Ticket #608: ticket-608-max-wildcard.patch

xapian-core/queryparser/queryparser.lemony

@@ -481 +481 @@
     return q;
 }
 
+struct TermF {
+    string term;
+    Xapian::termcount freq;
+
+    TermF(const string & t, Xapian::termcount f) : term(t), freq(f) { }
+};
+
+struct CompareFreq {
+    bool operator()(const TermF & a, const TermF & b) {
+        return a.freq > b.freq;
+    }
+};
+
+// Iterator shim to allow building a query from vector<TermF>.
+class TermFIterator {
+    vector<TermF>::const_iterator i;
+
+    Xapian::termpos pos;
+
+  public:
+    TermFIterator(vector<TermF>::const_iterator i_, Xapian::termpos pos_ = 0)
+        : i(i_), pos(pos_) { }
+
+    TermFIterator & operator++() {
+        ++i;
+        return *this;
+    }
+
+    const Xapian::Query operator*() const {
+        return Xapian::Query(i->term, 1, pos);
+    }
+
+    bool operator==(const TermFIterator & o) {
+        return i == o.i;
+    }
+
+    bool operator!=(const TermFIterator & o) {
+        return !(*this == o);
+    }
+
+    typedef std::input_iterator_tag iterator_category;
+    typedef Xapian::Query value_type;
+    typedef Xapian::termcount_diff difference_type;
+    typedef Xapian::Query * pointer;
+    typedef Xapian::Query & reference;
+};
+
 Query *
 Term::as_partial_query(State * state_) const
 {
     const Database & db = state_->get_database();
-    vector<Query> subqs_partial; // A synonym of all the partial terms.
     vector<Query> subqs_full; // A synonym of all the full terms.
 
+    Xapian::termcount max_k = state_->get_max_wildcard_expansion();
+    if (max_k == 0) max_k = 100;
+    Xapian::termcount min_freq = static_cast<Xapian::termcount>(-1);
+
+    vector<TermF> terms;
+    Xapian::termcount k = 0;
+
     const list<string> & prefixes = field_info->prefixes;
     list<string>::const_iterator piter;
     for (piter = prefixes.begin(); piter != prefixes.end(); ++piter) {
@@ -495 +548 @@
         root += name;
         TermIterator t = db.allterms_begin(root);
         while (t != db.allterms_end(root)) {
-            subqs_partial.push_back(Query(*t, 1, pos));
+            Xapian::termcount freq = t.get_termfreq();
+            if (k < max_k) {
+                terms.push_back(TermF(*t, freq));
+            } else {
+                if (k == max_k) {
+                    make_heap(terms.begin(), terms.end(), CompareFreq());
+                    min_freq = terms.front().freq;
+                }
+
+                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;
         }
         // Add the term, as it would normally be handled, as an alternative.
         subqs_full.push_back(Query(make_term(*piter), 1, pos));
     }
+
     Query * q = new Query(Query::OP_OR,
                           Query(Query::OP_SYNONYM,
-                                subqs_partial.begin(), subqs_partial.end()),
+                                TermFIterator(terms.begin(), pos),
+                                TermFIterator(terms.end())),
                           Query(Query::OP_SYNONYM,
                                 subqs_full.begin(), subqs_full.end()));
     delete this;