Merge pull request #156 from iaom/0301-dev

Add support for 'doc' file in file content search.

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/construct-document.cpp

@@ -122,6 +122,8 @@ void ConstructDocumentForContent::run()
         _doc_list_content = new QList<Document>;
     QString content;
     FileReader::getTextContent(m_path,content);
+    if(content.isEmpty())
+        return;
     QString uniqueterm = QString::fromStdString(FileUtils::makeDocUterm(m_path));
     QString upTerm = QString::fromStdString(FileUtils::makeDocUterm(m_path.section("/",0,-2,QString::SectionIncludeLeadingSep)));
 

libsearch/index/file-reader.cpp

@@ -19,6 +19,7 @@
  */
 #include "file-reader.h"
 #include "file-utils.h"
+#include "binary-parser.h"
 
 FileReader::FileReader(QObject *parent) : QObject(parent)
 {
@@ -27,13 +28,32 @@ FileReader::FileReader(QObject *parent) : QObject(parent)
 
 void FileReader::getTextContent(QString path, QString &textContent)
 {
-    //获取所有文件内容
-    //先分类
-    QString type =FileUtils::getMimetype(path,true);
-    if(type == "application/zip")
-         FileUtils::getDocxTextContent(path,textContent);
-    else if(type == "text/plain")
-         FileUtils::getTxtContent(path,textContent);
+    QMimeType type = FileUtils::getMimetype(path);
+    QString name = type.name();
+    QFileInfo file(path);
+    QString strsfx =  file.suffix();
+    if(name== "application/zip")
+    {
+        if(strsfx.endsWith( "docx"))
+            FileUtils::getDocxTextContent(path,textContent);
+    }
+    else if(name == "text/plain")
+    {
+        if(strsfx.endsWith( "txt"))
+            FileUtils::getTxtContent(path,textContent);
+    }
+    else if(type.inherits("application/msword"))
+    {
+        if (strsfx.endsWith("doc"))
+        {
+            KBinaryParser searchdata;
+            searchdata.RunParser(path,textContent);
+        }
+    }
+    else
+    {
+        qWarning()<<"Unsupport format:["<<path<<"]["<<type.name()<<"]";
+    }
 
     return;
 }

libsearch/index/file-reader.h

@@ -21,6 +21,7 @@
 #define FILEREADER_H
 
 #include <QObject>
+#include <QFileInfo>
 
 class FileReader : public QObject
 {

libsearch/index/first-index.h

@@ -63,7 +63,7 @@ private:
     QQueue<QVector<QString>>* q_index;
     QQueue<QString>* q_content_index;
     const QVector<QString> targetFileTypeVec ={
-//        QString(".doc"),
+        QString(".doc"),
         QString(".docx"),
 //        QString(".ppt"),
 //        QString(".pptx"),

libsearch/index/inotify-index.h

@@ -61,7 +61,7 @@ private:
 
     QMap<int, QString> currentPath;
     const QVector<QString> targetFileTypeVec ={
-//        QString(".doc"),
+        QString(".doc"),
         QString(".docx"),
 //        QString(".ppt"),
 //        QString(".pptx"),

libsearch/libsearch.pro

@@ -22,6 +22,7 @@ DEFINES += QT_DEPRECATED_WARNINGS
 #DEFINES += QT_DISABLE_DEPRECATED_BEFORE=0x060000    # disables all the APIs deprecated before Qt 6.0.0
 
 include(index/index.pri)
+include(parser/parser.pri))
 include(appsearch/appsearch.pri)
 include(settingsearch/settingsearch.pri))
 

libsearch/parser/binary-parser.h

@@ -0,0 +1,116 @@
+#ifndef SEARCHHELPER_H
+#define SEARCHHELPER_H
+#include <QtCore>
+#include <QtConcurrent/QtConcurrent>
+
+#define ULONG unsigned long
+#define UCHAR unsigned char
+#define USHORT unsigned short
+
+typedef enum
+{
+	Word = 0,
+	Excel,
+	Ppt
+}TYPE;
+
+/* Property Set Storage */
+typedef struct pps_tag
+{
+    ULONG	ulSB;
+    ULONG	ulSize;
+} ppsTag;
+
+typedef struct pps_info_tag
+{
+    ppsTag	tWordDocument;	/* Text stream */
+    ppsTag	tWorkBook;
+    ppsTag	tPPTDocument;
+    ppsTag	tData;		/* Data stream */
+    ppsTag	tTable;		/* Table stream */
+    ppsTag	tSummaryInfo;	/* Summary Information */
+    ppsTag	tDocSummaryInfo;/* Document Summary Information */
+    ppsTag	t0Table;	/* Table 0 stream */
+    ppsTag	t1Table;	/* Table 1 stream */
+    ppsTag	tCurrentUser;
+    TYPE		type;
+} ppsInfoType;
+
+/* Private type for Property Set Storage entries */
+typedef struct pps_entry_tag
+{
+    ULONG	ulNext;
+    ULONG	ulPrevious;
+    ULONG	ulDir;
+    ULONG	ulSB;
+    ULONG	ulSize;
+    int	iLevel;
+    char	szName[32];
+    UCHAR	ucType;
+} ppsEntryType;
+
+/* Excel Record Struct*/
+//typedef struct excelRecord
+//{
+//    excelRecord()
+//    {
+//        usLen = 0;
+//        usRichLen = 0;
+//        ulWLen = 0;
+//        bUni = false;
+//    }
+//    ushort usLen;
+//    ushort usRichLen;
+//    ulong ulWLen;
+//    bool bUni;
+//} excelRecord;
+
+typedef struct readDataParam
+{
+     readDataParam()
+     {
+         ulStBlk = 0;
+         pFile = NULL;
+         ulBBd = NULL;
+         tBBdLen = 0;
+         usBlkSize = 0;
+     }
+     ulong ulStBlk;
+     FILE *pFile;
+     ulong *ulBBd;
+     size_t tBBdLen;
+     ushort usBlkSize;
+}rdPara;
+
+class KBinaryParser :public QObject
+{
+	Q_OBJECT
+public:
+	KBinaryParser(QObject *parent=0);
+	~KBinaryParser();
+
+public:
+    bool RunParser(QString strFile,QString &content);
+
+private:
+	bool bGetPPS(FILE *pFile,
+		const ULONG *aulRootList, size_t tRootListLen, ppsInfoType *pPPS);
+
+	int readData(rdPara &readParam, uchar *aucBuffer, ulong ulOffset, size_t tToRead);
+
+    int InitDocOle(FILE *pFile,long lFilesize,QString &content);
+    bool read8DocText(FILE *pFile, const ppsInfoType *pPPS,
+        const ULONG *aulBBD, size_t tBBDLen,
+        const ULONG *aulSBD, size_t tSBDLen,
+        const UCHAR *aucHeader,QString &content);
+
+//	int readSSTRecord(readDataParam &rdParam, ppsInfoType, ulong &ulOff, ushort usPartLen);
+//	int read8BiffRecord(uchar uFlag, ulong ulOff, ulong &ulNext, readDataParam &rdParam, excelRecord &eR);
+
+//	ULONG readPPtRecord(FILE* pFile, ppsInfoType* PPS_info, ULONG* aulBBD,
+//                        size_t tBBDLen, ULONG ulPos);
+
+	QString m_strFileName;
+};
+
+#endif // SEARCHHELPER_H

libsearch/parser/common.h

@@ -0,0 +1,25 @@
+#ifndef COMMON_H
+#define COMMON_H
+#include <QtCore>
+#include <QtConcurrent/QtConcurrent>
+
+#define SERVER "Everything"
+
+#define LOG(a) \
+	//qWarning() << a;
+
+#define REHASH(a) \
+	if (sl_minus_1 < (int)sizeof(int) * CHAR_BIT)       \
+		hashHaystack -= (a) << sl_minus_1; \
+	hashHaystack <<= 1
+
+void* xmalloc(size_t tSize);
+
+void* xcalloc(size_t tNmemb, size_t tSize);
+
+void* xrealloc(void *pvArg, size_t tSize);
+
+void* xfree(void *pvArg);
+
+
+#endif // COMMON_H

libsearch/parser/parser.pri

@@ -0,0 +1,9 @@
+INCLUDEPATH += $$PWD
+
+HEADERS += \
+    $$PWD/common.h \
+    $$PWD/binary-parser.h 
+
+
+SOURCES += \
+    $$PWD/binary-parser.cpp