aspell checkpoint

src/rcldb/rcldb.cpp

@@ -1,5 +1,5 @@
 #ifndef lint
-static char rcsid[] = "@(#$Id: rcldb.cpp,v 1.79 2006-09-29 08:26:02 dockes Exp $ (C) 2004 J.F.Dockes";
+static char rcsid[] = "@(#$Id: rcldb.cpp,v 1.80 2006-10-09 16:37:08 dockes Exp $ (C) 2004 J.F.Dockes";
 #endif
 /*
  *   This program is free software; you can redistribute it and/or modify
@@ -890,7 +890,7 @@ static void stringToXapianQueries(const string &iq,
 		dumb_string(term, term1);
 		// Possibly perform stem compression/expansion
 		if (!nostemexp && (opts & Db::QO_STEM)) {
-		    exp = m_ndb->stemExpand(stemlang,term1);
+		    exp = m_ndb->stemExpand(stemlang, term1);
 		} else {
 		    exp.push_back(term1);
 		}
@@ -1117,6 +1117,60 @@ list<string> Db::completions(const string &root, const string &lang, int max)
     return res;
 }
 
+/** Term list walking. */
+class TermIter {
+public:
+    Xapian::TermIterator it;
+    Xapian::Database db;
+};
+TermIter *Db::termWalkOpen()
+{
+    if (!m_ndb || !m_ndb->m_isopen)
+	return 0;
+    TermIter *tit = new TermIter;
+    if (tit) {
+	tit->db = m_ndb->m_iswritable ? m_ndb->wdb: m_ndb->db;
+	tit->it = tit->db.allterms_begin();
+    }
+    return tit;
+}
+bool Db::termWalkNext(TermIter *tit, string &term)
+{
+    
+    if (tit && tit->it != tit->db.allterms_end()) {
+	term = *(tit->it)++;
+	return true;
+    }
+    return false;
+}
+void Db::termWalkClose(TermIter *tit)
+{
+    delete tit;
+}
+
+
+bool Db::termExists(const string& word)
+{
+    if (!m_ndb || !m_ndb->m_isopen)
+	return 0;
+    Xapian::Database db = m_ndb->m_iswritable ? m_ndb->wdb: m_ndb->db;
+    if (!db.term_exists(word))
+	return false;
+    return true;
+}
+
+bool Db::stemDiffers(const string& lang, const string& word, 
+		     const string& base)
+{
+    Xapian::Stem stemmer(lang);
+    if (!stemmer.stem_word(word).compare(stemmer.stem_word(base))) {
+	LOGDEB2(("Rcl::Db::stemDiffers: same for %s and %s\n", 
+		word.c_str(), base.c_str()));
+	return false;
+    }
+    return true;
+}
+
 bool Db::getQueryTerms(list<string>& terms)
 {
     if (!m_ndb)

src/rcldb/rcldb.h

@@ -16,7 +16,7 @@
  */
 #ifndef _DB_H_INCLUDED_
 #define _DB_H_INCLUDED_
-/* @(#$Id: rcldb.h,v 1.36 2006-09-13 13:53:35 dockes Exp $  (C) 2004 J.F.Dockes */
+/* @(#$Id: rcldb.h,v 1.37 2006-10-09 16:37:08 dockes Exp $  (C) 2004 J.F.Dockes */
 
 #include <string>
 #include <list>
@@ -105,6 +105,7 @@ class Doc {
 
 class AdvSearchData;
 class Native;
+class TermIter;
  
 /**
  * Wrapper class for the native database.
@@ -170,9 +171,22 @@ class Db {
     /** Get a list of existing stemming databases */
     std::list<std::string> getStemLangs();
 
+    /** Retrieve main database directory */
     string getDbDir();
+
+    /** Set parameters for synthetic abstract generation */
     void setAbstractParams(int idxTrunc, int synthLen, int syntCtxLen);
 
+    /** Whole term list walking. */
+    TermIter *termWalkOpen();
+    bool termWalkNext(TermIter *, string &term);
+    void termWalkClose(TermIter *);
+    /** Test term existence */
+    bool termExists(const string& term);
+    /** Test if terms stem to different roots. */
+    bool stemDiffers(const string& lang, const string& term, 
+		     const string& base);
+
 private:
 
     string m_filterTopDir; // Current query filter on subtree top directory 

src/rcldb/stemdb.cpp

@@ -1,5 +1,5 @@
 #ifndef lint
-static char rcsid[] = "@(#$Id: stemdb.cpp,v 1.4 2006-09-20 06:21:43 dockes Exp $ (C) 2005 J.F.Dockes";
+static char rcsid[] = "@(#$Id: stemdb.cpp,v 1.5 2006-10-09 16:37:08 dockes Exp $ (C) 2005 J.F.Dockes";
 #endif
 
 /**
@@ -254,7 +254,8 @@ list<string> stemExpand(const string& dbdir, const string& lang,
 	LOGDEB(("stemExpand: %s ->  %s\n", stem.c_str(),
 		stringlistdisp(explist).c_str()));
     } catch (...) {
-	LOGERR(("stemExpand: error accessing stem db\n"));
+	LOGERR(("stemExpand: error accessing stem db. dbdir [%s] lang [%s]\n",
+		dbdir.c_str(), lang.c_str()));
 	explist.push_back(term);
 	return explist;
     }