ferret 0.10.10 → 0.10.11

data/CHANGELOG

@@ -0,0 +1,6 @@
+Tue Oct 10 14:42:17 JST 2006
+  * Fixed MultiTermDocEnum bug introduced in version 0.10.10 during
+    performance enhancements. 
+  * Added Filter#bits(index_reader) method to C implemented filters so that
+    they can be used in Ruby.
+

data/ext/index.c

@@ -2415,9 +2415,9 @@ static TermDocEnum *mtde_next_tde(MultiTermDocEnum *mtde)
         return mtde->curr_tde = NULL;
     }
     else {
-        TermDocEnum *tde = mtde->curr_tde = mtde->irs_tde[mtde->ptr];
+        TermDocEnum *tde = mtde->irs_tde[mtde->ptr];
         mtde->base = mtde->starts[mtde->ptr];
-        return tde;
+        return mtde->curr_tde = tde;
     }
 }
 
@@ -2454,9 +2454,13 @@ static void mtde_seek(TermDocEnum *tde, int field_num, const char *term)
 {
     MultiTermDocEnum *mtde = MTDE(tde);
     TermEnum *te = mtde->te;
+    char *t;
     te->set_field(te, field_num);
-    te->skip_to(te, term);
-    mtde_seek_te(tde, te);
+    if ((t = te->skip_to(te, term)) != NULL && strcmp(term, t) == 0) {
+        mtde_seek_te(tde, te);
+    } else {
+        memset(mtde->state, 0, mtde->ir_cnt);
+    }
 }
 
 static int mtde_doc_num(TermDocEnum *tde)

data/ext/r_index.c

@@ -3234,7 +3234,6 @@ Init_IndexReader(void)
 {
     cIndexReader = rb_define_class_under(mIndex, "IndexReader", rb_cObject);
     rb_define_alloc_func(cIndexReader, frt_data_alloc);
-    /*rb_define_singleton_method(cIndexReader, "version", frt_class_ir_version, 0); */
     rb_define_method(cIndexReader, "initialize",    frt_ir_init, 1);
     rb_define_method(cIndexReader, "set_norm",      frt_ir_set_norm, 3);
     rb_define_method(cIndexReader, "norms",         frt_ir_norms, 1);

data/ext/r_search.c

@@ -1733,6 +1733,26 @@ frt_f_to_s(VALUE self)
     return rstr;
 }
 
+extern VALUE frt_get_bv(BitVector *bv);
+
+/*
+ *  call-seq:
+ *     filter.bits(index_reader) -> bit_vector
+ *
+ *  Get the bit_vector used by this filter. This method will usually be used
+ *  to group filters or apply filters to other filters.
+ */
+static VALUE
+frt_f_get_bits(VALUE self, VALUE rindex_reader)
+{
+    BitVector *bv;
+    IndexReader *ir;
+    GET_F();
+    Data_Get_Struct(rindex_reader, IndexReader, ir);
+    bv = filt_get_bv(f, ir);
+    return frt_get_bv(bv);
+}
+
 /****************************************************************************
  *
  * RangeFilter Methods
@@ -3610,6 +3630,7 @@ Init_Filter(void)
     frt_mark_cclass(cFilter);
     rb_define_alloc_func(cConstantScoreQuery, frt_data_alloc);
 
+    rb_define_method(cFilter, "bits", frt_f_get_bits, 1);
     rb_define_method(cFilter, "to_s", frt_f_to_s, 0);
 }
 

data/ext/r_utils.c

@@ -6,14 +6,36 @@
  *****************/
 static VALUE cBitVector;
 
+static void
+frt_bv_free(void *p)
+{
+    object_del(p);
+    bv_destroy((BitVector *)p);
+}
+
 static VALUE
 frt_bv_alloc(VALUE klass)
 {
-    return Data_Wrap_Struct(klass, NULL, &bv_destroy, bv_new());
+    BitVector *bv = bv_new();
+    VALUE rbv = Data_Wrap_Struct(klass, NULL, &frt_bv_free, bv);
+    object_add(bv, rbv);
+    return rbv;
 }
 
 #define GET_BV(bv, self) Data_Get_Struct(self, BitVector, bv)
 
+VALUE
+frt_get_bv(BitVector *bv)
+{
+    VALUE rbv;
+    if ((rbv = object_get(bv)) == Qnil) {
+        rbv = Data_Wrap_Struct(cBitVector, NULL, &frt_bv_free, bv);
+        REF(bv);
+        object_add(bv, rbv);
+    }
+    return rbv;
+}
+
 /*
  *  call-seq:
  *     BitVector.new() -> new_bv

data/ext/sort.c

@@ -265,6 +265,7 @@ static int sf_byte_compare(void *index, Hit *hit1, Hit *hit2)
 static void *sf_byte_create_index(int size)
 {
     int *index = ALLOC_AND_ZERO_N(int, size + 1);
+    index[0]++;
     return &index[1];
 }
 

data/lib/ferret_version.rb

@@ -1,3 +1,3 @@
 module Ferret
-  VERSION = '0.10.10'
+  VERSION = '0.10.11'
 end

data/test/unit/search/tc_filter.rb

@@ -68,6 +68,13 @@ class FilterTest < Test::Unit::TestCase
     do_test_top_docs(searcher, q, [0,1,2], rf)
     rf = RangeFilter.new(:int, :< => "2")
     do_test_top_docs(searcher, q, [0,1], rf)
+
+    bits = rf.bits(searcher.reader)
+    assert(bits[0])
+    assert(bits[1])
+    assert(!bits[2])
+    assert(!bits[3])
+    assert(!bits[4])
   end
 
   def test_range_filter_errors
@@ -85,6 +92,18 @@ class FilterTest < Test::Unit::TestCase
     do_test_top_docs(searcher, q, [0,2,4,6,8], qf)
     qf = QueryFilter.new(TermQuery.new(:switch, "off"))
     do_test_top_docs(searcher, q, [1,3,5,7,9], qf)
+
+    bits = qf.bits(searcher.reader)
+    assert(bits[1])
+    assert(bits[3])
+    assert(bits[5])
+    assert(bits[7])
+    assert(bits[9])
+    assert(!bits[0])
+    assert(!bits[2])
+    assert(!bits[4])
+    assert(!bits[6])
+    assert(!bits[8])
   end
 
   def test_filtered_query

metadata

@@ -3,8 +3,8 @@ rubygems_version: 0.9.0
 specification_version: 1
 name: ferret
 version: !ruby/object:Gem::Version 
-  version: 0.10.10
-date: 2006-10-08 00:00:00 +09:00
+  version: 0.10.11
+date: 2006-10-11 00:00:00 +09:00
 summary: Ruby indexing library.
 require_paths: 
 - lib
@@ -35,6 +35,7 @@ files:
 - README
 - MIT-LICENSE
 - TUTORIAL
+- CHANGELOG
 - ext/q_filtered_query.c
 - ext/priorityqueue.c
 - ext/hashset.c