picky 0.0.0 → 0.0.2
Sign up to get free protection for your applications and to get access to all the features.
- data/bin/picky +14 -0
- data/lib/bundling.rb +10 -0
- data/lib/constants.rb +9 -0
- data/lib/deployment.rb +212 -0
- data/lib/picky/application.rb +40 -0
- data/lib/picky/cacher/convenience.rb +3 -0
- data/lib/picky/cacher/generator.rb +17 -0
- data/lib/picky/cacher/partial/default.rb +7 -0
- data/lib/picky/cacher/partial/none.rb +19 -0
- data/lib/picky/cacher/partial/strategy.rb +7 -0
- data/lib/picky/cacher/partial/subtoken.rb +91 -0
- data/lib/picky/cacher/partial_generator.rb +15 -0
- data/lib/picky/cacher/similarity/default.rb +7 -0
- data/lib/picky/cacher/similarity/double_levenshtone.rb +73 -0
- data/lib/picky/cacher/similarity/none.rb +25 -0
- data/lib/picky/cacher/similarity/strategy.rb +7 -0
- data/lib/picky/cacher/similarity_generator.rb +15 -0
- data/lib/picky/cacher/weights/default.rb +7 -0
- data/lib/picky/cacher/weights/logarithmic.rb +39 -0
- data/lib/picky/cacher/weights/strategy.rb +7 -0
- data/lib/picky/cacher/weights_generator.rb +15 -0
- data/lib/picky/configuration/configuration.rb +13 -0
- data/lib/picky/configuration/field.rb +68 -0
- data/lib/picky/configuration/indexes.rb +60 -0
- data/lib/picky/configuration/queries.rb +32 -0
- data/lib/picky/configuration/type.rb +52 -0
- data/lib/picky/cores.rb +101 -0
- data/lib/picky/db/configuration.rb +23 -0
- data/lib/picky/ext/ruby19/extconf.rb +7 -0
- data/lib/picky/ext/ruby19/performant.c +339 -0
- data/lib/picky/extensions/array.rb +45 -0
- data/lib/picky/extensions/hash.rb +11 -0
- data/lib/picky/extensions/module.rb +15 -0
- data/lib/picky/extensions/symbol.rb +18 -0
- data/lib/picky/generator.rb +156 -0
- data/lib/picky/helpers/cache.rb +23 -0
- data/lib/picky/helpers/gc.rb +11 -0
- data/lib/picky/helpers/measuring.rb +45 -0
- data/lib/picky/helpers/search.rb +27 -0
- data/lib/picky/index/bundle.rb +328 -0
- data/lib/picky/index/category.rb +109 -0
- data/lib/picky/index/combined.rb +38 -0
- data/lib/picky/index/type.rb +30 -0
- data/lib/picky/indexers/base.rb +77 -0
- data/lib/picky/indexers/default.rb +3 -0
- data/lib/picky/indexers/field.rb +13 -0
- data/lib/picky/indexers/no_source_specified_error.rb +5 -0
- data/lib/picky/indexers/solr.rb +60 -0
- data/lib/picky/indexes.rb +180 -0
- data/lib/picky/initializers/ext.rb +6 -0
- data/lib/picky/initializers/mysql.rb +22 -0
- data/lib/picky/loader.rb +287 -0
- data/lib/picky/loggers/search.rb +19 -0
- data/lib/picky/performant/array.rb +23 -0
- data/lib/picky/query/allocation.rb +82 -0
- data/lib/picky/query/allocations.rb +131 -0
- data/lib/picky/query/base.rb +124 -0
- data/lib/picky/query/combination.rb +69 -0
- data/lib/picky/query/combinations.rb +106 -0
- data/lib/picky/query/combinator.rb +92 -0
- data/lib/picky/query/full.rb +15 -0
- data/lib/picky/query/live.rb +22 -0
- data/lib/picky/query/qualifiers.rb +73 -0
- data/lib/picky/query/solr.rb +77 -0
- data/lib/picky/query/token.rb +215 -0
- data/lib/picky/query/tokens.rb +102 -0
- data/lib/picky/query/weigher.rb +159 -0
- data/lib/picky/query/weights.rb +55 -0
- data/lib/picky/rack/harakiri.rb +37 -0
- data/lib/picky/results/base.rb +103 -0
- data/lib/picky/results/full.rb +19 -0
- data/lib/picky/results/live.rb +19 -0
- data/lib/picky/routing.rb +165 -0
- data/lib/picky/signals.rb +11 -0
- data/lib/picky/solr/schema_generator.rb +73 -0
- data/lib/picky/sources/base.rb +19 -0
- data/lib/picky/sources/csv.rb +30 -0
- data/lib/picky/sources/db.rb +77 -0
- data/lib/picky/tokenizers/base.rb +130 -0
- data/lib/picky/tokenizers/default.rb +3 -0
- data/lib/picky/tokenizers/index.rb +73 -0
- data/lib/picky/tokenizers/query.rb +70 -0
- data/lib/picky/umlaut_substituter.rb +21 -0
- data/lib/picky-tasks.rb +6 -0
- data/lib/picky.rb +18 -0
- data/lib/tasks/application.rake +5 -0
- data/lib/tasks/cache.rake +53 -0
- data/lib/tasks/framework.rake +4 -0
- data/lib/tasks/index.rake +29 -0
- data/lib/tasks/server.rake +48 -0
- data/lib/tasks/shortcuts.rake +13 -0
- data/lib/tasks/solr.rake +36 -0
- data/lib/tasks/spec.rake +11 -0
- data/lib/tasks/statistics.rake +13 -0
- data/lib/tasks/try.rake +29 -0
- data/prototype_project/Gemfile +23 -0
- data/prototype_project/Rakefile +1 -0
- data/prototype_project/app/README +6 -0
- data/prototype_project/app/application.rb +50 -0
- data/prototype_project/app/application.ru +29 -0
- data/prototype_project/app/db.yml +10 -0
- data/prototype_project/app/logging.rb +20 -0
- data/prototype_project/app/unicorn.ru +10 -0
- data/prototype_project/log/README +1 -0
- data/prototype_project/script/console +34 -0
- data/prototype_project/tmp/README +0 -0
- data/prototype_project/tmp/pids/README +0 -0
- data/spec/ext/performant_spec.rb +64 -0
- data/spec/lib/application_spec.rb +61 -0
- data/spec/lib/cacher/partial/subtoken_spec.rb +89 -0
- data/spec/lib/cacher/partial_generator_spec.rb +35 -0
- data/spec/lib/cacher/similarity/double_levenshtone_spec.rb +60 -0
- data/spec/lib/cacher/similarity/none_spec.rb +23 -0
- data/spec/lib/cacher/similarity_generator_spec.rb +22 -0
- data/spec/lib/cacher/weights/logarithmic_spec.rb +30 -0
- data/spec/lib/cacher/weights_generator_spec.rb +21 -0
- data/spec/lib/configuration/configuration_spec.rb +38 -0
- data/spec/lib/configuration/type_spec.rb +49 -0
- data/spec/lib/configuration_spec.rb +8 -0
- data/spec/lib/cores_spec.rb +65 -0
- data/spec/lib/extensions/array_spec.rb +37 -0
- data/spec/lib/extensions/hash_spec.rb +11 -0
- data/spec/lib/extensions/module_spec.rb +27 -0
- data/spec/lib/extensions/symbol_spec.rb +85 -0
- data/spec/lib/generator_spec.rb +135 -0
- data/spec/lib/helpers/cache_spec.rb +35 -0
- data/spec/lib/helpers/gc_spec.rb +71 -0
- data/spec/lib/helpers/measuring_spec.rb +18 -0
- data/spec/lib/helpers/search_spec.rb +50 -0
- data/spec/lib/index/bundle_partial_generation_speed_spec.rb +47 -0
- data/spec/lib/index/bundle_spec.rb +260 -0
- data/spec/lib/index/category_spec.rb +203 -0
- data/spec/lib/indexers/base_spec.rb +73 -0
- data/spec/lib/indexers/field_spec.rb +20 -0
- data/spec/lib/loader_spec.rb +48 -0
- data/spec/lib/loggers/search_spec.rb +19 -0
- data/spec/lib/performant/array_spec.rb +13 -0
- data/spec/lib/query/allocation_spec.rb +194 -0
- data/spec/lib/query/allocations_spec.rb +336 -0
- data/spec/lib/query/base_spec.rb +104 -0
- data/spec/lib/query/combination_spec.rb +90 -0
- data/spec/lib/query/combinations_spec.rb +83 -0
- data/spec/lib/query/combinator_spec.rb +112 -0
- data/spec/lib/query/full_spec.rb +22 -0
- data/spec/lib/query/live_spec.rb +61 -0
- data/spec/lib/query/qualifiers_spec.rb +31 -0
- data/spec/lib/query/solr_spec.rb +51 -0
- data/spec/lib/query/token_spec.rb +297 -0
- data/spec/lib/query/tokens_spec.rb +189 -0
- data/spec/lib/query/weights_spec.rb +47 -0
- data/spec/lib/results/base_spec.rb +233 -0
- data/spec/lib/routing_spec.rb +318 -0
- data/spec/lib/solr/schema_generator_spec.rb +42 -0
- data/spec/lib/sources/db_spec.rb +91 -0
- data/spec/lib/tokenizers/base_spec.rb +61 -0
- data/spec/lib/tokenizers/index_spec.rb +51 -0
- data/spec/lib/tokenizers/query_spec.rb +105 -0
- data/spec/lib/umlaut_substituter_spec.rb +84 -0
- data/spec/specific/speed_spec.rb +55 -0
- metadata +371 -15
- data/README.textile +0 -9
data/bin/picky
ADDED
@@ -0,0 +1,14 @@
|
|
1
|
+
#!/usr/bin/env ruby
|
2
|
+
#
|
3
|
+
|
4
|
+
begin
|
5
|
+
require 'picky/generator'
|
6
|
+
rescue LoadError => e
|
7
|
+
require 'rubygems'
|
8
|
+
picky_path = File.expand_path('../../lib', __FILE__)
|
9
|
+
$:.unshift(picky_path) if File.directory?(picky_path) && !$:.include?(picky_path)
|
10
|
+
require 'picky/generator'
|
11
|
+
end
|
12
|
+
|
13
|
+
generator = Picky::Generator.new
|
14
|
+
generator.generate ARGV
|
data/lib/bundling.rb
ADDED
data/lib/constants.rb
ADDED
data/lib/deployment.rb
ADDED
@@ -0,0 +1,212 @@
|
|
1
|
+
require File.expand_path(File.join(File.dirname(__FILE__), 'constants'))
|
2
|
+
|
3
|
+
module Picky
|
4
|
+
module Capistrano
|
5
|
+
|
6
|
+
# Include all
|
7
|
+
#
|
8
|
+
module All
|
9
|
+
def self.extended cap_config
|
10
|
+
|
11
|
+
cap_config.instance_eval do
|
12
|
+
|
13
|
+
# Executes a rake task on the server.
|
14
|
+
#
|
15
|
+
# Options:
|
16
|
+
# * env: The SEARCH_ENV. Will not set if set explicitly to false. Default: production.
|
17
|
+
# * All other options get passed on to the Capistrano run task.
|
18
|
+
#
|
19
|
+
def execute_rake_task name, options = {}, &block
|
20
|
+
env = options.delete :env
|
21
|
+
env = env == false ? '' : "SEARCH_ENV=#{env || 'production'}"
|
22
|
+
run "cd #{current_path}; rake #{name} #{env}", options, &block
|
23
|
+
end
|
24
|
+
|
25
|
+
end
|
26
|
+
|
27
|
+
cap_config.extend Standard
|
28
|
+
cap_config.extend Deploy
|
29
|
+
cap_config.extend Caching
|
30
|
+
cap_config.extend Overrides
|
31
|
+
|
32
|
+
end
|
33
|
+
end
|
34
|
+
|
35
|
+
# Removes unneeded Rails defaults.
|
36
|
+
#
|
37
|
+
module Overrides
|
38
|
+
def self.extended cap_config
|
39
|
+
cap_config.instance_eval do
|
40
|
+
|
41
|
+
namespace :deploy do
|
42
|
+
tasks.delete :check
|
43
|
+
tasks.delete :cold
|
44
|
+
tasks.delete :migrations
|
45
|
+
tasks.delete :migrate
|
46
|
+
tasks.delete :upload
|
47
|
+
|
48
|
+
namespace :web do
|
49
|
+
tasks.delete :enable
|
50
|
+
tasks.delete :disable
|
51
|
+
end
|
52
|
+
end
|
53
|
+
|
54
|
+
end
|
55
|
+
end
|
56
|
+
end
|
57
|
+
|
58
|
+
module Standard
|
59
|
+
def self.extended cap_config
|
60
|
+
cap_config.load 'standard'
|
61
|
+
cap_config.load 'deploy'
|
62
|
+
end
|
63
|
+
end
|
64
|
+
|
65
|
+
module Deploy
|
66
|
+
|
67
|
+
def self.extended cap_config
|
68
|
+
cap_config.instance_eval do
|
69
|
+
|
70
|
+
namespace :deploy do
|
71
|
+
%w(start stop).each do |action|
|
72
|
+
desc "#{action} the Servers"
|
73
|
+
task action.to_sym, :roles => :app do
|
74
|
+
execute_rake_task "server:#{action}"
|
75
|
+
end
|
76
|
+
end
|
77
|
+
desc "Restart the Servers sequentially"
|
78
|
+
task :restart, :roles => :app do
|
79
|
+
find_servers(:roles => :app).each do |server|
|
80
|
+
execute_rake_task "server:restart", :hosts => server.host
|
81
|
+
end
|
82
|
+
end
|
83
|
+
|
84
|
+
desc 'Hot deploy the code'
|
85
|
+
task 'hot', :roles => :app do
|
86
|
+
update
|
87
|
+
execute_rake_task 'server:usr1', :env => false # No env needed.
|
88
|
+
end
|
89
|
+
|
90
|
+
desc "Setup a GitHub-style deployment."
|
91
|
+
task :setup, :roles => :app do
|
92
|
+
cmd = "git clone #{repository} #{current_path}-clone-cache &&" +
|
93
|
+
"rm #{current_path} &&" +
|
94
|
+
"mv #{current_path}-clone-cache #{current_path}"
|
95
|
+
run cmd
|
96
|
+
end
|
97
|
+
|
98
|
+
desc "Deploy"
|
99
|
+
task :default, :roles => :app do
|
100
|
+
update
|
101
|
+
restart
|
102
|
+
end
|
103
|
+
|
104
|
+
desc "Update the deployed code."
|
105
|
+
task :update_code do # code needs to be updated with all servers
|
106
|
+
puts "updating code to branch #{branch}"
|
107
|
+
cmd = "cd #{current_path} &&" +
|
108
|
+
"git fetch origin &&" +
|
109
|
+
"(git checkout -f #{branch} || git checkout -b #{branch} origin/#{branch}) &&" +
|
110
|
+
"git pull;" +
|
111
|
+
"git branch"
|
112
|
+
run cmd
|
113
|
+
symlink
|
114
|
+
end
|
115
|
+
|
116
|
+
desc "Cleans up the git checkout"
|
117
|
+
task :cleanup, :roles => :app do
|
118
|
+
run "cd #{current_path} && git gc --aggressive"
|
119
|
+
end
|
120
|
+
|
121
|
+
desc "create the symlinks to the shared dirs"
|
122
|
+
task :symlink do
|
123
|
+
set :user, 'deploy'
|
124
|
+
run "rm -rf #{current_path}/log; ln -sf #{shared_path}/log #{current_path}/log"
|
125
|
+
run "rm -rf #{current_path}/index; ln -sf #{shared_path}/index #{current_path}/index"
|
126
|
+
# link database-config files
|
127
|
+
run "ln -sf #{shared_path}/config/base.yml #{current_path}/config/db/base.yml"
|
128
|
+
run "ln -sf #{shared_path}/config/source.yml #{current_path}/config/db/source.yml"
|
129
|
+
# link unicorn.ru
|
130
|
+
run "ln -sf #{shared_path}/config/unicorn.ru #{current_path}/config/unicorn.ru" # TODO change path
|
131
|
+
end
|
132
|
+
|
133
|
+
namespace :rollback do
|
134
|
+
desc "Rollback to last release."
|
135
|
+
task :default, :roles => :app do
|
136
|
+
set :branch, branches[-2]
|
137
|
+
puts "rolling back to branch #{branch}"
|
138
|
+
deploy.update_code
|
139
|
+
end
|
140
|
+
|
141
|
+
task :code, :roles => :app do
|
142
|
+
# implicit
|
143
|
+
end
|
144
|
+
end
|
145
|
+
end
|
146
|
+
|
147
|
+
end
|
148
|
+
end
|
149
|
+
|
150
|
+
end
|
151
|
+
|
152
|
+
module Caching
|
153
|
+
|
154
|
+
def self.extended cap_config
|
155
|
+
cap_config.instance_eval do
|
156
|
+
namespace :cache do
|
157
|
+
desc "check the index files if they are ready to be used"
|
158
|
+
task :check, :roles => :cache do
|
159
|
+
execute_rake_task 'cache:check'
|
160
|
+
end
|
161
|
+
end
|
162
|
+
namespace :cache do
|
163
|
+
namespace :structure do
|
164
|
+
desc "create the index cache structure"
|
165
|
+
task :create, :roles => :app do
|
166
|
+
execute_rake_task 'cache:structure:create'
|
167
|
+
end
|
168
|
+
end
|
169
|
+
end
|
170
|
+
namespace :solr do
|
171
|
+
desc "create the index cache structure"
|
172
|
+
task :index, :roles => :cache do
|
173
|
+
execute_rake_task 'solr:index'
|
174
|
+
end
|
175
|
+
%w|start stop restart|.collect(&:to_sym).each do |action|
|
176
|
+
desc "#{action} the solr server"
|
177
|
+
task action, :roles => :app do
|
178
|
+
execute_rake_task 'solr:start'
|
179
|
+
end
|
180
|
+
end
|
181
|
+
end
|
182
|
+
end
|
183
|
+
end
|
184
|
+
|
185
|
+
end
|
186
|
+
|
187
|
+
module Statistics
|
188
|
+
|
189
|
+
def self.extended cap_config
|
190
|
+
namespace :statistics do
|
191
|
+
desc 'Start the statistics server'
|
192
|
+
task :start, :roles => :statistics do
|
193
|
+
set :user, 'root'
|
194
|
+
run "daemonize -c #{current_path} -u deploy -v #{current_path}/script/statistics/start production"
|
195
|
+
end
|
196
|
+
desc 'Stop the statistics server'
|
197
|
+
task :stop, :roles => :statistics do
|
198
|
+
run "#{current_path}/script/statistics/stop production"
|
199
|
+
end
|
200
|
+
desc 'Restart the statistics server'
|
201
|
+
task :restart, :roles => :statistics do
|
202
|
+
stop
|
203
|
+
sleep 2
|
204
|
+
start
|
205
|
+
end
|
206
|
+
end
|
207
|
+
end
|
208
|
+
|
209
|
+
end
|
210
|
+
|
211
|
+
end
|
212
|
+
end
|
@@ -0,0 +1,40 @@
|
|
1
|
+
class Application
|
2
|
+
|
3
|
+
# An application simply delegates to the routing to handle a request.
|
4
|
+
#
|
5
|
+
def self.routing
|
6
|
+
@routing ||= Routing.new
|
7
|
+
end
|
8
|
+
def self.call env
|
9
|
+
routing.call env
|
10
|
+
end
|
11
|
+
|
12
|
+
# TODO Multiple indexes?
|
13
|
+
#
|
14
|
+
def self.indexes &block
|
15
|
+
indexes_configuration.instance_eval &block
|
16
|
+
# TODO Uglyyyyyy.
|
17
|
+
::Indexes.configuration = indexes_configuration
|
18
|
+
::Indexes.setup # TODO Think about setup/reload.
|
19
|
+
end
|
20
|
+
def self.indexes_configuration
|
21
|
+
@indexes || reset_indexes
|
22
|
+
end
|
23
|
+
def self.reset_indexes
|
24
|
+
@indexes = Configuration::Indexes.new # Is instance a problem?
|
25
|
+
end
|
26
|
+
|
27
|
+
# TODO Multiple Queries?
|
28
|
+
#
|
29
|
+
def self.queries &block
|
30
|
+
queries_configuration.instance_eval &block
|
31
|
+
routing.freeze
|
32
|
+
end
|
33
|
+
def self.queries_configuration
|
34
|
+
@queries || reset_queries
|
35
|
+
end
|
36
|
+
def self.reset_queries
|
37
|
+
@queries = Configuration::Queries.new routing # Is instance a problem?
|
38
|
+
end
|
39
|
+
|
40
|
+
end
|
@@ -0,0 +1,91 @@
|
|
1
|
+
module Cacher
|
2
|
+
|
3
|
+
module Partial
|
4
|
+
|
5
|
+
# The subtoken partial strategy.
|
6
|
+
#
|
7
|
+
# If given
|
8
|
+
# "florian"
|
9
|
+
# will index
|
10
|
+
# "floria"
|
11
|
+
# "flori"
|
12
|
+
# "flor"
|
13
|
+
# "flo"
|
14
|
+
# "fl"
|
15
|
+
# "f"
|
16
|
+
# Depending on what the given down_to value is. (Example with down_to == 1)
|
17
|
+
#
|
18
|
+
class Subtoken < Strategy
|
19
|
+
|
20
|
+
attr_reader :down_to, :starting_at
|
21
|
+
|
22
|
+
# Down to is how far it will go down in generating the subtokens.
|
23
|
+
#
|
24
|
+
# Examples:
|
25
|
+
# With :hello, and starting_at 0
|
26
|
+
# * down to == 1: [:hello, :hell, :hel, :he, :h]
|
27
|
+
# * down to == 4: [:hello, :hell]
|
28
|
+
#
|
29
|
+
# With :hello, and starting_at -1
|
30
|
+
# * down to == 1: [:hell, :hel, :he, :h]
|
31
|
+
# * down to == 4: [:hell]
|
32
|
+
#
|
33
|
+
def initialize options = {}
|
34
|
+
@down_to = options[:down_to] || 1
|
35
|
+
starting_at = options[:starting_at] || 0
|
36
|
+
@starting_at = starting_at.zero? ? 0 : starting_at - 1
|
37
|
+
end
|
38
|
+
|
39
|
+
# Generates a partial index from the given index.
|
40
|
+
#
|
41
|
+
def generate_from index
|
42
|
+
result = {}
|
43
|
+
|
44
|
+
# Generate for each key token the subtokens.
|
45
|
+
#
|
46
|
+
i = 5000
|
47
|
+
index.each_key do |token|
|
48
|
+
i -= 1
|
49
|
+
if i == 0
|
50
|
+
puts "#{Time.now}: Generating partial tokens for token #{token}. This appears every 5000 tokens."
|
51
|
+
i = 5000
|
52
|
+
end
|
53
|
+
generate_for token, index, result
|
54
|
+
end
|
55
|
+
|
56
|
+
# Remove duplicate ids.
|
57
|
+
#
|
58
|
+
# TODO If it is unique for a subtoken, it is
|
59
|
+
# unique for all derived longer tokens.
|
60
|
+
#
|
61
|
+
result.each_value &:uniq! # Removed because of the set combination operation below
|
62
|
+
|
63
|
+
result
|
64
|
+
end
|
65
|
+
|
66
|
+
private
|
67
|
+
|
68
|
+
# To each shortened token of :test
|
69
|
+
# :test, :tes, :te, :t
|
70
|
+
# add all ids of :test
|
71
|
+
#
|
72
|
+
# "token" here means just text.
|
73
|
+
#
|
74
|
+
# TODO Could be improved by appending the aforegoing ids?
|
75
|
+
#
|
76
|
+
def generate_for token, index, result
|
77
|
+
clipped_token = starting_at.zero? ? token : token[0..starting_at].to_sym
|
78
|
+
clipped_token.subtokens(down_to).each do |subtoken|
|
79
|
+
if result[subtoken]
|
80
|
+
result[subtoken] += index[token] # unique
|
81
|
+
else
|
82
|
+
result[subtoken] = index[token].dup
|
83
|
+
end
|
84
|
+
end
|
85
|
+
end
|
86
|
+
|
87
|
+
end
|
88
|
+
|
89
|
+
end
|
90
|
+
|
91
|
+
end
|
@@ -0,0 +1,15 @@
|
|
1
|
+
module Cacher
|
2
|
+
|
3
|
+
# The partial generator uses a subtoken(downto:1) generator as default.
|
4
|
+
#
|
5
|
+
class PartialGenerator < Generator
|
6
|
+
|
7
|
+
# Generate a similarity index based on the given index.
|
8
|
+
#
|
9
|
+
def generate strategy = Partial::Subtoken.new(:down_to => 1)
|
10
|
+
strategy.generate_from self.index
|
11
|
+
end
|
12
|
+
|
13
|
+
end
|
14
|
+
|
15
|
+
end
|
@@ -0,0 +1,73 @@
|
|
1
|
+
# encoding: utf-8
|
2
|
+
#
|
3
|
+
module Cacher
|
4
|
+
|
5
|
+
module Similarity
|
6
|
+
|
7
|
+
# DoubleLevensthone means that it's a combination of
|
8
|
+
# * DoubleMetaphone
|
9
|
+
# and
|
10
|
+
# * Levenshtein
|
11
|
+
# :)
|
12
|
+
#
|
13
|
+
class DoubleLevenshtone < Strategy
|
14
|
+
|
15
|
+
attr_reader :amount
|
16
|
+
|
17
|
+
#
|
18
|
+
#
|
19
|
+
def initialize amount = 10
|
20
|
+
@amount = amount
|
21
|
+
end
|
22
|
+
|
23
|
+
# Encodes the given symbol.
|
24
|
+
#
|
25
|
+
# Returns a symbol.
|
26
|
+
#
|
27
|
+
def encoded sym
|
28
|
+
codes = Text::Metaphone.double_metaphone sym.to_s
|
29
|
+
codes.first.to_sym unless codes.empty?
|
30
|
+
end
|
31
|
+
|
32
|
+
# Generates an index for the given index (in full index style).
|
33
|
+
#
|
34
|
+
# In the following form:
|
35
|
+
# [:meier, :mueller, :peter, :pater] => { :MR => [:meier], :MLR => [:mueller], :PTR => [:peter, :pater] }
|
36
|
+
#
|
37
|
+
def generate_from index
|
38
|
+
hash = hashify index.keys
|
39
|
+
sort hash
|
40
|
+
end
|
41
|
+
|
42
|
+
private
|
43
|
+
|
44
|
+
# Sorts the index values in place.
|
45
|
+
#
|
46
|
+
def sort index
|
47
|
+
index.each_pair.each do |code, ary|
|
48
|
+
ary.sort_by_levenshtein! code
|
49
|
+
ary.slice! amount, ary.size # size is not perfectly correct, but anyway
|
50
|
+
end
|
51
|
+
index
|
52
|
+
end
|
53
|
+
|
54
|
+
# Hashifies a list of symbols.
|
55
|
+
#
|
56
|
+
# Where:
|
57
|
+
# { encoded_sym => [syms] }
|
58
|
+
#
|
59
|
+
def hashify list
|
60
|
+
list.inject({}) do |total, element|
|
61
|
+
if code = encoded(element)
|
62
|
+
total[code] ||= []
|
63
|
+
total[code] << element
|
64
|
+
end
|
65
|
+
total
|
66
|
+
end
|
67
|
+
end
|
68
|
+
|
69
|
+
end
|
70
|
+
|
71
|
+
end
|
72
|
+
|
73
|
+
end
|
@@ -0,0 +1,25 @@
|
|
1
|
+
module Cacher
|
2
|
+
|
3
|
+
module Similarity
|
4
|
+
|
5
|
+
# Similarity strategy that does nothing.
|
6
|
+
#
|
7
|
+
class None < Strategy
|
8
|
+
|
9
|
+
# Does not encode text. Just returns nil.
|
10
|
+
#
|
11
|
+
def encoded text
|
12
|
+
nil
|
13
|
+
end
|
14
|
+
|
15
|
+
# Returns an empty index.
|
16
|
+
#
|
17
|
+
def generate_from index
|
18
|
+
{}
|
19
|
+
end
|
20
|
+
|
21
|
+
end
|
22
|
+
|
23
|
+
end
|
24
|
+
|
25
|
+
end
|
@@ -0,0 +1,15 @@
|
|
1
|
+
module Cacher
|
2
|
+
|
3
|
+
# Uses no similarity as default.
|
4
|
+
#
|
5
|
+
class SimilarityGenerator < Generator
|
6
|
+
|
7
|
+
# Generate a similarity index based on the given index.
|
8
|
+
#
|
9
|
+
def generate strategy = Similarity::None.new
|
10
|
+
strategy.generate_from self.index
|
11
|
+
end
|
12
|
+
|
13
|
+
end
|
14
|
+
|
15
|
+
end
|
@@ -0,0 +1,39 @@
|
|
1
|
+
module Cacher
|
2
|
+
|
3
|
+
module Weights
|
4
|
+
|
5
|
+
# Uses a logarithmic weight.
|
6
|
+
# If for a key k we have x ids, the weight is:
|
7
|
+
# w(x): log(x)
|
8
|
+
# Special case: If x < 1, then we use 0.
|
9
|
+
#
|
10
|
+
class Logarithmic < Strategy
|
11
|
+
|
12
|
+
# Generates a partial index from the given index.
|
13
|
+
#
|
14
|
+
def generate_from index
|
15
|
+
index.inject({}) do |hash, text_ids|
|
16
|
+
text, ids = *text_ids
|
17
|
+
weight = weight_for ids.size
|
18
|
+
hash[text] ||= weight.round(2) if weight
|
19
|
+
hash
|
20
|
+
end
|
21
|
+
end
|
22
|
+
|
23
|
+
# Sets the weight value.
|
24
|
+
#
|
25
|
+
# If the size is 0 or one, we would get -Infinity or 0.0.
|
26
|
+
# Thus we do not set a value if there is just one. The default, dynamically, is 0.
|
27
|
+
#
|
28
|
+
# BUT: We need the value, even if 0. To designate that there is a weight!
|
29
|
+
#
|
30
|
+
def weight_for amount
|
31
|
+
return 0 if amount < 1
|
32
|
+
Math.log amount
|
33
|
+
end
|
34
|
+
|
35
|
+
end
|
36
|
+
|
37
|
+
end
|
38
|
+
|
39
|
+
end
|
@@ -0,0 +1,15 @@
|
|
1
|
+
module Cacher
|
2
|
+
|
3
|
+
# Uses a logarithmic algorithm as default.
|
4
|
+
#
|
5
|
+
class WeightsGenerator < Generator
|
6
|
+
|
7
|
+
# Generate a weights index based on the given index.
|
8
|
+
#
|
9
|
+
def generate strategy = Weights::Logarithmic.new
|
10
|
+
strategy.generate_from self.index
|
11
|
+
end
|
12
|
+
|
13
|
+
end
|
14
|
+
|
15
|
+
end
|