fix: There are issues with searching for special characters and numbe…

…rs based on keywords in the file content Modify the word segmentation when establishing an index Log: There are issues with searching for special characters and numbers based on keywords in the file content Bug: https://pms.uniontech.com/bug-view-270025.html
linuxdeepin · Sep 14, 2024 · 597b5a1 · 597b5a1
1 parent 2170200
commit 597b5a1
Show file tree

Hide file tree

Showing 3 changed files with 20 additions and 36 deletions.
diff --git a/src/plugins/filemanager/dfmplugin-search/3rdparty/fulltext/chinesetokenizer.cpp b/src/plugins/filemanager/dfmplugin-search/3rdparty/fulltext/chinesetokenizer.cpp
@@ -80,7 +80,6 @@ bool ChineseTokenizer::incrementToken()
     length = 0;
     start = offset;
 
-    bool last_is_en = false, last_is_num = false;
     while (true) {
         wchar_t c;
         ++offset;
@@ -97,41 +96,13 @@ bool ChineseTokenizer::incrementToken()
             c = ioBuffer[bufferIndex++];
         }
 
-        if (UnicodeUtil::isLower(c) || UnicodeUtil::isUpper(c)) {
-            if (last_is_num) {
-                --bufferIndex;
-                --offset;
-                return flush();
-            }
-
-            push(c);
-            if (length == kMaxWordLen) {
-                return flush();
-            }
-            last_is_en = true;
-        } else if (UnicodeUtil::isDigit(c)) {
-            if (last_is_en) {
-                --bufferIndex;
-                --offset;
-                return flush();
-            }
-
-            push(c);
-            if (length == kMaxWordLen) {
-                return flush();
-            }
-            last_is_num = true;
-        } else if (UnicodeUtil::isOther(c)) {
-            if (length > 0) {
-                --bufferIndex;
-                --offset;
-                return flush();
-            }
-            push(c);
-            return flush();
-        } else if (length > 0) {
+        if (length > 0) {
+            --bufferIndex;
+            --offset;
             return flush();
         }
+        push(c);
+        return flush();
     }
 }
 

diff --git a/...plugins/filemanager/dfmplugin-search/searchmanager/searcher/fulltext/fulltextsearcher.cpp b/...plugins/filemanager/dfmplugin-search/searchmanager/searcher/fulltext/fulltextsearcher.cpp
@@ -394,7 +394,6 @@ bool FullTextSearcherPrivate::doSearch(const QString &path, const QString &keywo
     } catch (...) {
         fmWarning() << "Search failed!";
     }
-
     return true;
 }
 
@@ -436,6 +435,19 @@ QString FullTextSearcherPrivate::dealKeyword(const QString &keyword)
     return newStr.trimmed();
 }
 
+QString FullTextSearcherPrivate::dealKeywordEx(const QString &keyword)
+{
+    auto key = keyword;
+    for(int i = 0; i < key.length(); i++) {
+        if(QChar(key[i]).isPrint()) {
+            key.insert(i, "\\");
+            i++;
+        }
+    }
+
+    return key;
+}
+
 FullTextSearcher::FullTextSearcher(const QUrl &url, const QString &key, QObject *parent)
     : AbstractSearcher(url, key, parent),
       d(new FullTextSearcherPrivate(this))
@@ -477,7 +489,7 @@ bool FullTextSearcher::search()
         return false;
 
     const QString path = UrlRoute::urlToPath(searchUrl);
-    const QString key = d->dealKeyword(keyword);
+    const QString key = d->dealKeywordEx(keyword).trimmed();
     if (path.isEmpty() || key.isEmpty()) {
         d->status.storeRelease(kCompleted);
         return false;

diff --git a/...plugins/filemanager/dfmplugin-search/searchmanager/searcher/fulltext/fulltextsearcher_p.h b/...plugins/filemanager/dfmplugin-search/searchmanager/searcher/fulltext/fulltextsearcher_p.h
@@ -61,6 +61,7 @@ class FullTextSearcherPrivate : public QObject
 
     Lucene::DocumentPtr fileDocument(const QString &file);
     QString dealKeyword(const QString &keyword);
+    QString dealKeywordEx(const QString &keyword);
     void doIndexTask(const Lucene::IndexReaderPtr &reader, const Lucene::IndexWriterPtr &writer, const QString &path, TaskType type);
     void indexDocs(const Lucene::IndexWriterPtr &writer, const QString &file, IndexType type);
     bool checkUpdate(const Lucene::IndexReaderPtr &reader, const QString &file, IndexType &type);