the way we added the autophrase to the query (OR) created false matches. Use AND_MAYBE instead

src/python/recoll/pyrecoll.cpp

@@ -250,7 +250,7 @@ SearchData_addclause(recoll_SearchDataObject* self, PyObject *args,
     case 'S':
 	if (strcasecmp(tp, "sub"))
 	    goto defaultcase;
-	cl = new Rcl::SearchDataClauseSub(Rcl::SCLT_SUB, sub->sd);
+	cl = new Rcl::SearchDataClauseSub(sub->sd);
 	break;
     defaultcase:
     default:

src/query/docseqdb.cpp

@@ -161,7 +161,7 @@ bool DocSequenceDb::setFiltSpec(const DocSeqFiltSpec &fs)
 	m_fsdata = RefCntr<Rcl::SearchData>(
 	    new Rcl::SearchData(Rcl::SCLT_AND, m_sdata->getStemLang()));
 	Rcl::SearchDataClauseSub *cl = 
-	    new Rcl::SearchDataClauseSub(Rcl::SCLT_SUB, m_sdata);
+	    new Rcl::SearchDataClauseSub(m_sdata);
 	m_fsdata->addClause(cl);
     
 	for (unsigned int i = 0; i < fs.crits.size(); i++) {
@@ -181,8 +181,8 @@ bool DocSequenceDb::setFiltSpec(const DocSeqFiltSpec &fs)
 				    fs.values[i], reason);
 		if (sd)  {
 		    Rcl::SearchDataClauseSub *cl1 = 
-			new Rcl::SearchDataClauseSub(Rcl::SCLT_SUB, 
-						     RefCntr<Rcl::SearchData>(sd));
+			new Rcl::SearchDataClauseSub(
+			    RefCntr<Rcl::SearchData>(sd));
 		    m_fsdata->addClause(cl1);
 		}
 	    }

src/query/wasatorcl.cpp

@@ -241,8 +241,7 @@ static Rcl::SearchData *wasaQueryToRcl(const RclConfig *config,
 		continue;
 	    }
 	    nclause = 
-		new Rcl::SearchDataClauseSub(Rcl::SCLT_SUB, 
-					     RefCntr<Rcl::SearchData>(sub));
+		new Rcl::SearchDataClauseSub(RefCntr<Rcl::SearchData>(sub));
 	    if (nclause == 0) {
 		LOGERR(("wasaQueryToRcl: out of memory\n"));
 		reason = "Out of memory";

src/rcldb/searchdata.cpp

@@ -60,11 +60,18 @@ void SearchData::commoninit()
     m_maxSize = size_t(-1);
     m_minSize = size_t(-1);
     m_haveWildCards = false;
-    m_softmaxexpand = -1;
     m_autodiacsens = false;
     m_autocasesens = true;
     m_maxexp = 10000;
     m_maxcl = 100000;
+    m_softmaxexpand = -1;
+}
+
+SearchData::~SearchData() 
+{
+    LOGDEB0(("SearchData::~SearchData\n"));
+    for (qlist_it_t it = m_query.begin(); it != m_query.end(); it++)
+	delete *it;
 }
 
 // Expand categories and mime type wild card exps Categories are
@@ -259,6 +266,15 @@ bool SearchData::toNativeQuery(Rcl::Db &db, void *d)
         }
     }
 
+    // Add the autophrase if any
+    if (m_autophrase.isNotNull()) {
+	Xapian::Query apq;
+	if (m_autophrase->toNativeQuery(db, &apq)) {
+	    xq = xq.empty() ? apq : 
+		Xapian::Query(Xapian::Query::OP_AND_MAYBE, xq, apq);
+	}
+    }
+
     // Add the file type filtering clause if any
     if (!m_filetypes.empty()) {
 	expandFileTypes(db, m_filetypes);
@@ -376,27 +392,8 @@ bool SearchData::maybeAddAutoPhrase(Rcl::Db& db, double freqThreshold)
     // an actual user-entered phrase
     slack += 1 + nwords / 3;
     
-    SearchDataClauseDist *nclp = 
-	new SearchDataClauseDist(SCLT_PHRASE, swords, slack, field);
-
-    // If the toplevel conjunction is an OR, just OR the phrase, else 
-    // deepen the tree.
-    if (m_tp == SCLT_OR) {
-	addClause(nclp);
-    } else {
-	// My type is AND. Change it to OR and insert two queries, one
-	// being the original query as a subquery, the other the
-	// phrase.
-	SearchData *sd = new SearchData(m_tp, m_stemlang);
-	sd->m_query = m_query;
-	sd->m_stemlang = m_stemlang;
-	m_tp = SCLT_OR;
-	m_query.clear();
-	SearchDataClauseSub *oq = 
-	    new SearchDataClauseSub(SCLT_OR, RefCntr<SearchData>(sd));
-	addClause(oq);
-	addClause(nclp);
-    }
+    m_autophrase = RefCntr<SearchDataClauseDist>(
+	new SearchDataClauseDist(SCLT_PHRASE, swords, slack, field));
     return true;
 }
 
@@ -414,22 +411,6 @@ bool SearchData::addClause(SearchDataClause* cl)
     return true;
 }
 
-// Make me all new
-void SearchData::erase() 
-{
-    LOGDEB0(("SearchData::erase\n"));
-    m_tp = SCLT_AND;
-    for (qlist_it_t it = m_query.begin(); it != m_query.end(); it++)
-	delete *it;
-    m_query.clear();
-    m_filetypes.clear();
-    m_description.erase();
-    m_reason.erase();
-    m_haveDates = false;
-    m_minSize = size_t(-1);
-    m_maxSize = size_t(-1);
-}
-
 // Am I a file name only search ? This is to turn off term highlighting
 bool SearchData::fileNameOnly() 
 {

src/rcldb/searchdata.h

@@ -46,6 +46,7 @@ enum SClType {
 };
 
 class SearchDataClause;
+class SearchDataClauseDist;
 
 /** 
     A SearchData object represents a Recoll user query, for translation
@@ -89,13 +90,7 @@ public:
 	commoninit();
     }
     
-    ~SearchData() 
-    {
-	erase();
-    }
-
-    /** Make pristine */
-    void erase();
+    ~SearchData();
 
     /** Is there anything but a file name search in here ? */
     bool fileNameOnly();
@@ -175,7 +170,10 @@ private:
     std::vector<std::string>            m_filetypes; 
     // Excluded set of file types if not empty
     std::vector<std::string>            m_nfiletypes;
-
+    // Autophrase if set. Can't be part of the normal chain because
+    // it uses OP_AND_MAYBE
+    RefCntr<SearchDataClauseDist>   m_autophrase;
+    // 
     bool                      m_haveDates;
     DateInterval              m_dates; // Restrict to date interval
     size_t                    m_maxSize;
@@ -461,8 +459,8 @@ private:
 /** Subquery */
 class SearchDataClauseSub : public SearchDataClause {
 public:
-    SearchDataClauseSub(SClType tp, RefCntr<SearchData> sub) 
-	: SearchDataClause(tp), m_sub(sub) 
+    SearchDataClauseSub(RefCntr<SearchData> sub) 
+	: SearchDataClause(SCLT_SUB), m_sub(sub) 
     {
     }
     virtual bool toNativeQuery(Rcl::Db &db, void *p)