Search: allow negative directory filtering (all except from dir). Emit more explicit errors for other unallowed negative search clauses.

src/qtgui/ssearch_w.cpp

@@ -139,7 +139,7 @@ void SSearch::startSimpleSearch()
         else
             sdata = wasaStringToRcl(rclconfig, u8, reason);
 	if (sdata == 0) {
-	    QMessageBox::warning(0, "Recoll", tr("Bad query string") +
+	    QMessageBox::warning(0, "Recoll", tr("Bad query string") + ": " +
 				 QString::fromAscii(reason.c_str()));
 	    return;
 	}

src/query/wasatorcl.cpp

@@ -35,28 +35,15 @@ using std::list;
 #include "refcntr.h"
 #include "textsplit.h"
 
-Rcl::SearchData *wasaStringToRcl(RclConfig *config, 
+static Rcl::SearchData *wasaQueryToRcl(RclConfig *config, WasaQuery *wasa, 
-				 const string &qs, string &reason, 
+				       const string& autosuffs, string& reason)
-                                 const string& autosuffs)
 {
-    StringToWasaQuery parser;
+    if (wasa == 0) {
-    WasaQuery *wq = parser.stringToQuery(qs, reason);
+	reason = "NULL query";
-    if (wq == 0) 
-	return 0;
-    Rcl::SearchData *rq = wasaQueryToRcl(config, wq, autosuffs);
-    if (rq == 0) {
-	reason = "Failed translating xesam query structure to recoll";
 	return 0;
     }
-    return rq;
-}
-
-Rcl::SearchData *wasaQueryToRcl(RclConfig *config,
-				WasaQuery *wasa, const string& autosuffs)
-{
-    if (wasa == 0)
-	return 0;
     if (wasa->m_op != WasaQuery::OP_AND && wasa->m_op != WasaQuery::OP_OR) {
+	reason = "Top query neither AND nor OR ?";
 	LOGERR(("wasaQueryToRcl: top query neither AND nor OR!\n"));
 	return 0;
     }
@@ -70,44 +57,80 @@ Rcl::SearchData *wasaQueryToRcl(RclConfig *config,
     WasaQuery::subqlist_t::iterator it;
     Rcl::SearchDataClause *nclause;
 
+    // Walk the list of clauses. Some pseudo-field types need special
+    // processing, which results in setting data in the top struct
+    // instead of adding a clause. We check for these first
     for (it = wasa->m_subs.begin(); it != wasa->m_subs.end(); it++) {
+
+	if (!stringicmp("mime", (*it)->m_fieldspec) ||
+	    !stringicmp("format", (*it)->m_fieldspec)) {
+	    if ((*it)->m_op != WasaQuery::OP_LEAF) {
+		reason = "Negative mime/format clauses not supported yet";
+		return 0;
+	    }
+	    sdata->addFiletype((*it)->m_value);
+	    continue;
+	} 
+
+	// Xesam uses "type", we also support "rclcat", for broad
+	// categories like "audio", "presentation", etc.
+	if (!stringicmp("rclcat", (*it)->m_fieldspec) ||
+	    !stringicmp("type", (*it)->m_fieldspec)) {
+	    if ((*it)->m_op != WasaQuery::OP_LEAF) {
+		reason = "Negative rclcat/type clauses not supported yet";
+		return 0;
+	    }
+	    list<string> mtypes;
+	    if (config && config->getMimeCatTypes((*it)->m_value, mtypes)
+		&& !mtypes.empty()) {
+		for (list<string>::iterator mit = mtypes.begin();
+		     mit != mtypes.end(); mit++) {
+		    sdata->addFiletype(*mit);
+		}
+	    } else {
+		reason = "Unknown rclcat/type value: no mime types found";
+		return 0;
+	    }
+	    continue;
+	}
+
+	// Filtering on location
+	if (!stringicmp("dir", (*it)->m_fieldspec)) {
+	    sdata->setTopdir((*it)->m_value, (*it)->m_op == WasaQuery::OP_EXCL);
+	    continue;
+	} 
+
+	// Handle "date" spec
+	if (!stringicmp("date", (*it)->m_fieldspec)) {
+	    if ((*it)->m_op != WasaQuery::OP_LEAF) {
+		reason = "Negative date filtering not supported";
+		return 0;
+	    }
+	    DateInterval di;
+	    if (!parsedateinterval((*it)->m_value, &di)) {
+		LOGERR(("wasaQueryToRcl: bad date interval format\n"));
+		reason = "Bad date interval format";
+		return 0;
+	    }
+	    LOGDEB(("wasaQueryToRcl:: date span:  %d-%d-%d/%d-%d-%d\n",
+		    di.y1,di.m1,di.d1, di.y2,di.m2,di.d2));
+	    sdata->setDateSpan(&di);
+	    continue;
+	} 
+
+	// "Regular" processing follows:
 	switch ((*it)->m_op) {
 	case WasaQuery::OP_NULL:
 	case WasaQuery::OP_AND:
 	default:
-	    LOGINFO(("wasaQueryToRcl: found bad NULL or AND q type in list\n"));
+	    reason = "Found bad NULL or AND query type in list";
+	    LOGERR(("wasaQueryToRcl: found bad NULL or AND q type in list\n"));
 	    continue;
+
 	case WasaQuery::OP_LEAF: {
 	    LOGDEB2(("wasaQueryToRcl: leaf clause [%s]:[%s]\n", 
 		     (*it)->m_fieldspec.c_str(), (*it)->m_value.c_str()));
 
-	    // Special cases (mime, category, dir filter ...). Not pretty.
-
-	    if (!stringicmp("mime", (*it)->m_fieldspec) ||
-		!stringicmp("format", (*it)->m_fieldspec)
-		) {
-		sdata->addFiletype((*it)->m_value);
-		break;
-	    } 
-
-	    // Xesam uses "type", we also support "rclcat", for broad
-	    // categories like "audio", "presentation", etc.
-	    if (!stringicmp("rclcat", (*it)->m_fieldspec) ||
-		!stringicmp("type", (*it)->m_fieldspec)) {
-		list<string> mtypes;
-		if (config && config->getMimeCatTypes((*it)->m_value, mtypes)) {
-		    for (list<string>::iterator mit = mtypes.begin();
-			 mit != mtypes.end(); mit++) {
-			sdata->addFiletype(*mit);
-		    }
-		}
-		break;
-	    } 
-	    if (!stringicmp("dir", (*it)->m_fieldspec)) {
-		sdata->setTopdir((*it)->m_value);
-		break;
-	    } 
-
             // Change terms found in the "autosuffs" list into "ext"
             // field queries
             if ((*it)->m_fieldspec.empty() && !autosuffs.empty()) {
@@ -121,21 +144,6 @@ Rcl::SearchData *wasaQueryToRcl(RclConfig *config,
                 }
             }
 
-            // Handle "date" spec
-	    if (!stringicmp("date", (*it)->m_fieldspec)) {
-                DateInterval di;
-                if (!parsedateinterval((*it)->m_value, &di)) {
-                    LOGERR(("wasaQueryToRcl: bad date interval format\n"));
-                    // Process rest of query anyway ?
-                    break;
-                }
-                LOGDEB(("wasaQueryToRcl:: date span:  %d-%d-%d/%d-%d-%d\n",
-                        di.y1,di.m1,di.d1, di.y2,di.m2,di.d2))
-		sdata->setDateSpan(&di);
-		break;
-	    } 
-
-            // "Regular" processing follows:
 	    unsigned int mods = (unsigned int)(*it)->m_modifiers;
 
 	    if (TextSplit::hasVisibleWhite((*it)->m_value)) {
@@ -154,6 +162,7 @@ Rcl::SearchData *wasaQueryToRcl(RclConfig *config,
 							  (*it)->m_fieldspec);
 	    }
 	    if (nclause == 0) {
+		reason = "Out of memory";
 		LOGERR(("wasaQueryToRcl: out of memory\n"));
 		return 0;
 	    }
@@ -176,12 +185,14 @@ Rcl::SearchData *wasaQueryToRcl(RclConfig *config,
 	    // but should work. If there is actually a single
 	    // word, it will not be taken as a phrase, and
 	    // stem-expansion will work normally
+            // Have to do this because searchdata has nothing like and_not
 	    nclause = new Rcl::SearchDataClauseSimple(Rcl::SCLT_EXCL, 
 						      string("\"") + 
 						      (*it)->m_value + "\"",
 						      (*it)->m_fieldspec);
 	    
 	    if (nclause == 0) {
+		reason = "Out of memory";
 		LOGERR(("wasaQueryToRcl: out of memory\n"));
 		return 0;
 	    }
@@ -194,7 +205,8 @@ Rcl::SearchData *wasaQueryToRcl(RclConfig *config,
 	    LOGDEB2(("wasaQueryToRcl: OR clause [%s]:[%s]\n", 
 		     (*it)->m_fieldspec.c_str(), (*it)->m_value.c_str()));
 	    // Create a subquery.
-	    Rcl::SearchData *sub = wasaQueryToRcl(config, *it);
+	    Rcl::SearchData *sub = 
+		wasaQueryToRcl(config, *it, autosuffs, reason);
 	    if (sub == 0) {
 		continue;
 	    }
@@ -203,6 +215,7 @@ Rcl::SearchData *wasaQueryToRcl(RclConfig *config,
 					     RefCntr<Rcl::SearchData>(sub));
 	    if (nclause == 0) {
 		LOGERR(("wasaQueryToRcl: out of memory\n"));
+		reason = "Out of memory";
 		return 0;
 	    }
 	    if ((*it)->m_modifiers & WasaQuery::WQM_NOSTEM)
@@ -213,3 +226,14 @@ Rcl::SearchData *wasaQueryToRcl(RclConfig *config,
 
     return sdata;
 }
+
+Rcl::SearchData *wasaStringToRcl(RclConfig *config, 
+				 const string &qs, string &reason, 
+                                 const string& autosuffs)
+{
+    StringToWasaQuery parser;
+    WasaQuery *wq = parser.stringToQuery(qs, reason);
+    if (wq == 0) 
+	return 0;
+    return wasaQueryToRcl(config, wq, autosuffs, reason);
+}

src/query/wasatorcl.h

@@ -29,8 +29,4 @@ class RclConfig;
 extern Rcl::SearchData *wasaStringToRcl(RclConfig *, 
 					const string& query, string &reason,
 					const string& autosuffs = string());
-class WasaQuery;
-extern Rcl::SearchData *wasaQueryToRcl(RclConfig *, WasaQuery *wasa,
-                                       const string& autosuffs = string());
-
 #endif /* _WASATORCL_H_INCLUDED_ */

src/rcldb/searchdata.cpp

@@ -259,9 +259,10 @@ bool SearchData::toNativeQuery(Rcl::Db &db, void *d)
 	     it != vpath.end(); it++){
 	    pvpath.push_back(pathelt_prefix + *it);
 	}
-	xq = Xapian::Query(Xapian::Query::OP_FILTER, xq, 
+	xq = Xapian::Query(m_topdirexcl ? 
-			   Xapian::Query(Xapian::Query::OP_PHRASE, 
+			   Xapian::Query::OP_AND_NOT:Xapian::Query::OP_FILTER, 
-					 pvpath.begin(), pvpath.end()));
+			   xq, Xapian::Query(Xapian::Query::OP_PHRASE, 
+					     pvpath.begin(), pvpath.end()));
     }
 
     *((Xapian::Query *)d) = xq;
@@ -355,6 +356,7 @@ void SearchData::erase() {
     m_query.clear();
     m_filetypes.clear();
     m_topdir.erase();
+    m_topdirexcl = false;
     m_description.erase();
     m_reason.erase();
     m_haveDates = false;

src/rcldb/searchdata.h

@@ -72,7 +72,8 @@ class SearchDataClause;
 class SearchData {
 public:
     SearchData(SClType tp) 
-        : m_tp(tp), m_haveDates(false), m_haveWildCards(false) 
+        : m_tp(tp), m_topdirexcl(false), m_haveDates(false), 
+	  m_haveWildCards(false) 
     {
 	if (m_tp != SCLT_OR && m_tp != SCLT_AND) 
 	    m_tp = SCLT_OR;
@@ -102,8 +103,11 @@ public:
     bool maybeAddAutoPhrase();
 
     /** Set/get top subdirectory for filtering results */
-    void setTopdir(const string& t) {m_topdir = t;}
+    void setTopdir(const string& t, bool excl = false) 
-    string getTopdir() {return m_topdir;}
+    {
+	m_topdir = t;
+	m_topdirexcl = excl;
+    }
 
     /** Set date span for filtering results */
     void setDateSpan(DateInterval *dip) {m_dates = *dip; m_haveDates = true;}
@@ -138,6 +142,7 @@ private:
     vector<SearchDataClause*> m_query;
     vector<string>            m_filetypes; // Restrict to filetypes if set.
     string                    m_topdir; // Restrict to subtree.
+    bool                      m_topdirexcl; // Invert meaning
     bool                      m_haveDates;
     DateInterval              m_dates; // Restrict to date interval
     // Printable expanded version of the complete query, retrieved/set