Ticket #433: matchspy_api_changes_experiment.patch

matcher/multimatch.cc

@@ -190 +190 @@
      *
      *  This implementation calls all the spies in turn.
      */
-    void operator()(const Xapian::Document &doc, Xapian::weight wt);
+    double operator()(const Xapian::Document &doc, Xapian::weight wt);
 };
 
-void 
+double 
 MultipleMatchSpy::operator()(const Xapian::Document &doc, Xapian::weight wt) {
     LOGCALL_VOID(MATCH, "MultipleMatchSpy::operator()", doc << ", " << wt);
-    vector<Xapian::MatchSpy *>::const_iterator i;
-    for (i = spies.begin(); i != spies.end(); ++i) {
-        (**i)(doc, wt);
+    double min_min_wt = DBL_MAX;
+    vector<Xapian::MatchSpy *>::const_iterator i = spies.begin()
+    while (i != spies.end()) {
+        double min_wt = (**i)(doc, wt);
+        if (rare(min_wt == DBL_MAX)) {
+            i = spies.erase(i);
+            continue;
+        }
+        min_min_wt = std::min(min_min_wt, min_wt);
+        ++i;
     }
+    return min_min_wt;
 }
 
 ////////////////////////////////////
@@ -602 +610 @@
                     ++docs_matched;
                     if (!calculated_weight) wt = pl->get_weight();
                     if (matchspy) {
-                        matchspy->operator()(doc, wt);
+                        double matchspy_min_wt = matchspy->operator()(doc, wt);
+                        if (matchspy_min_wt == DBL_MAX) {
+                            matchspy = NULL;
+                        }
                     }
                     if (wt > greatest_wt) goto new_greatest_weight;
                     continue;
@@ -645 +656 @@
                         new_item.wt = wt;
                         calculated_weight = true;
                     }
-                    matchspy->operator()(doc, wt);
+                    double matchspy_min_wt = matchspy->operator()(doc, wt);
+                    if (matchspy_min_wt == DBL_MAX) {
+                        matchspy = NULL;
+                    }
                 }
             }
         }

include/xapian/matchspy.h

@@ -68 +68 @@
      *
      *  @param doc The document seen by the match spy.
      *  @param wt The weight of the document.
+     *
+     *  @return The minimum weight for documents which the matchspy is
+     *  interested in.
      */
-    virtual void operator()(const Xapian::Document &doc,
-                            Xapian::weight wt) = 0;
+    virtual double operator()(const Xapian::Document &doc,
+                              Xapian::weight wt) = 0;
 
     /** Clone the match spy.
      *
@@ -188 +191 @@
     /// Total number of documents seen by the match spy.
     Xapian::doccount total;
 
+    /// Maximum number of values to check.
+    Xapian::doccount check_max;
+
     /// The values seen so far, together with their frequency.
     std::map<std::string, Xapian::doccount> values;
 
   public:
     /// Construct an empty ValueCountMatchSpy.
-    ValueCountMatchSpy() : slot(Xapian::BAD_VALUENO), total(0) {}
+    ValueCountMatchSpy() : slot(Xapian::BAD_VALUENO), total(0), check_max(0) {}
 
-    /// Construct a MatchSpy which counts the values in a particular slot.
-    ValueCountMatchSpy(Xapian::valueno slot_)
-            : slot(slot_), total(0) {
+    /** Construct a MatchSpy which counts the values in a particular slot.
+     *
+     *  @param slot_ The slot number to check.
+     *  @param check_max_ The maximum number of values to check in the slot.
+     *
+     *  If check_max_ is 0, no limit is places on the number of values to check.
+     */
+    ValueCountMatchSpy(Xapian::valueno slot_, check_max_=0)
+            : slot(slot_), total(0), check_max(check_max_) {
     }
 
     /// Return the values seen in the slot.
@@ -226 +238 @@
      *
      *  This implementation tallies values for a matching document.
      */
-    void operator()(const Xapian::Document &doc, Xapian::weight wt);
+    double operator()(const Xapian::Document &doc, Xapian::weight wt);
 
     virtual MatchSpy * clone() const;
     virtual std::string name() const;

api/matchspy.cc

@@ -156 +156 @@
     }
 }
 
-void
+double
 ValueCountMatchSpy::operator()(const Document &doc, weight) {
     ++total;
     string val(doc.get_value(slot));
     if (!val.empty()) ++values[val];
+    if (check_max != 0 && rare(total >= check_max)) {
+        return DBL_MAX;
+    }
+    return 0.0;
 }
 
 void