Changeset 2ac444617a25eb599de0faec801ffa532425f554

Timestamp:

04/28/08 14:54:33 (8 months ago)

Author:

David Balmain <dbalmain@…>

Parents:

88af924cc987bc50098df966cc16b84e4213a7fe

Children:

8d89448795e517f0f1afd9bb633dc677fde09712

git-committer:

David Balmain <dbalmain@gmail.com> / 2008-04-28T14:54:33Z+1000

Message:

Full test coverage for PhraseQuery and its variants

Full test coverage for;
* PhraseQuery
* MultiPhraseQuery?
* PhraseQuery with slop
* MultiPhraseQuery? with slop

Files:

• c/.rake/gcov.history (modified) (previous)
• c/src/index.c (modified) (1 diff)
• c/src/q_phrase.c (modified) (1 diff)
• c/src/search.c (modified) (3 diffs)
• c/test/test_search.c (modified) (7 diffs)
• ruby/ext/r_search.c (modified) (1 diff)

c/src/index.c

@@ -5580 +5580 @@
                 while (NULL != (tk = ts->next(ts))) {
                     pos += tk->pos_inc;
+                    /* if for some reason pos gets set to some number less
+                     * than 0 the we'll start pos at 0 */
+                    if (pos < 0) {
+                        pos = 0;
+                    }
                     dw_add_posting(mp, curr_plists, fld_plists, doc_num,
                                    tk->text, tk->len, pos);

c/src/q_phrase.c

@@ -778 +778 @@
         if (tv_term) {
             TVPosEnum *tvpe = tvpe_new(tv_term->positions, tv_term->freq, 0);
-            if (tvpe_next(tvpe)) {
-                pq_push(tvpe_pq, tvpe);
-                total_positions += tv_term->freq;
-            }
-            else {
-                free(tvpe);
-            }
+            /* got tv_term so tvpe_next should always return true once here */
+            assert(tvpe_next(tvpe));
+            pq_push(tvpe_pq, tvpe);
+            total_positions += tv_term->freq;
         }
     }

c/src/search.c

@@ -540 +540 @@
     }
     self->matches[self->size].start = start;
-    self->matches[self->size].end = end;
-    self->matches[self->size++].score = 1.0;
+    self->matches[self->size].end   = end;
+    self->matches[self->size].score = 1.0;
+    self->size++;
     return self;
 }
@@ -1523 +1524 @@
     Query *rewritten_query = self->rewrite(self, query);
     /* terms get copied directly to df_map so no need to free here */
-    HashSet *terms = frt_hs_new((hash_ft)&frt_term_hash,
-                                (eq_ft)&frt_term_eq,
-                                (free_ft)NULL);
+    HashSet *terms = hs_new((hash_ft)&term_hash,
+                            (eq_ft)&term_eq,
+                            (free_ft)NULL);
     HashSetEntry *hse;
 
@@ -1801 +1802 @@
     int i = msea_get_searcher_index(self, doc_num);
     Searcher *s = msea->searchers[i];
-    return s->get_term_vector(s, doc_num - msea->starts[i],
-                              field);
+    return s->get_term_vector(s, doc_num - msea->starts[i], field);
 }
 

c/test/test_search.c

@@ -202 +202 @@
     IndexWriter *iw;
 
-    FieldInfos *fis = fis_new(STORE_YES, INDEX_YES, TERM_VECTOR_YES);
+    FieldInfos *fis = fis_new(STORE_YES,
+                              INDEX_YES,
+                              TERM_VECTOR_WITH_POSITIONS_OFFSETS);
     index_create(store, fis);
     fis_deref(fis);
@@ -514 +516 @@
 static void test_phrase_query(TestCase *tc, void *data)
 {
+    MatchVector *mv;
     Searcher *searcher = (Searcher *)data;
+    Explanation *explanation;
     Query *q;
     Query *phq = phq_new(field);
@@ -597 +601 @@
     phq_add_term(phq, "quick", 1);
     check_hits(tc, searcher, phq, "", -1);
+    explanation = searcher->explain(searcher, phq, 0);
+    Afequal(0.0, explanation->value);
+    expl_destroy(explanation);
     q_deref(phq);
 
-    /* test single term case, query is rewritten to TermQuery */
+    /* test single-term case, query is rewritten to TermQuery */
     phq = phq_new(field);
     phq_add_term(phq, "word2", 1);
@@ -605 +612 @@
     q = searcher_rewrite(searcher, phq);
     Aiequal(q->type, TERM_QUERY);
+    q_deref(q);
+
+    /* test single-position/multi-term query is rewritten as MultiTermQuery */
+    phq_append_multi_term(phq, "word3");
+    check_hits(tc, searcher, phq, "1,2,3,4,6,8,11,14", -1);
+    q = searcher_rewrite(searcher, phq);
+    Aiequal(q->type, MULTI_TERM_QUERY);
+    q_deref(q);
+
+    /* check boost doesn't break anything */;
+    phq_add_term(phq, "one", 1); /* make sure it won't be rewritten */
+    phq->boost = 10.0;
+    check_hits(tc, searcher, phq, "2,3", -1);
     q_deref(phq);
-    q_deref(q);
+
+    /* test get_matchv_i */
+    phq = phq_new(field);
+    phq_add_term(phq, "quick", 0);
+    phq_add_term(phq, "brown", 1);
+    check_hits(tc, searcher, phq, "1", -1);
+    mv = searcher_get_match_vector(searcher, phq, 1, field);
+    if (Aiequal(2, mv->size)) {
+        Aiequal(3, mv->matches[0].start);
+        Aiequal(4, mv->matches[0].end);
+        Aiequal(7, mv->matches[1].start);
+        Aiequal(8, mv->matches[1].end);
+    }
+    matchv_destroy(mv);
+    phq_set_slop(phq, 4);
+    check_hits(tc, searcher, phq, "1,16,17", -1);
+    mv = searcher_get_match_vector(searcher, phq, 1, field);
+    if (Aiequal(2, mv->size)) {
+        Aiequal(3, mv->matches[0].start);
+        Aiequal(4, mv->matches[0].end);
+        Aiequal(7, mv->matches[1].start);
+        Aiequal(8, mv->matches[1].end);
+    }
+    matchv_destroy(mv);
+    mv = searcher_get_match_vector(searcher, phq, 16, field);
+    if (Aiequal(1, mv->size)) {
+        Aiequal(2, mv->matches[0].start);
+        Aiequal(5, mv->matches[0].end);
+    }
+    matchv_destroy(mv);
+    phq_add_term(phq, "chicken", 1);
+    check_hits(tc, searcher, phq, "", -1);
+    mv = searcher_get_match_vector(searcher, phq, 16, field);
+    Aiequal(0, mv->size);
+    matchv_destroy(mv);
+    q_deref(phq);
 }
 
@@ -666 +721 @@
     Searcher *searcher = (Searcher *)data;
     Query *phq, *q;
+    MatchVector *mv;
 
     phq = phq_new(field);
-    phq_add_term(phq, "quick", 0);
+    /* ok to use append_multi_term to start */
+    phq_append_multi_term(phq, "quick");
     phq_append_multi_term(phq, "fast");
     check_hits(tc, searcher, phq, "1, 8, 11, 14, 16, 17", -1);
@@ -731 +788 @@
     q_deref(q);
 
+    /* test get_matchv_i */
+    phq = phq_new(field);
+    phq_add_term(phq, "quick", 0);
+    phq_add_term(phq, "brown", 1);
+    phq_append_multi_term(phq, "dirty");
+    phq_append_multi_term(phq, "red");
+    check_hits(tc, searcher, phq, "1,11", -1);
+    mv = searcher_get_match_vector(searcher, phq, 1, field);
+    if (Aiequal(2, mv->size)) {
+        Aiequal(3, mv->matches[0].start);
+        Aiequal(4, mv->matches[0].end);
+        Aiequal(7, mv->matches[1].start);
+        Aiequal(8, mv->matches[1].end);
+    }
+    matchv_destroy(mv);
+    phq_set_slop(phq, 1);
+    check_hits(tc, searcher, phq, "1,11,17", -1);
+    mv = searcher_get_match_vector(searcher, phq, 1, field);
+    if (Aiequal(2, mv->size)) {
+        Aiequal(3, mv->matches[0].start);
+        Aiequal(4, mv->matches[0].end);
+        Aiequal(7, mv->matches[1].start);
+        Aiequal(8, mv->matches[1].end);
+    }
+    matchv_destroy(mv);
+    mv = searcher_get_match_vector(searcher, phq, 17, field);
+    if (Aiequal(1, mv->size)) {
+        Aiequal(5, mv->matches[0].start);
+        Aiequal(7, mv->matches[0].end);
+    }
+    matchv_destroy(mv);
+    phq_add_term(phq, "chicken", 1);
+    phq_append_multi_term(phq, "turtle");
+    check_hits(tc, searcher, phq, "", -1);
+    mv = searcher_get_match_vector(searcher, phq, 17, field);
+    Aiequal(0, mv->size);
+    matchv_destroy(mv);
+    q_deref(phq);
 }
 
@@ -1406 +1501 @@
     int i;
     IndexWriter *iw;
-    FieldInfos *fis = fis_new(STORE_YES, INDEX_YES, TERM_VECTOR_YES);
+    FieldInfos *fis = fis_new(STORE_YES,
+                              INDEX_YES,
+                              TERM_VECTOR_WITH_POSITIONS_OFFSETS);
     index_create(store, fis);
     fis_deref(fis);

ruby/ext/r_search.c

@@ -494 +494 @@
 {
     VALUE rterms = rb_ary_new();
-    HashSet *terms = term_set_new();
+    HashSet *terms = hs_new((hash_ft)&term_hash,
+                            (eq_ft)&term_eq,
+                            (free_ft)term_destroy);
     HashSetEntry *hse;
     GET_Q();