From b955f42655311dfdf43d4f536011b0c19653c0c6 Mon Sep 17 00:00:00 2001
From: dockes <none@none>
Date: Fri, 7 Apr 2006 13:10:22 +0000
Subject: [PATCH] check for and forbid pure negative query

---
 src/qtgui/advsearch.ui.h | 13 +++++++++
 src/rcldb/rcldb.cpp      | 63 ++++++++++++++++++++++++----------------
 2 files changed, 51 insertions(+), 25 deletions(-)

diff --git a/src/qtgui/advsearch.ui.h b/src/qtgui/advsearch.ui.h
index 7561ccc8..4081373e 100644
--- a/src/qtgui/advsearch.ui.h
+++ b/src/qtgui/advsearch.ui.h
@@ -27,6 +27,7 @@
 *****************************************************************************/
 
 #include <qfiledialog.h>
+#include <qmessagebox.h>
 
 #include <list>
 #include <string>
@@ -127,11 +128,23 @@ void advsearch::restrictFtCB_toggled(bool on)
 void advsearch::searchPB_clicked()
 {
     Rcl::AdvSearchData mydata;
+
     mydata.allwords = string((const char*)(andWordsLE->text().utf8()));
     mydata.phrase  = string((const char*)(phraseLE->text().utf8()));
     mydata.orwords = string((const char*)(orWordsLE->text().utf8()));
     mydata.nowords = string((const char*)(noWordsLE->text().utf8()));
     mydata.filename = string((const char*)(fileNameLE->text().utf8()));
+
+    if (mydata.allwords.empty() && mydata.phrase.empty() && 
+	mydata.orwords.empty() && mydata.filename.empty()) {
+	if (mydata.nowords.empty())
+	    return;
+	QMessageBox::warning(0, "Recoll",
+			     tr("Cannot execute pure negative query. "
+				"Please enter common terms in the 'any words' field")); 
+	return;
+    }
+
     if (restrictFtCB->isOn() && noFiltypsLB->count() > 0) {
 	for (unsigned int i = 0; i < yesFiltypsLB->count(); i++) {
 	    QCString ctext = yesFiltypsLB->item(i)->text().utf8();
diff --git a/src/rcldb/rcldb.cpp b/src/rcldb/rcldb.cpp
index a213939f..6664d566 100644
--- a/src/rcldb/rcldb.cpp
+++ b/src/rcldb/rcldb.cpp
@@ -1,5 +1,5 @@
 #ifndef lint
-static char rcsid[] = "@(#$Id: rcldb.cpp,v 1.62 2006-04-05 13:39:07 dockes Exp $ (C) 2004 J.F.Dockes";
+static char rcsid[] = "@(#$Id: rcldb.cpp,v 1.63 2006-04-07 13:10:22 dockes Exp $ (C) 2004 J.F.Dockes";
 #endif
 /*
  *   This program is free software; you can redistribute it and/or modify
@@ -545,13 +545,6 @@ bool Db::add(const string &fn, const Doc &idoc,
     // Mime type
     newdocument.add_term("T" + doc.mimetype);
 
-    // Path name term. This is used for existence/uptodate checks
-    string hash;
-    pathHash(fn, hash, PATHHASHLEN);
-    LOGDEB2(("Db::add: pathhash [%s]\n", hash.c_str()));
-    string pathterm  = "P" + hash;
-    newdocument.add_term(pathterm);
-
     // Simple file name. This is used for file name searches only. We index
     // it with a term prefix. utf8fn used to be the full path, but it's now
     // the simple file name.
@@ -560,10 +553,23 @@ bool Db::add(const string &fn, const Doc &idoc,
 	newdocument.add_term(noacc);
     }
 
-    // Internal path: with path, makes unique identifier for documents
+    // Pathname/ipath terms. This is used for file existence/uptodate
+    // checks, and unique id for the replace_document() call 
+
+    // Truncate the filepath part to a reasonable length and
+    // replace the truncated part with a hopefully unique hash
+    string hash;
+    pathHash(fn, hash, PATHHASHLEN);
+    LOGDEB2(("Db::add: pathhash [%s]\n", hash.c_str()));
+    string pathterm = "P" + hash;
+    newdocument.add_term(pathterm);
+    
+    // Unique term: with path, makes unique identifier for documents
     // inside multidocument files.
     string uniterm;
-    if (!doc.ipath.empty()) {
+    if (doc.ipath.empty()) {
+	uniterm = pathterm;
+    } else {
 	uniterm  = "Q" + hash + "|" + doc.ipath;
 	newdocument.add_term(uniterm);
     }
@@ -617,8 +623,7 @@ bool Db::add(const string &fn, const Doc &idoc,
     // Add db entry or update existing entry:
     try {
 	Xapian::docid did = 
-	    m_ndb->wdb.replace_document(uniterm.empty() ? pathterm : uniterm, 
-				      newdocument);
+	    m_ndb->wdb.replace_document(uniterm, newdocument);
 	if (did < m_ndb->updated.size()) {
 	    m_ndb->updated[did] = true;
 	    LOGDEB(("Db::add: docid %d updated [%s , %s]\n", did, fnc,
@@ -1216,19 +1221,6 @@ bool Db::setQuery(AdvSearchData &sdata, QueryOpts opts,
 	}
     }
 
-    // We do no stem expansion on 'No' words. Should we ?
-    if (!sdata.nowords.empty()) {
-	stringToXapianQueries(sdata.nowords, stemlang, m_ndb, pqueries);
-	if (!pqueries.empty()) {
-	    Xapian::Query nq;
-	    nq = Xapian::Query(Xapian::Query::OP_OR, pqueries.begin(),
-			       pqueries.end());
-	    xq = xq.empty() ? nq :
-		Xapian::Query(Xapian::Query::OP_AND_NOT, xq, nq);
-	    pqueries.clear();
-	}
-    }
-
     if (!sdata.phrase.empty()) {
 	Xapian::Query nq;
 	string s = string("\"") + sdata.phrase + string("\"");
@@ -1253,6 +1245,27 @@ bool Db::setQuery(AdvSearchData &sdata, QueryOpts opts,
 	xq = xq.empty() ? tq : Xapian::Query(Xapian::Query::OP_FILTER, xq, tq);
     }
 
+    // "And not" part. Must come last, as we have to check it's not
+    // the only term in the query.  We do no stem expansion on 'No'
+    // words. Should we ?
+    if (!sdata.nowords.empty()) {
+	stringToXapianQueries(sdata.nowords, stemlang, m_ndb, pqueries);
+	if (!pqueries.empty()) {
+	    Xapian::Query nq;
+	    nq = Xapian::Query(Xapian::Query::OP_OR, pqueries.begin(),
+			       pqueries.end());
+	    if (xq.empty()) {
+		// Xapian cant do this currently. Have to have a positive 
+		// part!
+		sdata.description = "Error: pure negative query\n";
+		LOGERR(("Rcl::Db::setQuery: error: pure negative query\n"));
+		return false;
+	    }
+	    xq = Xapian::Query(Xapian::Query::OP_AND_NOT, xq, nq);
+	    pqueries.clear();
+	}
+    }
+
     m_ndb->query = xq;
     delete m_ndb->enquire;
     m_ndb->enquire = new Xapian::Enquire(m_ndb->db);