Allow stem expansion for several (all) stemming languages at a time

src/qtgui/guiutils.h

@@ -17,7 +17,7 @@
 #ifndef _GUIUTILS_H_INCLUDED_
 #define _GUIUTILS_H_INCLUDED_
 /* 
- * @(#$Id: guiutils.h,v 1.23 2007-07-12 10:13:37 dockes Exp $  (C) 2005 Jean-Francois Dockes 
+ * @(#$Id: guiutils.h,v 1.24 2007-08-01 10:04:53 dockes Exp $  (C) 2005 Jean-Francois Dockes 
  *                         jean-francois.dockes@wanadoo.fr
  *
  *   This program is free software; you can redistribute it and/or modify
@@ -144,4 +144,6 @@ extern PrefsPack prefs;
 /** Read write settings from disk file */
 extern void rwSettings(bool dowrite);
 
+extern QString g_stringAllStem, g_stringNoStem;
+
 #endif /* _GUIUTILS_H_INCLUDED_ */

src/qtgui/rclmain_w.cpp

@@ -1,5 +1,5 @@
 #ifndef lint
-static char rcsid[] = "@(#$Id: rclmain_w.cpp,v 1.37 2007-08-01 07:55:03 dockes Exp $ (C) 2005 J.F.Dockes";
+static char rcsid[] = "@(#$Id: rclmain_w.cpp,v 1.38 2007-08-01 10:04:53 dockes Exp $ (C) 2005 J.F.Dockes";
 #endif
 /*
  *   This program is free software; you can redistribute it and/or modify
@@ -76,6 +76,7 @@ using std::pair;
 #include "moc_rclmain_w.cpp"
 
 extern "C" int XFlush(void *);
+QString g_stringAllStem, g_stringNoStem;
 
 // Taken from qt designer. Don't know why it's needed.
 #if (QT_VERSION < 0x040000)
@@ -108,10 +109,13 @@ void RclMain::init()
     }
 
     // Stemming language menu
-    m_idNoStem = preferencesMenu->insertItem(tr("(no stemming)"));
+    g_stringNoStem = tr("(no stemming)");
-    m_stemLangToId[tr("(no stemming)")] = m_idNoStem;
+    g_stringAllStem = tr("(all languages)");
+    m_idNoStem = preferencesMenu->insertItem(g_stringNoStem);
+    m_stemLangToId[g_stringNoStem] = m_idNoStem;
+    m_idAllStem = preferencesMenu->insertItem(g_stringAllStem);
+    m_stemLangToId[g_stringAllStem] = m_idAllStem;
 
-    LOGDEB(("idNoStem: %d\n", m_idNoStem));
     // Can't get the stemming languages from the db at this stage as
     // db not open yet (the case where it does not even exist makes
     // things complicated). So get the languages from the config
@@ -124,7 +128,8 @@ void RclMain::init()
 	QMessageBox::warning(0, "Recoll", 
 			     tr("error retrieving stemming languages"));
     }
-    int curid = m_idNoStem, id; // Menu ids are negative integers
+    int curid = prefs.queryStemLang == "ALL" ? m_idAllStem : m_idNoStem;
+    int id; 
     for (list<string>::const_iterator it = langs.begin(); 
 	 it != langs.end(); it++) {
 	QString qlang = QString::fromAscii(it->c_str(), it->length());
@@ -248,6 +253,8 @@ void RclMain::setStemLang(int id)
     QString lang;
     if (id == m_idNoStem) {
 	lang = "";
+    } else if (id == m_idAllStem) {
+	lang = "ALL";
     } else {
 	lang = preferencesMenu->text(id);
     }
@@ -265,6 +272,8 @@ void RclMain::setStemLang(const QString& lang)
     int id;
     if (lang == "") {
 	id = m_idNoStem;
+    } else if (lang == "ALL") {
+	id = m_idAllStem;
     } else {
 	map<QString, int>::iterator it = m_stemLangToId.find(lang);
 	if (it == m_stemLangToId.end()) 
@@ -399,7 +408,13 @@ void RclMain::startSearch(RefCntr<Rcl::SearchData> sdata)
     if (!prefs.queryStemLang.length() == 0)
 	qopts |= Rcl::Db::QO_STEM;
     QApplication::setOverrideCursor(QCursor(Qt::WaitCursor));
-    if (!rcldb->setQuery(sdata, qopts, prefs.queryStemLang.ascii())) {
+
+    string stemLang = (const char *)prefs.queryStemLang.ascii();
+    if (stemLang == "ALL") {
+	rclconfig->getConfParam("indexstemminglanguages", stemLang);
+    }
+
+    if (!rcldb->setQuery(sdata, qopts, stemLang)) {
 	QMessageBox::warning(0, "Recoll", tr("Cant start query: ") +
 			     QString::fromAscii(rcldb->getReason().c_str()));
 	return;

src/qtgui/rclmain_w.h

@@ -126,6 +126,7 @@ private:
     int                      m_searchId; 
     map<QString, int>        m_stemLangToId;
     int                      m_idNoStem;
+    int                      m_idAllStem;
 
     virtual void init();
     virtual void previewPrevOrNextInTab(Preview *, int sid, int docnum, 

src/qtgui/ssearch_w.cpp

@@ -1,5 +1,5 @@
 #ifndef lint
-static char rcsid[] = "@(#$Id: ssearch_w.cpp,v 1.21 2007-06-12 10:33:48 dockes Exp $ (C) 2006 J.F.Dockes";
+static char rcsid[] = "@(#$Id: ssearch_w.cpp,v 1.22 2007-08-01 10:04:53 dockes Exp $ (C) 2006 J.F.Dockes";
 #endif
 /*
  *   This program is free software; you can redistribute it and/or modify
@@ -215,8 +215,12 @@ void SSearch::completion()
     // Query database
     const int max = 100;
     list<Rcl::TermMatchEntry> strs;
-    if (!rcldb->termMatch(Rcl::Db::ET_WILD, prefs.queryStemLang.ascii(),
+    string stemLang = (const char *)prefs.queryStemLang.ascii();
-			  s, strs, max) || strs.size() == 0) {
+    if (stemLang == "ALL") {
+	rclconfig->getConfParam("indexstemminglanguages", stemLang);
+    }
+    if (!rcldb->termMatch(Rcl::Db::ET_WILD, stemLang, s, strs, max) || 
+	strs.size() == 0) {
 	QApplication::beep();
 	return;
     }

src/qtgui/uiprefs_w.cpp

@@ -1,5 +1,5 @@
 #ifndef lint
-static char rcsid[] = "@(#$Id: uiprefs_w.cpp,v 1.20 2007-08-01 07:55:03 dockes Exp $ (C) 2005 J.F.Dockes";
+static char rcsid[] = "@(#$Id: uiprefs_w.cpp,v 1.21 2007-08-01 10:04:53 dockes Exp $ (C) 2005 J.F.Dockes";
 #endif
 /*
  *   This program is free software; you can redistribute it and/or modify
@@ -121,25 +121,22 @@ void UIPrefsDialog::setFromPrefs()
 
     // Stemming language combobox
     stemLangCMB->clear();
-    stemLangCMB->insertItem(tr("(no stemming)"));
+    stemLangCMB->insertItem(g_stringNoStem);
+    stemLangCMB->insertItem(g_stringAllStem);
     list<string> langs;
     if (!getStemLangs(langs)) {
 	QMessageBox::warning(0, "Recoll", 
 			     tr("error retrieving stemming languages"));
     }
-    int i = 0, cur = -1;
+    int cur = prefs.queryStemLang == ""  ? 0 : 1;
     for (list<string>::const_iterator it = langs.begin(); 
 	 it != langs.end(); it++) {
 	stemLangCMB->
 	    insertItem(QString::fromAscii(it->c_str(), it->length()));
-	i++;
+	if (cur == 0 && !strcmp(prefs.queryStemLang.ascii(), it->c_str())) {
-	if (cur == -1) {
+	    cur = stemLangCMB->count();
-	    if (!strcmp(prefs.queryStemLang.ascii(), it->c_str()))
-		cur = i;
 	}
     }
-    if (cur < 0)
-	cur = 0;
     stemLangCMB->setCurrentItem(cur);
 
     autoPhraseCB->setChecked(prefs.ssearchAutoPhrase);
@@ -191,6 +188,8 @@ void UIPrefsDialog::accept()
 
     if (stemLangCMB->currentItem() == 0) {
 	prefs.queryStemLang = "";
+    } else if (stemLangCMB->currentItem() == 1) {
+	prefs.queryStemLang = "ALL";
     } else {
 	prefs.queryStemLang = stemLangCMB->currentText();
     }
@@ -238,6 +237,8 @@ void UIPrefsDialog::setStemLang(const QString& lang)
     int cur = 0;
     if (lang == "") {
 	cur = 0;
+    } else if (lang == "ALL") {
+	cur = 1;
     } else {
 	for (int i = 1; i < stemLangCMB->count(); i++) {
 	    if (lang == stemLangCMB->text(i)) {

src/qtgui/uiprefs_w.h

@@ -1,4 +1,4 @@
-/* @(#$Id: uiprefs_w.h,v 1.9 2007-08-01 07:55:03 dockes Exp $  (C) 2006 J.F.Dockes */
+/* @(#$Id: uiprefs_w.h,v 1.10 2007-08-01 10:04:53 dockes Exp $  (C) 2006 J.F.Dockes */
 /*
  *   This program is free software; you can redistribute it and/or modify
  *   it under the terms of the GNU General Public License as published by
@@ -17,7 +17,7 @@
  */
 #ifndef _UIPREFS_W_H_INCLUDED_
 #define _UIPREFS_W_H_INCLUDED_
-/* @(#$Id: uiprefs_w.h,v 1.9 2007-08-01 07:55:03 dockes Exp $  (C) 2005 J.F.Dockes */
+/* @(#$Id: uiprefs_w.h,v 1.10 2007-08-01 10:04:53 dockes Exp $  (C) 2005 J.F.Dockes */
 #include <qvariant.h>
 #include <qdialog.h>
 
@@ -89,5 +89,4 @@ private:
 
 };
 
-
 #endif /* _UIPREFS_W_H_INCLUDED_ */

src/rcldb/stemdb.cpp

@@ -1,5 +1,5 @@
 #ifndef lint
-static char rcsid[] = "@(#$Id: stemdb.cpp,v 1.9 2007-05-24 09:35:02 dockes Exp $ (C) 2005 J.F.Dockes";
+static char rcsid[] = "@(#$Id: stemdb.cpp,v 1.10 2007-08-01 10:04:53 dockes Exp $ (C) 2005 J.F.Dockes";
 #endif
 
 /**
@@ -218,12 +218,13 @@ static string stringlistdisp(const list<string>& sl)
 }
 
 /**
- * Expand term to list of all terms which stem to the same term.
+ * Expand term to list of all terms which stem to the same term, for one
+ * expansion language
  */
-bool stemExpand(const std::string& dbdir, 
+static bool stemExpandOne(const std::string& dbdir, 
-		const std::string& lang,
+			  const std::string& lang,
-		const std::string& term,
+			  const std::string& term,
-		list<string>& result)
+			  list<string>& result)
 {
     try {
 	Xapian::Stem stemmer(lang);
@@ -282,5 +283,29 @@ bool stemExpand(const std::string& dbdir,
     return true;
 }
     
+/**
+ * Expand term to list of all terms which stem to the same term, add the
+ * expansion sets for possibly multiple expansion languages
+ */
+bool stemExpand(const std::string& dbdir, 
+		const std::string& langs,
+		const std::string& term,
+		list<string>& result)
+{
+
+    list<string> llangs;
+    stringToStrings(langs, llangs);
+    for (list<string>::const_iterator it = llangs.begin();
+	 it != llangs.end(); it++) {
+	list<string> oneexp;
+	stemExpandOne(dbdir, *it, term, oneexp);
+	result.insert(result.end(), oneexp.begin(), oneexp.end());
+    }
+    result.sort();
+    result.unique();
+    return true;
+}
+
+
 }
 }

src/rcldb/stemdb.h

@@ -1,6 +1,6 @@
 #ifndef _STEMDB_H_INCLUDED_
 #define _STEMDB_H_INCLUDED_
-/* @(#$Id: stemdb.h,v 1.3 2006-12-19 12:11:21 dockes Exp $  (C) 2004 J.F.Dockes */
+/* @(#$Id: stemdb.h,v 1.4 2007-08-01 10:04:53 dockes Exp $  (C) 2004 J.F.Dockes */
 /// Stem database code
 /// 
 /// Stem databases list stems and the set of index terms they expand to. They 
@@ -30,7 +30,7 @@ extern bool createDb(Xapian::Database& xdb,
 		     const std::string& dbdir, const std::string& lang);
 /// Expand term to stem siblings
 extern bool stemExpand(const std::string& dbdir, 
-		       const std::string& lang,
+		       const std::string& langs,
 		       const std::string& term,
 		       list<string>& result);
 #ifndef NO_NAMESPACES