searchkick 4.4.0 → 5.3.1
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/CHANGELOG.md +160 -3
- data/LICENSE.txt +1 -1
- data/README.md +567 -421
- data/lib/searchkick/bulk_reindex_job.rb +12 -8
- data/lib/searchkick/controller_runtime.rb +40 -0
- data/lib/searchkick/index.rb +167 -74
- data/lib/searchkick/index_cache.rb +30 -0
- data/lib/searchkick/index_options.rb +465 -404
- data/lib/searchkick/indexer.rb +15 -8
- data/lib/searchkick/log_subscriber.rb +57 -0
- data/lib/searchkick/middleware.rb +9 -2
- data/lib/searchkick/model.rb +50 -51
- data/lib/searchkick/process_batch_job.rb +9 -25
- data/lib/searchkick/process_queue_job.rb +4 -3
- data/lib/searchkick/query.rb +106 -77
- data/lib/searchkick/record_data.rb +1 -1
- data/lib/searchkick/record_indexer.rb +136 -51
- data/lib/searchkick/reindex_queue.rb +51 -9
- data/lib/searchkick/reindex_v2_job.rb +10 -34
- data/lib/searchkick/relation.rb +247 -0
- data/lib/searchkick/relation_indexer.rb +155 -0
- data/lib/searchkick/results.rb +131 -96
- data/lib/searchkick/version.rb +1 -1
- data/lib/searchkick/where.rb +11 -0
- data/lib/searchkick.rb +202 -96
- data/lib/tasks/searchkick.rake +14 -10
- metadata +18 -85
- data/CONTRIBUTING.md +0 -53
- data/lib/searchkick/bulk_indexer.rb +0 -173
- data/lib/searchkick/logging.rb +0 -246
data/lib/searchkick/query.rb
CHANGED
@@ -1,5 +1,6 @@
|
|
1
1
|
module Searchkick
|
2
2
|
class Query
|
3
|
+
include Enumerable
|
3
4
|
extend Forwardable
|
4
5
|
|
5
6
|
@@metric_aggs = [:avg, :cardinality, :max, :min, :sum]
|
@@ -12,11 +13,12 @@ module Searchkick
|
|
12
13
|
:took, :error, :model_name, :entry_name, :total_count, :total_entries,
|
13
14
|
:current_page, :per_page, :limit_value, :padding, :total_pages, :num_pages,
|
14
15
|
:offset_value, :offset, :previous_page, :prev_page, :next_page, :first_page?, :last_page?,
|
15
|
-
:out_of_range?, :hits, :response, :to_a, :first, :scroll
|
16
|
+
:out_of_range?, :hits, :response, :to_a, :first, :scroll, :highlights, :with_highlights,
|
17
|
+
:with_score, :misspellings?, :scroll_id, :clear_scroll, :missing_records, :with_hit
|
16
18
|
|
17
19
|
def initialize(klass, term = "*", **options)
|
18
20
|
unknown_keywords = options.keys - [:aggs, :block, :body, :body_options, :boost,
|
19
|
-
:boost_by, :boost_by_distance, :boost_by_recency, :boost_where, :conversions, :conversions_term, :debug, :emoji, :exclude, :
|
21
|
+
:boost_by, :boost_by_distance, :boost_by_recency, :boost_where, :conversions, :conversions_term, :debug, :emoji, :exclude, :explain,
|
20
22
|
:fields, :highlight, :includes, :index_name, :indices_boost, :limit, :load,
|
21
23
|
:match, :misspellings, :models, :model_includes, :offset, :operator, :order, :padding, :page, :per_page, :profile,
|
22
24
|
:request_params, :routing, :scope_results, :scroll, :select, :similar, :smart_aggs, :suggest, :total_entries, :track, :type, :where]
|
@@ -25,7 +27,7 @@ module Searchkick
|
|
25
27
|
term = term.to_s
|
26
28
|
|
27
29
|
if options[:emoji]
|
28
|
-
term = EmojiParser.parse_unicode(term) { |e| " #{e.name} " }.strip
|
30
|
+
term = EmojiParser.parse_unicode(term) { |e| " #{e.name.tr('_', ' ')} " }.strip
|
29
31
|
end
|
30
32
|
|
31
33
|
@klass = klass
|
@@ -73,7 +75,8 @@ module Searchkick
|
|
73
75
|
elsif searchkick_index
|
74
76
|
searchkick_index.name
|
75
77
|
else
|
76
|
-
|
78
|
+
# fixes warning about accessing system indices
|
79
|
+
"*,-.*"
|
77
80
|
end
|
78
81
|
|
79
82
|
params = {
|
@@ -109,7 +112,12 @@ module Searchkick
|
|
109
112
|
request_params = query.except(:index, :type, :body)
|
110
113
|
|
111
114
|
# no easy way to tell which host the client will use
|
112
|
-
host =
|
115
|
+
host =
|
116
|
+
if Searchkick.client.transport.respond_to?(:transport)
|
117
|
+
Searchkick.client.transport.transport.hosts.first
|
118
|
+
else
|
119
|
+
Searchkick.client.transport.hosts.first
|
120
|
+
end
|
113
121
|
credentials = host[:user] || host[:password] ? "#{host[:user]}:#{host[:password]}@" : nil
|
114
122
|
params = ["pretty"]
|
115
123
|
request_params.each do |k, v|
|
@@ -140,9 +148,6 @@ module Searchkick
|
|
140
148
|
}
|
141
149
|
|
142
150
|
if options[:debug]
|
143
|
-
# can remove when minimum Ruby version is 2.5
|
144
|
-
require "pp"
|
145
|
-
|
146
151
|
puts "Searchkick Version: #{Searchkick::VERSION}"
|
147
152
|
puts "Elasticsearch Version: #{Searchkick.server_version}"
|
148
153
|
puts
|
@@ -182,11 +187,11 @@ module Searchkick
|
|
182
187
|
end
|
183
188
|
|
184
189
|
# set execute for multi search
|
185
|
-
@execute =
|
190
|
+
@execute = Results.new(searchkick_klass, response, opts)
|
186
191
|
end
|
187
192
|
|
188
193
|
def retry_misspellings?(response)
|
189
|
-
@misspellings_below &&
|
194
|
+
@misspellings_below && response["error"].nil? && Results.new(searchkick_klass, response).total_count < @misspellings_below
|
190
195
|
end
|
191
196
|
|
192
197
|
private
|
@@ -194,7 +199,11 @@ module Searchkick
|
|
194
199
|
def handle_error(e)
|
195
200
|
status_code = e.message[1..3].to_i
|
196
201
|
if status_code == 404
|
197
|
-
|
202
|
+
if e.message.include?("No search context found for id")
|
203
|
+
raise MissingIndexError, "No search context found for id"
|
204
|
+
else
|
205
|
+
raise MissingIndexError, "Index missing - run #{reindex_command}"
|
206
|
+
end
|
198
207
|
elsif status_code == 500 && (
|
199
208
|
e.message.include?("IllegalArgumentException[minimumSimilarity >= 1]") ||
|
200
209
|
e.message.include?("No query registered for [multi_match]") ||
|
@@ -202,14 +211,14 @@ module Searchkick
|
|
202
211
|
e.message.include?("No query registered for [function_score]")
|
203
212
|
)
|
204
213
|
|
205
|
-
raise UnsupportedVersionError
|
214
|
+
raise UnsupportedVersionError
|
206
215
|
elsif status_code == 400
|
207
216
|
if (
|
208
217
|
e.message.include?("bool query does not support [filter]") ||
|
209
218
|
e.message.include?("[bool] filter does not support [filter]")
|
210
219
|
)
|
211
220
|
|
212
|
-
raise UnsupportedVersionError
|
221
|
+
raise UnsupportedVersionError
|
213
222
|
elsif e.message =~ /analyzer \[searchkick_.+\] not found/
|
214
223
|
raise InvalidQueryError, "Bad mapping - run #{reindex_command}"
|
215
224
|
else
|
@@ -225,7 +234,14 @@ module Searchkick
|
|
225
234
|
end
|
226
235
|
|
227
236
|
def execute_search
|
228
|
-
|
237
|
+
name = searchkick_klass ? "#{searchkick_klass.name} Search" : "Search"
|
238
|
+
event = {
|
239
|
+
name: name,
|
240
|
+
query: params
|
241
|
+
}
|
242
|
+
ActiveSupport::Notifications.instrument("search.searchkick", event) do
|
243
|
+
Searchkick.client.search(params)
|
244
|
+
end
|
229
245
|
end
|
230
246
|
|
231
247
|
def prepare
|
@@ -239,9 +255,15 @@ module Searchkick
|
|
239
255
|
default_limit = searchkick_options[:deep_paging] ? 1_000_000_000 : 10_000
|
240
256
|
per_page = (options[:limit] || options[:per_page] || default_limit).to_i
|
241
257
|
padding = [options[:padding].to_i, 0].max
|
242
|
-
offset = options[:offset] || (page - 1) * per_page + padding
|
258
|
+
offset = (options[:offset] || (page - 1) * per_page + padding).to_i
|
243
259
|
scroll = options[:scroll]
|
244
260
|
|
261
|
+
max_result_window = searchkick_options[:max_result_window]
|
262
|
+
if max_result_window
|
263
|
+
offset = max_result_window if offset > max_result_window
|
264
|
+
per_page = max_result_window - offset if offset + per_page > max_result_window
|
265
|
+
end
|
266
|
+
|
245
267
|
# model and eager loading
|
246
268
|
load = options[:load].nil? ? true : options[:load]
|
247
269
|
|
@@ -260,9 +282,10 @@ module Searchkick
|
|
260
282
|
should = []
|
261
283
|
|
262
284
|
if options[:similar]
|
285
|
+
like = options[:similar] == true ? term : options[:similar]
|
263
286
|
query = {
|
264
287
|
more_like_this: {
|
265
|
-
like:
|
288
|
+
like: like,
|
266
289
|
min_doc_freq: 1,
|
267
290
|
min_term_freq: 1,
|
268
291
|
analyzer: "searchkick_search2"
|
@@ -350,11 +373,11 @@ module Searchkick
|
|
350
373
|
field_misspellings = misspellings && (!misspellings_fields || misspellings_fields.include?(base_field(field)))
|
351
374
|
|
352
375
|
if field == "_all" || field.end_with?(".analyzed")
|
353
|
-
shared_options[:cutoff_frequency] = 0.001 unless operator.to_s == "and" || field_misspellings == false || (!below73? && !track_total_hits?)
|
376
|
+
shared_options[:cutoff_frequency] = 0.001 unless operator.to_s == "and" || field_misspellings == false || (!below73? && !track_total_hits?) || match_type == :match_phrase || !below80? || Searchkick.opensearch?
|
354
377
|
qs << shared_options.merge(analyzer: "searchkick_search")
|
355
378
|
|
356
|
-
# searchkick_search and searchkick_search2 are the same for
|
357
|
-
unless %w(japanese korean polish ukrainian vietnamese).include?(searchkick_options[:language])
|
379
|
+
# searchkick_search and searchkick_search2 are the same for some languages
|
380
|
+
unless %w(japanese japanese2 korean polish ukrainian vietnamese).include?(searchkick_options[:language])
|
358
381
|
qs << shared_options.merge(analyzer: "searchkick_search2")
|
359
382
|
end
|
360
383
|
exclude_analyzer = "searchkick_search2"
|
@@ -375,11 +398,6 @@ module Searchkick
|
|
375
398
|
|
376
399
|
if field.start_with?("*.")
|
377
400
|
q2 = qs.map { |q| {multi_match: q.merge(fields: [field], type: match_type == :match_phrase ? "phrase" : "best_fields")} }
|
378
|
-
if below61?
|
379
|
-
q2.each do |q|
|
380
|
-
q[:multi_match].delete(:fuzzy_transpositions)
|
381
|
-
end
|
382
|
-
end
|
383
401
|
else
|
384
402
|
q2 = qs.map { |q| {match_type => {field => q}} }
|
385
403
|
end
|
@@ -431,7 +449,7 @@ module Searchkick
|
|
431
449
|
payload = {}
|
432
450
|
|
433
451
|
# type when inheritance
|
434
|
-
where = (options[:where] || {}).dup
|
452
|
+
where = ensure_permitted(options[:where] || {}).dup
|
435
453
|
if searchkick_options[:inheritance] && (options[:type] || (klass != searchkick_klass && searchkick_index))
|
436
454
|
where[:type] = [options[:type] || klass].flatten.map { |v| searchkick_index.klass_document_type(v, true) }
|
437
455
|
end
|
@@ -491,7 +509,7 @@ module Searchkick
|
|
491
509
|
set_highlights(payload, fields) if options[:highlight]
|
492
510
|
|
493
511
|
# timeout shortly after client times out
|
494
|
-
payload[:timeout] ||= "#{Searchkick.search_timeout + 1}
|
512
|
+
payload[:timeout] ||= "#{((Searchkick.search_timeout + 1) * 1000).round}ms"
|
495
513
|
|
496
514
|
# An empty array will cause only the _id and _type for each hit to be returned
|
497
515
|
# https://www.elastic.co/guide/en/elasticsearch/reference/current/search-request-source-filtering.html
|
@@ -688,9 +706,9 @@ module Searchkick
|
|
688
706
|
def set_boost_by(multiply_filters, custom_filters)
|
689
707
|
boost_by = options[:boost_by] || {}
|
690
708
|
if boost_by.is_a?(Array)
|
691
|
-
boost_by =
|
709
|
+
boost_by = boost_by.to_h { |f| [f, {factor: 1}] }
|
692
710
|
elsif boost_by.is_a?(Hash)
|
693
|
-
multiply_by, boost_by = boost_by.partition { |_, v| v.delete(:boost_mode) == "multiply" }.map
|
711
|
+
multiply_by, boost_by = boost_by.partition { |_, v| v.delete(:boost_mode) == "multiply" }.map(&:to_h)
|
694
712
|
end
|
695
713
|
boost_by[options[:boost]] = {factor: 1} if options[:boost]
|
696
714
|
|
@@ -755,7 +773,7 @@ module Searchkick
|
|
755
773
|
|
756
774
|
def set_highlights(payload, fields)
|
757
775
|
payload[:highlight] = {
|
758
|
-
fields:
|
776
|
+
fields: fields.to_h { |f| [f, {}] },
|
759
777
|
fragment_size: 0
|
760
778
|
}
|
761
779
|
|
@@ -789,7 +807,7 @@ module Searchkick
|
|
789
807
|
aggs = options[:aggs]
|
790
808
|
payload[:aggs] = {}
|
791
809
|
|
792
|
-
aggs =
|
810
|
+
aggs = aggs.to_h { |f| [f, {}] } if aggs.is_a?(Array) # convert to more advanced syntax
|
793
811
|
aggs.each do |field, agg_options|
|
794
812
|
size = agg_options[:limit] ? agg_options[:limit] : 1_000
|
795
813
|
shared_agg_options = agg_options.except(:limit, :field, :ranges, :date_ranges, :where)
|
@@ -828,8 +846,9 @@ module Searchkick
|
|
828
846
|
end
|
829
847
|
|
830
848
|
where = {}
|
831
|
-
where = (options[:where] || {}).reject { |k| k == field } unless options[:smart_aggs] == false
|
832
|
-
|
849
|
+
where = ensure_permitted(options[:where] || {}).reject { |k| k == field } unless options[:smart_aggs] == false
|
850
|
+
agg_where = ensure_permitted(agg_options[:where] || {})
|
851
|
+
agg_filters = where_filters(where.merge(agg_where))
|
833
852
|
|
834
853
|
# only do one level comparison for simplicity
|
835
854
|
filters.select! do |filter|
|
@@ -864,19 +883,17 @@ module Searchkick
|
|
864
883
|
}
|
865
884
|
end
|
866
885
|
|
867
|
-
# TODO id transformation for arrays
|
868
886
|
def set_order(payload)
|
869
|
-
|
870
|
-
id_field = :_id
|
871
|
-
payload[:sort] = order.is_a?(Array) ? order : Hash[order.map { |k, v| [k.to_s == "id" ? id_field : k, v] }]
|
887
|
+
payload[:sort] = options[:order].is_a?(Enumerable) ? options[:order] : {options[:order] => :asc}
|
872
888
|
end
|
873
889
|
|
874
|
-
|
875
|
-
|
876
|
-
|
877
|
-
|
878
|
-
|
890
|
+
# provides *very* basic protection from unfiltered parameters
|
891
|
+
# this is not meant to be comprehensive and may be expanded in the future
|
892
|
+
def ensure_permitted(obj)
|
893
|
+
obj.to_h
|
894
|
+
end
|
879
895
|
|
896
|
+
def where_filters(where)
|
880
897
|
filters = []
|
881
898
|
(where || {}).each do |field, value|
|
882
899
|
field = :_id if field.to_s == "id"
|
@@ -896,12 +913,7 @@ module Searchkick
|
|
896
913
|
else
|
897
914
|
# expand ranges
|
898
915
|
if value.is_a?(Range)
|
899
|
-
|
900
|
-
if value.end.nil? || (value.end.respond_to?(:infinite?) && value.end.infinite?)
|
901
|
-
value = {gte: value.first}
|
902
|
-
else
|
903
|
-
value = {gte: value.first, (value.exclude_end? ? :lt : :lte) => value.last}
|
904
|
-
end
|
916
|
+
value = expand_range(value)
|
905
917
|
end
|
906
918
|
|
907
919
|
value = {in: value} if value.is_a?(Array)
|
@@ -953,7 +965,7 @@ module Searchkick
|
|
953
965
|
}
|
954
966
|
}
|
955
967
|
}
|
956
|
-
when :like
|
968
|
+
when :like, :ilike
|
957
969
|
# based on Postgres
|
958
970
|
# https://www.postgresql.org/docs/current/functions-matching.html
|
959
971
|
# % matches zero or more characters
|
@@ -961,13 +973,22 @@ module Searchkick
|
|
961
973
|
# \ is escape character
|
962
974
|
# escape Lucene reserved characters
|
963
975
|
# https://www.elastic.co/guide/en/elasticsearch/reference/current/regexp-syntax.html#regexp-optional-operators
|
964
|
-
reserved = %w(. ? + * | { } [ ] ( ) "
|
976
|
+
reserved = %w(\\ . ? + * | { } [ ] ( ) ")
|
965
977
|
regex = op_value.dup
|
966
978
|
reserved.each do |v|
|
967
|
-
regex.gsub!(v, "
|
979
|
+
regex.gsub!(v, "\\\\" + v)
|
968
980
|
end
|
969
981
|
regex = regex.gsub(/(?<!\\)%/, ".*").gsub(/(?<!\\)_/, ".").gsub("\\%", "%").gsub("\\_", "_")
|
970
|
-
|
982
|
+
|
983
|
+
if op == :ilike
|
984
|
+
if below710?
|
985
|
+
raise ArgumentError, "ilike requires Elasticsearch 7.10+"
|
986
|
+
else
|
987
|
+
filters << {regexp: {field => {value: regex, flags: "NONE", case_insensitive: true}}}
|
988
|
+
end
|
989
|
+
else
|
990
|
+
filters << {regexp: {field => {value: regex, flags: "NONE"}}}
|
991
|
+
end
|
971
992
|
when :prefix
|
972
993
|
filters << {prefix: {field => {value: op_value}}}
|
973
994
|
when :regexp # support for regexp queries without using a regexp ruby object
|
@@ -994,7 +1015,7 @@ module Searchkick
|
|
994
1015
|
when :lte
|
995
1016
|
{to: op_value, include_upper: true}
|
996
1017
|
else
|
997
|
-
raise "Unknown where operator: #{op.inspect}"
|
1018
|
+
raise ArgumentError, "Unknown where operator: #{op.inspect}"
|
998
1019
|
end
|
999
1020
|
# issue 132
|
1000
1021
|
if (existing = filters.find { |f| f[:range] && f[:range][field] })
|
@@ -1022,40 +1043,37 @@ module Searchkick
|
|
1022
1043
|
elsif value.nil?
|
1023
1044
|
{bool: {must_not: {exists: {field: field}}}}
|
1024
1045
|
elsif value.is_a?(Regexp)
|
1025
|
-
if value.casefold?
|
1026
|
-
Searchkick.warn("Case-insensitive flag does not work with Elasticsearch")
|
1027
|
-
end
|
1028
|
-
|
1029
1046
|
source = value.source
|
1030
|
-
|
1031
|
-
|
1032
|
-
Searchkick.warn("Regular expressions are always anchored in Elasticsearch")
|
1033
|
-
end
|
1047
|
+
|
1048
|
+
# TODO handle other regexp options
|
1034
1049
|
|
1035
1050
|
# TODO handle other anchor characters, like ^, $, \Z
|
1036
1051
|
if source.start_with?("\\A")
|
1037
1052
|
source = source[2..-1]
|
1038
1053
|
else
|
1039
|
-
|
1040
|
-
# source = ".*#{source}"
|
1054
|
+
source = ".*#{source}"
|
1041
1055
|
end
|
1042
1056
|
|
1043
1057
|
if source.end_with?("\\z")
|
1044
1058
|
source = source[0..-3]
|
1045
1059
|
else
|
1046
|
-
|
1047
|
-
# source = "#{source}.*"
|
1060
|
+
source = "#{source}.*"
|
1048
1061
|
end
|
1049
1062
|
|
1050
|
-
|
1063
|
+
if below710?
|
1064
|
+
if value.casefold?
|
1065
|
+
raise ArgumentError, "Case-insensitive flag does not work with Elasticsearch < 7.10"
|
1066
|
+
end
|
1067
|
+
{regexp: {field => {value: source, flags: "NONE"}}}
|
1068
|
+
else
|
1069
|
+
{regexp: {field => {value: source, flags: "NONE", case_insensitive: value.casefold?}}}
|
1070
|
+
end
|
1051
1071
|
else
|
1052
1072
|
# TODO add this for other values
|
1053
1073
|
if value.as_json.is_a?(Enumerable)
|
1054
1074
|
# query will fail, but this is better
|
1055
1075
|
# same message as Active Record
|
1056
|
-
|
1057
|
-
# raise InvalidQueryError for backward compatibility
|
1058
|
-
raise Searchkick::InvalidQueryError, "can't cast #{value.class.name}"
|
1076
|
+
raise TypeError, "can't cast #{value.class.name}"
|
1059
1077
|
end
|
1060
1078
|
|
1061
1079
|
{term: {field => {value: value}}}
|
@@ -1118,26 +1136,29 @@ module Searchkick
|
|
1118
1136
|
end
|
1119
1137
|
end
|
1120
1138
|
|
1139
|
+
def expand_range(range)
|
1140
|
+
expanded = {}
|
1141
|
+
expanded[:gte] = range.begin if range.begin
|
1142
|
+
|
1143
|
+
if range.end && !(range.end.respond_to?(:infinite?) && range.end.infinite?)
|
1144
|
+
expanded[range.exclude_end? ? :lt : :lte] = range.end
|
1145
|
+
end
|
1146
|
+
|
1147
|
+
expanded
|
1148
|
+
end
|
1149
|
+
|
1121
1150
|
def base_field(k)
|
1122
1151
|
k.sub(/\.(analyzed|word_start|word_middle|word_end|text_start|text_middle|text_end|exact)\z/, "")
|
1123
1152
|
end
|
1124
1153
|
|
1125
1154
|
def track_total_hits?
|
1126
|
-
|
1155
|
+
searchkick_options[:deep_paging] || body_options[:track_total_hits]
|
1127
1156
|
end
|
1128
1157
|
|
1129
1158
|
def body_options
|
1130
1159
|
options[:body_options] || {}
|
1131
1160
|
end
|
1132
1161
|
|
1133
|
-
def below61?
|
1134
|
-
Searchkick.server_below?("6.1.0")
|
1135
|
-
end
|
1136
|
-
|
1137
|
-
def below70?
|
1138
|
-
Searchkick.server_below?("7.0.0")
|
1139
|
-
end
|
1140
|
-
|
1141
1162
|
def below73?
|
1142
1163
|
Searchkick.server_below?("7.3.0")
|
1143
1164
|
end
|
@@ -1145,5 +1166,13 @@ module Searchkick
|
|
1145
1166
|
def below75?
|
1146
1167
|
Searchkick.server_below?("7.5.0")
|
1147
1168
|
end
|
1169
|
+
|
1170
|
+
def below710?
|
1171
|
+
Searchkick.server_below?("7.10.0")
|
1172
|
+
end
|
1173
|
+
|
1174
|
+
def below80?
|
1175
|
+
Searchkick.server_below?("8.0.0")
|
1176
|
+
end
|
1148
1177
|
end
|
1149
1178
|
end
|
@@ -25,6 +25,7 @@ module Searchkick
|
|
25
25
|
{delete: record_data}
|
26
26
|
end
|
27
27
|
|
28
|
+
# custom id can be useful for load: false
|
28
29
|
def search_id
|
29
30
|
id = record.respond_to?(:search_document_id) ? record.search_document_id : record.id
|
30
31
|
id.is_a?(Numeric) ? id : id.to_s
|
@@ -39,7 +40,6 @@ module Searchkick
|
|
39
40
|
_index: index.name,
|
40
41
|
_id: search_id
|
41
42
|
}
|
42
|
-
data[:_type] = document_type if Searchkick.server_below7?
|
43
43
|
data[:routing] = record.search_routing if record.respond_to?(:search_routing)
|
44
44
|
data
|
45
45
|
end
|
@@ -1,78 +1,163 @@
|
|
1
1
|
module Searchkick
|
2
2
|
class RecordIndexer
|
3
|
-
attr_reader :
|
3
|
+
attr_reader :index
|
4
4
|
|
5
|
-
def initialize(
|
6
|
-
@
|
7
|
-
@index = record.class.searchkick_index
|
5
|
+
def initialize(index)
|
6
|
+
@index = index
|
8
7
|
end
|
9
8
|
|
10
|
-
def reindex(
|
11
|
-
|
12
|
-
|
13
|
-
|
14
|
-
|
15
|
-
mode ||= Searchkick.callbacks_value || index.options[:callbacks] || true
|
9
|
+
def reindex(records, mode:, method_name:, full: false, single: false)
|
10
|
+
# prevents exists? check if records is a relation
|
11
|
+
records = records.to_a
|
12
|
+
return if records.empty?
|
16
13
|
|
17
14
|
case mode
|
15
|
+
when :async
|
16
|
+
unless defined?(ActiveJob)
|
17
|
+
raise Error, "Active Job not found"
|
18
|
+
end
|
19
|
+
|
20
|
+
# we could likely combine ReindexV2Job, BulkReindexJob, and ProcessBatchJob
|
21
|
+
# but keep them separate for now
|
22
|
+
if single
|
23
|
+
record = records.first
|
24
|
+
|
25
|
+
# always pass routing in case record is deleted
|
26
|
+
# before the async job runs
|
27
|
+
if record.respond_to?(:search_routing)
|
28
|
+
routing = record.search_routing
|
29
|
+
end
|
30
|
+
|
31
|
+
Searchkick::ReindexV2Job.perform_later(
|
32
|
+
record.class.name,
|
33
|
+
record.id.to_s,
|
34
|
+
method_name ? method_name.to_s : nil,
|
35
|
+
routing: routing,
|
36
|
+
index_name: index.name
|
37
|
+
)
|
38
|
+
else
|
39
|
+
Searchkick::BulkReindexJob.perform_later(
|
40
|
+
class_name: records.first.class.searchkick_options[:class_name],
|
41
|
+
record_ids: records.map { |r| r.id.to_s },
|
42
|
+
index_name: index.name,
|
43
|
+
method_name: method_name ? method_name.to_s : nil
|
44
|
+
)
|
45
|
+
end
|
18
46
|
when :queue
|
19
47
|
if method_name
|
20
|
-
raise
|
48
|
+
raise Error, "Partial reindex not supported with queue option"
|
21
49
|
end
|
22
50
|
|
23
|
-
|
24
|
-
|
25
|
-
|
26
|
-
|
27
|
-
|
51
|
+
index.reindex_queue.push_records(records)
|
52
|
+
when true, :inline
|
53
|
+
index_records, other_records = records.partition { |r| index_record?(r) }
|
54
|
+
import_inline(index_records, !full ? other_records : [], method_name: method_name, single: single)
|
55
|
+
else
|
56
|
+
raise ArgumentError, "Invalid value for mode"
|
57
|
+
end
|
28
58
|
|
29
|
-
|
30
|
-
|
31
|
-
|
32
|
-
index.reindex_queue.push(value)
|
33
|
-
when :async
|
34
|
-
unless defined?(ActiveJob)
|
35
|
-
raise Searchkick::Error, "Active Job not found"
|
36
|
-
end
|
59
|
+
# return true like model and relation reindex for now
|
60
|
+
true
|
61
|
+
end
|
37
62
|
|
38
|
-
|
39
|
-
|
40
|
-
|
41
|
-
routing = record.search_routing
|
42
|
-
end
|
63
|
+
def reindex_items(klass, items, method_name:, single: false)
|
64
|
+
routing = items.to_h { |r| [r[:id], r[:routing]] }
|
65
|
+
record_ids = routing.keys
|
43
66
|
|
44
|
-
|
45
|
-
|
46
|
-
|
47
|
-
|
48
|
-
routing: routing
|
49
|
-
)
|
50
|
-
else # bulk, inline/true/nil
|
51
|
-
reindex_record(method_name)
|
67
|
+
relation = Searchkick.load_records(klass, record_ids)
|
68
|
+
# call search_import even for single records for nested associations
|
69
|
+
relation = relation.search_import if relation.respond_to?(:search_import)
|
70
|
+
records = relation.select(&:should_index?)
|
52
71
|
|
53
|
-
|
54
|
-
|
72
|
+
# determine which records to delete
|
73
|
+
delete_ids = record_ids - records.map { |r| r.id.to_s }
|
74
|
+
delete_records =
|
75
|
+
delete_ids.map do |id|
|
76
|
+
construct_record(klass, id, routing[id])
|
77
|
+
end
|
78
|
+
|
79
|
+
import_inline(records, delete_records, method_name: method_name, single: single)
|
55
80
|
end
|
56
81
|
|
57
82
|
private
|
58
83
|
|
59
|
-
def
|
60
|
-
|
84
|
+
def index_record?(record)
|
85
|
+
record.persisted? && !record.destroyed? && record.should_index?
|
61
86
|
end
|
62
87
|
|
63
|
-
|
64
|
-
|
65
|
-
|
66
|
-
|
67
|
-
|
68
|
-
|
88
|
+
# import in single request with retries
|
89
|
+
def import_inline(index_records, delete_records, method_name:, single:)
|
90
|
+
return if index_records.empty? && delete_records.empty?
|
91
|
+
|
92
|
+
maybe_bulk(index_records, delete_records, method_name, single) do
|
93
|
+
if index_records.any?
|
94
|
+
if method_name
|
95
|
+
index.bulk_update(index_records, method_name)
|
96
|
+
else
|
97
|
+
index.bulk_index(index_records)
|
98
|
+
end
|
69
99
|
end
|
100
|
+
|
101
|
+
if delete_records.any?
|
102
|
+
index.bulk_delete(delete_records)
|
103
|
+
end
|
104
|
+
end
|
105
|
+
end
|
106
|
+
|
107
|
+
def maybe_bulk(index_records, delete_records, method_name, single)
|
108
|
+
if Searchkick.callbacks_value == :bulk
|
109
|
+
yield
|
70
110
|
else
|
71
|
-
|
72
|
-
|
73
|
-
|
74
|
-
|
111
|
+
# set action and data
|
112
|
+
action =
|
113
|
+
if single && index_records.empty?
|
114
|
+
"Remove"
|
115
|
+
elsif method_name
|
116
|
+
"Update"
|
117
|
+
else
|
118
|
+
single ? "Store" : "Import"
|
119
|
+
end
|
120
|
+
record = index_records.first || delete_records.first
|
121
|
+
name = record.class.searchkick_klass.name
|
122
|
+
message = lambda do |event|
|
123
|
+
event[:name] = "#{name} #{action}"
|
124
|
+
if single
|
125
|
+
event[:id] = index.search_id(record)
|
126
|
+
else
|
127
|
+
event[:count] = index_records.size + delete_records.size
|
128
|
+
end
|
129
|
+
end
|
130
|
+
|
131
|
+
with_retries do
|
132
|
+
Searchkick.callbacks(:bulk, message: message) do
|
133
|
+
yield
|
134
|
+
end
|
135
|
+
end
|
136
|
+
end
|
137
|
+
end
|
138
|
+
|
139
|
+
def construct_record(klass, id, routing)
|
140
|
+
record = klass.new
|
141
|
+
record.id = id
|
142
|
+
if routing
|
143
|
+
record.define_singleton_method(:search_routing) do
|
144
|
+
routing
|
145
|
+
end
|
146
|
+
end
|
147
|
+
record
|
148
|
+
end
|
149
|
+
|
150
|
+
def with_retries
|
151
|
+
retries = 0
|
152
|
+
|
153
|
+
begin
|
154
|
+
yield
|
155
|
+
rescue Faraday::ClientError => e
|
156
|
+
if retries < 1
|
157
|
+
retries += 1
|
158
|
+
retry
|
75
159
|
end
|
160
|
+
raise e
|
76
161
|
end
|
77
162
|
end
|
78
163
|
end
|