lunar 0.5.5 → 0.6.0
Sign up to get free protection for your applications and to get access to all the features.
- data/VERSION +1 -1
- data/lib/lunar.rb +12 -5
- data/lib/lunar/fuzzy_matches.rb +3 -1
- data/lib/lunar/index.rb +93 -40
- data/lib/lunar/keyword_matches.rb +4 -2
- data/lib/lunar/number_matches.rb +33 -0
- data/lib/lunar/scoring.rb +3 -3
- data/lib/lunar/stopwords.rb +3 -3
- data/lib/lunar/words.rb +30 -5
- data/lunar.gemspec +9 -7
- data/test/test_index.rb +64 -9
- data/test/test_lunar_words.rb +14 -0
- data/test/test_number_search.rb +69 -0
- data/test/test_scoring.rb +1 -1
- metadata +16 -8
- data/lib/lunar/lunar_nest.rb +0 -19
- data/test/test_lunar_nest.rb +0 -46
data/VERSION
CHANGED
@@ -1 +1 @@
|
|
1
|
-
0.
|
1
|
+
0.6.0
|
data/lib/lunar.rb
CHANGED
@@ -4,15 +4,15 @@ require 'nest'
|
|
4
4
|
require 'text'
|
5
5
|
|
6
6
|
module Lunar
|
7
|
-
VERSION = '0.
|
7
|
+
VERSION = '0.6.0'
|
8
8
|
|
9
9
|
autoload :Connection, "lunar/connection"
|
10
|
-
autoload :LunarNest, "lunar/lunar_nest"
|
11
10
|
autoload :Index, "lunar/index"
|
12
11
|
autoload :Scoring, "lunar/scoring"
|
13
12
|
autoload :Words, "lunar/words"
|
14
13
|
autoload :Stopwords, "lunar/stopwords"
|
15
14
|
autoload :FuzzyWord, "lunar/fuzzy_word"
|
15
|
+
autoload :NumberMatches, "lunar/number_matches"
|
16
16
|
autoload :KeywordMatches, "lunar/keyword_matches"
|
17
17
|
autoload :RangeMatches, "lunar/range_matches"
|
18
18
|
autoload :FuzzyMatches, "lunar/fuzzy_matches"
|
@@ -36,7 +36,7 @@ module Lunar
|
|
36
36
|
# end
|
37
37
|
#
|
38
38
|
# @see Lunar::Index#initialize
|
39
|
-
# @param [String, Symbol, Class] namespace the namespace
|
39
|
+
# @param [String, Symbol, Class] namespace the namespace of this document.
|
40
40
|
# @yield [Lunar::Index] an instance of Lunar::Index.
|
41
41
|
# @return [Lunar::Index] returns the yielded Lunar::Index.
|
42
42
|
def self.index(namespace)
|
@@ -95,12 +95,12 @@ module Lunar
|
|
95
95
|
|
96
96
|
# @private abstraction of how encoding should be done for Lunar.
|
97
97
|
def self.encode(word)
|
98
|
-
Base64.encode64(word).
|
98
|
+
Base64.encode64(word).gsub("\n", "")
|
99
99
|
end
|
100
100
|
|
101
101
|
# @private convenience method for getting a scoped Nest.
|
102
102
|
def self.nest
|
103
|
-
|
103
|
+
Nest.new(:Lunar, redis)
|
104
104
|
end
|
105
105
|
|
106
106
|
private
|
@@ -115,6 +115,13 @@ private
|
|
115
115
|
end
|
116
116
|
}
|
117
117
|
sets.push(*fuzzy_matches.compact)
|
118
|
+
elsif key == :numbers
|
119
|
+
number_matches = value.map { |num_key, num_value|
|
120
|
+
unless num_value.to_s.empty?
|
121
|
+
matches = NumberMatches.new(nest[namespace], num_key, num_value)
|
122
|
+
sets << matches.distkey if matches.distkey
|
123
|
+
end
|
124
|
+
}
|
118
125
|
else
|
119
126
|
unless value.to_s.empty?
|
120
127
|
sets << KeywordMatches.new(nest[namespace], key, value).distkey
|
data/lib/lunar/fuzzy_matches.rb
CHANGED
data/lib/lunar/index.rb
CHANGED
@@ -6,7 +6,27 @@ module Lunar
|
|
6
6
|
# @see Lunar::index
|
7
7
|
# @see Lunar::delete
|
8
8
|
class Index
|
9
|
+
# This constant is in place to maintain a certain level of performance.
|
10
|
+
# Fuzzy searching internally stores an index per letter i.e. for Quentin
|
11
|
+
# q
|
12
|
+
# qu
|
13
|
+
# que
|
14
|
+
# quen
|
15
|
+
# quent
|
16
|
+
# quenti
|
17
|
+
# quentin
|
18
|
+
#
|
19
|
+
# This can become pretty unweildy very fast, so we limit the length
|
20
|
+
# of all fuzzy fields.
|
9
21
|
FUZZY_MAX_LENGTH = 100
|
22
|
+
|
23
|
+
# The following are all used to construct redis keys
|
24
|
+
TEXT = :Text
|
25
|
+
METAPHONES = :Metaphones
|
26
|
+
NUMBERS = :Numbers
|
27
|
+
SORTABLES = :Sortables
|
28
|
+
FUZZIES = :Fuzzies
|
29
|
+
FIELDS = :Fields
|
10
30
|
|
11
31
|
MissingID = Class.new(StandardError)
|
12
32
|
FuzzyFieldTooLong = Class.new(StandardError)
|
@@ -16,6 +36,7 @@ module Lunar
|
|
16
36
|
attr :numbers
|
17
37
|
attr :sortables
|
18
38
|
attr :fuzzies
|
39
|
+
attr :fields
|
19
40
|
|
20
41
|
# This is actually wrapped by `Lunar.index` and is not inteded to be
|
21
42
|
# used directly.
|
@@ -24,10 +45,11 @@ module Lunar
|
|
24
45
|
# @return [Lunar::Index]
|
25
46
|
def initialize(namespace)
|
26
47
|
@nest = Lunar.nest[namespace]
|
27
|
-
@metaphones = @nest[
|
28
|
-
@numbers = @nest[
|
29
|
-
@sortables = @nest[
|
30
|
-
@fuzzies = @nest[
|
48
|
+
@metaphones = @nest[METAPHONES]
|
49
|
+
@numbers = @nest[NUMBERS]
|
50
|
+
@sortables = @nest[SORTABLES]
|
51
|
+
@fuzzies = @nest[FUZZIES]
|
52
|
+
@fields = @nest[FIELDS]
|
31
53
|
end
|
32
54
|
|
33
55
|
# Get / Set the id of the document
|
@@ -81,23 +103,16 @@ module Lunar
|
|
81
103
|
#
|
82
104
|
# @return [Array<String>] all the metaphones added for the document.
|
83
105
|
def text(att, value)
|
84
|
-
|
85
|
-
new = []
|
106
|
+
clear_text_field(att)
|
86
107
|
|
87
108
|
Scoring.new(value).scores.each do |word, score|
|
88
109
|
metaphone = Lunar.metaphone(word)
|
89
110
|
|
90
111
|
nest[att][metaphone].zadd(score, id)
|
91
112
|
metaphones[id][att].sadd(metaphone)
|
92
|
-
new << metaphone
|
93
113
|
end
|
94
114
|
|
95
|
-
(
|
96
|
-
nest[att][metaphone].zrem(id)
|
97
|
-
metaphones[id][att].srem(metaphone)
|
98
|
-
end
|
99
|
-
|
100
|
-
return new
|
115
|
+
fields[TEXT].sadd(att)
|
101
116
|
end
|
102
117
|
|
103
118
|
# Adds a numeric index for `att` with `value`.
|
@@ -117,8 +132,20 @@ module Lunar
|
|
117
132
|
# @param [Numeric] value the numeric value of `att`.
|
118
133
|
#
|
119
134
|
# @return [Boolean] whether or not the value was added
|
120
|
-
def number(att, value)
|
135
|
+
def number(att, value, purge = true)
|
136
|
+
if value.kind_of?(Enumerable)
|
137
|
+
clear_number_field(att)
|
138
|
+
|
139
|
+
value.each { |v| number(att, v, false) } and return
|
140
|
+
end
|
141
|
+
|
142
|
+
clear_number_field(att) if purge
|
143
|
+
|
121
144
|
numbers[att].zadd(value, id)
|
145
|
+
numbers[att][value].zadd(1, id)
|
146
|
+
numbers[id][att].sadd(value)
|
147
|
+
|
148
|
+
fields[NUMBERS].sadd att
|
122
149
|
end
|
123
150
|
|
124
151
|
# Adds a sortable index for `att` with `value`.
|
@@ -156,6 +183,8 @@ module Lunar
|
|
156
183
|
# @return [String] the response from the redis server.
|
157
184
|
def sortable(att, value)
|
158
185
|
sortables[id][att].set(value)
|
186
|
+
|
187
|
+
fields[SORTABLES].sadd att
|
159
188
|
end
|
160
189
|
|
161
190
|
# Deletes everything related to an existing document given its `id`.
|
@@ -168,63 +197,80 @@ module Lunar
|
|
168
197
|
delete_numbers
|
169
198
|
delete_sortables
|
170
199
|
delete_fuzzies
|
200
|
+
|
201
|
+
delete_field_meta
|
171
202
|
end
|
172
203
|
|
173
204
|
def fuzzy(att, value)
|
174
|
-
|
175
|
-
raise FuzzyFieldTooLong,
|
176
|
-
"#{att} has a value #{value} exceeding the max #{FUZZY_MAX_LENGTH}"
|
177
|
-
end
|
205
|
+
assert_valid_fuzzy(value, att)
|
178
206
|
|
179
|
-
|
207
|
+
clear_fuzzy_field(att)
|
208
|
+
|
209
|
+
words = Words.new(value, [:downcase, :uniq])
|
180
210
|
|
181
211
|
fuzzy_words_and_parts(words) do |word, parts|
|
182
|
-
parts.each
|
183
|
-
fuzzies[att][encoded].zadd(1, id)
|
184
|
-
end
|
212
|
+
parts.each { |part, encoded| fuzzies[att][encoded].zadd(1, id) }
|
185
213
|
fuzzies[id][att].sadd word
|
186
214
|
end
|
187
215
|
|
188
|
-
|
216
|
+
fields[FUZZIES].sadd att
|
189
217
|
end
|
190
218
|
|
191
219
|
private
|
220
|
+
def assert_valid_fuzzy(value, att)
|
221
|
+
if value.to_s.length > FUZZY_MAX_LENGTH
|
222
|
+
raise FuzzyFieldTooLong,
|
223
|
+
"#{att} has a value #{value} exceeding the max #{FUZZY_MAX_LENGTH}"
|
224
|
+
end
|
225
|
+
end
|
226
|
+
|
192
227
|
def delete_metaphones
|
193
|
-
|
194
|
-
|
195
|
-
|
196
|
-
|
228
|
+
fields[TEXT].smembers.each do |att|
|
229
|
+
clear_text_field(att)
|
230
|
+
end
|
231
|
+
end
|
197
232
|
|
198
|
-
|
233
|
+
def clear_text_field(att)
|
234
|
+
metaphones[id][att].smembers.each do |metaphone|
|
235
|
+
nest[att][metaphone].zrem id
|
199
236
|
end
|
237
|
+
|
238
|
+
metaphones[id][att].del
|
200
239
|
end
|
201
240
|
|
202
241
|
def delete_numbers
|
203
|
-
|
242
|
+
fields[NUMBERS].smembers.each do |att|
|
243
|
+
clear_number_field(att)
|
244
|
+
end
|
245
|
+
end
|
246
|
+
|
247
|
+
def clear_number_field(att)
|
248
|
+
numbers[id][att].smembers.each do |number|
|
249
|
+
numbers[att][number].zrem(id)
|
204
250
|
numbers[att].zrem(id)
|
205
251
|
end
|
252
|
+
|
253
|
+
numbers[id][att].del
|
206
254
|
end
|
207
255
|
|
256
|
+
|
208
257
|
def delete_sortables
|
209
|
-
|
210
|
-
Lunar.redis.del key
|
211
|
-
end
|
258
|
+
fields[SORTABLES].smembers.each { |att| sortables[id][att].del }
|
212
259
|
end
|
213
260
|
|
214
261
|
def delete_fuzzies
|
215
|
-
|
216
|
-
|
217
|
-
key.del
|
262
|
+
fields[FUZZIES].smembers.each do |att|
|
263
|
+
clear_fuzzy_field(att)
|
218
264
|
end
|
219
265
|
end
|
220
|
-
|
221
|
-
def
|
222
|
-
fuzzy_words_and_parts(
|
266
|
+
|
267
|
+
def clear_fuzzy_field(att)
|
268
|
+
fuzzy_words_and_parts(fuzzies[id][att].smembers) do |word, parts|
|
223
269
|
parts.each do |part, encoded|
|
224
|
-
next if existing_words.grep(/^#{part}/u).any?
|
225
270
|
fuzzies[att][encoded].zrem(id)
|
226
271
|
end
|
227
|
-
|
272
|
+
|
273
|
+
fuzzies[id][att].del
|
228
274
|
end
|
229
275
|
end
|
230
276
|
|
@@ -238,5 +284,12 @@ module Lunar
|
|
238
284
|
yield word, partials
|
239
285
|
end
|
240
286
|
end
|
287
|
+
|
288
|
+
def delete_field_meta
|
289
|
+
fields[TEXT].del
|
290
|
+
fields[NUMBERS].del
|
291
|
+
fields[SORTABLES].del
|
292
|
+
fields[FUZZIES].del
|
293
|
+
end
|
241
294
|
end
|
242
295
|
end
|
@@ -21,14 +21,16 @@ module Lunar
|
|
21
21
|
protected
|
22
22
|
def keys
|
23
23
|
if att == :q
|
24
|
-
metaphones.map { |m|
|
24
|
+
metaphones.map { |m|
|
25
|
+
nest[Index::FIELDS][Index::TEXT].smembers.map { |att| nest[att][m] }
|
26
|
+
}
|
25
27
|
else
|
26
28
|
metaphones.map { |m| nest[att][m] }
|
27
29
|
end
|
28
30
|
end
|
29
31
|
|
30
32
|
def metaphones
|
31
|
-
Words.new(value).map { |word| Lunar.metaphone(word) }
|
33
|
+
Words.new(value, [:stopwords]).map { |word| Lunar.metaphone(word) }
|
32
34
|
end
|
33
35
|
end
|
34
36
|
end
|
@@ -0,0 +1,33 @@
|
|
1
|
+
module Lunar
|
2
|
+
# @private Used internally by Lunar::search to get all the fuzzy matches
|
3
|
+
# given `nest`, `att` and it's `val`.
|
4
|
+
class NumberMatches
|
5
|
+
attr :nest
|
6
|
+
attr :att
|
7
|
+
attr :values
|
8
|
+
|
9
|
+
def initialize(nest, att, value)
|
10
|
+
@nest = nest
|
11
|
+
@att = att
|
12
|
+
@values = value.kind_of?(Enumerable) ? value : [value]
|
13
|
+
end
|
14
|
+
|
15
|
+
def distkey
|
16
|
+
case keys.size
|
17
|
+
when 0 then nil
|
18
|
+
when 1 then keys.first
|
19
|
+
else
|
20
|
+
nest[{ att => values }.hash].tap do |dk|
|
21
|
+
dk.zunionstore keys.flatten
|
22
|
+
end
|
23
|
+
end
|
24
|
+
end
|
25
|
+
|
26
|
+
protected
|
27
|
+
def keys
|
28
|
+
values.
|
29
|
+
reject { |v| v.to_s.empty? }.
|
30
|
+
map { |v| nest[:Numbers][att][v] }
|
31
|
+
end
|
32
|
+
end
|
33
|
+
end
|
data/lib/lunar/scoring.rb
CHANGED
@@ -3,11 +3,11 @@ module Lunar
|
|
3
3
|
# of a given text.
|
4
4
|
class Scoring
|
5
5
|
def initialize(words)
|
6
|
-
@words = Words.new(words)
|
6
|
+
@words = Words.new(words, [:stopwords, :downcase])
|
7
7
|
end
|
8
8
|
|
9
9
|
def scores
|
10
|
-
@words.inject(Hash.new(0)) { |a, w| a[w
|
10
|
+
@words.inject(Hash.new(0)) { |a, w| a[w] += 1 and a }
|
11
11
|
end
|
12
12
|
end
|
13
|
-
end
|
13
|
+
end
|
data/lib/lunar/stopwords.rb
CHANGED
@@ -3,15 +3,15 @@ module Lunar
|
|
3
3
|
# common words like an, the, etc.
|
4
4
|
module Stopwords
|
5
5
|
def include?(word)
|
6
|
-
stopwords.include?(word)
|
6
|
+
stopwords.include?(word.downcase)
|
7
7
|
end
|
8
|
+
module_function :include?
|
8
9
|
|
9
10
|
private
|
10
11
|
def stopwords
|
11
12
|
%w(an and are as at be but by for if in into is it no not of on or s
|
12
13
|
such t that the their then there these they this to was will with)
|
13
14
|
end
|
14
|
-
|
15
|
-
module_function :stopwords, :include?
|
15
|
+
module_function :stopwords
|
16
16
|
end
|
17
17
|
end
|
data/lib/lunar/words.rb
CHANGED
@@ -4,22 +4,47 @@ module Lunar
|
|
4
4
|
# @private Internally used to determine the words given some str.
|
5
5
|
# i.e. Words.new("the quick brown") == %w(the quick brown)
|
6
6
|
class Words < Array
|
7
|
+
UnknownFilter = Class.new(ArgumentError)
|
8
|
+
|
7
9
|
SEPARATOR = /\s+/
|
10
|
+
FILTERS = [:stopwords, :downcase, :uniq]
|
8
11
|
|
9
|
-
def initialize(str,
|
12
|
+
def initialize(str, filters = [])
|
10
13
|
words = str.split(SEPARATOR).
|
11
14
|
reject { |w| w.to_s.strip.empty? }.
|
12
15
|
map { |w| sanitize(w) }
|
13
|
-
|
14
|
-
words
|
16
|
+
|
17
|
+
apply_filters(words, filters)
|
15
18
|
|
16
19
|
super(words)
|
17
20
|
end
|
18
21
|
|
19
22
|
private
|
23
|
+
def apply_filters(words, filters)
|
24
|
+
filters.each do |filter|
|
25
|
+
unless FILTERS.include?(filter)
|
26
|
+
raise UnknownFilter, "Unknown filter: #{ filter }"
|
27
|
+
end
|
28
|
+
|
29
|
+
send(filter, words)
|
30
|
+
end
|
31
|
+
end
|
32
|
+
|
33
|
+
def stopwords(words)
|
34
|
+
words.reject! { |w| Stopwords.include?(w) }
|
35
|
+
end
|
36
|
+
|
37
|
+
def downcase(words)
|
38
|
+
words.each { |w| w.downcase! }
|
39
|
+
end
|
40
|
+
|
41
|
+
def uniq(words)
|
42
|
+
words.uniq!
|
43
|
+
end
|
44
|
+
|
20
45
|
def sanitize(str)
|
21
|
-
Iconv.iconv('UTF-8//IGNORE', 'UTF-8', str)[0].to_s.
|
22
|
-
gsub(/[
|
46
|
+
Iconv.iconv('UTF-8//IGNORE//TRANSLIT', 'UTF-8', str)[0].to_s.
|
47
|
+
gsub(/["'\.,@!$%\^&\*\(\)\[\]\+\-\_\:\;\<\>\\\/\?\`\~]/, '')
|
23
48
|
end
|
24
49
|
end
|
25
50
|
end
|
data/lunar.gemspec
CHANGED
@@ -5,11 +5,11 @@
|
|
5
5
|
|
6
6
|
Gem::Specification.new do |s|
|
7
7
|
s.name = %q{lunar}
|
8
|
-
s.version = "0.
|
8
|
+
s.version = "0.6.0"
|
9
9
|
|
10
10
|
s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
|
11
11
|
s.authors = ["Cyril David"]
|
12
|
-
s.date = %q{2010-
|
12
|
+
s.date = %q{2010-06-30}
|
13
13
|
s.description = %q{Features full text searching via metaphones, range querying for numbers, fuzzy searching and sorting based on custom fields}
|
14
14
|
s.email = %q{cyx.ucron@gmail.com}
|
15
15
|
s.extra_rdoc_files = [
|
@@ -28,7 +28,7 @@ Gem::Specification.new do |s|
|
|
28
28
|
"lib/lunar/fuzzy_word.rb",
|
29
29
|
"lib/lunar/index.rb",
|
30
30
|
"lib/lunar/keyword_matches.rb",
|
31
|
-
"lib/lunar/
|
31
|
+
"lib/lunar/number_matches.rb",
|
32
32
|
"lib/lunar/range_matches.rb",
|
33
33
|
"lib/lunar/result_set.rb",
|
34
34
|
"lib/lunar/scoring.rb",
|
@@ -40,13 +40,14 @@ Gem::Specification.new do |s|
|
|
40
40
|
"test/test_index.rb",
|
41
41
|
"test/test_lunar.rb",
|
42
42
|
"test/test_lunar_fuzzy_word.rb",
|
43
|
-
"test/
|
43
|
+
"test/test_lunar_words.rb",
|
44
|
+
"test/test_number_search.rb",
|
44
45
|
"test/test_scoring.rb"
|
45
46
|
]
|
46
47
|
s.homepage = %q{http://github.com/sinefunc/lunar}
|
47
48
|
s.rdoc_options = ["--charset=UTF-8"]
|
48
49
|
s.require_paths = ["lib"]
|
49
|
-
s.rubygems_version = %q{1.3.
|
50
|
+
s.rubygems_version = %q{1.3.7}
|
50
51
|
s.summary = %q{A redis based full text search engine}
|
51
52
|
s.test_files = [
|
52
53
|
"test/helper.rb",
|
@@ -54,7 +55,8 @@ Gem::Specification.new do |s|
|
|
54
55
|
"test/test_index.rb",
|
55
56
|
"test/test_lunar.rb",
|
56
57
|
"test/test_lunar_fuzzy_word.rb",
|
57
|
-
"test/
|
58
|
+
"test/test_lunar_words.rb",
|
59
|
+
"test/test_number_search.rb",
|
58
60
|
"test/test_scoring.rb"
|
59
61
|
]
|
60
62
|
|
@@ -62,7 +64,7 @@ Gem::Specification.new do |s|
|
|
62
64
|
current_version = Gem::Specification::CURRENT_SPECIFICATION_VERSION
|
63
65
|
s.specification_version = 3
|
64
66
|
|
65
|
-
if Gem::Version.new(Gem::
|
67
|
+
if Gem::Version.new(Gem::VERSION) >= Gem::Version.new('1.2.0') then
|
66
68
|
s.add_runtime_dependency(%q<redis>, [">= 2.0.0"])
|
67
69
|
s.add_runtime_dependency(%q<nest>, [">= 0.0.4"])
|
68
70
|
s.add_runtime_dependency(%q<text>, [">= 0"])
|
data/test/test_index.rb
CHANGED
@@ -22,6 +22,8 @@ class IndexTest < Test::Unit::TestCase
|
|
22
22
|
|
23
23
|
assert_equal %w{APL IFN KS SMRTFN},
|
24
24
|
Lunar.nest[:Gadget][:Metaphones][1001][:title].smembers.sort
|
25
|
+
|
26
|
+
assert_equal %w(title), Lunar.redis.smembers('Lunar:Gadget:Fields:Text')
|
25
27
|
end
|
26
28
|
|
27
29
|
test "deleting non-repeating words scenario" do
|
@@ -38,6 +40,8 @@ class IndexTest < Test::Unit::TestCase
|
|
38
40
|
assert_equal 0, nest('smartphone').zcard
|
39
41
|
|
40
42
|
assert_equal 0, Lunar.nest[:Gadget][1001][:title].scard
|
43
|
+
|
44
|
+
assert ! Lunar.redis.exists('Lunar:Gadget:Fields:Text')
|
41
45
|
end
|
42
46
|
|
43
47
|
test "with multiple word instances and stopwords" do
|
@@ -86,22 +90,65 @@ class IndexTest < Test::Unit::TestCase
|
|
86
90
|
assert_equal ['1001'], nest('apple').zrangebyscore(1, 1)
|
87
91
|
assert_equal ['1001'], nest('iphone').zrangebyscore(1, 1)
|
88
92
|
assert_equal ['1001'], nest('3g').zrangebyscore(1, 1)
|
89
|
-
|
93
|
+
|
90
94
|
assert nest('3gs').zrange(0, -1).empty?
|
91
95
|
assert nest('smartphone').zrange(0, -1).empty?
|
96
|
+
|
97
|
+
assert_equal %w(title), Lunar.redis.smembers("Lunar:Gadget:Fields:Text")
|
92
98
|
end
|
93
99
|
end
|
94
100
|
|
95
101
|
describe "indexing numbers" do
|
102
|
+
def numbers
|
103
|
+
Lunar.nest[:Gadget][:Numbers]
|
104
|
+
end
|
105
|
+
|
96
106
|
test "works for integers and floats" do
|
97
107
|
Lunar.index :Gadget do |i|
|
98
108
|
i.id 1001
|
99
109
|
i.number :price, 200
|
100
110
|
i.number :score, 25.5
|
111
|
+
i.number :category_ids, %w(100 101 102)
|
101
112
|
end
|
102
113
|
|
103
|
-
assert_equal '200',
|
104
|
-
assert_equal '25.5',
|
114
|
+
assert_equal '200', numbers[:price].zscore(1001)
|
115
|
+
assert_equal '25.5', numbers[:score].zscore(1001)
|
116
|
+
|
117
|
+
assert_equal '1', numbers[:price]["200"].zscore(1001)
|
118
|
+
|
119
|
+
assert_equal '1', numbers[:category_ids]["100"].zscore(1001)
|
120
|
+
assert_equal '1', numbers[:category_ids]["101"].zscore(1001)
|
121
|
+
assert_equal '1', numbers[:category_ids]["102"].zscore(1001)
|
122
|
+
|
123
|
+
assert_equal %w(category_ids price score),
|
124
|
+
Lunar.redis.smembers("Lunar:Gadget:Fields:Numbers").sort
|
125
|
+
end
|
126
|
+
|
127
|
+
test "reindexing" do
|
128
|
+
Lunar.index :Gadget do |i|
|
129
|
+
i.id 1001
|
130
|
+
i.number :price, 200
|
131
|
+
i.number :author_ids, [10, 20, 30]
|
132
|
+
end
|
133
|
+
|
134
|
+
Lunar.index :Gadget do |i|
|
135
|
+
i.id 1001
|
136
|
+
i.number :price, 150
|
137
|
+
i.number :author_ids, [40, 50, 60]
|
138
|
+
end
|
139
|
+
|
140
|
+
assert_nil numbers[:price]["200"].zrank(1001)
|
141
|
+
assert_equal '1', numbers[:price]["150"].zscore(1001)
|
142
|
+
|
143
|
+
assert_nil numbers[:author_ids]["10"].zrank(1001)
|
144
|
+
assert_nil numbers[:author_ids]["20"].zrank(1001)
|
145
|
+
assert_nil numbers[:author_ids]["30"].zrank(1001)
|
146
|
+
assert_equal '1', numbers[:author_ids]["40"].zscore(1001)
|
147
|
+
assert_equal '1', numbers[:author_ids]["50"].zscore(1001)
|
148
|
+
assert_equal '1', numbers[:author_ids]["60"].zscore(1001)
|
149
|
+
|
150
|
+
assert_equal %w(author_ids price),
|
151
|
+
Lunar.redis.smembers("Lunar:Gadget:Fields:Numbers").sort
|
105
152
|
end
|
106
153
|
|
107
154
|
test "allows deletion" do
|
@@ -113,8 +160,11 @@ class IndexTest < Test::Unit::TestCase
|
|
113
160
|
|
114
161
|
Lunar.delete :Gadget, 1001
|
115
162
|
|
116
|
-
assert_nil
|
117
|
-
assert_nil
|
163
|
+
assert_nil numbers[:price].zrank(1001)
|
164
|
+
assert_nil numbers[:score].zrank(1001)
|
165
|
+
assert_nil numbers[:price]["200"].zrank(1001)
|
166
|
+
|
167
|
+
assert ! Lunar.redis.exists("Lunar:Gadget:Fields:Numbers")
|
118
168
|
end
|
119
169
|
end
|
120
170
|
|
@@ -130,6 +180,9 @@ class IndexTest < Test::Unit::TestCase
|
|
130
180
|
assert_equal 'iphone', Lunar.nest[:Gadget][:Sortables][1001][:name].get
|
131
181
|
assert_equal '200', Lunar.nest[:Gadget][:Sortables][1001][:price].get
|
132
182
|
assert_equal '25.5', Lunar.nest[:Gadget][:Sortables][1001][:score].get
|
183
|
+
|
184
|
+
assert_equal %w(name price score),
|
185
|
+
Lunar.redis.smembers("Lunar:Gadget:Fields:Sortables").sort
|
133
186
|
end
|
134
187
|
|
135
188
|
test "deletes sortable fields" do
|
@@ -142,9 +195,11 @@ class IndexTest < Test::Unit::TestCase
|
|
142
195
|
|
143
196
|
Lunar.delete :Gadget, 1001
|
144
197
|
|
145
|
-
assert_nil Lunar.nest[:Gadget][1001][:name].get
|
146
|
-
assert_nil Lunar.nest[:Gadget][1001][:price].get
|
147
|
-
assert_nil Lunar.nest[:Gadget][1001][:score].get
|
198
|
+
assert_nil Lunar.nest[:Gadget][:Sortables][1001][:name].get
|
199
|
+
assert_nil Lunar.nest[:Gadget][:Sortables][1001][:price].get
|
200
|
+
assert_nil Lunar.nest[:Gadget][:Sortables][1001][:score].get
|
201
|
+
|
202
|
+
assert ! Lunar.redis.exists("Lunar:Gadget:Fields:Sortables")
|
148
203
|
end
|
149
204
|
end
|
150
|
-
end
|
205
|
+
end
|
@@ -0,0 +1,14 @@
|
|
1
|
+
# encoding: UTF-8
|
2
|
+
|
3
|
+
require "helper"
|
4
|
+
|
5
|
+
class LunarWordsTest < Test::Unit::TestCase
|
6
|
+
test "german words" do
|
7
|
+
str = "Der schnelle braune Fuchs springt über den faulen Hund"
|
8
|
+
metaphones = %w(TR SXNL BRN FXS SPRNKT BR TN FLN HNT)
|
9
|
+
|
10
|
+
words = Lunar::Words.new(str)
|
11
|
+
|
12
|
+
assert_equal metaphones, words.map { |w| Lunar.metaphone(w) }
|
13
|
+
end
|
14
|
+
end
|
@@ -0,0 +1,69 @@
|
|
1
|
+
require "helper"
|
2
|
+
|
3
|
+
class NumberSearchTest < Test::Unit::TestCase
|
4
|
+
class Gadget < Struct.new(:id)
|
5
|
+
def self.[](id)
|
6
|
+
new(id)
|
7
|
+
end
|
8
|
+
end
|
9
|
+
|
10
|
+
setup do
|
11
|
+
Lunar.index Gadget do |i|
|
12
|
+
i.id 1001
|
13
|
+
i.number :price, 200
|
14
|
+
i.number :category_ids, %w(100 101 102)
|
15
|
+
end
|
16
|
+
end
|
17
|
+
|
18
|
+
test "doing straight searches for numbers" do
|
19
|
+
search = Lunar.search Gadget, numbers: { price: 200 }
|
20
|
+
|
21
|
+
assert_equal ["1001"], search.map(&:id)
|
22
|
+
end
|
23
|
+
|
24
|
+
test "also doing range searches" do
|
25
|
+
search = Lunar.search Gadget, price: 150..200
|
26
|
+
|
27
|
+
assert_equal ["1001"], search.map(&:id)
|
28
|
+
end
|
29
|
+
|
30
|
+
test "searching both" do
|
31
|
+
search = Lunar.search Gadget, price: 150..200, numbers: { price: 200 }
|
32
|
+
|
33
|
+
assert_equal ["1001"], search.map(&:id)
|
34
|
+
end
|
35
|
+
|
36
|
+
test "searching for a multi-id individually" do
|
37
|
+
search100 = Lunar.search Gadget, numbers: { category_ids: 100 }
|
38
|
+
search101 = Lunar.search Gadget, numbers: { category_ids: 101 }
|
39
|
+
search102 = Lunar.search Gadget, numbers: { category_ids: 102 }
|
40
|
+
|
41
|
+
assert_equal ["1001"], search100.map(&:id)
|
42
|
+
assert_equal ["1001"], search101.map(&:id)
|
43
|
+
assert_equal ["1001"], search102.map(&:id)
|
44
|
+
end
|
45
|
+
|
46
|
+
test "searching for a valid multi-id at the same time" do
|
47
|
+
search = Lunar.search Gadget, numbers: { category_ids: %w(100 101 102) }
|
48
|
+
|
49
|
+
assert_equal ["1001"], search.map(&:id)
|
50
|
+
end
|
51
|
+
|
52
|
+
test "searching with one invalid multi-id at the same time" do
|
53
|
+
search = Lunar.search Gadget, numbers: { category_ids: %w(100 101 102 103) }
|
54
|
+
|
55
|
+
assert_equal ["1001"], search.map(&:id)
|
56
|
+
end
|
57
|
+
|
58
|
+
test "searching with all invalid multi-id" do
|
59
|
+
search = Lunar.search Gadget, numbers: { category_ids: %w(103 104 105) }
|
60
|
+
assert_equal 0, search.size
|
61
|
+
end
|
62
|
+
|
63
|
+
test "searching with valid category_ids and an empty price array" do
|
64
|
+
search = Lunar.search(Gadget, numbers: { category_ids: %w(100 101 102),
|
65
|
+
price: [""]})
|
66
|
+
|
67
|
+
assert_equal ["1001"], search.map(&:id)
|
68
|
+
end
|
69
|
+
end
|
data/test/test_scoring.rb
CHANGED
metadata
CHANGED
@@ -4,9 +4,9 @@ version: !ruby/object:Gem::Version
|
|
4
4
|
prerelease: false
|
5
5
|
segments:
|
6
6
|
- 0
|
7
|
-
-
|
8
|
-
-
|
9
|
-
version: 0.
|
7
|
+
- 6
|
8
|
+
- 0
|
9
|
+
version: 0.6.0
|
10
10
|
platform: ruby
|
11
11
|
authors:
|
12
12
|
- Cyril David
|
@@ -14,13 +14,14 @@ autorequire:
|
|
14
14
|
bindir: bin
|
15
15
|
cert_chain: []
|
16
16
|
|
17
|
-
date: 2010-
|
17
|
+
date: 2010-06-30 00:00:00 +08:00
|
18
18
|
default_executable:
|
19
19
|
dependencies:
|
20
20
|
- !ruby/object:Gem::Dependency
|
21
21
|
name: redis
|
22
22
|
prerelease: false
|
23
23
|
requirement: &id001 !ruby/object:Gem::Requirement
|
24
|
+
none: false
|
24
25
|
requirements:
|
25
26
|
- - ">="
|
26
27
|
- !ruby/object:Gem::Version
|
@@ -35,6 +36,7 @@ dependencies:
|
|
35
36
|
name: nest
|
36
37
|
prerelease: false
|
37
38
|
requirement: &id002 !ruby/object:Gem::Requirement
|
39
|
+
none: false
|
38
40
|
requirements:
|
39
41
|
- - ">="
|
40
42
|
- !ruby/object:Gem::Version
|
@@ -49,6 +51,7 @@ dependencies:
|
|
49
51
|
name: text
|
50
52
|
prerelease: false
|
51
53
|
requirement: &id003 !ruby/object:Gem::Requirement
|
54
|
+
none: false
|
52
55
|
requirements:
|
53
56
|
- - ">="
|
54
57
|
- !ruby/object:Gem::Version
|
@@ -61,6 +64,7 @@ dependencies:
|
|
61
64
|
name: contest
|
62
65
|
prerelease: false
|
63
66
|
requirement: &id004 !ruby/object:Gem::Requirement
|
67
|
+
none: false
|
64
68
|
requirements:
|
65
69
|
- - ">="
|
66
70
|
- !ruby/object:Gem::Version
|
@@ -90,7 +94,7 @@ files:
|
|
90
94
|
- lib/lunar/fuzzy_word.rb
|
91
95
|
- lib/lunar/index.rb
|
92
96
|
- lib/lunar/keyword_matches.rb
|
93
|
-
- lib/lunar/
|
97
|
+
- lib/lunar/number_matches.rb
|
94
98
|
- lib/lunar/range_matches.rb
|
95
99
|
- lib/lunar/result_set.rb
|
96
100
|
- lib/lunar/scoring.rb
|
@@ -102,7 +106,8 @@ files:
|
|
102
106
|
- test/test_index.rb
|
103
107
|
- test/test_lunar.rb
|
104
108
|
- test/test_lunar_fuzzy_word.rb
|
105
|
-
- test/
|
109
|
+
- test/test_lunar_words.rb
|
110
|
+
- test/test_number_search.rb
|
106
111
|
- test/test_scoring.rb
|
107
112
|
has_rdoc: true
|
108
113
|
homepage: http://github.com/sinefunc/lunar
|
@@ -114,6 +119,7 @@ rdoc_options:
|
|
114
119
|
require_paths:
|
115
120
|
- lib
|
116
121
|
required_ruby_version: !ruby/object:Gem::Requirement
|
122
|
+
none: false
|
117
123
|
requirements:
|
118
124
|
- - ">="
|
119
125
|
- !ruby/object:Gem::Version
|
@@ -121,6 +127,7 @@ required_ruby_version: !ruby/object:Gem::Requirement
|
|
121
127
|
- 0
|
122
128
|
version: "0"
|
123
129
|
required_rubygems_version: !ruby/object:Gem::Requirement
|
130
|
+
none: false
|
124
131
|
requirements:
|
125
132
|
- - ">="
|
126
133
|
- !ruby/object:Gem::Version
|
@@ -130,7 +137,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
|
|
130
137
|
requirements: []
|
131
138
|
|
132
139
|
rubyforge_project:
|
133
|
-
rubygems_version: 1.3.
|
140
|
+
rubygems_version: 1.3.7
|
134
141
|
signing_key:
|
135
142
|
specification_version: 3
|
136
143
|
summary: A redis based full text search engine
|
@@ -140,5 +147,6 @@ test_files:
|
|
140
147
|
- test/test_index.rb
|
141
148
|
- test/test_lunar.rb
|
142
149
|
- test/test_lunar_fuzzy_word.rb
|
143
|
-
- test/
|
150
|
+
- test/test_lunar_words.rb
|
151
|
+
- test/test_number_search.rb
|
144
152
|
- test/test_scoring.rb
|
data/lib/lunar/lunar_nest.rb
DELETED
@@ -1,19 +0,0 @@
|
|
1
|
-
module Lunar
|
2
|
-
# @private Provides convenience methods to look up keys.
|
3
|
-
# Since the Redis KEYS command is actually pretty slow,
|
4
|
-
# i'm considering an alternative approach of manually maintaining
|
5
|
-
# the sets to manage all the groups of keys.
|
6
|
-
class LunarNest < Nest
|
7
|
-
def keys
|
8
|
-
redis.keys self
|
9
|
-
end
|
10
|
-
|
11
|
-
def matches
|
12
|
-
regex = Regexp.new(self.gsub('*', '(.*)'))
|
13
|
-
keys.map { |key|
|
14
|
-
match = key.match(regex)
|
15
|
-
[LunarNest.new(key, redis), *match[1, match.size - 1]]
|
16
|
-
}
|
17
|
-
end
|
18
|
-
end
|
19
|
-
end
|
data/test/test_lunar_nest.rb
DELETED
@@ -1,46 +0,0 @@
|
|
1
|
-
require "helper"
|
2
|
-
|
3
|
-
class LunarNestTest < Test::Unit::TestCase
|
4
|
-
test "retrieving keys for a pattern" do
|
5
|
-
Lunar.redis.set("Foo:1:Bar", 1)
|
6
|
-
Lunar.redis.set("Foo:2:Bar", 2)
|
7
|
-
|
8
|
-
nest = Lunar::LunarNest.new("Foo", Lunar.redis)['*']["Bar"]
|
9
|
-
|
10
|
-
assert_equal ['Foo:1:Bar', 'Foo:2:Bar'], nest.keys.sort
|
11
|
-
end
|
12
|
-
|
13
|
-
test "retrieving keys and their matches" do
|
14
|
-
Lunar.redis.set("Foo:1:Bar", 1)
|
15
|
-
Lunar.redis.set("Foo:2:Bar", 2)
|
16
|
-
|
17
|
-
nest = Lunar::LunarNest.new("Foo", Lunar.redis)['*']["Bar"]
|
18
|
-
|
19
|
-
matches = []
|
20
|
-
|
21
|
-
nest.matches.each do |key, m|
|
22
|
-
matches << m
|
23
|
-
end
|
24
|
-
|
25
|
-
assert_equal ['1', '2'], matches.sort
|
26
|
-
end
|
27
|
-
|
28
|
-
test "retrieving keys and their matches when more than one *" do
|
29
|
-
Lunar.redis.set("Foo:1:Bar:3:Baz", 1)
|
30
|
-
Lunar.redis.set("Foo:2:Bar:4:Baz", 2)
|
31
|
-
|
32
|
-
nest = Lunar::LunarNest.new("Foo", Lunar.redis)['*']["Bar"]["*"]["Baz"]
|
33
|
-
|
34
|
-
matches1 = []
|
35
|
-
matches2 = []
|
36
|
-
|
37
|
-
nest.matches.each do |key, m1, m2|
|
38
|
-
matches1 << m1
|
39
|
-
matches2 << m2
|
40
|
-
end
|
41
|
-
|
42
|
-
assert_equal ['1', '2'], matches1.sort
|
43
|
-
assert_equal ['3', '4'], matches2.sort
|
44
|
-
end
|
45
|
-
|
46
|
-
end
|