groonga 0.0.5 → 0.0.6
Sign up to get free protection for your applications and to get access to all the features.
- data/NEWS.ja.rdoc +13 -0
- data/NEWS.rdoc +13 -0
- data/README.ja.rdoc +2 -1
- data/README.rdoc +2 -1
- data/Rakefile +0 -1
- data/TUTORIAL.ja.rdoc +49 -4
- data/benchmark/read-write-many-small-items.rb +5 -4
- data/benchmark/write-many-small-items.rb +5 -4
- data/example/bookmark.rb +30 -1
- data/example/index-html.rb +79 -0
- data/example/search/config.ru +182 -0
- data/example/search/public/css/groonga.css +122 -0
- data/ext/rb-grn-array.c +12 -9
- data/ext/rb-grn-column.c +13 -8
- data/ext/rb-grn-context.c +34 -16
- data/ext/rb-grn-database.c +3 -2
- data/ext/rb-grn-expression-builder.c +8 -2
- data/ext/rb-grn-expression.c +3 -3
- data/ext/rb-grn-hash.c +13 -10
- data/ext/rb-grn-object.c +127 -19
- data/ext/rb-grn-patricia-trie.c +8 -7
- data/ext/rb-grn-table-cursor.c +2 -40
- data/ext/rb-grn-table.c +154 -42
- data/ext/rb-grn-type.c +18 -10
- data/ext/rb-grn-utils.c +22 -20
- data/ext/rb-grn.h +9 -12
- data/extconf.rb +1 -1
- data/html/developer.html +1 -1
- data/html/index.html +2 -2
- data/lib/groonga/expression-builder.rb +133 -63
- data/lib/groonga/schema.rb +229 -37
- data/test/groonga-test-utils.rb +1 -1
- data/test/test-array.rb +1 -0
- data/test/test-context.rb +7 -1
- data/test/test-database.rb +11 -3
- data/test/test-expression-builder.rb +26 -2
- data/test/test-fix-size-column.rb +2 -1
- data/test/test-hash.rb +6 -1
- data/test/test-record.rb +2 -1
- data/test/test-schema.rb +85 -10
- data/test/test-table.rb +99 -3
- data/test/test-type.rb +3 -2
- data/test/test-variable-size-column.rb +2 -1
- data/test-unit/Rakefile +6 -1
- data/test-unit/lib/test/unit/autorunner.rb +26 -3
- data/test-unit/lib/test/unit/priority.rb +21 -1
- data/test-unit/lib/test/unit/testcase.rb +101 -36
- data/test-unit/lib/test/unit/ui/console/testrunner.rb +7 -4
- data/test-unit/test/{test_testcase.rb → test-testcase.rb} +30 -1
- data/test-unit/test/test_assertions.rb +1 -1
- metadata +9 -6
data/NEWS.ja.rdoc
CHANGED
@@ -1,5 +1,18 @@
|
|
1
1
|
= お知らせ
|
2
2
|
|
3
|
+
== 0.0.6: 2009-07-31
|
4
|
+
|
5
|
+
* groonga 0.1.1対応
|
6
|
+
* ドキュメントの修正 [id:mat_aki]
|
7
|
+
* Groonga::Table#selectでのg式対応
|
8
|
+
* APIの追加
|
9
|
+
* Groonga::Table#union!
|
10
|
+
* Groonga::Table#intersect!
|
11
|
+
* Groonga::Table#differene!
|
12
|
+
* Groonga::Table#merge!
|
13
|
+
* tar.gzも提供 [id:m_seki]
|
14
|
+
* メモリリークの修正
|
15
|
+
|
3
16
|
== 0.0.3: 2009-07-18
|
4
17
|
|
5
18
|
* [#26145] Groonga::TableKeySupport#has_key?の追加 [Tasuku SUENAGA]
|
data/NEWS.rdoc
CHANGED
@@ -1,5 +1,18 @@
|
|
1
1
|
= NEWS
|
2
2
|
|
3
|
+
== 0.0.6: 2009-07-31
|
4
|
+
|
5
|
+
* Supported groonga 0.1.1.
|
6
|
+
* Fixed documents [id:mat_aki]
|
7
|
+
* Supported groonga expression for searching.
|
8
|
+
* Added API
|
9
|
+
* Groonga::Table#union!
|
10
|
+
* Groonga::Table#intersect!
|
11
|
+
* Groonga::Table#differene!
|
12
|
+
* Groonga::Table#merge!
|
13
|
+
* Provided tar.gz [id:m_seki]
|
14
|
+
* Fixed memory leaks
|
15
|
+
|
3
16
|
== 0.0.3: 2009-07-18
|
4
17
|
|
5
18
|
* [#26145] Added Groonga::TableKeySupport#has_key? [Tasuku SUENAGA]
|
data/README.ja.rdoc
CHANGED
@@ -32,7 +32,7 @@ pkg-config.rbはrcairoに付属しているもので、これはRubyライセ
|
|
32
32
|
== 依存ソフトウェア
|
33
33
|
|
34
34
|
* Ruby >= 1.8 (1.9.1対応)
|
35
|
-
* groonga >= 0.
|
35
|
+
* groonga >= 0.1.1
|
36
36
|
|
37
37
|
== インストール
|
38
38
|
|
@@ -54,3 +54,4 @@ http://lists.sourceforge.jp/mailman/listinfo/groonga-dev
|
|
54
54
|
* グニャラくん: バグレポートしてくれました。
|
55
55
|
* にくさん: バグレポートしてくれました。
|
56
56
|
* daraさん: テストを書いてくれました。
|
57
|
+
* id:mat_akiさん: チュートリアルのバグを教えてくれました。
|
data/README.rdoc
CHANGED
@@ -33,7 +33,7 @@ license/GPL for details.
|
|
33
33
|
== Dependencies
|
34
34
|
|
35
35
|
* Ruby >= 1.8 (including 1.9.1)
|
36
|
-
* groonga >= 0.
|
36
|
+
* groonga >= 0.1.1
|
37
37
|
|
38
38
|
== Install
|
39
39
|
|
@@ -55,3 +55,4 @@ http://rubyforge.org/mailman/listinfo/groonga-users-en
|
|
55
55
|
* Tasuku SUENAGA: sent bug reports.
|
56
56
|
* niku: sent bug reports.
|
57
57
|
* dara: wrote tests.
|
58
|
+
* id:mat_aki: sent bug reports.
|
data/Rakefile
CHANGED
@@ -119,7 +119,6 @@ Hoe.spec('groonga') do |_project|
|
|
119
119
|
description = cleanup_white_space(entries[entries.index("Description") + 1])
|
120
120
|
project.summary, project.description, = description.split(/\n\n+/, 3)
|
121
121
|
|
122
|
-
project.need_tar = false
|
123
122
|
project.remote_rdoc_dir = "groonga"
|
124
123
|
end
|
125
124
|
|
data/TUTORIAL.ja.rdoc
CHANGED
@@ -187,9 +187,9 @@ http://qwik.jp/senna/senna2.files/rect4605.png
|
|
187
187
|
|
188
188
|
次に、<tt><comments></tt>テーブルを追加します。
|
189
189
|
|
190
|
-
>> comments = Groonga::
|
191
|
-
|
192
|
-
=> #<Groonga::
|
190
|
+
>> comments = Groonga::Array.create(:name => "<comments>",
|
191
|
+
:key_type => "<shorttext>")
|
192
|
+
=> #<Groonga::Array ...>
|
193
193
|
>> comments.define_column("item", items)
|
194
194
|
=> #<Groonga::FixSizeColumn ..>
|
195
195
|
>> comments.define_column("author", users)
|
@@ -347,4 +347,49 @@ http://qwik.jp/senna/senna2.files/rect4605.png
|
|
347
347
|
|
348
348
|
== 少し複雑な検索
|
349
349
|
|
350
|
-
|
350
|
+
↓はまだ動かない!!!
|
351
|
+
|
352
|
+
さらに実用的な検索について考えてみましょう。
|
353
|
+
|
354
|
+
ブックマークが大量に蓄積されるに従って、より的確に適合度を算
|
355
|
+
出する必要性に迫られます。
|
356
|
+
|
357
|
+
今のところ検索対象として利用できるのは<tt><items>.title</tt>
|
358
|
+
と<tt><comments>.content</tt>ですが、<tt><items>.title</tt>は
|
359
|
+
元ページから得られるやや信頼できる情報なのに対して、
|
360
|
+
<tt><comments>.content</tt>はブックマークユーザが任意に設定で
|
361
|
+
きる情報で、やや信憑性に乏しいと言えます。しかし、再現率を確
|
362
|
+
保するためにはユーザのコメントも是非対象に含めたいところです。
|
363
|
+
|
364
|
+
そこで、以下のようなポリシーで検索を行うことにします。
|
365
|
+
|
366
|
+
* <tt><items>.title</tt>か<tt><comments>.content</tt>のいずれ
|
367
|
+
かにマッチするitemを検索する。
|
368
|
+
* ただし、<tt><items>.title</tt>にマッチしたレコードはスコア
|
369
|
+
を10倍重み付けする。
|
370
|
+
* 同一のitemに対して、キーワードにマッチする<tt>comment</tt>
|
371
|
+
が複数存在した場合は、それぞれの<tt>comment</tt>のスコアの
|
372
|
+
和を、該当するitemのスコアとする。
|
373
|
+
|
374
|
+
以下のようにして、commentとitemとそれぞれに対する検索結果を求
|
375
|
+
めます。
|
376
|
+
|
377
|
+
>> ruby_comments = @comments.select {|record| record["content"] =~ "Ruby"}
|
378
|
+
#<Groonga::Hash ..., size: <2>>
|
379
|
+
>> ruby_items = @items.select("*W1:50 title:%Ruby")
|
380
|
+
#<Groonga::Hash ..., size: <2>>
|
381
|
+
|
382
|
+
_ruby_comments_の結果をitem毎にグループ化し、_ruby_items_と
|
383
|
+
unionして出力します。
|
384
|
+
|
385
|
+
>> ruby_items = ruby_comments.group([".item"]).union!(ruby_items)
|
386
|
+
#<Groonga::Hash ..., size: <4>>
|
387
|
+
>> ruby_items.sort([{:key => ".:score", :order => "descendant"}]).each do |record|
|
388
|
+
>> p [record[".:score"], record[".title"]]
|
389
|
+
>> end
|
390
|
+
[1, "るびま"]
|
391
|
+
[1, "オブジェクトスクリプト言語Ruby"]
|
392
|
+
[1, "Ruby"]
|
393
|
+
[1, "ラングバ"]
|
394
|
+
|
395
|
+
これで目的の結果が得られました。(FIXME: 得られていない!)
|
@@ -55,10 +55,11 @@ begin
|
|
55
55
|
|
56
56
|
require 'groonga'
|
57
57
|
@database = Groonga::Database.create
|
58
|
+
value_type = Groonga::Type.new("Text8", :size => 8)
|
58
59
|
|
59
60
|
item("groonga: Hash: memory") do
|
60
61
|
@hash = Groonga::Hash.create(:key_type => "<shorttext>",
|
61
|
-
:
|
62
|
+
:value_type => value_type)
|
62
63
|
values.each do |value|
|
63
64
|
@hash[value] = value
|
64
65
|
@hash[value]
|
@@ -67,7 +68,7 @@ begin
|
|
67
68
|
|
68
69
|
item("groonga: Trie: memory") do
|
69
70
|
@hash = Groonga::PatriciaTrie.create(:key_type => "<shorttext>",
|
70
|
-
:
|
71
|
+
:value_type => value_type)
|
71
72
|
values.each do |value|
|
72
73
|
@hash[value] = value
|
73
74
|
@hash[value]
|
@@ -77,7 +78,7 @@ begin
|
|
77
78
|
@hash_file = Tempfile.new("groonga-hash")
|
78
79
|
item("groonga: Hash: file") do
|
79
80
|
@hash = Groonga::Hash.create(:key_type => "<shorttext>",
|
80
|
-
:
|
81
|
+
:value_type => value_type,
|
81
82
|
:path => @hash_file.path)
|
82
83
|
values.each do |value|
|
83
84
|
@hash[value] = value
|
@@ -88,7 +89,7 @@ begin
|
|
88
89
|
trie_file = Tempfile.new("groonga-trie")
|
89
90
|
item("groonga: Trie: file") do
|
90
91
|
@hash = Groonga::PatriciaTrie.create(:key_type => "<shorttext>",
|
91
|
-
:
|
92
|
+
:value_type => value_type,
|
92
93
|
:path => trie_file.path)
|
93
94
|
values.each do |value|
|
94
95
|
@hash[value] = value
|
@@ -54,10 +54,11 @@ begin
|
|
54
54
|
|
55
55
|
require 'groonga'
|
56
56
|
@database = Groonga::Database.create
|
57
|
+
value_type = Groonga::Type.new("Text8", :size => 8)
|
57
58
|
|
58
59
|
item("groonga: Hash: memory") do
|
59
60
|
@hash = Groonga::Hash.create(:key_type => "<shorttext>",
|
60
|
-
:
|
61
|
+
:value_type => value_type)
|
61
62
|
values.each do |value|
|
62
63
|
@hash[value] = value
|
63
64
|
end
|
@@ -65,7 +66,7 @@ begin
|
|
65
66
|
|
66
67
|
item("groonga: Trie: memory") do
|
67
68
|
@hash = Groonga::PatriciaTrie.create(:key_type => "<shorttext>",
|
68
|
-
:
|
69
|
+
:value_type => value_type)
|
69
70
|
values.each do |value|
|
70
71
|
@hash[value] = value
|
71
72
|
end
|
@@ -74,7 +75,7 @@ begin
|
|
74
75
|
@hash_file = Tempfile.new("groonga-hash")
|
75
76
|
item("groonga: Hash: file") do
|
76
77
|
@hash = Groonga::Hash.create(:key_type => "<shorttext>",
|
77
|
-
:
|
78
|
+
:value_type => value_type,
|
78
79
|
:path => @hash_file.path)
|
79
80
|
values.each do |value|
|
80
81
|
@hash[value] = value
|
@@ -84,7 +85,7 @@ begin
|
|
84
85
|
trie_file = Tempfile.new("groonga-trie")
|
85
86
|
item("groonga: Trie: file") do
|
86
87
|
@hash = Groonga::PatriciaTrie.create(:key_type => "<shorttext>",
|
87
|
-
:
|
88
|
+
:value_type => value_type,
|
88
89
|
:path => trie_file.path)
|
89
90
|
values.each do |value|
|
90
91
|
@hash[value] = value
|
data/example/bookmark.rb
CHANGED
@@ -14,6 +14,12 @@ rescue LoadError
|
|
14
14
|
require "groonga"
|
15
15
|
end
|
16
16
|
|
17
|
+
require 'time'
|
18
|
+
|
19
|
+
# Groonga::Logger.register(:level => :debug) do |level, time, title, message, location|
|
20
|
+
# p [level, time, title, message, location]
|
21
|
+
# end
|
22
|
+
|
17
23
|
$KCODE = "UTF-8"
|
18
24
|
Groonga::Context.default_options = {:encoding => :utf8}
|
19
25
|
|
@@ -43,10 +49,13 @@ p(terms = Groonga::Hash.create(:name => "<terms>",
|
|
43
49
|
:default_tokenizer => "<token:bigram>"))
|
44
50
|
p terms.define_index_column("item_title", items,
|
45
51
|
:persistent => persistent,
|
52
|
+
:with_weight => true,
|
53
|
+
:with_section => true,
|
54
|
+
:with_position => true,
|
46
55
|
:source => "<items>.title")
|
47
56
|
|
48
57
|
p items.find("http://ja.wikipedia.org/wiki/Ruby")["title"] = "Ruby"
|
49
|
-
p items.find("http://www.ruby-lang.org/")["title"] = "
|
58
|
+
p items.find("http://www.ruby-lang.org/")["title"] = "オブジェクト指向スクリプト言語Ruby"
|
50
59
|
|
51
60
|
p(users = Groonga::Hash.create(:name => "<users>",
|
52
61
|
:key_type => "<shorttext>",
|
@@ -63,6 +72,9 @@ p comments.define_column("issued", "<time>")
|
|
63
72
|
|
64
73
|
p terms.define_index_column("comment_content", comments,
|
65
74
|
:persistent => persistent,
|
75
|
+
:with_weight => true,
|
76
|
+
:with_section => true,
|
77
|
+
:with_position => true,
|
66
78
|
:source => "<comments>.content")
|
67
79
|
|
68
80
|
p users.add("moritan", :name => "モリタン")
|
@@ -96,6 +108,15 @@ p add_bookmark("http://d.hatena.ne.jp/higepon/20070815/1187192864",
|
|
96
108
|
p add_bookmark("http://practical-scheme.net/docs/cont-j.html",
|
97
109
|
"なんでも継続", "taporobo", "トランポリン LISP continuation",
|
98
110
|
1187568692)
|
111
|
+
p add_bookmark("http://jp.rubyist.net/managinze",
|
112
|
+
"るびま", "moritan", "Ruby ドキュメント",
|
113
|
+
Time.now)
|
114
|
+
p add_bookmark("http://jp.rubyist.net/managinze",
|
115
|
+
"るびま", "taporobo", "Ruby 雑誌",
|
116
|
+
Time.now)
|
117
|
+
p add_bookmark("http://groonga.rubyforge.org/",
|
118
|
+
"ラングバ", "moritan", "Ruby groonga",
|
119
|
+
Time.parse("2009-07-19"))
|
99
120
|
|
100
121
|
|
101
122
|
records = comments.select do |record|
|
@@ -128,3 +149,11 @@ records.group([".item"]).each do |record|
|
|
128
149
|
item.key,
|
129
150
|
item[".title"]]
|
130
151
|
end
|
152
|
+
|
153
|
+
p ruby_comments = @comments.select {|record| record["content"] =~ "Ruby"}
|
154
|
+
p ruby_items = @items.select("*W1:50 title:%Ruby")
|
155
|
+
|
156
|
+
p ruby_items = ruby_comments.group([".item"]).union!(ruby_items)
|
157
|
+
ruby_items.sort([{:key => ".:score", :order => "descending"}]).each do |record|
|
158
|
+
p [record[".:score"], record[".title"]]
|
159
|
+
end
|
@@ -0,0 +1,79 @@
|
|
1
|
+
#!/usr/bin/env ruby
|
2
|
+
|
3
|
+
if ARGV.size < 2 or ARGV.find {|option| option == "-h" or option == "--help"}
|
4
|
+
puts "Usage: #{$0} DATABASE_FILE [FILE_OR_DIRECTORY ...]"
|
5
|
+
exit
|
6
|
+
end
|
7
|
+
|
8
|
+
require 'pathname'
|
9
|
+
|
10
|
+
base_directory = Pathname(__FILE__).dirname + ".."
|
11
|
+
$LOAD_PATH.unshift((base_directory + "ext").to_s)
|
12
|
+
$LOAD_PATH.unshift((base_directory + "lib").to_s)
|
13
|
+
|
14
|
+
require 'rubygems'
|
15
|
+
require 'groonga'
|
16
|
+
require 'nokogiri'
|
17
|
+
|
18
|
+
database_file, *targets = ARGV
|
19
|
+
|
20
|
+
database_file = Pathname(database_file)
|
21
|
+
database_directory = database_file.dirname
|
22
|
+
database_directory.mkpath unless database_directory.exist?
|
23
|
+
|
24
|
+
if database_file.exist?
|
25
|
+
Groonga::Database.open(database_file.to_s)
|
26
|
+
else
|
27
|
+
Groonga::Database.create(:path => database_file.to_s)
|
28
|
+
Groonga::Schema.define do |schema|
|
29
|
+
schema.create_table("documents") do |table|
|
30
|
+
table.string("title")
|
31
|
+
table.text("content")
|
32
|
+
table.string("path")
|
33
|
+
table.time("last-modified")
|
34
|
+
end
|
35
|
+
|
36
|
+
schema.create_table("terms",
|
37
|
+
:type => :patricia_trie,
|
38
|
+
:key_normalize => true,
|
39
|
+
:default_tokenizer => "TokenBigram") do |table|
|
40
|
+
table.index("documents.title")
|
41
|
+
table.index("documents.content")
|
42
|
+
end
|
43
|
+
end
|
44
|
+
end
|
45
|
+
|
46
|
+
documents = Groonga::Context.default["documents"]
|
47
|
+
|
48
|
+
targets.each do |target|
|
49
|
+
target = Pathname(target)
|
50
|
+
target.find do |path|
|
51
|
+
throw :prune if path.basename.to_s == ".svn"
|
52
|
+
if path.file? and path.extname == ".html"
|
53
|
+
path.open do |html|
|
54
|
+
values = {:path => path.relative_path_from(target).to_s}
|
55
|
+
_documents = documents.select do |record|
|
56
|
+
record["path"] == values[:path]
|
57
|
+
end
|
58
|
+
if _documents.size.zero?
|
59
|
+
document = documents.add
|
60
|
+
else
|
61
|
+
document = _documents.to_a[0].key
|
62
|
+
end
|
63
|
+
|
64
|
+
html_document = Nokogiri::HTML(html)
|
65
|
+
html_document.css("title").each do |title|
|
66
|
+
values[:title] = title.text
|
67
|
+
end
|
68
|
+
html_document.css("body").each do |body|
|
69
|
+
values[:content] = body.text
|
70
|
+
end
|
71
|
+
values["last-modified"] = path.mtime
|
72
|
+
|
73
|
+
values.each do |key, value|
|
74
|
+
document[key] = value
|
75
|
+
end
|
76
|
+
end
|
77
|
+
end
|
78
|
+
end
|
79
|
+
end
|
@@ -0,0 +1,182 @@
|
|
1
|
+
# -*- mode: ruby; coding: utf-8 -*-
|
2
|
+
|
3
|
+
base_dir = File.join(File.dirname(__FILE__), "..", "..")
|
4
|
+
$LOAD_PATH.unshift(File.join(base_dir, "ext"))
|
5
|
+
$LOAD_PATH.unshift(File.join(base_dir, "lib"))
|
6
|
+
|
7
|
+
require 'rubygems'
|
8
|
+
require 'rack'
|
9
|
+
require 'groonga'
|
10
|
+
|
11
|
+
use Rack::CommonLogger
|
12
|
+
use Rack::Static, :urls => ["/css", "/images"], :root => "public"
|
13
|
+
|
14
|
+
Groonga::Database.new("data/database")
|
15
|
+
|
16
|
+
class Searcher
|
17
|
+
include Rack::Utils
|
18
|
+
|
19
|
+
def initialize
|
20
|
+
@documents = Groonga::Context.default["documents"]
|
21
|
+
end
|
22
|
+
|
23
|
+
def call(env)
|
24
|
+
request = Rack::Request.new(env)
|
25
|
+
response = Rack::Response.new
|
26
|
+
response["Content-Type"] = "text/html"
|
27
|
+
|
28
|
+
response.write(<<-EOH)
|
29
|
+
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
|
30
|
+
"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
|
31
|
+
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="ja" lang="ja">
|
32
|
+
<head>
|
33
|
+
<meta http-equiv="content-type" content="text/html;charset=UTF-8" />
|
34
|
+
<meta name="robot" content="noindex,nofollow" />
|
35
|
+
<title>groongaで全文検索</title>
|
36
|
+
<link rel="stylesheet" href="css/groonga.css" type="text/css" media="all" />
|
37
|
+
</head>
|
38
|
+
<body>
|
39
|
+
<div class="header">
|
40
|
+
<h1>groongaで全文検索</h1>
|
41
|
+
</div>
|
42
|
+
|
43
|
+
<div class="content">
|
44
|
+
EOH
|
45
|
+
|
46
|
+
render_search_box(request, response)
|
47
|
+
render_search_result(request, response)
|
48
|
+
|
49
|
+
response.write(<<-EOF)
|
50
|
+
</div>
|
51
|
+
|
52
|
+
<div class="footer">
|
53
|
+
<p class="powered-by-groonga">
|
54
|
+
<a href="http://groonga.org/">groonga</a>
|
55
|
+
#{Groonga::VERSION.join('.')}を使っています。
|
56
|
+
</p>
|
57
|
+
<p class="powered-by-ruby-groonga">
|
58
|
+
<a href="http://groonga.rubyforge.org/">Ruby/groonga</a>
|
59
|
+
#{Groonga::BINDINGS_VERSION.join('.')}を使っています。
|
60
|
+
</p>
|
61
|
+
</div>
|
62
|
+
|
63
|
+
</body>
|
64
|
+
</html>
|
65
|
+
EOF
|
66
|
+
response.to_a
|
67
|
+
end
|
68
|
+
|
69
|
+
private
|
70
|
+
def query(request)
|
71
|
+
request['query'] || ''
|
72
|
+
end
|
73
|
+
|
74
|
+
def words(request)
|
75
|
+
query(request).split
|
76
|
+
end
|
77
|
+
|
78
|
+
def render_search_box(request, response)
|
79
|
+
response.write(<<-EOF)
|
80
|
+
<form method="get" action="./">
|
81
|
+
<p>
|
82
|
+
<a href="."><img src="images/mini-groonga.png" alt="groonga" /></a>
|
83
|
+
<input name="query" type="text" value="#{escape_html(query(request))}" />
|
84
|
+
<input type="submit" value="検索" />
|
85
|
+
</p>
|
86
|
+
</form>
|
87
|
+
EOF
|
88
|
+
end
|
89
|
+
|
90
|
+
def render_search_result(request, response)
|
91
|
+
_words = words(request)
|
92
|
+
if _words.empty?
|
93
|
+
records = []
|
94
|
+
response.write(<<-EOS)
|
95
|
+
<div class='search-summary'>
|
96
|
+
<p>Rubyでgroonga使って全文検索</p>
|
97
|
+
</div>
|
98
|
+
EOS
|
99
|
+
else
|
100
|
+
offset = 0
|
101
|
+
options = {}
|
102
|
+
before = Time.now
|
103
|
+
records = @documents.select do |record|
|
104
|
+
expression = nil
|
105
|
+
_words.each do |word|
|
106
|
+
sub_expression = record["content"] =~ word
|
107
|
+
if expression.nil?
|
108
|
+
expression = sub_expression
|
109
|
+
else
|
110
|
+
expression &= sub_expression
|
111
|
+
end
|
112
|
+
end
|
113
|
+
expression
|
114
|
+
end
|
115
|
+
total_records = records.size
|
116
|
+
records = records.sort([[".:score", "descending"],
|
117
|
+
[".last-modified", "descending"]],
|
118
|
+
:limit => 20)
|
119
|
+
elapsed = Time.now - before
|
120
|
+
|
121
|
+
response.write(<<-EOS)
|
122
|
+
<div class='search-summary'>
|
123
|
+
<p>
|
124
|
+
<span class="keyword">#{escape_html(query(request))}</span>の検索結果:
|
125
|
+
<span class="total-entries">#{total_records}</span>件中
|
126
|
+
<span class="display-range">
|
127
|
+
#{total_records.zero? ? 0 : offset + 1}
|
128
|
+
-
|
129
|
+
#{offset + records.size}
|
130
|
+
</span>
|
131
|
+
件(#{elapsed}秒)
|
132
|
+
</p>
|
133
|
+
</div>
|
134
|
+
EOS
|
135
|
+
end
|
136
|
+
|
137
|
+
response.write(" <div class='records'>\n")
|
138
|
+
records.each do |record|
|
139
|
+
render_record(request, response, record)
|
140
|
+
end
|
141
|
+
response.write(" </div>\n")
|
142
|
+
end
|
143
|
+
|
144
|
+
def render_record(request, response, record)
|
145
|
+
response.write(" <div class='record'>\n")
|
146
|
+
href = escape_html(record['.path'])
|
147
|
+
title = escape_html(record['.title'])
|
148
|
+
last_modified = escape_html(record['.last-modified'].iso8601)
|
149
|
+
score = record.score
|
150
|
+
response.write(" <h2><a href='#{href}'>#{title}</a>(#{score})</h2>\n")
|
151
|
+
render_snippet(request, response, record)
|
152
|
+
response.write(<<-EOM)
|
153
|
+
<p class="metadata">
|
154
|
+
<span class="url">#{unescape(href)}</span>
|
155
|
+
-
|
156
|
+
<span class="last-modified">#{last_modified}</span>
|
157
|
+
</p>
|
158
|
+
EOM
|
159
|
+
response.write(" </div>\n")
|
160
|
+
end
|
161
|
+
|
162
|
+
def render_snippet(request, response, record)
|
163
|
+
open_tag = "<span class=\"keyword\">"
|
164
|
+
close_tag = "</span>"
|
165
|
+
snippet = Groonga::Snippet.new(:width => 100,
|
166
|
+
:default_open_tag => open_tag,
|
167
|
+
:default_close_tag => close_tag,
|
168
|
+
:html_escape => true,
|
169
|
+
:normalize => true)
|
170
|
+
words(request).each do |word|
|
171
|
+
snippet.add_keyword(word)
|
172
|
+
end
|
173
|
+
separator = "\n<span class='separator'>...</span>\n"
|
174
|
+
response.write(<<-EOS)
|
175
|
+
<p class="snippet">
|
176
|
+
#{snippet.execute(record[".content"]).join(separator)}
|
177
|
+
</p>
|
178
|
+
EOS
|
179
|
+
end
|
180
|
+
end
|
181
|
+
|
182
|
+
run Searcher.new
|
@@ -0,0 +1,122 @@
|
|
1
|
+
body
|
2
|
+
{
|
3
|
+
background-color: #fff;
|
4
|
+
color: #000;
|
5
|
+
}
|
6
|
+
|
7
|
+
div.header
|
8
|
+
{
|
9
|
+
display: none;
|
10
|
+
}
|
11
|
+
|
12
|
+
div.search-box
|
13
|
+
{
|
14
|
+
width: 100%;
|
15
|
+
text-align: center;
|
16
|
+
}
|
17
|
+
|
18
|
+
div.re-search
|
19
|
+
{
|
20
|
+
}
|
21
|
+
|
22
|
+
div.search-summary
|
23
|
+
{
|
24
|
+
font-size: small;
|
25
|
+
border-top: 1px solid #fc3;
|
26
|
+
background-color: #ffd;
|
27
|
+
}
|
28
|
+
|
29
|
+
div.search-summary p
|
30
|
+
{
|
31
|
+
padding: 3px;
|
32
|
+
margin: 0;
|
33
|
+
}
|
34
|
+
|
35
|
+
div.records
|
36
|
+
{
|
37
|
+
font-size: smaller;
|
38
|
+
padding-top: 10px;
|
39
|
+
padding-left: 30px;
|
40
|
+
margin-right: 20%;
|
41
|
+
margin-bottom: 30px;
|
42
|
+
}
|
43
|
+
|
44
|
+
div.records h2
|
45
|
+
{
|
46
|
+
padding: 0;
|
47
|
+
margin-bottom: 5px;
|
48
|
+
}
|
49
|
+
|
50
|
+
div.records p
|
51
|
+
{
|
52
|
+
}
|
53
|
+
|
54
|
+
p.snippet
|
55
|
+
{
|
56
|
+
padding: 0;
|
57
|
+
margin: 0;
|
58
|
+
}
|
59
|
+
|
60
|
+
p.snippet span.separator
|
61
|
+
{
|
62
|
+
font-weight: bold;
|
63
|
+
}
|
64
|
+
|
65
|
+
span.keyword,
|
66
|
+
span.total-entries,
|
67
|
+
span.display-range
|
68
|
+
{
|
69
|
+
font-weight: bold;
|
70
|
+
}
|
71
|
+
|
72
|
+
div.pagination
|
73
|
+
{
|
74
|
+
text-align: center;
|
75
|
+
margin-bottom: 20px;
|
76
|
+
}
|
77
|
+
|
78
|
+
p.metadata
|
79
|
+
{
|
80
|
+
padding: 0;
|
81
|
+
margin: 0;
|
82
|
+
margin-top: 5px;
|
83
|
+
}
|
84
|
+
|
85
|
+
span.url
|
86
|
+
{
|
87
|
+
color: green;
|
88
|
+
}
|
89
|
+
|
90
|
+
pre
|
91
|
+
{
|
92
|
+
overflow: auto;
|
93
|
+
margin: 20px 30px;
|
94
|
+
padding: 10px;
|
95
|
+
border-top: 1px solid #fc3;
|
96
|
+
background-color: #ffd;
|
97
|
+
}
|
98
|
+
|
99
|
+
a img
|
100
|
+
{
|
101
|
+
border-style: none;
|
102
|
+
}
|
103
|
+
|
104
|
+
ul.sources li
|
105
|
+
{
|
106
|
+
display: inline;
|
107
|
+
}
|
108
|
+
|
109
|
+
div.footer
|
110
|
+
{
|
111
|
+
border-top: 1px solid #fc3;
|
112
|
+
background-color: #ffd;
|
113
|
+
margin-top: 5px;
|
114
|
+
text-align: right;
|
115
|
+
}
|
116
|
+
|
117
|
+
div.footer p
|
118
|
+
{
|
119
|
+
font-size: small;
|
120
|
+
padding: 0;
|
121
|
+
margin: 0;
|
122
|
+
}
|