mongoid-fts 1.1.1 → 2.0.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +8 -8
- data/README.md +33 -16
- data/lib/mongoid-fts.rb +97 -642
- data/lib/mongoid-fts/able.rb +68 -0
- data/lib/mongoid-fts/error.rb +5 -0
- data/lib/mongoid-fts/index.rb +293 -0
- data/lib/mongoid-fts/rails.rb +11 -0
- data/lib/mongoid-fts/raw.rb +16 -0
- data/lib/mongoid-fts/results.rb +195 -0
- data/lib/mongoid-fts/stemming.rb +81 -0
- data/lib/mongoid-fts/stemming/stopwords/english.txt +32 -0
- data/lib/mongoid-fts/stemming/stopwords/extended_english.txt +216 -0
- data/lib/mongoid-fts/stemming/stopwords/full_danish.txt +94 -0
- data/lib/mongoid-fts/stemming/stopwords/full_dutch.txt +101 -0
- data/lib/mongoid-fts/stemming/stopwords/full_english.txt +174 -0
- data/lib/mongoid-fts/stemming/stopwords/full_finnish.txt +0 -0
- data/lib/mongoid-fts/stemming/stopwords/full_french.txt +155 -0
- data/lib/mongoid-fts/stemming/stopwords/full_german.txt +231 -0
- data/lib/mongoid-fts/stemming/stopwords/full_italian.txt +279 -0
- data/lib/mongoid-fts/stemming/stopwords/full_norwegian.txt +176 -0
- data/lib/mongoid-fts/stemming/stopwords/full_portuguese.txt +203 -0
- data/lib/mongoid-fts/stemming/stopwords/full_russian.txt +101 -0
- data/lib/mongoid-fts/stemming/stopwords/full_russiankoi8_r.txt +101 -0
- data/lib/mongoid-fts/stemming/stopwords/full_spanish.txt +313 -0
- data/lib/mongoid-fts/util.rb +322 -0
- data/mongoid-fts.gemspec +37 -3
- data/test/helper.rb +44 -0
- data/test/mongoid-fts_test.rb +177 -0
- data/test/testing.rb +196 -0
- metadata +69 -2
@@ -0,0 +1,68 @@
|
|
1
|
+
module Mongoid
|
2
|
+
module FTS
|
3
|
+
module Able
|
4
|
+
def Able.code
|
5
|
+
@code ||= proc do
|
6
|
+
class << self
|
7
|
+
def fts_search(*args, &block)
|
8
|
+
options = Map.options_for!(args)
|
9
|
+
|
10
|
+
options[:model] = self
|
11
|
+
|
12
|
+
args.push(options)
|
13
|
+
|
14
|
+
FTS.search(*args, &block)
|
15
|
+
end
|
16
|
+
alias :search :fts_search
|
17
|
+
|
18
|
+
def _fts_search(*args, &block)
|
19
|
+
options = Map.options_for!(args)
|
20
|
+
|
21
|
+
options[:model] = self
|
22
|
+
|
23
|
+
args.push(options)
|
24
|
+
|
25
|
+
FTS.search(*args, &block)
|
26
|
+
end
|
27
|
+
alias :_search :_fts_search
|
28
|
+
end
|
29
|
+
|
30
|
+
after_save do |model|
|
31
|
+
FTS::Index.add(model)
|
32
|
+
end
|
33
|
+
|
34
|
+
after_destroy do |model|
|
35
|
+
FTS::Index.remove(model) rescue nil
|
36
|
+
end
|
37
|
+
|
38
|
+
has_one(:fts_index, :as => :context, :class_name => '::Mongoid::FTS::Index')
|
39
|
+
end
|
40
|
+
end
|
41
|
+
|
42
|
+
def Able.included(other)
|
43
|
+
unless other.is_a?(Able)
|
44
|
+
begin
|
45
|
+
super
|
46
|
+
ensure
|
47
|
+
other.module_eval(&Able.code)
|
48
|
+
|
49
|
+
FTS.models.dup.each do |model|
|
50
|
+
FTS.models.delete(model) if model.name == other.name
|
51
|
+
end
|
52
|
+
|
53
|
+
FTS.models.push(other)
|
54
|
+
FTS.models.uniq!
|
55
|
+
end
|
56
|
+
end
|
57
|
+
end
|
58
|
+
end
|
59
|
+
|
60
|
+
def FTS.included(other)
|
61
|
+
other.send(:include, FTS::Able)
|
62
|
+
end
|
63
|
+
|
64
|
+
def FTS.able
|
65
|
+
Able
|
66
|
+
end
|
67
|
+
end
|
68
|
+
end
|
@@ -0,0 +1,293 @@
|
|
1
|
+
module Mongoid
|
2
|
+
module FTS
|
3
|
+
class Index
|
4
|
+
#
|
5
|
+
include Mongoid::Document
|
6
|
+
|
7
|
+
#
|
8
|
+
belongs_to(:context, :polymorphic => true)
|
9
|
+
|
10
|
+
#
|
11
|
+
field(:literals, :type => Array)
|
12
|
+
|
13
|
+
field(:title, :type => Array)
|
14
|
+
|
15
|
+
field(:keywords, :type => Array)
|
16
|
+
|
17
|
+
field(:fuzzy, :type => Array)
|
18
|
+
|
19
|
+
field(:fulltext, :type => Array)
|
20
|
+
|
21
|
+
#
|
22
|
+
index(
|
23
|
+
{
|
24
|
+
:context_type => 1,
|
25
|
+
:context_id => 1
|
26
|
+
},
|
27
|
+
|
28
|
+
{
|
29
|
+
:unique => true,
|
30
|
+
:sparse => true
|
31
|
+
}
|
32
|
+
)
|
33
|
+
|
34
|
+
# FIXME - this hack gets around https://github.com/mongoid/mongoid/issues/3080
|
35
|
+
#
|
36
|
+
index_options[
|
37
|
+
|
38
|
+
normalize_spec(
|
39
|
+
:context_type => 1,
|
40
|
+
|
41
|
+
:literals => 'text',
|
42
|
+
|
43
|
+
:title => 'text',
|
44
|
+
|
45
|
+
:keywords => 'text',
|
46
|
+
|
47
|
+
:fuzzy => 'text',
|
48
|
+
|
49
|
+
:fulltext => 'text'
|
50
|
+
)
|
51
|
+
|
52
|
+
] =
|
53
|
+
|
54
|
+
normalize_index_options(
|
55
|
+
:name => 'search_index',
|
56
|
+
|
57
|
+
:default_language => 'none',
|
58
|
+
|
59
|
+
:weights => {
|
60
|
+
:literals => 200,
|
61
|
+
|
62
|
+
:title => 90,
|
63
|
+
|
64
|
+
:keywords => 50,
|
65
|
+
|
66
|
+
:fulltext => 1
|
67
|
+
}
|
68
|
+
)
|
69
|
+
#
|
70
|
+
#
|
71
|
+
|
72
|
+
before_validation do |index|
|
73
|
+
index.normalize
|
74
|
+
end
|
75
|
+
|
76
|
+
before_upsert do |index|
|
77
|
+
index.normalize
|
78
|
+
end
|
79
|
+
|
80
|
+
before_save do |index|
|
81
|
+
index.normalize
|
82
|
+
end
|
83
|
+
|
84
|
+
validates_presence_of(:context_type)
|
85
|
+
|
86
|
+
def normalize
|
87
|
+
if !defined?(@normalized) or !@normalized
|
88
|
+
normalize!
|
89
|
+
end
|
90
|
+
end
|
91
|
+
|
92
|
+
def normalize!
|
93
|
+
index = self
|
94
|
+
|
95
|
+
%w( literals title keywords fulltext ).each do |attr|
|
96
|
+
index[attr] = FTS.list_of_strings(index[attr])
|
97
|
+
end
|
98
|
+
ensure
|
99
|
+
@normalized = true
|
100
|
+
end
|
101
|
+
|
102
|
+
def inspect(*args, &block)
|
103
|
+
Map.for(as_document).inspect(*args, &block)
|
104
|
+
end
|
105
|
+
|
106
|
+
def Index.teardown!
|
107
|
+
Index.remove_indexes
|
108
|
+
Index.destroy_all
|
109
|
+
end
|
110
|
+
|
111
|
+
def Index.setup!
|
112
|
+
Index.create_indexes
|
113
|
+
end
|
114
|
+
|
115
|
+
def Index.reset!
|
116
|
+
teardown!
|
117
|
+
setup!
|
118
|
+
end
|
119
|
+
|
120
|
+
def Index.rebuild!
|
121
|
+
batches = Hash.new{|h,k| h[k] = []}
|
122
|
+
|
123
|
+
each do |index|
|
124
|
+
context_type, context_id = index.context_type, index.context_id
|
125
|
+
next unless context_type && context_id
|
126
|
+
(batches[context_type] ||= []).push(context_id)
|
127
|
+
end
|
128
|
+
|
129
|
+
models = FTS.find_in_batches(batches)
|
130
|
+
|
131
|
+
reset!
|
132
|
+
|
133
|
+
models.each{|model| add(model)}
|
134
|
+
end
|
135
|
+
|
136
|
+
def Index.add!(model)
|
137
|
+
to_search = Index.to_search(model)
|
138
|
+
|
139
|
+
literals = to_search.has_key?(:literals) ? Coerce.list_of_strings(to_search[:literals]) : nil
|
140
|
+
|
141
|
+
title = to_search.has_key?(:title) ? Coerce.string(to_search[:title]) : nil
|
142
|
+
|
143
|
+
keywords = to_search.has_key?(:keywords) ? Coerce.list_of_strings(to_search[:keywords]) : nil
|
144
|
+
|
145
|
+
fuzzy = to_search.has_key?(:fuzzy) ? Coerce.list_of_strings(to_search[:fuzzy]) : nil
|
146
|
+
|
147
|
+
fulltext = to_search.has_key?(:fulltext) ? Coerce.string(to_search[:fulltext]) : nil
|
148
|
+
|
149
|
+
context_type = model.class.name.to_s
|
150
|
+
context_id = model.id
|
151
|
+
|
152
|
+
conditions = {
|
153
|
+
:context_type => context_type,
|
154
|
+
:context_id => context_id
|
155
|
+
}
|
156
|
+
|
157
|
+
attributes = {
|
158
|
+
:literals => literals,
|
159
|
+
:title => title,
|
160
|
+
:keywords => keywords,
|
161
|
+
:fuzzy => fuzzy,
|
162
|
+
:fulltext => fulltext
|
163
|
+
}
|
164
|
+
|
165
|
+
index = nil
|
166
|
+
n = 42
|
167
|
+
|
168
|
+
n.times do |i|
|
169
|
+
index = where(conditions).first
|
170
|
+
break if index
|
171
|
+
|
172
|
+
begin
|
173
|
+
index = create!(conditions)
|
174
|
+
break if index
|
175
|
+
rescue Object
|
176
|
+
nil
|
177
|
+
end
|
178
|
+
|
179
|
+
sleep(rand) if i < (n - 1)
|
180
|
+
end
|
181
|
+
|
182
|
+
if index
|
183
|
+
begin
|
184
|
+
index.update_attributes!(attributes)
|
185
|
+
rescue Object
|
186
|
+
raise Error.new("failed to update index for #{ conditions.inspect }")
|
187
|
+
end
|
188
|
+
else
|
189
|
+
raise Error.new("failed to create index for #{ conditions.inspect }")
|
190
|
+
end
|
191
|
+
|
192
|
+
index
|
193
|
+
end
|
194
|
+
|
195
|
+
def Index.add(*args, &block)
|
196
|
+
add!(*args, &block)
|
197
|
+
end
|
198
|
+
|
199
|
+
def Index.remove!(*args, &block)
|
200
|
+
options = args.extract_options!.to_options!
|
201
|
+
models = args.flatten.compact
|
202
|
+
|
203
|
+
model_ids = {}
|
204
|
+
|
205
|
+
models.each do |model|
|
206
|
+
model_name = model.class.name.to_s
|
207
|
+
model_ids[model_name] ||= []
|
208
|
+
model_ids[model_name].push(model.id)
|
209
|
+
end
|
210
|
+
|
211
|
+
conditions = model_ids.map do |model_name, model_ids|
|
212
|
+
{:context_type => model_name, :context_id.in => model_ids}
|
213
|
+
end
|
214
|
+
|
215
|
+
any_of(conditions).destroy_all
|
216
|
+
end
|
217
|
+
|
218
|
+
def Index.remove(*args, &block)
|
219
|
+
remove!(*args, &block)
|
220
|
+
end
|
221
|
+
|
222
|
+
def Index.to_search(model)
|
223
|
+
#
|
224
|
+
to_search = nil
|
225
|
+
|
226
|
+
#
|
227
|
+
if model.respond_to?(:to_search)
|
228
|
+
to_search = Map.for(model.to_search)
|
229
|
+
else
|
230
|
+
to_search = Map.new
|
231
|
+
|
232
|
+
to_search[:literals] =
|
233
|
+
%w( id ).map do |attr|
|
234
|
+
model.send(attr) if model.respond_to?(attr)
|
235
|
+
end
|
236
|
+
|
237
|
+
to_search[:title] =
|
238
|
+
%w( title ).map do |attr|
|
239
|
+
model.send(attr) if model.respond_to?(attr)
|
240
|
+
end
|
241
|
+
|
242
|
+
to_search[:keywords] =
|
243
|
+
%w( keywords tags ).map do |attr|
|
244
|
+
model.send(attr) if model.respond_to?(attr)
|
245
|
+
end
|
246
|
+
|
247
|
+
to_search[:fulltext] =
|
248
|
+
%w( fulltext text content body description ).map do |attr|
|
249
|
+
model.send(attr) if model.respond_to?(attr)
|
250
|
+
end
|
251
|
+
end
|
252
|
+
|
253
|
+
#
|
254
|
+
required = %w( literals title keywords fuzzy fulltext )
|
255
|
+
actual = to_search.keys
|
256
|
+
|
257
|
+
=begin
|
258
|
+
missing = required - actual
|
259
|
+
unless missing.empty?
|
260
|
+
raise ArgumentError, "#{ model.class.inspect }#to_search missing keys #{ missing.inspect }"
|
261
|
+
end
|
262
|
+
=end
|
263
|
+
|
264
|
+
invalid = actual - required
|
265
|
+
unless invalid.empty?
|
266
|
+
raise ArgumentError, "#{ model.class.inspect }#to_search invalid keys #{ invalid.inspect }"
|
267
|
+
end
|
268
|
+
|
269
|
+
#
|
270
|
+
literals = FTS.normalized_array(to_search[:literals])
|
271
|
+
title = FTS.normalized_array(to_search[:title])
|
272
|
+
keywords = FTS.normalized_array(to_search[:keywords])
|
273
|
+
fuzzy = FTS.normalized_array(to_search[:fuzzy])
|
274
|
+
fulltext = FTS.normalized_array(to_search[:fulltext])
|
275
|
+
|
276
|
+
#
|
277
|
+
if to_search[:fuzzy].nil?
|
278
|
+
fuzzy = [title, keywords]
|
279
|
+
end
|
280
|
+
|
281
|
+
#
|
282
|
+
to_search[:literals] = FTS.literals_for(literals).uniq
|
283
|
+
to_search[:title] = (title + FTS.terms_for(title)).uniq
|
284
|
+
to_search[:keywords] = (keywords + FTS.terms_for(keywords)).uniq
|
285
|
+
to_search[:fuzzy] = FTS.fuzzy_for(fuzzy).uniq
|
286
|
+
to_search[:fulltext] = (FTS.terms_for(fulltext, :subterms => true)).uniq
|
287
|
+
|
288
|
+
#
|
289
|
+
to_search
|
290
|
+
end
|
291
|
+
end
|
292
|
+
end
|
293
|
+
end
|
@@ -0,0 +1,16 @@
|
|
1
|
+
module Mongoid
|
2
|
+
module FTS
|
3
|
+
class Raw < ::Array
|
4
|
+
attr_accessor :_search
|
5
|
+
attr_accessor :_text
|
6
|
+
attr_accessor :_limit
|
7
|
+
attr_accessor :_models
|
8
|
+
|
9
|
+
def initialize(_searches, options = {})
|
10
|
+
replace(_searches)
|
11
|
+
ensure
|
12
|
+
options.each{|k, v| send("#{ k }=", v)}
|
13
|
+
end
|
14
|
+
end
|
15
|
+
end
|
16
|
+
end
|
@@ -0,0 +1,195 @@
|
|
1
|
+
module Mongoid
|
2
|
+
module FTS
|
3
|
+
class Results < ::Array
|
4
|
+
attr_accessor :_searches
|
5
|
+
attr_accessor :_models
|
6
|
+
|
7
|
+
def initialize(_searches)
|
8
|
+
@_searches = _searches
|
9
|
+
@_models = []
|
10
|
+
_denormalize!
|
11
|
+
@page = 1
|
12
|
+
@per = size
|
13
|
+
@num_pages = 1
|
14
|
+
end
|
15
|
+
|
16
|
+
def paginate(*args)
|
17
|
+
results = self
|
18
|
+
options = Map.options_for!(args)
|
19
|
+
|
20
|
+
page = Integer(args.shift || options[:page] || @page)
|
21
|
+
per = args.shift || options[:per] || options[:size]
|
22
|
+
|
23
|
+
if per.nil?
|
24
|
+
return Promise.new(results, page)
|
25
|
+
else
|
26
|
+
per = Integer(per)
|
27
|
+
end
|
28
|
+
|
29
|
+
@page = [page.abs, 1].max
|
30
|
+
@per = [per.abs, 1].max
|
31
|
+
@num_pages = (size.to_f / @per).ceil
|
32
|
+
|
33
|
+
offset = (@page - 1) * @per
|
34
|
+
length = @per
|
35
|
+
|
36
|
+
slice = Array(@_models[offset, length])
|
37
|
+
|
38
|
+
replace(slice)
|
39
|
+
|
40
|
+
self
|
41
|
+
end
|
42
|
+
|
43
|
+
class Promise
|
44
|
+
attr_accessor :results
|
45
|
+
attr_accessor :page
|
46
|
+
|
47
|
+
def initialize(results, page)
|
48
|
+
@results = results
|
49
|
+
@page = page
|
50
|
+
end
|
51
|
+
|
52
|
+
def per(per)
|
53
|
+
results.per(:page => page, :per => per)
|
54
|
+
end
|
55
|
+
end
|
56
|
+
|
57
|
+
def page(*args)
|
58
|
+
if args.empty?
|
59
|
+
return @page
|
60
|
+
else
|
61
|
+
options = Map.options_for!(args)
|
62
|
+
page = args.shift || options[:page]
|
63
|
+
options[:page] = page
|
64
|
+
paginate(options)
|
65
|
+
end
|
66
|
+
end
|
67
|
+
|
68
|
+
alias_method(:current_page, :page)
|
69
|
+
|
70
|
+
def per(*args)
|
71
|
+
if args.empty?
|
72
|
+
return @per
|
73
|
+
else
|
74
|
+
options = Map.options_for!(args)
|
75
|
+
per = args.shift || options[:per]
|
76
|
+
options[:per] = per
|
77
|
+
paginate(options)
|
78
|
+
end
|
79
|
+
end
|
80
|
+
|
81
|
+
def num_pages
|
82
|
+
@num_pages
|
83
|
+
end
|
84
|
+
|
85
|
+
def total_pages
|
86
|
+
num_pages
|
87
|
+
end
|
88
|
+
|
89
|
+
# TODO - text sorting more...
|
90
|
+
#
|
91
|
+
def _denormalize!
|
92
|
+
#
|
93
|
+
collection = self
|
94
|
+
|
95
|
+
collection.clear
|
96
|
+
@_models = []
|
97
|
+
|
98
|
+
return self if @_searches.empty?
|
99
|
+
|
100
|
+
#
|
101
|
+
_models = @_searches._models
|
102
|
+
|
103
|
+
_position = proc do |model|
|
104
|
+
_models.index(model) or raise("no position for #{ model.inspect }!?")
|
105
|
+
end
|
106
|
+
|
107
|
+
results =
|
108
|
+
@_searches.map do |_search|
|
109
|
+
_search['results'] ||= []
|
110
|
+
|
111
|
+
_search['results'].each do |result|
|
112
|
+
result['_model'] = _search._model
|
113
|
+
result['_position'] = _position[_search._model]
|
114
|
+
end
|
115
|
+
|
116
|
+
_search['results']
|
117
|
+
end
|
118
|
+
|
119
|
+
results.flatten!
|
120
|
+
results.compact!
|
121
|
+
|
122
|
+
=begin
|
123
|
+
results.sort! do |a, b|
|
124
|
+
score = Float(b['score']) <=> Float(a['score'])
|
125
|
+
|
126
|
+
case score
|
127
|
+
when 0
|
128
|
+
a['_position'] <=> b['_position']
|
129
|
+
else
|
130
|
+
score
|
131
|
+
end
|
132
|
+
end
|
133
|
+
=end
|
134
|
+
|
135
|
+
#
|
136
|
+
batches = Hash.new{|h,k| h[k] = []}
|
137
|
+
|
138
|
+
results.each do |entry|
|
139
|
+
obj = entry['obj']
|
140
|
+
|
141
|
+
context_type, context_id = obj['context_type'], obj['context_id']
|
142
|
+
|
143
|
+
batches[context_type].push(context_id)
|
144
|
+
end
|
145
|
+
|
146
|
+
#
|
147
|
+
models = FTS.find_in_batches(batches)
|
148
|
+
|
149
|
+
#
|
150
|
+
result_index = {}
|
151
|
+
|
152
|
+
results.each do |result|
|
153
|
+
context_type = result['obj']['context_type'].to_s
|
154
|
+
context_id = result['obj']['context_id'].to_s
|
155
|
+
key = [context_type, context_id]
|
156
|
+
|
157
|
+
result_index[key] = result
|
158
|
+
end
|
159
|
+
|
160
|
+
#
|
161
|
+
models.each do |model|
|
162
|
+
context_type = model.class.name.to_s
|
163
|
+
context_id = model.id.to_s
|
164
|
+
key = [context_type, context_id]
|
165
|
+
|
166
|
+
result = result_index[key]
|
167
|
+
model['_fts_index'] = result
|
168
|
+
end
|
169
|
+
|
170
|
+
#
|
171
|
+
models.sort! do |model_a, model_b|
|
172
|
+
a = model_a['_fts_index']
|
173
|
+
b = model_b['_fts_index']
|
174
|
+
|
175
|
+
score = Float(b['score']) <=> Float(a['score'])
|
176
|
+
|
177
|
+
case score
|
178
|
+
when 0
|
179
|
+
a['_position'] <=> b['_position']
|
180
|
+
else
|
181
|
+
score
|
182
|
+
end
|
183
|
+
end
|
184
|
+
|
185
|
+
#
|
186
|
+
limit = @_searches._limit
|
187
|
+
|
188
|
+
#
|
189
|
+
replace(@_models = models[0 ... limit])
|
190
|
+
|
191
|
+
self
|
192
|
+
end
|
193
|
+
end
|
194
|
+
end
|
195
|
+
end
|