rroonga 4.0.0 → 4.0.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 8fdfe9138a4546d1c3d4f7475d92f57d664f23a8
-  data.tar.gz: 39eab083625dc558c5287aa34eb202b2f7e418b7
+  metadata.gz: 53910c50882ceba0c8c1fd10471076e0c9adb48e
+  data.tar.gz: 97c9676d4436ce853b7d684e3c8a70fabb9bf023
 SHA512:
-  metadata.gz: d345881451aecc7c1fefdfeb15f62f7df6a01de07c478b372496828aec5dc74789369fac4b4185c514ac44d8fa43914974cfc966fb57736ea19169e31706c388
-  data.tar.gz: 30216dcfe16899e6647cb084dcdff4b5742c8ae67859bb79c4bd4221e3af123388e68680a1a222a0658e1cb36b79a8b712ae4d62cced0fdb1748db0fb55501c4
+  metadata.gz: 05cd87977c479c0c95d4fcab3bb0d6e41b2f9f798dda965f671066d55e3b495bcbf5e41677bccdb539f1041739528ed5f520bb8c738ac6f088973cc56ab900da
+  data.tar.gz: 2b359a701d8d188b155e7dbc2a334930221df42d243811468b60f67e3acd89da3c2e7c1bb51d14233c6bab3d60dec58894405863bb53f949954065794717741b

data/.yardopts

@@ -0,0 +1,6 @@
+--output-dir doc/reference/en
+--markup textile
+lib/**/*.rb
+ext/**/*.c
+-
+doc/text/*

data/doc/text/news.textile

@@ -1,5 +1,31 @@
 h1. NEWS
 
+h2(#4-0-1). 4.0.1: 2014-04-04
+
+h3. Improvements
+
+* Supported Groonga 4.0.1. Groonga 4.0.0 or older are not supported.
+* Supported no weight match column case.
+  <pre>
+  table.select do |record|
+    match_target = record.match_target do |target|
+      target.column
+    end
+    match_target =~ keyword
+  end
+  </pre>
+* Supported weight vector.
+* grndump: Changed to use @--normalizer@ instead of @KEY_NORMALIZE@.
+  Old Groonga can't restore dumped database.
+* {Groonga::IndexClumn#search}: Added @mode@ option.
+* {Groonga::IndexClumn#search}: Added @weight@ option.
+* Accepted String as option key.
+
+h3. Fixes
+
+* Fixed a bug that index dump drops index options in Ruby syntax.
+
+
 h2(#4-0-0). 4.0.0: 2014-02-09
 
 h3. Improvements

data/ext/groonga/rb-grn-column.c

@@ -1,7 +1,7 @@
 /* -*- coding: utf-8; mode: C; indent-tabs-mode: nil; c-basic-offset: 4 -*- */
 /* vim: set sts=4 sw=4 ts=8 noet: */
 /*
-  Copyright (C) 2009-2011  Kouhei Sutou <kou@clear-code.com>
+  Copyright (C) 2009-2014  Kouhei Sutou <kou@clear-code.com>
 
   This library is free software; you can redistribute it and/or
   modify it under the terms of the GNU Lesser General Public
@@ -71,10 +71,29 @@ rb_grn_column_bind (RbGrnColumn *rb_column,
                     grn_ctx *context, grn_obj *column)
 {
     RbGrnObject *rb_grn_object;
+    int column_type;
+    unsigned char value_type;
 
     rb_grn_object = RB_GRN_OBJECT(rb_column);
     rb_grn_named_object_bind(RB_GRN_NAMED_OBJECT(rb_column), context, column);
-    rb_column->value = grn_obj_open(context, GRN_BULK, 0,
+
+    column_type = (column->header.flags & GRN_OBJ_COLUMN_TYPE_MASK);
+    if (column_type == GRN_OBJ_COLUMN_VECTOR) {
+        switch (rb_grn_object->range->header.type) {
+        case GRN_TABLE_HASH_KEY:
+        case GRN_TABLE_PAT_KEY:
+        case GRN_TABLE_DAT_KEY:
+        case GRN_TABLE_NO_KEY:
+            value_type = GRN_UVECTOR;
+            break;
+        default:
+            value_type = GRN_VECTOR;
+            break;
+        }
+    } else {
+        value_type = GRN_BULK;
+    }
+    rb_column->value = grn_obj_open(context, value_type, 0,
                                     rb_grn_object->range_id);
 }
 
@@ -674,6 +693,24 @@ rb_grn_column_scalar_p (VALUE self)
     }
 }
 
+/*
+ * @overload with_weight?
+ * @return [Boolean] @true@ if the column is vector and created with
+ *   @:with_weight => true@ flag, @false@ otherwise.
+ * @since 4.0.1
+ */
+static VALUE
+rb_grn_column_with_weight_p(VALUE self)
+{
+    grn_obj *column;
+
+    rb_grn_column_deconstruct(SELF(self), &column, NULL,
+                              NULL, NULL,
+                              NULL, NULL, NULL);
+
+    return CBOOL2RVAL(column->header.flags & GRN_OBJ_WITH_WEIGHT);
+}
+
 /*
  * _operator_ を実行できる _column_ のインデックスを返す。
  * @since 1.0.9
@@ -772,6 +809,9 @@ rb_grn_init_column (VALUE mGrn)
     rb_define_method(rb_cGrnColumn, "vector?", rb_grn_column_vector_p, 0);
     rb_define_method(rb_cGrnColumn, "scalar?", rb_grn_column_scalar_p, 0);
 
+    rb_define_method(rb_cGrnColumn, "with_weight?",
+                     rb_grn_column_with_weight_p, 0);
+
     rb_define_method(rb_cGrnColumn, "indexes", rb_grn_column_get_indexes, -1);
 
     rb_define_method(rb_cGrnColumn, "rename", rb_grn_column_rename, 1);

data/ext/groonga/rb-grn-index-column.c

@@ -1,6 +1,6 @@
 /* -*- coding: utf-8; mode: C; indent-tabs-mode: nil; c-basic-offset: 4 -*- */
 /*
-  Copyright (C) 2009-2013  Kouhei Sutou <kou@clear-code.com>
+  Copyright (C) 2009-2014  Kouhei Sutou <kou@clear-code.com>
 
   This library is free software; you can redistribute it and/or
   modify it under the terms of the GNU Lesser General Public
@@ -44,6 +44,7 @@ rb_grn_index_column_finalizer (grn_ctx *context, grn_obj *object,
     grn_obj_unlink(context, rb_grn_index_column->id_query);
     grn_obj_unlink(context, rb_grn_index_column->string_query);
     grn_obj_unlink(context, rb_grn_index_column->old_value);
+    grn_obj_unlink(context, rb_grn_index_column->set_value);
 
     rb_grn_column_finalizer(context, object, RB_GRN_COLUMN(rb_grn_index_column));
 }
@@ -59,6 +60,9 @@ rb_grn_index_column_bind (RbGrnIndexColumn *rb_grn_index_column,
 
     rb_grn_index_column->old_value = grn_obj_open(context, GRN_BULK, 0,
                                                   rb_grn_object->range_id);
+    rb_grn_index_column->set_value =
+        grn_obj_open(context, GRN_VECTOR, 0,
+                     rb_grn_object->range->header.domain);
 
     rb_grn_index_column->id_query = grn_obj_open(context, GRN_BULK, 0,
                                                  rb_grn_object->domain_id);
@@ -75,6 +79,7 @@ rb_grn_index_column_deconstruct (RbGrnIndexColumn *rb_grn_index_column,
                                  grn_obj **domain,
                                  grn_obj **value,
                                  grn_obj **old_value,
+                                 grn_obj **set_value,
                                  grn_id *range_id,
                                  grn_obj **range,
                                  grn_obj **id_query,
@@ -89,138 +94,14 @@ rb_grn_index_column_deconstruct (RbGrnIndexColumn *rb_grn_index_column,
 
     if (old_value)
         *old_value = rb_grn_index_column->old_value;
+    if (set_value)
+        *set_value = rb_grn_index_column->set_value;
     if (id_query)
         *id_query = rb_grn_index_column->id_query;
     if (string_query)
         *string_query = rb_grn_index_column->string_query;
 }
 
-/*
- * IDが _id_ であるレコードを高速に全文検索するため転置索引を作
- * 成する。多くの場合、 {Groonga::Table#define_index_column} で
- * +:source+ オプションを指定することにより、自動的に全文検索
- * 用の索引は更新されるので、明示的にこのメソッドを使うこと
- * は少ない。
- *
- * @example 記事の段落毎に索引を作成する。
- *   articles = Groonga::Array.create(:name => "<articles>")
- *   articles.define_column("title", "ShortText")
- *   articles.define_column("content", "Text")
- *
- *   terms = Groonga::Hash.create(:name => "<terms>",
- *                                :default_tokenizer => "TokenBigram")
- *   content_index = terms.define_index_column("content", articles,
- *                                             :with_section => true)
- *
- *   content = <<-EOC
- *   groonga は組み込み型の全文検索エンジンライブラリです。
- *   DBMSやスクリプト言語処理系等に組み込むことによって、その
- *   全文検索機能を強化することができます。また、リレーショナ
- *   ルモデルに基づくデータストア機能を内包しており、groonga
- *   単体でも高速なデータストアサーバとして使用することができ
- *   ます。
- *
- *   ■全文検索方式
- *   転置索引型の全文検索エンジンです。転置索引は圧縮されてファ
- *   イルに格納され、検索時のディスク読み出し量を小さく、かつ
- *   局所的に抑えるように設計されています。用途に応じて以下の
- *   索引タイプを選択できます。
- *   EOC
- *
- *   groonga = articles.add(:title => "groonga", :content => content)
- *
- *   content.split(/\n{2,}/).each_with_index do |sentence, i|
- *     content_index[groonga] = {:value => sentence, :section => i + 1}
- *   end
- *
- *   content_index.search("エンジン").collect do |record|
- *     p record.key["title"] # -> "groonga"
- *   end
- *
- * @overload []=(id, value)
- *   @param [String] value 新しい値
- * @overload []=(id, options)
- *   _options_ を指定することにより、 _value_ を指定したときよりも索引の作
- *   成を制御できる。
- *   @param [::Hash] options The name and value
- *     pairs. Omitted names are initialized as the default value
- *   @option options :section
- *     段落番号を指定する。省略した場合は1を指定したとみなされ
- *     る。
- *     {Groonga::Table#define_index_column} で
- *     @{:with_section => true}@ を指定していなければい
- *     けない。
- *   @option options :old_value
- *     以前の値を指定する。省略した場合は現在の値が用いられる。
- *     通常は指定する必要はない。
- *   @option options :value
- *     新しい値を指定する。 _value_ を指定した場合と _options_ で
- *     @{:value => value}@ を指定した場合は同じ動作とな
- *     る。
- *
- * @deprecated Since 3.0.2. Use {#add}, {#delete} or {#update} instead.
- */
-static VALUE
-rb_grn_index_column_array_set (VALUE self, VALUE rb_id, VALUE rb_value)
-{
-    grn_ctx *context = NULL;
-    grn_obj *column, *range;
-    grn_rc rc;
-    grn_id id;
-    unsigned int section;
-    grn_obj *old_value, *new_value;
-    VALUE original_rb_value, rb_section, rb_old_value, rb_new_value;
-
-    original_rb_value = rb_value;
-
-    rb_grn_index_column_deconstruct(SELF(self), &column, &context,
-                                    NULL, NULL,
-                                    &new_value, &old_value,
-                                    NULL, &range,
-                                    NULL, NULL);
-
-    id = RVAL2GRNID(rb_id, context, range, self);
-
-    if (!RVAL2CBOOL(rb_obj_is_kind_of(rb_value, rb_cHash))) {
-        VALUE hash_value;
-        hash_value = rb_hash_new();
-        rb_hash_aset(hash_value, RB_GRN_INTERN("value"), rb_value);
-        rb_value = hash_value;
-    }
-
-    rb_grn_scan_options(rb_value,
-                        "section", &rb_section,
-                        "old_value", &rb_old_value,
-                        "value", &rb_new_value,
-                        NULL);
-
-    if (NIL_P(rb_section))
-        section = 1;
-    else
-        section = NUM2UINT(rb_section);
-
-    if (NIL_P(rb_old_value)) {
-        old_value = NULL;
-    } else {
-        GRN_BULK_REWIND(old_value);
-        RVAL2GRNBULK(rb_old_value, context, old_value);
-    }
-
-    if (NIL_P(rb_new_value)) {
-        new_value = NULL;
-    } else {
-        GRN_BULK_REWIND(new_value);
-        RVAL2GRNBULK(rb_new_value, context, new_value);
-    }
-
-    rc = grn_column_index_update(context, column,
-                                 id, section, old_value, new_value);
-    rb_grn_context_check(context, self);
-    rb_grn_rc_check(rc, self);
-
-    return original_rb_value;
-}
-
 /*
  * Adds a record that has @value@ content to inverted index for fast
  * fulltext serach. Normally, this method is not used
@@ -304,7 +185,7 @@ rb_grn_index_column_add (int argc, VALUE *argv, VALUE self)
 
     rb_grn_index_column_deconstruct(SELF(self), &column, &context,
                                     NULL, NULL,
-                                    &new_value, NULL,
+                                    &new_value, NULL, NULL,
                                     NULL, &range,
                                     NULL, NULL);
 
@@ -418,7 +299,7 @@ rb_grn_index_column_delete (int argc, VALUE *argv, VALUE self)
 
     rb_grn_index_column_deconstruct(SELF(self), &column, &context,
                                     NULL, NULL,
-                                    NULL, &old_value,
+                                    NULL, &old_value, NULL,
                                     NULL, &range,
                                     NULL, NULL);
 
@@ -526,7 +407,7 @@ rb_grn_index_column_update (int argc, VALUE *argv, VALUE self)
 
     rb_grn_index_column_deconstruct(SELF(self), &column, &context,
                                     NULL, NULL,
-                                    &new_value, &old_value,
+                                    &new_value, &old_value, NULL,
                                     NULL, &range,
                                     NULL, NULL);
 
@@ -583,7 +464,7 @@ rb_grn_index_column_get_sources (VALUE self)
 
     rb_grn_index_column_deconstruct(SELF(self), &column, &context,
                                     NULL, NULL,
-                                    NULL, NULL, NULL, NULL,
+                                    NULL, NULL, NULL, NULL, NULL,
                                     NULL, NULL);
 
     GRN_OBJ_INIT(&sources, GRN_BULK, 0, GRN_ID_NIL);
@@ -702,7 +583,7 @@ rb_grn_index_column_set_sources (VALUE self, VALUE rb_sources)
 
     rb_grn_index_column_deconstruct(SELF(self), &column, &context,
                                     NULL, NULL,
-                                    NULL, NULL,
+                                    NULL, NULL, NULL,
                                     &range_id, NULL,
                                     NULL, NULL);
 
@@ -785,15 +666,16 @@ rb_grn_index_column_search (int argc, VALUE *argv, VALUE self)
     grn_obj *query = NULL, *id_query = NULL, *string_query = NULL;
     grn_obj *result;
     grn_operator operator;
+    grn_search_optarg options;
     grn_rc rc;
-    VALUE rb_query, options, rb_result, rb_operator;
+    VALUE rb_query, rb_options, rb_result, rb_operator, rb_mode, rb_weight;
 
     rb_grn_index_column_deconstruct(SELF(self), &column, &context,
                                     NULL, NULL,
-                                    NULL, NULL, NULL, &range,
+                                    NULL, NULL, NULL, NULL, &range,
                                     &id_query, &string_query);
 
-    rb_scan_args(argc, argv, "11", &rb_query, &options);
+    rb_scan_args(argc, argv, "11", &rb_query, &rb_options);
 
     if (CBOOL2RVAL(rb_obj_is_kind_of(rb_query, rb_cInteger))) {
         grn_id id;
@@ -807,9 +689,11 @@ rb_grn_index_column_search (int argc, VALUE *argv, VALUE self)
         query = string_query;
     }
 
-    rb_grn_scan_options(options,
+    rb_grn_scan_options(rb_options,
                         "result", &rb_result,
                         "operator", &rb_operator,
+                        "mode", &rb_mode,
+                        "weight", &rb_weight,
                         NULL);
 
     if (NIL_P(rb_result)) {
@@ -824,7 +708,23 @@ rb_grn_index_column_search (int argc, VALUE *argv, VALUE self)
 
     operator = RVAL2GRNOPERATOR(rb_operator);
 
-    rc = grn_obj_search(context, column, query, result, operator, NULL);
+    if (NIL_P(rb_mode)) {
+        options.mode = GRN_OP_EXACT;
+    } else {
+        options.mode = RVAL2GRNOPERATOR(rb_mode);
+    }
+    options.similarity_threshold = 0;
+    options.max_interval = 0;
+    options.weight_vector = NULL;
+    if (NIL_P(rb_weight)) {
+        options.vector_size = 1;
+    } else {
+        options.vector_size = NUM2UINT(rb_weight);
+    }
+    options.proc = NULL;
+    options.max_size = 0;
+
+    rc = grn_obj_search(context, column, query, result, operator, &options);
     rb_grn_rc_check(rc, self);
 
     return rb_result;
@@ -842,7 +742,7 @@ rb_grn_index_column_with_section_p (VALUE self)
 
     rb_grn_index_column_deconstruct(SELF(self), &column, NULL,
                                     NULL, NULL,
-                                    NULL, NULL, NULL, NULL,
+                                    NULL, NULL, NULL, NULL, NULL,
                                     NULL, NULL);
 
     return CBOOL2RVAL(column->header.flags & GRN_OBJ_WITH_SECTION);
@@ -860,7 +760,7 @@ rb_grn_index_column_with_weight_p (VALUE self)
 
     rb_grn_index_column_deconstruct(SELF(self), &column, NULL,
                                     NULL, NULL,
-                                    NULL, NULL, NULL, NULL,
+                                    NULL, NULL, NULL, NULL, NULL,
                                     NULL, NULL);
 
     return CBOOL2RVAL(column->header.flags & GRN_OBJ_WITH_WEIGHT);
@@ -878,7 +778,7 @@ rb_grn_index_column_with_position_p (VALUE self)
 
     rb_grn_index_column_deconstruct(SELF(self), &column, NULL,
                                     NULL, NULL,
-                                    NULL, NULL, NULL, NULL,
+                                    NULL, NULL, NULL, NULL, NULL,
                                     NULL, NULL);
 
     return CBOOL2RVAL(column->header.flags & GRN_OBJ_WITH_POSITION);
@@ -906,24 +806,24 @@ rb_grn_index_column_with_position_p (VALUE self)
 static VALUE
 rb_grn_index_column_open_cursor (int argc, VALUE *argv, VALUE self)
 {
-    grn_ctx *context;
-    grn_obj *column;
-    grn_obj *range_object;
+    grn_ctx          *context;
+    grn_obj          *column;
+    grn_obj          *range_object;
     grn_table_cursor *table_cursor;
-    grn_id rid_min = GRN_ID_NIL;
-    grn_id rid_max = GRN_ID_MAX;
-    int flags = 0;
-    grn_obj *index_cursor;
-    VALUE rb_table_cursor;
-    VALUE options;
-    VALUE rb_with_section, rb_with_weight, rb_with_position;
-    VALUE rb_table;
-    VALUE rb_lexicon;
-    VALUE rb_cursor;
+    grn_id            rid_min = GRN_ID_NIL;
+    grn_id            rid_max = GRN_ID_MAX;
+    int               flags   = 0;
+    grn_obj          *index_cursor;
+    VALUE             rb_table_cursor;
+    VALUE             options;
+    VALUE             rb_with_section, rb_with_weight, rb_with_position;
+    VALUE             rb_table;
+    VALUE             rb_lexicon;
+    VALUE             rb_cursor;
 
     rb_grn_index_column_deconstruct(SELF(self), &column, &context,
                                     NULL, NULL,
-                                    NULL, NULL,
+                                    NULL, NULL, NULL,
                                     NULL, &range_object,
                                     NULL, NULL);
 
@@ -935,8 +835,8 @@ rb_grn_index_column_open_cursor (int argc, VALUE *argv, VALUE self)
                         NULL);
 
     table_cursor = RVAL2GRNTABLECURSOR(rb_table_cursor, NULL);
-    rb_table = GRNOBJECT2RVAL(Qnil, context, range_object, GRN_FALSE);
-    rb_lexicon = rb_iv_get(rb_table_cursor, "@table");
+    rb_table     = GRNOBJECT2RVAL(Qnil, context, range_object, GRN_FALSE);
+    rb_lexicon   = rb_iv_get(rb_table_cursor, "@table");
 
     if (NIL_P(rb_with_section)) {
         flags |= column->header.flags & GRN_OBJ_WITH_SECTION;
@@ -973,9 +873,6 @@ rb_grn_init_index_column (VALUE mGrn)
     rb_cGrnIndexColumn =
         rb_define_class_under(mGrn, "IndexColumn", rb_cGrnColumn);
 
-    rb_define_method(rb_cGrnIndexColumn, "[]=",
-                     rb_grn_index_column_array_set, 2);
-
     rb_define_method(rb_cGrnIndexColumn, "add",
                      rb_grn_index_column_add, -1);
     rb_define_method(rb_cGrnIndexColumn, "delete",
@@ -999,6 +896,7 @@ rb_grn_init_index_column (VALUE mGrn)
                      rb_grn_index_column_with_weight_p, 0);
     rb_define_method(rb_cGrnIndexColumn, "with_position?",
                      rb_grn_index_column_with_position_p, 0);
+
     rb_define_method(rb_cGrnIndexColumn, "open_cursor",
                      rb_grn_index_column_open_cursor, -1);
 }