修复中文分词接口编译问题；

2023-03-20 15:34:05 +08:00 · 2023-03-20 15:34:05 +08:00 · 79527082f9
parent 73cccc4083
commit 79527082f9
1 changed files with 1 additions and 3 deletions
--- a/libsearch/index/file-content-indexer.cpp
+++ b/libsearch/index/file-content-indexer.cpp
@ -41,9 +41,7 @@ bool fileContentIndexer::index()
    }

    m_document.setData(content);
-    //'\xEF\xBC\x8C' is "，" "\xE3\x80\x82" is "。"  use three " " to replace ,to ensure the offset info.
-    content = content.replace("\t", " ").replace("\xEF\xBC\x8C", "   ").replace("\xE3\x80\x82", "   ");
-    std::vector<KeyWord> term = ChineseSegmentation::getInstance()->callSegment(content.left(20480000).toStdString());
+    std::vector<KeyWord> term = ChineseSegmentation::getInstance()->callSegment(content);
    content.clear();
    content.squeeze();