tris/recoll
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

*** empty log message ***

Browse Source
This commit is contained in:
dockes 2005-01-24 13:17:59 +00:00
parent 0dde2934d6
commit 5210139b85
14 changed files with 821 additions and 17 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
src/common/Makefile Normal file
View File

@ -0,0 +1,24 @@
CXXFLAGS = -Wall -g -I. -I../unac -I../utils -I/usr/local/include
CFLAGS = -g -I. -I../unac -I/usr/local/include -DUNAC_VERSION=\"1.0.7\"
PROGS = unacpp
all: $(PROGS)
UNACPP_OBJS= trunacpp.o unacpp.o unac.o readfile.o
unacpp : $(UNACPP_OBJS)
$(CXX) $(CXXFLAGS) -o unacpp $(UNACPP_OBJS) \
-L/usr/local/lib -liconv
unac.o : ../unac/unac.c ../unac/unac.h
$(CC) $(CFLAGS) -c -o unac.o ../unac/unac.c
readfile.o : ../utils/readfile.cpp ../utils/readfile.h
$(CXX) $(CXXFLAGS) -c -o readfile.o ../utils/readfile.cpp
trunacpp.o : unacpp.cpp unacpp.h
$(CXX) $(CXXFLAGS) -DTEST_UNACPP -c -o trunacpp.o unacpp.cpp
clean:
rm -f *.o $(PROGS)

4
src/common/rclconfig.cpp
View File

@ -1,5 +1,5 @@
#ifndef lint #ifndef lint
static char rcsid[] = "@(#$Id: rclconfig.cpp,v 1.2 2004-12-15 15:00:36 dockes Exp $ (C) 2004 J.F.Dockes"; static char rcsid[] = "@(#$Id: rclconfig.cpp,v 1.3 2005-01-24 13:17:58 dockes Exp $ (C) 2004 J.F.Dockes";
#endif #endif
#include <iostream> #include <iostream>
@ -54,7 +54,7 @@ RclConfig::RclConfig()
cerr << "No mime conf file" << endl; cerr << "No mime conf file" << endl;
return; return;
} }
mimeconf->list(); // mimeconf->list();
m_ok = true; m_ok = true;
return; return;
} }

4
src/common/textsplit.h
View File

@ -1,13 +1,13 @@
#ifndef _TEXTSPLIT_H_INCLUDED_ #ifndef _TEXTSPLIT_H_INCLUDED_
#define _TEXTSPLIT_H_INCLUDED_ #define _TEXTSPLIT_H_INCLUDED_
/* @(#$Id: textsplit.h,v 1.2 2004-12-17 13:01:01 dockes Exp $ (C) 2004 J.F.Dockes */ /* @(#$Id: textsplit.h,v 1.3 2005-01-24 13:17:58 dockes Exp $ (C) 2004 J.F.Dockes */
#include <string> #include <string>
/** /**
* Split text into words. * Split text into words.
* See comments at top of .cpp for more explanations. * See comments at top of .cpp for more explanations.
* This used a callback function. It could be done with an iterator instead, * This uses a callback function. It could be done with an iterator instead,
* but 'ts much simpler this way... * but 'ts much simpler this way...
*/ */
class TextSplit { class TextSplit {

3
src/common/unacpp.cpp
View File

@ -1,5 +1,5 @@
#ifndef lint #ifndef lint
static char rcsid[] = "@(#$Id: unacpp.cpp,v 1.1 2004-12-17 15:36:13 dockes Exp $ (C) 2004 J.F.Dockes"; static char rcsid[] = "@(#$Id: unacpp.cpp,v 1.2 2005-01-24 13:17:58 dockes Exp $ (C) 2004 J.F.Dockes";
#endif #endif
#ifndef TEST_UNACPP #ifndef TEST_UNACPP
@ -26,6 +26,7 @@ bool unac_cpp(const std::string &in, std::string &out, const char *encoding)
return false; return false;
} }
out.assign(cout, out_len); out.assign(cout, out_len);
free(cout);
return true; return true;
} }

73
src/index/Makefile Normal file
View File

@ -0,0 +1,73 @@
CXXFLAGS = -Wall -g -I. -I../utils -I../common -I/usr/local/include
PROGS = transcode textsplit idxdriver csguess
all: $(PROGS)
IDXDRIVER_OBJS= idxdriver.o pathut.o conftree.o rclconfig.o fstreewalk.o \
mimetype.o rcldb.o readfile.o transcode.o csguess.o \
textsplit.o unac.o unacpp.o
idxdriver : $(IDXDRIVER_OBJS)
$(CXX) $(CXXFLAGS) -o idxdriver $(IDXDRIVER_OBJS) \
-L/usr/local/lib -lxapian -liconv
TEXTSPLIT_OBJS= trtextsplit.o textsplit.o readfile.o
textsplit : $(TEXTSPLIT_OBJS)
$(CXX) $(CXXFLAGS) -o textsplit $(TEXTSPLIT_OBJS)
trtextsplit.o : ../common/textsplit.cpp ../common/textsplit.h
$(CXX) $(CXXFLAGS) -DTEST_TEXTSPLIT -c -o trtextsplit.o \
../common/textsplit.cpp
pathut.o : ../utils/pathut.cpp ../utils/pathut.h
$(CXX) $(CXXFLAGS) -c -o pathut.o ../utils/pathut.cpp
conftree.o : ../utils/conftree.cpp ../utils/conftree.h
$(CXX) $(CXXFLAGS) -c -o conftree.o ../utils/conftree.cpp
rclconfig.o : ../common/rclconfig.cpp ../common/rclconfig.h
$(CXX) $(CXXFLAGS) -c -o rclconfig.o ../common/rclconfig.cpp
fstreewalk.o : ../utils/fstreewalk.cpp ../utils/fstreewalk.h
$(CXX) $(CXXFLAGS) -c -o fstreewalk.o ../utils/fstreewalk.cpp
readfile.o : ../utils/readfile.cpp ../utils/readfile.h
$(CXX) $(CXXFLAGS) -c -o readfile.o ../utils/readfile.cpp
unacpp.o : ../common/unacpp.cpp ../common/unacpp.h
$(CXX) $(CXXFLAGS) -I../unac -c -o unacpp.o ../common/unacpp.cpp
rcldb.o : ../common/rcldb.cpp ../common/rcldb.h
$(CXX) $(CXXFLAGS) -c -o rcldb.o ../common/rcldb.cpp
textsplit.o : ../common/textsplit.cpp ../common/textsplit.h
$(CXX) $(CXXFLAGS) -c -o textsplit.o ../common/textsplit.cpp
CSGUESS_OBJS= trcsguess.o csguess.o readfile.o
csguess : $(CSGUESS_OBJS)
$(CXX) $(CXXFLAGS) -o csguess $(CSGUESS_OBJS) \
-L/usr/local/lib -liconv
trcsguess.o : csguess.cpp csguess.h
$(CXX) $(CXXFLAGS) -DTEST_CSGUESS -c -o trcsguess.o \
csguess.cpp
TRANSCODE_OBJS= trtranscode.o transcode.o readfile.o
transcode : $(TRANSCODE_OBJS)
$(CXX) $(CXXFLAGS) -o transcode $(TRANSCODE_OBJS) \
-L/usr/local/lib -liconv
trtranscode.o : transcode.cpp transcode.h
$(CXX) $(CXXFLAGS) -DTEST_TRANSCODE -c -o trtranscode.o \
transcode.cpp
CFLAGS = -g -I. -I../unac -I/usr/local/include -DUNAC_VERSION=\"1.0.7\"
unac.o : ../unac/unac.c ../unac/unac.h
$(CC) $(CFLAGS) -c -o unac.o ../unac/unac.c
clean:
rm -f *.o $(PROGS)
alldeps:depend
depend:
$(CXX) $(CXXFLAGS) -M *.cpp > alldeps
include alldeps

245
src/qtgui/form1.ui Normal file
View File

@ -0,0 +1,245 @@
<!DOCTYPE UI><UI version="3.3" stdsetdef="1">
<class>Form1</class>
<widget class="QMainWindow">
<property name="name">
<cstring>Form1</cstring>
</property>
<property name="geometry">
<rect>
<x>0</x>
<y>0</y>
<width>774</width>
<height>619</height>
</rect>
</property>
<property name="caption">
<string>Form1</string>
</property>
<widget class="QLayoutWidget">
<property name="name">
<cstring>layout7</cstring>
</property>
<property name="geometry">
<rect>
<x>11</x>
<y>11</y>
<width>752</width>
<height>41</height>
</rect>
</property>
<hbox>
<property name="name">
<cstring>unnamed</cstring>
</property>
<widget class="QLineEdit">
<property name="name">
<cstring>queryText</cstring>
</property>
<property name="frameShape">
<enum>LineEditPanel</enum>
</property>
<property name="frameShadow">
<enum>Sunken</enum>
</property>
</widget>
<widget class="QPushButton">
<property name="name">
<cstring>Search</cstring>
</property>
<property name="text">
<string>pushButton1</string>
</property>
</widget>
<spacer>
<property name="name">
<cstring>spacer1</cstring>
</property>
<property name="orientation">
<enum>Horizontal</enum>
</property>
<property name="sizeType">
<enum>Expanding</enum>
</property>
<property name="sizeHint">
<size>
<width>40</width>
<height>20</height>
</size>
</property>
</spacer>
</hbox>
</widget>
<widget class="QSplitter">
<property name="name">
<cstring>splitter9</cstring>
</property>
<property name="geometry">
<rect>
<x>11</x>
<y>58</y>
<width>752</width>
<height>491</height>
</rect>
</property>
<property name="orientation">
<enum>Horizontal</enum>
</property>
<widget class="QTextEdit">
<property name="name">
<cstring>resTextEdit</cstring>
</property>
</widget>
<widget class="QSplitter">
<property name="name">
<cstring>splitter8</cstring>
</property>
<property name="orientation">
<enum>Vertical</enum>
</property>
<widget class="QTextEdit">
<property name="name">
<cstring>textEdit12</cstring>
</property>
</widget>
<widget class="QTextEdit">
<property name="name">
<cstring>textEdit13</cstring>
</property>
</widget>
</widget>
</widget>
</widget>
<menubar>
<property name="name">
<cstring>MenuBar</cstring>
</property>
<item text="&amp;File" name="fileMenu">
<separator/>
<separator/>
<action name="fileExitAction"/>
</item>
<item text="&amp;Help" name="helpMenu">
<action name="helpContentsAction"/>
<action name="helpIndexAction"/>
<separator/>
<action name="helpAboutAction"/>
</item>
</menubar>
<toolbars>
</toolbars>
<actions>
<action>
<property name="name">
<cstring>fileExitAction</cstring>
</property>
<property name="text">
<string>Exit</string>
</property>
<property name="menuText">
<string>E&amp;xit</string>
</property>
<property name="accel">
<string></string>
</property>
</action>
<action>
<property name="name">
<cstring>helpContentsAction</cstring>
</property>
<property name="text">
<string>Contents</string>
</property>
<property name="menuText">
<string>&amp;Contents...</string>
</property>
<property name="accel">
<string></string>
</property>
</action>
<action>
<property name="name">
<cstring>helpIndexAction</cstring>
</property>
<property name="text">
<string>Index</string>
</property>
<property name="menuText">
<string>&amp;Index...</string>
</property>
<property name="accel">
<string></string>
</property>
</action>
<action>
<property name="name">
<cstring>helpAboutAction</cstring>
</property>
<property name="text">
<string>About</string>
</property>
<property name="menuText">
<string>&amp;About</string>
</property>
<property name="accel">
<string></string>
</property>
</action>
</actions>
<connections>
<connection>
<sender>fileExitAction</sender>
<signal>activated()</signal>
<receiver>Form1</receiver>
<slot>fileExit()</slot>
</connection>
<connection>
<sender>helpIndexAction</sender>
<signal>activated()</signal>
<receiver>Form1</receiver>
<slot>helpIndex()</slot>
</connection>
<connection>
<sender>helpContentsAction</sender>
<signal>activated()</signal>
<receiver>Form1</receiver>
<slot>helpContents()</slot>
</connection>
<connection>
<sender>helpAboutAction</sender>
<signal>activated()</signal>
<receiver>Form1</receiver>
<slot>helpAbout()</slot>
</connection>
<connection>
<sender>resTextEdit</sender>
<signal>clicked(int,int)</signal>
<receiver>Form1</receiver>
<slot>resTextEdit_clicked(int,int)</slot>
</connection>
<connection>
<sender>resTextEdit</sender>
<signal>returnPressed()</signal>
<receiver>Form1</receiver>
<slot>resTextEdit_returnPressed()</slot>
</connection>
<connection>
<sender>fileExitAction</sender>
<signal>activated()</signal>
<receiver>Form1</receiver>
<slot>fileExit()</slot>
</connection>
</connections>
<includes>
<include location="local" impldecl="in implementation">form1.ui.h</include>
</includes>
<slots>
<slot>fileExit()</slot>
<slot>helpIndex()</slot>
<slot>helpContents()</slot>
<slot>helpAbout()</slot>
<slot>resTextEdit_clicked( int par, int car )</slot>
<slot>resTextEdit_returnPressed()</slot>
</slots>
<pixmapinproject/>
<layoutdefaults spacing="6" margin="11"/>
</UI>

60
src/qtgui/form1.ui.h Normal file
View File

@ -0,0 +1,60 @@
/****************************************************************************
** ui.h extension file, included from the uic-generated form implementation.
**
** If you want to add, delete, or rename functions or slots, use
** Qt Designer to update this file, preserving your code.
**
** You should not define a constructor or destructor in this file.
** Instead, write your code in functions called init() and destroy().
** These will automatically be called by the form's constructor and
** destructor.
*****************************************************************************/
void Form1::fileExit()
{
exit(0);
}
void Form1::helpIndex()
{
}
void Form1::helpContents()
{
}
void Form1::helpAbout()
{
}
void Form1::resTextEdit_clicked( int par, int car )
{
fprintf(stderr, "Clicked at paragraph %d, char %d\n", par, car);
}
#include "qfontdialog.h"
#define BS 200000
void Form1::resTextEdit_returnPressed()
{
fprintf(stderr, "ReturnPressed()\n");
resTextEdit->setFont( QFontDialog::getFont( 0, resTextEdit->font() ) );
const char *fname = "utf8.txt";
FILE *fp = fopen(fname, "r");
if (fp) {
char buf[BS];
memset(buf,0, sizeof(buf));
int n = fread(buf, 1, BS-1, fp);
fclose(fp);
QString str = QString::fromUtf8(buf, n);
resTextEdit->setTextFormat(RichText);
resTextEdit->setText(str);
}
}

11
src/qtgui/main.cpp Normal file
View File

@ -0,0 +1,11 @@
#include <qapplication.h>
#include "form1.h"
int main( int argc, char ** argv )
{
QApplication a( argc, argv );
Form1 w;
w.show();
a.connect( &a, SIGNAL( lastWindowClosed() ), &a, SLOT( quit() ) );
return a.exec();
}

38
src/query/Makefile
View File

@ -2,7 +2,7 @@
CXXFLAGS = -Wall -g -I. -I../index -I../utils -I../common -I/usr/local/include CXXFLAGS = -Wall -g -I. -I../index -I../utils -I../common -I/usr/local/include
PROGS = xadump PROGS = qtry qxtry xadump
all: $(PROGS) all: $(PROGS)
XADUMP_OBJS= xadump.o transcode.o XADUMP_OBJS= xadump.o transcode.o
@ -10,8 +10,44 @@ xadump : $(XADUMP_OBJS)
$(CXX) $(CXXFLAGS) -o xadump $(XADUMP_OBJS) \ $(CXX) $(CXXFLAGS) -o xadump $(XADUMP_OBJS) \
-L/usr/local/lib -lxapian -liconv -L/usr/local/lib -lxapian -liconv
QXTRY_OBJS= qxtry.o transcode.o
qxtry : $(QXTRY_OBJS)
$(CXX) $(CXXFLAGS) -o qxtry $(QXTRY_OBJS) \
-L/usr/local/lib -lxapian -liconv
QTRY_OBJS= qtry.o conftree.o rclconfig.o \
rcldb.o transcode.o \
textsplit.o unac.o unacpp.o pathut.o
qtry : $(QTRY_OBJS)
$(CXX) $(CXXFLAGS) -o qtry $(QTRY_OBJS) \
-L/usr/local/lib -lxapian -liconv
transcode.o : ../index/transcode.cpp ../index/transcode.h transcode.o : ../index/transcode.cpp ../index/transcode.h
$(CXX) $(CXXFLAGS) -c -o transcode.o ../index/transcode.cpp $(CXX) $(CXXFLAGS) -c -o transcode.o ../index/transcode.cpp
conftree.o : ../utils/conftree.cpp ../utils/conftree.h
$(CXX) $(CXXFLAGS) -c -o conftree.o ../utils/conftree.cpp
rclconfig.o : ../common/rclconfig.cpp ../common/rclconfig.h
$(CXX) $(CXXFLAGS) -c -o rclconfig.o ../common/rclconfig.cpp
unacpp.o : ../common/unacpp.cpp ../common/unacpp.h
$(CXX) $(CXXFLAGS) -I../unac -c -o unacpp.o ../common/unacpp.cpp
pathut.o : ../utils/pathut.cpp ../utils/pathut.h
$(CXX) $(CXXFLAGS) -c -o pathut.o ../utils/pathut.cpp
rcldb.o : ../common/rcldb.cpp ../common/rcldb.h
$(CXX) $(CXXFLAGS) -c -o rcldb.o ../common/rcldb.cpp
textsplit.o : ../common/textsplit.cpp ../common/textsplit.h
$(CXX) $(CXXFLAGS) -c -o textsplit.o ../common/textsplit.cpp
CFLAGS = -g -I. -I../unac -I/usr/local/include -DUNAC_VERSION=\"1.0.7\"
unac.o : ../unac/unac.c ../unac/unac.h
$(CC) $(CFLAGS) -c -o unac.o ../unac/unac.c
clean: clean:
rm -f *.o $(PROGS) rm -f *.o $(PROGS)
alldeps:depend
depend:
$(CXX) $(CXXFLAGS) -M *.cpp > alldeps
include alldeps

105
src/query/qtry.cpp Normal file
View File

@ -0,0 +1,105 @@
#ifndef lint
static char rcsid[] = "@(#$Id: qtry.cpp,v 1.1 2005-01-24 13:17:58 dockes Exp $ (C) 2004 J.F.Dockes";
#endif
// Tests with the query interface
#include <strings.h>
#include <iostream>
#include <string>
#include <vector>
#include "conftree.h"
#include "rclconfig.h"
#include "rcldb.h"
#include "transcode.h"
using namespace std;
#include "rcldb.h"
static string thisprog;
static string usage =
" -d <dbdir> -e <interface encoding> term [term] ..."
" \n\n"
;
static void
Usage(void)
{
cerr << thisprog << ": usage:\n" << usage;
exit(1);
}
static int op_flags;
#define OPT_e 0x2
int main(int argc, char **argv)
{
string encoding = "ISO8859-1";
thisprog = argv[0];
argc--; argv++;
while (argc > 0 && **argv == '-') {
(*argv)++;
if (!(**argv))
/* Cas du "adb - core" */
Usage();
while (**argv)
switch (*(*argv)++) {
case 'e': op_flags |= OPT_e; if (argc < 2) Usage();
encoding = *(++argv);
argc--;
goto b1;
default: Usage(); break;
}
b1: argc--; argv++;
}
if (argc < 1)
Usage();
RclConfig *config = new RclConfig;
if (!config->ok())
cerr << "Config could not be built" << endl;
string dbdir;
if (config->getConfParam(string("dbdir"), dbdir) == 0) {
cerr << "No database directory in configuration" << endl;
exit(1);
}
Rcl::Db *db = new Rcl::Db;
if (!db->open(dbdir, Rcl::Db::DbRO)) {
fprintf(stderr, "Could not open database\n");
exit(1);
}
// TOBEDONE: query syntax. Yes it's mighty stupid to cat terms.
string query;
while (argc--)
query += string(*argv++) + " " ;
db->setQuery(query);
int i = 0;
Rcl::Doc doc;
while (db->getDoc(i++, doc)) {
cout << "Url: " << doc.url << endl;
cout << "Mimetype: " << doc.mimetype << endl;
cout << "Mtime: " << doc.mtime << endl;
cout << "Origcharset: " << doc.origcharset << endl;
cout << "Title: " << doc.title << endl;
cout << "Text: " << doc.text << endl;
cout << "Keywords: " << doc.keywords << endl;
cout << "Abstract: " << doc.abstract << endl;
cout << endl;
doc.erase();
}
delete db;
cerr << "Exiting" << endl;
exit(0);
}

109
src/query/qxtry.cpp Normal file
View File

@ -0,0 +1,109 @@
#ifndef lint
static char rcsid[] = "@(#$Id: qxtry.cpp,v 1.1 2005-01-24 13:17:58 dockes Exp $ (C) 2004 J.F.Dockes";
#endif
// Tests with direct xapian questions
#include <strings.h>
#include <iostream>
#include <string>
#include <vector>
#include "transcode.h"
using namespace std;
#include "xapian.h"
static string thisprog;
static string usage =
" -d <dbdir> -e <interface encoding> term [term] ..."
" \n\n"
;
static void
Usage(void)
{
cerr << thisprog << ": usage:\n" << usage;
exit(1);
}
static int op_flags;
#define OPT_d 0x1
#define OPT_e 0x2
Xapian::Database db;
int main(int argc, char **argv)
{
string dbdir = "/home/dockes/tmp/xapiandb";
string encoding = "ISO8859-1";
thisprog = argv[0];
argc--; argv++;
while (argc > 0 && **argv == '-') {
(*argv)++;
if (!(**argv))
/* Cas du "adb - core" */
Usage();
while (**argv)
switch (*(*argv)++) {
case 'd': op_flags |= OPT_e; if (argc < 2) Usage();
dbdir = *(++argv);
argc--;
goto b1;
case 'e': op_flags |= OPT_d; if (argc < 2) Usage();
encoding = *(++argv);
argc--;
goto b1;
default: Usage(); break;
}
b1: argc--; argv++;
}
if (argc < 1)
Usage();
vector<string> qterms;
while (argc--) {
qterms.push_back(*argv++);
}
try {
db = Xapian::Auto::open(dbdir, Xapian::DB_OPEN);
cout << "DB: ndocs " << db.get_doccount() << " lastdocid " <<
db.get_lastdocid() << " avglength " << db.get_avlength() << endl;
Xapian::Enquire enquire(db);
Xapian::Query query(Xapian::Query::OP_OR, qterms.begin(),
qterms.end());
cout << "Performing query `" <<
query.get_description() << "'" << endl;
enquire.set_query(query);
Xapian::MSet matches = enquire.get_mset(0, 10);
cout << "Estimated results: " << matches.get_matches_lower_bound() <<
endl;
Xapian::MSetIterator i;
for (i = matches.begin(); i != matches.end(); ++i) {
cout << "Document ID " << *i << "\t";
cout << i.get_percent() << "% ";
Xapian::Document doc = i.get_document();
cout << "[" << doc.get_data() << "]" << endl;
}
} catch (const Xapian::Error &e) {
cout << "Exception: " << e.get_msg() << endl;
} catch (const string &s) {
cout << "Exception: " << s << endl;
} catch (const char *s) {
cout << "Exception: " << s << endl;
} catch (...) {
cout << "Caught unknown exception" << endl;
}
exit(0);
}

100
src/rcldb/rcldb.cpp
View File

@ -1,5 +1,5 @@
#ifndef lint #ifndef lint
static char rcsid[] = "@(#$Id: rcldb.cpp,v 1.5 2004-12-17 15:50:48 dockes Exp $ (C) 2004 J.F.Dockes"; static char rcsid[] = "@(#$Id: rcldb.cpp,v 1.6 2005-01-24 13:17:58 dockes Exp $ (C) 2004 J.F.Dockes";
#endif #endif
#include <sys/stat.h> #include <sys/stat.h>
@ -14,18 +14,23 @@ using namespace std;
#include "textsplit.h" #include "textsplit.h"
#include "transcode.h" #include "transcode.h"
#include "unacpp.h" #include "unacpp.h"
#include "conftree.h"
#include "xapian.h" #include "xapian.h"
// Data for a xapian database // Data for a xapian database. There could actually be 2 different ones for
// indexing or query as there is not much in common.
class Native { class Native {
public: public:
bool isopen; bool isopen;
bool iswritable; bool iswritable;
class Xapian::Database db; // Indexing
class Xapian::WritableDatabase wdb; Xapian::WritableDatabase wdb;
vector<bool> updated; vector<bool> updated;
// Querying
Xapian::Database db;
Xapian::Query query;
Native() : isopen(false), iswritable(false) {} Native() : isopen(false), iswritable(false) {}
}; };
@ -37,6 +42,7 @@ Rcl::Db::Db()
Rcl::Db::~Db() Rcl::Db::~Db()
{ {
cerr << "Rcl::Db::~Db" << endl;
if (pdata == 0) if (pdata == 0)
return; return;
Native *ndb = (Native *)pdata; Native *ndb = (Native *)pdata;
@ -45,6 +51,7 @@ Rcl::Db::~Db()
try { try {
// There is nothing to do for an ro db. // There is nothing to do for an ro db.
if (ndb->isopen == false || ndb->iswritable == false) { if (ndb->isopen == false || ndb->iswritable == false) {
cerr << "Deleting native database" << endl;
delete ndb; delete ndb;
return; return;
} }
@ -82,8 +89,8 @@ bool Rcl::Db::open(const string& dir, OpenMode mode)
case DbRO: case DbRO:
default: default:
ndb->iswritable = false; ndb->iswritable = false;
cerr << "Not ready to open RO yet" << endl; ndb->db = Xapian::Auto::open(dir, Xapian::DB_OPEN);
exit(1); break;
} }
ndb->isopen = true; ndb->isopen = true;
return true; return true;
@ -142,7 +149,8 @@ class wsData {
{} {}
}; };
bool splitCb(void *cdata, const std::string &term, int pos) // Callback for the document to word splitting class during indexation
static bool splitCb(void *cdata, const std::string &term, int pos)
{ {
wsData *data = (wsData*)cdata; wsData *data = (wsData*)cdata;
@ -172,7 +180,7 @@ bool dumb_string(const string &in, string &out)
out.erase(); out.erase();
if (!unac_cpp(in, inter)) if (!unac_cpp(in, inter))
return false; return false;
out.resize(inter.length()); out.reserve(inter.length());
for (unsigned int i = 0; i < inter.length(); i++) { for (unsigned int i = 0; i < inter.length(); i++) {
if (inter[i] >= 'A' && inter[i] <= 'Z') if (inter[i] >= 'A' && inter[i] <= 'Z')
out += inter[i] + 'a' - 'A'; out += inter[i] + 'a' - 'A';
@ -239,8 +247,10 @@ bool Rcl::Db::add(const string &fn, const Rcl::Doc &doc)
// If this document has already been indexed, update the existing // If this document has already been indexed, update the existing
// entry. // entry.
try { try {
Xapian::docid did = ndb->wdb.replace_document(pathterm, #if 0
newdocument); Xapian::docid did =
#endif
ndb->wdb.replace_document(pathterm, newdocument);
#if 0 #if 0
if (did < updated.size()) { if (did < updated.size()) {
updated[did] = true; updated[did] = true;
@ -297,3 +307,73 @@ bool Rcl::Db::needUpdate(const string &filename, const struct stat *stp)
return true; return true;
} }
#include <vector>
class wsQData {
public:
vector<string> terms;
};
// Callback for the document to word splitting class during indexation
static bool splitQCb(void *cdata, const std::string &term, int )
{
wsQData *data = (wsQData*)cdata;
cerr << "splitQCb: term '" << term << "'" << endl;
cerr << "splitQCb: term length: " << term.length() << endl;
//string printable;
//transcode(term, printable, "UTF-8", "ISO8859-1");
//cerr << "Adding " << printable << endl;
data->terms.push_back(term);
return true;
}
bool Rcl::Db::setQuery(const std::string &querystring)
{
wsQData splitData;
TextSplit splitter(splitQCb, &splitData);
string noacc;
if (!dumb_string(querystring, noacc)) {
return false;
}
// noacc = querystring;
splitter.text_to_words(noacc);
Native *ndb = (Native *)pdata;
// splitData.terms.resize(0);
// splitData.terms.push_back(string("le"));
ndb->query = Xapian::Query(Xapian::Query::OP_OR, splitData.terms.begin(),
splitData.terms.end());
return true;
}
bool Rcl::Db::getDoc(int i, Doc &doc)
{
// cerr << "Rcl::Db::getDoc: " << i << endl;
Native *ndb = (Native *)pdata;
Xapian::Enquire enquire(ndb->db);
enquire.set_query(ndb->query);
Xapian::MSet matches = enquire.get_mset(i, 1);
// cerr << "Query `" << ndb->query.get_description() << "'" <<
// "Estimated results: " << matches.get_matches_lower_bound() << endl;
if (matches.empty())
return false;
Xapian::Document xdoc = matches.begin().get_document();
// Parse xapian document's data and populate doc fields
string data = xdoc.get_data();
ConfSimple parms(&data);
parms.get(string("mtype"), doc.mimetype);
parms.get(string("mtime"), doc.mtime);
parms.get(string("url"), doc.url);
return true;
}

39
src/rcldb/rcldb.h
View File

@ -1,9 +1,22 @@
#ifndef _DB_H_INCLUDED_ #ifndef _DB_H_INCLUDED_
#define _DB_H_INCLUDED_ #define _DB_H_INCLUDED_
/* @(#$Id: rcldb.h,v 1.3 2004-12-17 13:01:01 dockes Exp $ (C) 2004 J.F.Dockes */ /* @(#$Id: rcldb.h,v 1.4 2005-01-24 13:17:58 dockes Exp $ (C) 2004 J.F.Dockes */
#include <string> #include <string>
// rcldb defines an interface for a 'real' text database. The current
// implementation uses xapian only, and xapian-related code is in rcldb.cpp
// If support was added for other backend, the xapian code would be moved in
// rclxapian.cpp, another file would be created for the new backend, and the
// configuration/compile/link code would be adjusted to allow choosing. There
// is no plan for supporting multiple different backends.
//
// In no case does this try to implement a useful virtualized text-db interface
// The main goal is simplicity and good matching to usage inside the recoll
// user interface. In other words, this is not exhaustive or well-designed or
// reusable.
struct stat; struct stat;
namespace Rcl { namespace Rcl {
@ -13,6 +26,7 @@ namespace Rcl {
*/ */
class Doc { class Doc {
public: public:
string url;
string mimetype; string mimetype;
string mtime; // Modification time as decimal ascii string mtime; // Modification time as decimal ascii
string origcharset; string origcharset;
@ -20,6 +34,16 @@ class Doc {
string text; string text;
string keywords; string keywords;
string abstract; string abstract;
void erase() {
url.erase();
mimetype.erase();
mtime.erase();
origcharset.erase();
title.erase();
text.erase();
keywords.erase();
abstract.erase();
}
}; };
/** /**
@ -27,14 +51,27 @@ class Doc {
*/ */
class Db { class Db {
void *pdata; void *pdata;
Doc curdoc;
public: public:
Db(); Db();
~Db(); ~Db();
enum OpenMode {DbRO, DbUpd, DbTrunc}; enum OpenMode {DbRO, DbUpd, DbTrunc};
bool open(const std::string &dbdir, OpenMode mode); bool open(const std::string &dbdir, OpenMode mode);
bool close(); bool close();
// Update-related functions
bool add(const string &filename, const Doc &doc); bool add(const string &filename, const Doc &doc);
bool needUpdate(const string &filename, const struct stat *stp); bool needUpdate(const string &filename, const struct stat *stp);
// Query-related functions
// Parse query string and initialize query
bool setQuery(const std::string &q);
// Get document at rank i. This is probably vastly inferior to the type
// of interface in Xapian, but we have to start with something simple
// to experiment with the GUI
bool getDoc(int i, Doc &doc);
}; };

23
src/utils/Makefile Normal file
View File

@ -0,0 +1,23 @@
CXXFLAGS = -I.
PROGS = trfstreewalk pathut execmd
all: $(PROGS)
FSTREEWALK_OBJS= trfstreewalk.o fstreewalk.o pathut.o
trfstreewalk : $(FSTREEWALK_OBJS)
$(CXX) -o trfstreewalk $(FSTREEWALK_OBJS)
trfstreewalk.o : fstreewalk.cpp fstreewalk.h
$(CXX) -o trfstreewalk.o -c $(CXXFLAGS) \
-DTEST_FSTREEWALK fstreewalk.cpp
PATHUT_OBJS= trpathut.o pathut.o
trpathut : $(PATHUT_OBJS)
$(CXX) -o trpathut $(PATHUT_OBJS)
trpathut.o : pathut.cpp pathut.h
$(CXX) -o trpathut.o -c $(CXXFLAGS) \
-DTEST_PATHUT pathut.cpp
execmd: pathut.o
$(CXX) -o execmd $(CXXFLAGS) execmd.cpp pathut.o
clean:
rm -f *.o $(PROGS)