Ticket #608: x.patch

queryparser.lemony

@@ -412 +412 @@
     return q;
 }
 
+class TermF
+{
+    public:
+        string term;
+        Xapian::termcount freq;
+
+        TermF(string t, Xapian::termcount f) {
+            term = t;
+            freq = f;
+        }
+};
+
+struct CompareFreq {
+    bool operator()(const TermF a, const TermF b) {
+    return a.freq > b.freq;
+    }
+};
+
 Query *
 Term::as_partial_query(State * state_) const
 {
@@ -419 +437 @@
     vector<Query> subqs_partial; // A synonym of all the partial terms.
     vector<Query> subqs_full; // A synonym of all the full terms.
 
+    unsigned int maxK = state_->get_max_wildcard_expansion();
+    if(maxK == 0) maxK = 100;
+    unsigned int K = 0;
+    Xapian::termcount min_freq = 1000000;
+    Xapian::termcount freq = 0;
+
     const list<string> & prefixes = prefix_info->prefixes;
     list<string>::const_iterator piter;
+
     for (piter = prefixes.begin(); piter != prefixes.end(); ++piter) {
         string root = *piter;
         root += name;
@@ -425 +450 @@
         string root = *piter;
         root += name;
         TermIterator t = db.allterms_begin(root);
+        vector<TermF> terms;
+
         while (t != db.allterms_end(root)) {
-            subqs_partial.push_back(Query(*t, 1, pos));
+            freq = t.get_termfreq();
+            if(K < maxK) {
+                if(min_freq > freq)
+                    min_freq = freq;
+
+                terms.push_back(TermF(*t, freq));
+            }
+            else {
+                if(K == maxK) {
+                    make_heap( terms.begin(), terms.end(), CompareFreq() );
+                }
+
+                if(min_freq < freq) {
+                    pop_heap( terms.begin(),terms.end(), CompareFreq() ); terms.pop_back();
+                    terms.push_back( TermF(*t, freq) ); push_heap( terms.begin(), terms.end(), CompareFreq() );
+                    min_freq = terms.front().freq;
+                }
+            }
+
+            ++K;
             ++t;
         }
+
+        for(unsigned int j = 0; j < terms.size(); j++)
+        {
+            subqs_partial.push_back(Query(terms[j].term, 1, pos));
+        }
+
         // Add the term, as it would normally be handled, as an alternative.
         subqs_full.push_back(Query(make_term(*piter), 1, pos));
     }