From 9ee1b37693f5ce6890285ae9289e0da473cc054d Mon Sep 17 00:00:00 2001
From: zhangpengfei <zhangpengfei@kylinos.com>
Date: Sat, 2 Jan 2021 17:21:38 +0800
Subject: [PATCH] Store the content text for search and generate snippets when
 performing text content search.

---
 libsearch/index/document.cpp        |  9 +++--
 libsearch/index/document.h          |  3 +-
 libsearch/index/file-searcher.cpp   | 54 ++++++++++++++++++++++++++---
 libsearch/index/file-searcher.h     |  6 ++--
 libsearch/index/filetypefilter.cpp  |  4 +--
 libsearch/index/filetypefilter.h    | 15 ++++----
 libsearch/index/index-generator.cpp |  7 ++--
 src/main.cpp                        |  4 +--
 8 files changed, 77 insertions(+), 25 deletions(-)

diff --git a/libsearch/index/document.cpp b/libsearch/index/document.cpp
index 0c8bde4..980ca67 100644
--- a/libsearch/index/document.cpp
+++ b/libsearch/index/document.cpp
@@ -23,11 +23,14 @@ void Document::setData(QString data)
     m_document->set_data(data.toStdString());
 }
 
-void Document::addterm(std::string term, int weight)
+void Document::addterm(std::string term,QVector<size_t> offset, int weight)
 {
-    if(term == "")
+    if(term == ""||term.length() > 240)
         return;
-    m_document->add_term(term,weight);
+    for(size_t i : offset)
+    {
+        m_document->add_posting(term,i,weight);
+    }
 }
 
 void Document::addValue(QString value)
diff --git a/libsearch/index/document.h b/libsearch/index/document.h
index 807aecc..96d5ee1 100644
--- a/libsearch/index/document.h
+++ b/libsearch/index/document.h
@@ -4,6 +4,7 @@
 #include <xapian.h>
 #include <QString>
 #include <QStringList>
+#include <QVector>
 
 class Document
 {
@@ -11,7 +12,7 @@ public:
     Document();
     ~Document();
     void setData(QString data);
-    void addterm(std::string term,int weight =1);
+    void addterm(std::string term, QVector<size_t> offset, int weight =1);
     void addValue(QString value);
     void setUniqueTerm(QString term);
     std::string getUniqueTerm();
diff --git a/libsearch/index/file-searcher.cpp b/libsearch/index/file-searcher.cpp
index 3e744e6..2641f15 100644
--- a/libsearch/index/file-searcher.cpp
+++ b/libsearch/index/file-searcher.cpp
@@ -75,7 +75,7 @@ void FileSearcher::onKeywordSearch(QString keyword, int begin, int num)
 
 void FileSearcher::onKeywordSearchContent(QString keyword, int begin, int num)
 {
-    QStringList searchResult;
+    QMap<QString,QStringList> searchResult = QMap<QString,QStringList>();
     try
     {
         qDebug()<<"--content search start--";
@@ -83,11 +83,11 @@ void FileSearcher::onKeywordSearchContent(QString keyword, int begin, int num)
         Xapian::Database db(CONTENT_INDEX_PATH);
         Xapian::Enquire enquire(db);
         Xapian::QueryParser qp;
-        qp.set_default_op(Xapian::Query::OP_PHRASE);
+//        qp.set_default_op(Xapian::Query::OP_PHRASE);
         qp.set_database(db);
 
         //Creat a query
-        Xapian::Query queryPhrase = qp.parse_query(keyword.toStdString(),Xapian::QueryParser::FLAG_PHRASE);
+        Xapian::Query queryPhrase = qp.parse_query(keyword.toStdString());
 
         qDebug()<<QString::fromStdString(queryPhrase.get_description());
 
@@ -95,7 +95,7 @@ void FileSearcher::onKeywordSearchContent(QString keyword, int begin, int num)
         //dir result
         Xapian::MSet result = enquire.get_mset(begin, begin+num);
         qDebug()<< "find results count=" <<static_cast<int>(result.get_matches_estimated());
-        searchResult = getResult(result);
+        searchResult = getContentResult(result,keyword);
 
         qDebug()<< "--content search finish--";
     }
@@ -118,6 +118,7 @@ QStringList FileSearcher::getResult(Xapian::MSet &result)
     QStringList searchResult = QStringList();
     if(result.size() == 0)
         return searchResult;
+
     for (auto it = result.begin(); it != result.end(); ++it)
     {
         Xapian::Document doc = it.get_document();
@@ -135,10 +136,53 @@ QStringList FileSearcher::getResult(Xapian::MSet &result)
         {
             searchResult.append(QString::fromStdString(data));
         }
-
         qDebug()<< "doc="<< QString::fromStdString(data) << ",weight=" <<docScoreWeight << ",percent=" << docScorePercent;
     }
     //        if(!pathTobeDelete->isEmpty())
     //            deleteAllIndex(pathTobeDelete)
     return searchResult;
 }
+
+QMap<QString,QStringList> FileSearcher::getContentResult(Xapian::MSet &result, QString &keyWord)
+{
+    //QStringList *pathTobeDelete = new QStringList;
+    //Delete those path doc which is not already exist.
+
+    int size = keyWord.size();
+    QMap<QString,QStringList> searchResult;
+    if(result.size() == 0)
+        return searchResult;
+
+    for (auto it = result.begin(); it != result.end(); ++it)
+    {
+        Xapian::Document doc = it.get_document();
+        std::string data = doc.get_data();
+        double docScoreWeight = it.get_weight();
+        Xapian::percent docScorePercent = it.get_percent();
+        QString path = QString::fromStdString(doc.get_value(1));
+        QFileInfo *info = new QFileInfo(path);
+
+        if(!info->exists())
+        {
+            //                pathTobeDelete->append(QString::fromStdString(data));
+            qDebug()<<path<<"is not exist!!";
+            continue;
+        }
+        // Construct snippets containing keyword.
+        QStringList snippets;
+        auto term = doc.termlist_begin();
+        term.skip_to(keyWord.toStdString());
+        for(auto pos = term.positionlist_begin();pos != term.positionlist_end();++pos)
+        {
+            QByteArray snippetByte = QByteArray::fromStdString(data);
+            QString snippet = "..."+QString(snippetByte.left(*pos)).right(size +5) + QString(snippetByte.mid(*pos,-1)).left(size+5) + "...";
+//            qDebug()<<snippet;
+            snippets.append(snippet);
+        }
+        searchResult.insert(path,snippets);
+        qDebug()<< "path="<< path << ",weight=" <<docScoreWeight << ",percent=" << docScorePercent;
+    }
+    //        if(!pathTobeDelete->isEmpty())
+    //            deleteAllIndex(pathTobeDelete)
+    return searchResult;
+}
diff --git a/libsearch/index/file-searcher.h b/libsearch/index/file-searcher.h
index 8ca2577..dfe7fb7 100644
--- a/libsearch/index/file-searcher.h
+++ b/libsearch/index/file-searcher.h
@@ -5,6 +5,7 @@
 #include <xapian.h>
 #include <QStandardPaths>
 #include <QVector>
+#include <QMap>
 #define INDEX_PATH (QStandardPaths::writableLocation(QStandardPaths::HomeLocation)+"/.config/org.ukui/index_data").toStdString()
 #define CONTENT_INDEX_PATH (QStandardPaths::writableLocation(QStandardPaths::HomeLocation)+"/.config/org.ukui/content_index_data").toStdString()
 
@@ -20,10 +21,11 @@ public Q_SLOTS:
     void onKeywordSearchContent(QString keyword, int begin = 0, int num = 20);
 
 Q_SIGNALS:
-    void result(QVector<QStringList> resultV);
-    void contentResult(QStringList resultL);
+    void result(QVector<QStringList> resultP);
+    void contentResult(QMap<QString,QStringList> resultC);
 private:
     QStringList getResult(Xapian::MSet &result);
+    QMap<QString,QStringList> getContentResult(Xapian::MSet &result,QString &keyWord);
 };
 
 #endif // FILESEARCHER_H
diff --git a/libsearch/index/filetypefilter.cpp b/libsearch/index/filetypefilter.cpp
index 9b908d2..f53f0b6 100644
--- a/libsearch/index/filetypefilter.cpp
+++ b/libsearch/index/filetypefilter.cpp
@@ -21,7 +21,7 @@ void FileTypeFilter::DoSomething(const QFileInfo& fileInfo){
 //    qDebug() << qmt.preferredSuffix();
     for (auto i : this->targetFileTypeVec){
         if (fileInfo.fileName().endsWith(i)){
-//            qDebug() << fileInfo.fileName();
+            qDebug() << fileInfo.fileName();
             this->result->append(fileInfo.absoluteFilePath());
         }
     }
@@ -36,7 +36,7 @@ QList<QString>* FileTypeFilter::getTargetFileAbsolutePath(){
 void FileTypeFilter::Test(){
     IndexGenerator* ig = IndexGenerator::getInstance();
 //    this->result = new QList<QString>();
-//    this->result->append(QString("/home/zpf/桌面/DOCX 文档.docx"));
+//    this->result->append(QString("/home/zpf/桌面/DOCX 文档(1).docx"));
 
         ig->creatAllIndex(this->result);
 }
diff --git a/libsearch/index/filetypefilter.h b/libsearch/index/filetypefilter.h
index 4f517ba..cdefc09 100644
--- a/libsearch/index/filetypefilter.h
+++ b/libsearch/index/filetypefilter.h
@@ -18,13 +18,14 @@ public:
     void Test();
 Q_SIGNALS:
 private:
-    const QVector<QString> targetFileTypeVec ={ /*QString(".doc"),*/
-                                                QString(".docx")/*,*/
-                                                /*QString(".ppt"),
-                                                QString(".pptx"),
-                                                QString(".xls"),
-                                                QString(".xlsx"),
-                                                QString(".txt")*/};
+    const QVector<QString> targetFileTypeVec ={
+//        QString(".doc"),
+        QString(".docx"),
+/*        QString(".ppt"),
+        QString(".pptx"),
+        QString(".xls"),
+        QString(".xlsx"),
+        QString(".txt")*/};
     QList<QString>* result;
 
 };
diff --git a/libsearch/index/index-generator.cpp b/libsearch/index/index-generator.cpp
index 009feda..32642dc 100644
--- a/libsearch/index/index-generator.cpp
+++ b/libsearch/index/index-generator.cpp
@@ -201,7 +201,6 @@ Document IndexGenerator::GenerateDocument(const QVector<QString> &list)
     doc.setData(sourcePath);
     doc.setUniqueTerm(uniqueterm);
     doc.addValue(list.at(2));
-    if(list.at(2) == QString("1"))
     QStringList temp;
     temp.append(index_text);
 //    temp.append(pinyin_text_list);
@@ -219,11 +218,12 @@ Document IndexGenerator::GenerateContentDocument(const QString &path)
     QString uniqueterm = QString::fromStdString(FileUtils::makeDocUterm(path));
     QVector<SKeyWord> term = ChineseSegmentation::callSegement(content);
     Document doc;
-    doc.setData(path);
+    doc.setData(*content);
     doc.setUniqueTerm(uniqueterm);
+    doc.addValue(path);
     for(int i = 0;i<term.size();++i)
     {
-        doc.addterm(term.at(i).word,static_cast<int>(term.at(i).weight));
+        doc.addterm(term.at(i).word,term.at(i).offsets,static_cast<int>(term.at(i).weight));
 
     }
     return doc;
@@ -322,6 +322,7 @@ bool IndexGenerator::deleteAllIndex(QStringList *pathlist)
         {
             qDebug()<<"--delete start--";
             m_datebase_path->delete_document(uniqueterm);
+            m_database_content->delete_document(uniqueterm);
             qDebug()<<"delete md5"<<QString::fromStdString(uniqueterm);
             m_datebase_path->commit();
             qDebug()<< "--delete finish--";
diff --git a/src/main.cpp b/src/main.cpp
index 3602e42..8328cb1 100644
--- a/src/main.cpp
+++ b/src/main.cpp
@@ -69,8 +69,8 @@ int main(int argc, char *argv[])
     /*-------------InotyifyRefact Test End-----------------*/
 
     /*-------------文本搜索 Test start-----------------*/
-//    FileSearcher *search = new FileSearcher();
-//    search->onKeywordSearchContent("麒麟");
+    FileSearcher *search = new FileSearcher();
+    search->onKeywordSearchContent("测试");
     /*-------------文本搜索 Test End-----------------*/
 
     qRegisterMetaType<QVector<QStringList>>("QVector<QStringList>");