(postingCount)).toStdString(),postingCount);
         ++postingCount;
     }
+    int i = 0;
+    for (QString& s : pinyin_text_list)
+    {
+        i = 0;
+        while(i < s.size())
+        {
+            doc.addPosting(QUrl::toPercentEncoding(s.at(i)).toStdString(),postingCount);
+            ++postingCount;
+            ++i;
+        }
+    }
 
 //    QMetaObject::invokeMethod(m_indexGenerator,"appendDocListPath",Q_ARG(Document,doc));
     _mutex_doc_list_path.lock();

libsearch/index/document.cpp

@@ -40,6 +40,16 @@ void Document::addPosting(std::string term,QVector<size_t> offset, int weight)
     }
 }
 
+void Document::addPosting(std::string term, unsigned int offset, int weight)
+{
+    if(term == "")
+        return;
+    if(term.length() > 240)
+        term = QString::fromStdString(term).left(30).toStdString();
+
+    m_document.add_posting(term,offset,weight);
+}
+
 void Document::addTerm(QString term)
 {
     if(term.isEmpty())

libsearch/index/document.h

@@ -42,6 +42,7 @@ public:
     }
     void setData(QString data);
     void addPosting(std::string term, QVector<size_t> offset, int weight =1);
+    void addPosting(std::string term, unsigned int offset, int weight =1);
     void addTerm(QString term);
     void addValue(QString value);
     void setUniqueTerm(QString term);

libsearch/index/file-searcher.cpp

@@ -246,11 +246,11 @@ Xapian::Query FileSearcher::creatQueryForFileSearch(QString keyword, Xapian::Dat
         //        qDebug()<<QString::fromStdString(Xapian::Query(QString(userInput.at(i)).toStdString()).get_description());
     }
     Xapian::Query queryPhrase =Xapian::Query(Xapian::Query::OP_PHRASE, v.begin(), v.end());
-    Xapian::Query queryNear =Xapian::Query(Xapian::Query::OP_NEAR, v.begin(), v.end());
+//    Xapian::Query queryNear =Xapian::Query(Xapian::Query::OP_NEAR, v.begin(), v.end());
 
-    Xapian::Query query = Xapian::Query(Xapian::Query::OP_AND,queryNear,queryPhrase);
+//    Xapian::Query query = Xapian::Query(Xapian::Query::OP_AND,queryNear,queryPhrase);
     //    qDebug()<<QString::fromStdString(query.get_description());
-    return query;
+    return queryPhrase;
 }
 
 Xapian::Query FileSearcher::creatQueryForContentSearch(QString keyword, Xapian::Database &db)

libsearch/index/index-generator.cpp

@@ -194,17 +194,17 @@ void IndexGenerator::insertIntoDatabase(Document& doc)
 {
 //    qDebug()<< "--index start--";
     Xapian::Document document = doc.getXapianDocument();
-    m_indexer.set_document(document);
+//    m_indexer.set_document(document);
 //    qDebug()<<doc.getIndexText();
 
-    for(auto i : doc.getIndexText()){
+//    for(auto i : doc.getIndexText()){
-        m_indexer.index_text(i.toStdString());
+//        m_indexer.index_text(i.toStdString());
-    }
+//    }
-    qDebug()<<"insert m_database_path:"<<QString::fromStdString(document.get_data());
+//    qDebug()<<"insert m_database_path:"<<QString::fromStdString(document.get_data());
-    qDebug()<<"document:"<<QString::fromStdString(document.get_description());
+//    qDebug()<<"document:"<<QString::fromStdString(document.get_description());
 
     Xapian::docid innerId= m_database_path->replace_document(doc.getUniqueTerm(),document);
-    qDebug()<<"replace doc docid="<<static_cast<int>(innerId);
+//    qDebug()<<"replace doc docid="<<static_cast<int>(innerId);
 //    qDebug()<< "--index finish--";
     return;
 }