Optimized file content parser.

libsearch/file-utils.cpp

@@ -28,8 +28,6 @@
 #include "quazip/quazip.h"
 #include <quazip/quazipfile.h>
 #include <QDomDocument>
-#include <QMimeDatabase>
-#include <QMimeType>
 #include <QQueue>
 #include "uchardet/uchardet.h"
 
@@ -211,14 +209,12 @@ void FileUtils::loadHanziTable(const QString &fileName)
     return;
 }
 
-QString FileUtils::getMimetype(QString &path, bool getsuffix)
+QMimeType FileUtils::getMimetype(QString &path)
 {
     QMimeDatabase mdb;
     QMimeType type = mdb.mimeTypeForFile(path,QMimeDatabase::MatchContent);
-    if(getsuffix)
-        return type.name();
-    else
-        return type.preferredSuffix();
+
+    return type;
 }
 
 //aborted
@@ -559,7 +555,7 @@ void FileUtils::getTxtContent(QString &path, QString &textcontent)
     const char *codec = uchardet_get_charset(chardet);
 
     if(QTextCodec::codecForName(codec) == 0)
-        qWarning()<<"Unsupported Text encoding format"<<path<<QString::fromLocal8Bit(codec)<<"zpf666";
+        qWarning()<<"Unsupported Text encoding format"<<path<<QString::fromLocal8Bit(codec);
 
     QTextStream stream(encodedString,QIODevice::ReadOnly);
     stream.setCodec(codec);

libsearch/file-utils.h

@@ -32,6 +32,8 @@
 #include <QCryptographicHash>
 #include <QIcon>
 #include <QMap>
+#include <QMimeDatabase>
+#include <QMimeType>
 #include "libsearch_global.h"
 //#define INITIAL_STATE 0
 //#define CREATING_INDEX 1
@@ -59,7 +61,7 @@ public:
     static void loadHanziTable(const QString&);
 
     //parse text,docx.....
-    static QString getMimetype(QString &path, bool getsuffix = false);
+    static QMimeType getMimetype(QString &path);
     static void getDocxTextContent(QString &path, QString &textcontent);
     static void getTxtContent(QString &path, QString &textcontent);
     static size_t _max_index_count;

libsearch/index/file-reader.cpp

@@ -28,20 +28,21 @@ FileReader::FileReader(QObject *parent) : QObject(parent)
 
 void FileReader::getTextContent(QString path, QString &textContent)
 {
-    QString type =FileUtils::getMimetype(path,true);
+    QMimeType type = FileUtils::getMimetype(path);
+    QString name = type.name();
     QFileInfo file(path);
     QString strsfx =  file.suffix();
-    if(type == "application/zip")
+    if(name== "application/zip")
     {
         if(strsfx.endsWith( "docx"))
             FileUtils::getDocxTextContent(path,textContent);
     }
-    else if(type == "text/plain")
+    else if(name == "text/plain")
     {
         if(strsfx.endsWith( "txt"))
             FileUtils::getTxtContent(path,textContent);
     }
-    else if(type == "application/x-ole-storage")
+    else if(type.inherits("application/msword"))
     {
         if (strsfx.endsWith("doc"))
         {
@@ -49,6 +50,10 @@ void FileReader::getTextContent(QString path, QString &textContent)
             searchdata.RunParser(path,textContent);
         }
     }
+    else
+    {
+        qWarning()<<"Unsupport format:["<<path<<"]["<<type.name()<<"]";
+    }
 
     return;
 }

libsearch/parser/binary-parser.cpp

@@ -5060,11 +5060,13 @@ bool KBinaryParser::read8DocText(FILE *pFile, const ppsInfoType *pPPS,
 				{
 					ushort* usAucData = (ushort*)ptaucBytes;
                     content.append(QString::fromUtf16(usAucData).replace("\r",""));
+                    usAucData = (ushort*)xfree((void*)usAucData);
 				}
 				else
 				{
                     //need more format document
-                    qWarning()<<"Parser error:";
+                    ptaucBytes = (UCHAR*)xfree((void*)ptaucBytes);
+                    qWarning()<<"Parser error:"<<m_strFileName;
 				}
             }
         }
@@ -5393,6 +5395,10 @@ int KBinaryParser::InitDocOle(FILE* pFile,long lFilesize,QString &content)
 						aulBBD, tBBDLen, aulSBD, tSBDLen,
                         aucHeader,content);
 	}
+    else
+    {
+        qWarning()<<"Unsupport doc type:"<<m_strFileName;
+    }
 //	else if (PPS_info.type == Excel)
 //	{
 //		readParam.ulStBlk = PPS_info.tWorkBook.ulSB;