Ticket #608: 608-lazy-wildcard.patch

xapian-core/api/queryinternal.cc

@@ -166 +166 @@
     /// Select the best set_size postlists from the last out_of added.
     void select_elite_set(size_t set_size, size_t out_of);
 
+    /// Select the set_size postlists with the highest term frequency.
+    void select_most_frequent(size_t set_size);
+
     PostList * postlist(QueryOptimiser* qopt);
     PostList * postlist_max(QueryOptimiser* qopt);
 };
@@ -183 +186 @@
     pls.resize(pls.size() - out_of + set_size);
 }
 
+void
+OrContext::select_most_frequent(size_t set_size)
+{
+    vector<PostList*>::iterator begin = pls.begin();
+    nth_element(begin, begin + set_size - 1, pls.end(),
+                ComparePostListTermFreqAscending());
+    for_each(begin + set_size, pls.end(), delete_ptr<PostList>());
+    pls.resize(set_size);
+}
+
 PostList *
 OrContext::postlist(QueryOptimiser* qopt)
 {
@@ -854 +867 @@
     }
     OrContext ctx(0);
     AutoPtr<TermList> t(qopt->db.open_allterms(pattern));
+#if 0
     Xapian::termcount expansions_left = max_expansion;
     // If there's no expansion limit, set expansions_left to the maximum
     // value Xapian::termcount can hold.
     if (expansions_left == 0)
         --expansions_left;
+#endif
     while (true) {
         t->next();
         if (t->at_end())
             break;
+#if 0
         if (expansions_left-- == 0) {
             string msg("Wildcard ");
             msg += pattern;
@@ -871 +887 @@
             msg += " terms";
             throw Xapian::WildcardError(msg);
         }
+#endif
         const string & term = t->get_termname();
         ctx.add_postlist(qopt->open_lazy_post_list(term, 1, or_factor));
     }
 
+    if (max_expansion && ctx.size() > max_expansion)
+        ctx.select_most_frequent(max_expansion);
+
     if (factor != 0.0) {
         if (op != Query::OP_SYNONYM) {
             qopt->set_total_subqs(qopt->get_total_subqs() + ctx.size());