From a7e296ca0c6239bd0c36bf256c600b09e59df543 Mon Sep 17 00:00:00 2001
From: baijunjie <baijunjie@kylinos.cn>
Date: Fri, 20 Aug 2021 10:10:46 +0800
Subject: [PATCH] fix the question that the keyword will be cut off in the
 resultPage of the file content search.

---
 libsearch/file-utils.cpp           | 66 ++++++++++++++++++++++++++++++
 libsearch/file-utils.h             |  3 ++
 libsearch/index/search-manager.cpp | 34 ++++++++-------
 3 files changed, 88 insertions(+), 15 deletions(-)

diff --git a/libsearch/file-utils.cpp b/libsearch/file-utils.cpp
index 113fbf5..98789a0 100644
--- a/libsearch/file-utils.cpp
+++ b/libsearch/file-utils.cpp
@@ -773,3 +773,69 @@ void FileUtils::getTxtContent(QString &path, QString &textcontent) {
 
     return;
 }
+
+QString FileUtils::chineseSubString(const std::string &data, int start, int length)
+{
+    std::string afterSub = "";
+    //越界保护
+    if(start < 0 || length < 0){
+        return " ";
+    }
+    if (length >= data.length()) {
+        return QString::fromStdString(data);
+    }
+
+    QString snippet = "";
+    QFont ft(QApplication::font().family(),QApplication::font().pointSize());
+    QFontMetrics fm (ft);
+
+    if (start + length <= data.length()) {
+        afterSub = data.substr(start,length);    //截取
+        snippet = QString::fromStdString(afterSub);    //转QString
+
+        if(start + length < data.length()){
+           snippet.replace(snippet.length()-3,3,"...");    //替换后三位
+        }
+        snippet = fm.elidedText(snippet, Qt::ElideRight, 2*366);    //超过两行则省略
+    } else {
+        int newStart = data.length()-length;    //更新截取位置
+        int bodyByteCount = 0;
+        int letterCount = 0;
+
+        for(int i = newStart;i > 0;i--){
+            if(data[i] & 0x80) {
+                char tmp = data[i] << 1;
+                if(tmp & 0x80){
+                    break;
+                } else {
+                    bodyByteCount++;
+                }
+
+            } else {
+                letterCount++;
+                if(letterCount == 3) {
+                    break;
+                }
+
+            }
+        }
+        int correctionValue = letterCount + (bodyByteCount);
+        afterSub = data.substr(newStart - correctionValue, length + correctionValue);
+        snippet=QString::fromStdString(afterSub);
+
+        switch(letterCount)
+        {
+        case 0:
+            snippet.replace(0,1,".").prepend("..");
+            break;
+        case 1:
+            snippet.replace(0,2,"..").prepend(".");
+            break;
+        default:
+            snippet.replace(0,3,"...");
+        }
+
+        snippet = fm.elidedText(snippet, Qt::ElideLeft, 2*500);
+    }
+    return snippet;
+}
diff --git a/libsearch/file-utils.h b/libsearch/file-utils.h
index f0a9085..b3ef3ae 100644
--- a/libsearch/file-utils.h
+++ b/libsearch/file-utils.h
@@ -35,6 +35,8 @@
 #include <QMap>
 #include <QDomDocument>
 #include <QQueue>
+#include <QApplication>
+#include <QFontMetrics>
 
 #include <quazip/quazipfile.h>
 #include <stdio.h>
@@ -83,6 +85,7 @@ public:
     static void getXlsxTextContent(QString &path, QString &textcontent);
     static void getPdfTextContent(QString &path, QString &textcontent);
     static void getTxtContent(QString &path, QString &textcontent);
+    static QString chineseSubString(const std::string &data, int start, int length);
     static size_t _max_index_count;
     static size_t _current_index_count; //this one has been Abandoned,do not use it.
     static unsigned short _index_status;
diff --git a/libsearch/index/search-manager.cpp b/libsearch/index/search-manager.cpp
index c501594..09edbf1 100644
--- a/libsearch/index/search-manager.cpp
+++ b/libsearch/index/search-manager.cpp
@@ -338,22 +338,26 @@ int FileContentSearch::getResult(Xapian::MSet &result, std::string &keyWord) {
 //        qWarning()<<QString::fromStdString(s);
         auto term = doc.termlist_begin();
         std::string wordTobeFound = QString::fromStdString(keyWord).section(" ", 0, 0).toStdString();
-        int size = wordTobeFound.length();
+//        int size = wordTobeFound.length();
         term.skip_to(wordTobeFound);
-        int count = 0;
-        for(auto pos = term.positionlist_begin(); pos != term.positionlist_end() && count < 6; ++pos) {
-            std::string s = data.substr((*pos < 60) ? 0 : (*pos  - 60), size + 120);
-            QString snippet = QString::fromStdString(s);
-            if(snippet.size() > 6 + QString::fromStdString(keyWord).size()) {
-                snippet.replace(0, 3, "...").replace(snippet.size() - 3, 3, "...");
-            } else {
-                snippet.append("...").prepend("...");
-            }
-            snippets.append(snippet);
-            QString().swap(snippet);
-            std::string().swap(s);
-            ++count;
-        }
+        auto pos = term.positionlist_begin();
+        int length = 120;
+        QString snippet = FileUtils::chineseSubString(data,*pos,length);
+//        int count = 0;
+//        for(auto pos = term.positionlist_begin(); pos != term.positionlist_end() && count < 6; ++pos) {
+//            std::string s = data.substr((*pos < 60) ? 0 : (*pos  - 60), size + 120);
+//            QString snippet = QString::fromStdString(s);
+//            if(snippet.size() > 6 + QString::fromStdString(keyWord).size()) {
+//                    snippet.replace(0, 3, "...").replace(snippet.size() - 3, 3, "...");
+//            } else {
+//                snippet.append("...").prepend("...");
+//            }
+//            snippets.append(snippet);
+//            QString().swap(snippet);
+//            std::string().swap(s);
+//            ++count;
+//        }
+        snippets.append(snippet);
         std::string().swap(data);
 
 //        for(QString i : QString::fromStdString(keyWord).split(" ",QString::SkipEmptyParts))