Changeset 73644644566a2ef9791d80ee31ab66770be83046

Timestamp:

04/27/08 21:33:05 (8 months ago)

Author:

David Balmain <dbalmain@…>

Parents:

15dee3043d3e551973defb68d67022123989c1fa

Children:

56424f8d1cfd0e927e5e8f9e0c4329531221a966

git-committer:

David Balmain <dbalmain@gmail.com> / 2008-04-27T21:33:05Z+1000

Message:

Fixed bug in sloppy PhraseQuery

Phrase positions wheren't being sorted consistantly causing inconsistant search
results.

Location:

c

Files:

• src/q_phrase.c (modified) (3 diffs)
• test/test.c (modified) (1 diff)
• test/test_search.c (modified) (1 diff)

c/src/q_phrase.c

@@ -117 +117 @@
 static bool pp_less_than(const PhPos *pp1, const PhPos *pp2)
 {
-    /* docs will all be equal when this method is used */
-    return pp1->position < pp2->position;
-    /*
-    if (PP(p)->doc == PP(p)->doc) {
-        return PP(p)->position < PP(p)->position;
+    if (pp1->position == pp2->position) {
+        return pp1->offset > pp2->offset;
     }
     else {
-        return PP(p)->doc < PP(p)->doc;
-    }
-    */
+        return pp1->position < pp2->position;
+    }
 }
 
@@ -235 +231 @@
     return raw_score * sim_decode_norm(
         self->similarity,
-        phsc->norms[phsc->phrase_pos[phsc->pp_first_idx]->doc]);
+        phsc->norms[self->doc]);
 }
 
@@ -279 +275 @@
     phsc_skip_to(self, doc_num);
 
-    phrase_freq = (self->doc == doc_num) ? phsc->freq : (float)0.0;
+    phrase_freq = (self->doc == doc_num) ? phsc->freq : 0.0f;
     return expl_new(sim_tf(self->similarity, phrase_freq),
                     "tf(phrase_freq=%f)", phrase_freq);

c/test/test.c

@@ -259 +259 @@
 
     dptr = suite->head;
-    fprintf(stdout, "%-15s\t\tTotal\tFail\tFailed %%\n", "Failed Tests");
+    fprintf(stdout, "%-24sTotal\tFail\tFailed %%\n", "Failed Tests");
     fprintf(stdout, "===================================================\n");
     while (dptr != NULL) {

c/test/test_search.c

@@ -162 +162 @@
     {"20050930", "word1",
         "cat1/",                ".123"},
-    {"20051001", "word1 word2 the quick brown fox",
+    {"20051001", "word1 word2 the quick brown fox the quick brown fox",
         "cat1/sub1",            "0.954"},
     {"20051002", "word1 word3",