sunspot_padrino 0.1.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.document +5 -0
- data/.rspec +1 -0
- data/Gemfile +14 -0
- data/Gemfile.lock +74 -0
- data/LICENSE.txt +20 -0
- data/README.md +5 -0
- data/Rakefile +54 -0
- data/VERSION +1 -0
- data/generators/sunspot.rb +57 -0
- data/generators/templates/sunspot.yml +22 -0
- data/lib/sunspot/padrino.rb +73 -0
- data/lib/sunspot/padrino/adapters.rb +90 -0
- data/lib/sunspot/padrino/configuration.rb +404 -0
- data/lib/sunspot/padrino/init.rb +8 -0
- data/lib/sunspot/padrino/request_lifecycle.rb +36 -0
- data/lib/sunspot/padrino/searchable.rb +497 -0
- data/lib/sunspot/padrino/server.rb +99 -0
- data/lib/sunspot/padrino/solr_instrumentation.rb +21 -0
- data/lib/sunspot/padrino/spec_helper.rb +26 -0
- data/lib/sunspot/padrino/stub_session_proxy.rb +168 -0
- data/lib/sunspot/padrino/tasks.rb +67 -0
- data/lib/sunspot_padrino.rb +14 -0
- data/spec/spec_helper.rb +29 -0
- data/spec/sunspot_padrino_spec.rb +7 -0
- metadata +143 -0
@@ -0,0 +1,497 @@
|
|
1
|
+
module Sunspot #:nodoc:
|
2
|
+
module Padrino #:nodoc:
|
3
|
+
#
|
4
|
+
# This module adds Sunspot functionality to ActiveRecord models. As well as
|
5
|
+
# providing class and instance methods, it optionally adds lifecycle hooks
|
6
|
+
# to automatically add and remove models from the Solr index as they are
|
7
|
+
# created and destroyed.
|
8
|
+
#
|
9
|
+
module Searchable
|
10
|
+
class <<self
|
11
|
+
def included(base) #:nodoc:
|
12
|
+
base.module_eval do
|
13
|
+
extend(ActsAsMethods)
|
14
|
+
end
|
15
|
+
end
|
16
|
+
end
|
17
|
+
|
18
|
+
module ActsAsMethods
|
19
|
+
#
|
20
|
+
# Makes a class searchable if it is not already, or adds search
|
21
|
+
# configuration if it is. Note that the options passed in are only used
|
22
|
+
# the first time this method is called for a particular class; so,
|
23
|
+
# search should be defined before activating any mixins that extend
|
24
|
+
# search configuration.
|
25
|
+
#
|
26
|
+
# The block passed into this method is evaluated by the
|
27
|
+
# <code>Sunspot.setup</code> method. See the Sunspot documentation for
|
28
|
+
# complete information on the functionality provided by that method.
|
29
|
+
#
|
30
|
+
# ==== Options (+options+)
|
31
|
+
#
|
32
|
+
# :auto_index<Boolean>::
|
33
|
+
# Automatically index models in Solr when they are saved.
|
34
|
+
# Default: true
|
35
|
+
# :auto_remove<Boolean>::
|
36
|
+
# Automatically remove models from the Solr index when they are
|
37
|
+
# destroyed. <b>Setting this option to +false+ is not recommended
|
38
|
+
# </b>(see the README).
|
39
|
+
# :if<Mixed>::
|
40
|
+
# Only index models in Solr if the method, proc or string evaluates
|
41
|
+
# to true (e.g. <code>:if => :should_index?</code> or <code>:if =>
|
42
|
+
# proc { |model| model.foo > 2 }</code>). Models that do not match
|
43
|
+
# the constraint will be removed from the index upon save. Multiple
|
44
|
+
# constraints can be specified by passing an array (e.g. <code>:if =>
|
45
|
+
# [:method1, :method2]</code>).
|
46
|
+
# :ignore_attribute_changes_of<Array>::
|
47
|
+
# Define attributes, that should not trigger a reindex of that
|
48
|
+
# object. Usual suspects are updated_at or counters.
|
49
|
+
# :only_reindex_attribute_changes_of<Array>::
|
50
|
+
# Define attributes, that are the only attributes that should
|
51
|
+
# trigger a reindex of that object. Useful if there are a small
|
52
|
+
# number of searchable attributes and a large number of attributes
|
53
|
+
# to ignore.
|
54
|
+
# :include<Mixed>::
|
55
|
+
# Define default ActiveRecord includes, set this to allow ActiveRecord
|
56
|
+
# to load required associations when indexing. See ActiveRecord's
|
57
|
+
# documentation on eager-loading for examples on how to set this
|
58
|
+
# Default: []
|
59
|
+
# :unless<Mixed>::
|
60
|
+
# Only index models in Solr if the method, proc or string evaluates
|
61
|
+
# to false (e.g. <code>:unless => :should_not_index?</code> or <code>
|
62
|
+
# :unless => proc { |model| model.foo <= 2 }</code>). Models that do
|
63
|
+
# not match the constraint will be removed from the index upon save.
|
64
|
+
# Multiple constraints can be specified by passing an array (e.g.
|
65
|
+
# <code>:unless => [:method1, :method2]</code>).
|
66
|
+
#
|
67
|
+
# ==== Example
|
68
|
+
#
|
69
|
+
# class Post < ActiveRecord::Base
|
70
|
+
# searchable do
|
71
|
+
# text :title, :body
|
72
|
+
# string :sort_title do
|
73
|
+
# title.downcase.sub(/^(an?|the)/, '')
|
74
|
+
# end
|
75
|
+
# integer :blog_id
|
76
|
+
# time :updated_at
|
77
|
+
# end
|
78
|
+
# end
|
79
|
+
#
|
80
|
+
def searchable(options = {}, &block)
|
81
|
+
Sunspot.setup(self, &block)
|
82
|
+
|
83
|
+
if searchable?
|
84
|
+
sunspot_options[:include].concat(Util::Array(options[:include]))
|
85
|
+
else
|
86
|
+
extend ClassMethods
|
87
|
+
include InstanceMethods
|
88
|
+
|
89
|
+
class_attribute :sunspot_options
|
90
|
+
|
91
|
+
unless options[:auto_index] == false
|
92
|
+
before_save :mark_for_auto_indexing_or_removal
|
93
|
+
after_save :perform_index_tasks
|
94
|
+
end
|
95
|
+
|
96
|
+
unless options[:auto_remove] == false
|
97
|
+
after_destroy do |searchable|
|
98
|
+
searchable.remove_from_index
|
99
|
+
end
|
100
|
+
end
|
101
|
+
options[:include] = Util::Array(options[:include])
|
102
|
+
|
103
|
+
self.sunspot_options = options
|
104
|
+
end
|
105
|
+
end
|
106
|
+
|
107
|
+
#
|
108
|
+
# This method is defined on all ActiveRecord::Base subclasses. It
|
109
|
+
# is false for classes on which #searchable has not been called, and
|
110
|
+
# true for classes on which #searchable has been called.
|
111
|
+
#
|
112
|
+
# ==== Returns
|
113
|
+
#
|
114
|
+
# +false+
|
115
|
+
#
|
116
|
+
def searchable?
|
117
|
+
false
|
118
|
+
end
|
119
|
+
end
|
120
|
+
|
121
|
+
module ClassMethods
|
122
|
+
def self.extended(base) #:nodoc:
|
123
|
+
class <<base
|
124
|
+
alias_method :search, :solr_search unless method_defined? :search
|
125
|
+
alias_method :search_ids, :solr_search_ids unless method_defined? :search_ids
|
126
|
+
alias_method :remove_all_from_index, :solr_remove_all_from_index unless method_defined? :remove_all_from_index
|
127
|
+
alias_method :remove_all_from_index!, :solr_remove_all_from_index! unless method_defined? :remove_all_from_index!
|
128
|
+
alias_method :reindex, :solr_reindex unless method_defined? :reindex
|
129
|
+
alias_method :index, :solr_index unless method_defined? :index
|
130
|
+
alias_method :index_orphans, :solr_index_orphans unless method_defined? :index_orphans
|
131
|
+
alias_method :clean_index_orphans, :solr_clean_index_orphans unless method_defined? :clean_index_orphans
|
132
|
+
end
|
133
|
+
end
|
134
|
+
#
|
135
|
+
# Search for instances of this class in Solr. The block is delegated to
|
136
|
+
# the Sunspot.search method - see the Sunspot documentation for the full
|
137
|
+
# API.
|
138
|
+
#
|
139
|
+
# ==== Example
|
140
|
+
#
|
141
|
+
# Post.search(:include => [:blog]) do
|
142
|
+
# keywords 'best pizza'
|
143
|
+
# with :blog_id, 1
|
144
|
+
# order :updated_at, :desc
|
145
|
+
# facet :category_ids
|
146
|
+
# end
|
147
|
+
#
|
148
|
+
# ==== Options
|
149
|
+
#
|
150
|
+
# :include:: Specify associations to eager load
|
151
|
+
# :select:: Specify columns to select from database when loading results
|
152
|
+
#
|
153
|
+
# ==== Returns
|
154
|
+
#
|
155
|
+
# Sunspot::Search:: Object containing results, totals, facets, etc.
|
156
|
+
#
|
157
|
+
def solr_search(options = {}, &block)
|
158
|
+
solr_execute_search(options) do
|
159
|
+
Sunspot.new_search(self, &block)
|
160
|
+
end
|
161
|
+
end
|
162
|
+
|
163
|
+
#
|
164
|
+
# Get IDs of matching results without loading the result objects from
|
165
|
+
# the database. This method may be useful if search is used as an
|
166
|
+
# intermediate step in a larger find operation. The block is the same
|
167
|
+
# as the block provided to the #search method.
|
168
|
+
#
|
169
|
+
# ==== Returns
|
170
|
+
#
|
171
|
+
# Array:: Array of IDs, in the order returned by the search
|
172
|
+
#
|
173
|
+
def solr_search_ids(&block)
|
174
|
+
solr_execute_search_ids do
|
175
|
+
solr_search(&block)
|
176
|
+
end
|
177
|
+
end
|
178
|
+
|
179
|
+
#
|
180
|
+
# Remove instances of this class from the Solr index.
|
181
|
+
#
|
182
|
+
def solr_remove_all_from_index
|
183
|
+
Sunspot.remove_all(self)
|
184
|
+
end
|
185
|
+
|
186
|
+
#
|
187
|
+
# Remove all instances of this class from the Solr index and immediately
|
188
|
+
# commit.
|
189
|
+
#
|
190
|
+
#
|
191
|
+
def solr_remove_all_from_index!
|
192
|
+
Sunspot.remove_all!(self)
|
193
|
+
end
|
194
|
+
|
195
|
+
#
|
196
|
+
# Completely rebuild the index for this class. First removes all
|
197
|
+
# instances from the index, then loads records and indexes them.
|
198
|
+
#
|
199
|
+
# See #index for information on options, etc.
|
200
|
+
#
|
201
|
+
def solr_reindex(options = {})
|
202
|
+
solr_remove_all_from_index
|
203
|
+
solr_index(options)
|
204
|
+
end
|
205
|
+
|
206
|
+
#
|
207
|
+
# Add/update all existing records in the Solr index. The
|
208
|
+
# +batch_size+ argument specifies how many records to load out of the
|
209
|
+
# database at a time. The default batch size is 50; if nil is passed,
|
210
|
+
# records will not be indexed in batches. By default, a commit is issued
|
211
|
+
# after each batch; passing +false+ for +batch_commit+ will disable
|
212
|
+
# this, and only issue a commit at the end of the process. If associated
|
213
|
+
# objects need to indexed also, you can specify +include+ in format
|
214
|
+
# accepted by ActiveRecord to improve your sql select performance
|
215
|
+
#
|
216
|
+
# ==== Options (passed as a hash)
|
217
|
+
#
|
218
|
+
# batch_size<Integer>:: Batch size with which to load records. Passing
|
219
|
+
# 'nil' will skip batches. Default is 50.
|
220
|
+
# batch_commit<Boolean>:: Flag signalling if a commit should be done after
|
221
|
+
# after each batch is indexed, default is 'true'
|
222
|
+
# include<Mixed>:: include option to be passed to the ActiveRecord find,
|
223
|
+
# used for including associated objects that need to be
|
224
|
+
# indexed with the parent object, accepts all formats
|
225
|
+
# ActiveRecord::Base.find does
|
226
|
+
# first_id:: The lowest possible ID for this class. Defaults to 0, which
|
227
|
+
# is fine for integer IDs; string primary keys will need to
|
228
|
+
# specify something reasonable here.
|
229
|
+
#
|
230
|
+
# ==== Examples
|
231
|
+
#
|
232
|
+
# # index in batches of 50, commit after each
|
233
|
+
# Post.index
|
234
|
+
#
|
235
|
+
# # index all rows at once, then commit
|
236
|
+
# Post.index(:batch_size => nil)
|
237
|
+
#
|
238
|
+
# # index in batches of 50, commit when all batches complete
|
239
|
+
# Post.index(:batch_commit => false)
|
240
|
+
#
|
241
|
+
# # include the associated +author+ object when loading to index
|
242
|
+
# Post.index(:include => :author)
|
243
|
+
#
|
244
|
+
def solr_index(opts={})
|
245
|
+
options = {
|
246
|
+
:batch_size => Sunspot.config.indexing.default_batch_size,
|
247
|
+
:batch_commit => true,
|
248
|
+
:include => self.sunspot_options[:include],
|
249
|
+
:start => opts.delete(:first_id)
|
250
|
+
}.merge(opts)
|
251
|
+
|
252
|
+
if options[:batch_size].to_i > 0
|
253
|
+
batch_counter = 0
|
254
|
+
self.includes(options[:include]).find_in_batches(options.slice(:batch_size, :start)) do |records|
|
255
|
+
|
256
|
+
solr_benchmark(options[:batch_size], batch_counter += 1) do
|
257
|
+
Sunspot.index(records.select { |model| model.indexable? })
|
258
|
+
Sunspot.commit if options[:batch_commit]
|
259
|
+
end
|
260
|
+
|
261
|
+
options[:progress_bar].increment!(records.length) if options[:progress_bar]
|
262
|
+
end
|
263
|
+
else
|
264
|
+
Sunspot.index! self.includes(options[:include]).select(&:indexable?)
|
265
|
+
end
|
266
|
+
|
267
|
+
# perform a final commit if not committing in batches
|
268
|
+
Sunspot.commit unless options[:batch_commit]
|
269
|
+
end
|
270
|
+
|
271
|
+
#
|
272
|
+
# Return the IDs of records of this class that are indexed in Solr but
|
273
|
+
# do not exist in the database. Under normal circumstances, this should
|
274
|
+
# never happen, but this method is provided in case something goes
|
275
|
+
# wrong. Usually you will want to rectify the situation by calling
|
276
|
+
# #clean_index_orphans or #reindex
|
277
|
+
#
|
278
|
+
# ==== Options (passed as a hash)
|
279
|
+
#
|
280
|
+
# batch_size<Integer>:: Batch size with which to load records. Passing
|
281
|
+
# Default is 1000 (from ActiveRecord).
|
282
|
+
#
|
283
|
+
# ==== Returns
|
284
|
+
#
|
285
|
+
# Array:: Collection of IDs that exist in Solr but not in the database
|
286
|
+
def solr_index_orphans(opts={})
|
287
|
+
batch_size = opts[:batch_size] || Sunspot.config.indexing.default_batch_size
|
288
|
+
|
289
|
+
solr_page = 0
|
290
|
+
solr_ids = []
|
291
|
+
while (solr_page = solr_page.next)
|
292
|
+
ids = solr_search_ids { paginate(:page => solr_page, :per_page => 1000) }.to_a
|
293
|
+
break if ids.empty?
|
294
|
+
solr_ids.concat ids
|
295
|
+
end
|
296
|
+
|
297
|
+
return solr_ids - self.connection.select_values("SELECT id FROM #{quoted_table_name}").collect(&:to_i)
|
298
|
+
end
|
299
|
+
|
300
|
+
#
|
301
|
+
# Find IDs of records of this class that are indexed in Solr but do not
|
302
|
+
# exist in the database, and remove them from Solr. Under normal
|
303
|
+
# circumstances, this should not be necessary; this method is provided
|
304
|
+
# in case something goes wrong.
|
305
|
+
#
|
306
|
+
# ==== Options (passed as a hash)
|
307
|
+
#
|
308
|
+
# batch_size<Integer>:: Batch size with which to load records
|
309
|
+
# Default is 50
|
310
|
+
#
|
311
|
+
def solr_clean_index_orphans(opts={})
|
312
|
+
solr_index_orphans(opts).each do |id|
|
313
|
+
new do |fake_instance|
|
314
|
+
fake_instance.id = id
|
315
|
+
end.solr_remove_from_index
|
316
|
+
end
|
317
|
+
end
|
318
|
+
|
319
|
+
#
|
320
|
+
# Classes that have been defined as searchable return +true+ for this
|
321
|
+
# method.
|
322
|
+
#
|
323
|
+
# ==== Returns
|
324
|
+
#
|
325
|
+
# +true+
|
326
|
+
#
|
327
|
+
def searchable?
|
328
|
+
true
|
329
|
+
end
|
330
|
+
|
331
|
+
def solr_execute_search(options = {})
|
332
|
+
options.assert_valid_keys(:include, :select)
|
333
|
+
search = yield
|
334
|
+
unless options.empty?
|
335
|
+
search.build do |query|
|
336
|
+
if options[:include]
|
337
|
+
query.data_accessor_for(self).include = options[:include]
|
338
|
+
end
|
339
|
+
if options[:select]
|
340
|
+
query.data_accessor_for(self).select = options[:select]
|
341
|
+
end
|
342
|
+
end
|
343
|
+
end
|
344
|
+
search.execute
|
345
|
+
end
|
346
|
+
|
347
|
+
def solr_execute_search_ids(options = {})
|
348
|
+
search = yield
|
349
|
+
search.raw_results.map { |raw_result| raw_result.primary_key.to_i }
|
350
|
+
end
|
351
|
+
|
352
|
+
protected
|
353
|
+
|
354
|
+
#
|
355
|
+
# Does some logging for benchmarking indexing performance
|
356
|
+
#
|
357
|
+
def solr_benchmark(batch_size, counter, &block)
|
358
|
+
start = Time.now
|
359
|
+
logger.info("[#{Time.now}] Start Indexing")
|
360
|
+
yield
|
361
|
+
elapsed = Time.now-start
|
362
|
+
logger.info("[#{Time.now}] Completed Indexing. Rows indexed #{counter * batch_size}. Rows/sec: #{batch_size/elapsed.to_f} (Elapsed: #{elapsed} sec.)")
|
363
|
+
end
|
364
|
+
|
365
|
+
end
|
366
|
+
|
367
|
+
module InstanceMethods
|
368
|
+
def self.included(base) #:nodoc:
|
369
|
+
base.module_eval do
|
370
|
+
alias_method :index, :solr_index unless method_defined? :index
|
371
|
+
alias_method :index!, :solr_index! unless method_defined? :index!
|
372
|
+
alias_method :remove_from_index, :solr_remove_from_index unless method_defined? :remove_from_index
|
373
|
+
alias_method :remove_from_index!, :solr_remove_from_index! unless method_defined? :remove_from_index!
|
374
|
+
alias_method :more_like_this, :solr_more_like_this unless method_defined? :more_like_this
|
375
|
+
alias_method :more_like_this_ids, :solr_more_like_this_ids unless method_defined? :more_like_this_ids
|
376
|
+
end
|
377
|
+
end
|
378
|
+
#
|
379
|
+
# Index the model in Solr. If the model is already indexed, it will be
|
380
|
+
# updated. Using the defaults, you will usually not need to call this
|
381
|
+
# method, as models are indexed automatically when they are created or
|
382
|
+
# updated. If you have disabled automatic indexing (see
|
383
|
+
# ClassMethods#searchable), this method allows you to manage indexing
|
384
|
+
# manually.
|
385
|
+
#
|
386
|
+
def solr_index
|
387
|
+
Sunspot.index(self)
|
388
|
+
end
|
389
|
+
|
390
|
+
#
|
391
|
+
# Index the model in Solr and immediately commit. See #index
|
392
|
+
#
|
393
|
+
def solr_index!
|
394
|
+
Sunspot.index!(self)
|
395
|
+
end
|
396
|
+
|
397
|
+
#
|
398
|
+
# Remove the model from the Solr index. Using the defaults, this should
|
399
|
+
# not be necessary, as models will automatically be removed from the
|
400
|
+
# index when they are destroyed. If you disable automatic removal
|
401
|
+
# (which is not recommended!), you can use this method to manage removal
|
402
|
+
# manually.
|
403
|
+
#
|
404
|
+
def solr_remove_from_index
|
405
|
+
Sunspot.remove(self)
|
406
|
+
end
|
407
|
+
|
408
|
+
#
|
409
|
+
# Remove the model from the Solr index and commit immediately. See
|
410
|
+
# #remove_from_index
|
411
|
+
#
|
412
|
+
def solr_remove_from_index!
|
413
|
+
Sunspot.remove!(self)
|
414
|
+
end
|
415
|
+
|
416
|
+
def solr_more_like_this(*args, &block)
|
417
|
+
options = args.extract_options!
|
418
|
+
self.class.solr_execute_search(options) do
|
419
|
+
Sunspot.new_more_like_this(self, *args, &block)
|
420
|
+
end
|
421
|
+
end
|
422
|
+
|
423
|
+
def solr_more_like_this_ids(&block)
|
424
|
+
self.class.solr_execute_search_ids do
|
425
|
+
solr_more_like_this(&block)
|
426
|
+
end
|
427
|
+
end
|
428
|
+
|
429
|
+
def indexable?
|
430
|
+
# options[:if] is not specified or they successfully pass
|
431
|
+
if_passes = self.class.sunspot_options[:if].nil? ||
|
432
|
+
constraint_passes?(self.class.sunspot_options[:if])
|
433
|
+
|
434
|
+
# options[:unless] is not specified or they successfully pass
|
435
|
+
unless_passes = self.class.sunspot_options[:unless].nil? ||
|
436
|
+
!constraint_passes?(self.class.sunspot_options[:unless])
|
437
|
+
|
438
|
+
if_passes and unless_passes
|
439
|
+
end
|
440
|
+
|
441
|
+
private
|
442
|
+
|
443
|
+
def constraint_passes?(constraint)
|
444
|
+
case constraint
|
445
|
+
when Symbol
|
446
|
+
self.__send__(constraint)
|
447
|
+
when String
|
448
|
+
self.__send__(constraint.to_sym)
|
449
|
+
when Enumerable
|
450
|
+
# All constraints must pass
|
451
|
+
constraint.all? { |inner_constraint| constraint_passes?(inner_constraint) }
|
452
|
+
else
|
453
|
+
if constraint.respond_to?(:call) # could be a Proc or anything else that responds to call
|
454
|
+
constraint.call(self)
|
455
|
+
else
|
456
|
+
raise ArgumentError, "Unknown constraint type: #{constraint} (#{constraint.class})"
|
457
|
+
end
|
458
|
+
end
|
459
|
+
end
|
460
|
+
|
461
|
+
def mark_for_auto_indexing_or_removal
|
462
|
+
if indexable?
|
463
|
+
# :if/:unless constraints pass or were not present
|
464
|
+
|
465
|
+
@marked_for_auto_indexing =
|
466
|
+
if !new_record? && ignore_attributes = self.class.sunspot_options[:ignore_attribute_changes_of]
|
467
|
+
!(changed.map { |attr| attr.to_sym } - ignore_attributes).blank?
|
468
|
+
elsif !new_record? && only_attributes = self.class.sunspot_options[:only_reindex_attribute_changes_of]
|
469
|
+
!(changed.map { |attr| attr.to_sym } & only_attributes).blank?
|
470
|
+
else
|
471
|
+
true
|
472
|
+
end
|
473
|
+
|
474
|
+
@marked_for_auto_removal = false
|
475
|
+
else
|
476
|
+
# :if/:unless constraints did not pass; do not auto index and
|
477
|
+
# actually go one step further by removing it from the index
|
478
|
+
@marked_for_auto_indexing = false
|
479
|
+
@marked_for_auto_removal = true
|
480
|
+
end
|
481
|
+
|
482
|
+
true
|
483
|
+
end
|
484
|
+
|
485
|
+
def perform_index_tasks
|
486
|
+
if @marked_for_auto_indexing
|
487
|
+
solr_index
|
488
|
+
remove_instance_variable(:@marked_for_auto_indexing)
|
489
|
+
elsif @marked_for_auto_removal
|
490
|
+
solr_remove_from_index
|
491
|
+
remove_instance_variable(:@marked_for_auto_removal)
|
492
|
+
end
|
493
|
+
end
|
494
|
+
end
|
495
|
+
end
|
496
|
+
end
|
497
|
+
end
|