Mailing-List: contact commits-help@lucene.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@lucene.apache.org
Message-ID: <20050301003722.48084.qmail@minotaur.apache.org>
Content-Type: text/plain; charset="utf-8"
MIME-Version: 1.0
Content-Transfer-Encoding: quoted-printable
Date: Tue, 01 Mar 2005 00:37:22 -0000
Subject: svn commit: r155723 - in incubator/lucene4c/trunk:
 src/search/scorer.c
 test/search/scorer_test.c
To: lucene4c-cvs@incubator.apache.org
From: rooneg@apache.org

Author: rooneg
Date: Mon Feb 28 16:37:21 2005
New Revision: 155723

URL: http://svn.apache.org/viewcvs?view=3Drev&rev=3D155723
Log:
Finish fixing up boolean queries containing should occurances.

If we have should queries then we can't just bail the first time one of
the scorers returns EOF, if it was a should scorer then we remove that
scorer from our set of candidates and move on, until we run out of should
scorers, at which point we're done.

* src/search/scorer.c
  (boolean_search_scorer_t): add should_left member to track the number
   of should scorers left.
  (pick_scorer): deal with the fact that should scorers can be NULL.
  (boolean_scorer_find_doc): handle EOF errors more carefully, deal
   with the fact that should scorers can be NULL.
  (lcn_boolean_scorer_create): initialize bsb->should_left.

* test/search/scorer_test.c
  (test_boolean_scorer): remove out of date comment.

Modified:
    incubator/lucene4c/trunk/src/search/scorer.c
    incubator/lucene4c/trunk/test/search/scorer_test.c

Modified: incubator/lucene4c/trunk/src/search/scorer.c
URL: http://svn.apache.org/viewcvs/incubator/lucene4c/trunk/src/search/scor=
er.c?view=3Ddiff&r1=3D155722&r2=3D155723
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D
--- incubator/lucene4c/trunk/src/search/scorer.c (original)
+++ incubator/lucene4c/trunk/src/search/scorer.c Mon Feb 28 16:37:21 2005
@@ -65,6 +65,9 @@
   apr_array_header_t *should;
   apr_array_header_t *must_not;
=20
+  /* the number of should scorers we have left in the array */
+  int should_left;
+
   /* the scorer we used last time through boolean_scorer_find_doc's main
    * loop to get our initial document for comparisons. */
   lcn_scorer_t *last_scorer;
@@ -102,6 +105,8 @@
               lcn_scorer_t *scorer =3D APR_ARRAY_IDX (bsb->should,
                                                     i,
                                                     lcn_scorer_t *);
+              if (! scorer)
+                continue;
=20
               apr_uint32_t doc =3D lcn_scorer_doc (scorer);
=20
@@ -139,7 +144,39 @@
       /* if we have a last scorer (i.e. we already came through this
        * function already) move it along to the next document. */
       if (bsb->last_scorer !=3D NULL)
-        LCN_ERR (lcn_scorer_next (bsb->last_scorer));
+        {
+          lcn_error_t *err =3D lcn_scorer_next (bsb->last_scorer);
+
+          /* If we get an EOF we need to be careful...  If there are any
+           * should scorers then we have to check to see if this was one
+           * of them, and if it was we don't bail out, instead we remove
+           * it from the array and reduce our count, if we hit zero then
+           * we're done with all the shoulds and that means we are done,
+           * otherwise clear the error and continue looping. */
+
+          if (err && err->apr_err =3D=3D APR_EOF && bsb->should_left > 0)
+            {
+              lcn_boolean_t its_a_should =3D FALSE;
+
+              for (i =3D 0; i < bsb->should->nelts; ++i)
+                {
+                  if (APR_ARRAY_IDX (bsb->should, i, lcn_scorer_t *)
+                      =3D=3D bsb->last_scorer)
+                    {
+                      APR_ARRAY_IDX (bsb->should, i, lcn_scorer_t *) =3D N=
ULL;
+                      bsb->should_left--;
+                      its_a_should =3D TRUE;
+                    }
+                }
+
+              if (bsb->should_left =3D=3D 0 || its_a_should =3D=3D FALSE)
+                return err;
+              else
+                lcn_error_clear (err);
+            }
+          else if (err)
+            return err;
+        }
=20
       /* when we add support for should scorers the selection of this
        * scorer becomes more complex, but for now we can just pick an
@@ -158,6 +195,8 @@
           lcn_scorer_t *should_scorer =3D APR_ARRAY_IDX (bsb->should,
                                                        i,
                                                        lcn_scorer_t *);
+          if (! should_scorer)
+            continue;
=20
           apr_uint32_t otherdoc =3D lcn_scorer_doc (should_scorer);
=20
@@ -267,6 +306,8 @@
                                 sizeof (lcn_scorer_t *));
=20
   LCN_ERR (fill_scorers_array (should, bsb->should, index, pool));
+
+  bsb->should_left =3D should->nelts;
=20
   bsb->must_not =3D apr_array_make (pool,
                                   must_not->nelts,

Modified: incubator/lucene4c/trunk/test/search/scorer_test.c
URL: http://svn.apache.org/viewcvs/incubator/lucene4c/trunk/test/search/sco=
rer_test.c?view=3Ddiff&r1=3D155722&r2=3D155723
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D
--- incubator/lucene4c/trunk/test/search/scorer_test.c (original)
+++ incubator/lucene4c/trunk/test/search/scorer_test.c Mon Feb 28 16:37:21 =
2005
@@ -174,7 +174,6 @@
=20
   ABTS_INT_EQUAL (tc, APR_EOF, err->apr_err);
=20
-  /* expected to fail for now, should queries aren't quite working yet */
   ABTS_INT_EQUAL (tc, 113, count);
=20
   apr_pool_clear (p);