LittleWeasel 3.0.4 → 5.0.1
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +5 -5
- data/.github/workflows/codeql-analysis.yml +72 -0
- data/.gitignore +19 -17
- data/.reek.yml +17 -0
- data/.rspec +4 -2
- data/.rubocop.yml +187 -0
- data/.ruby-version +1 -1
- data/.yardopts +2 -0
- data/CHANGELOG.md +21 -1
- data/Gemfile +3 -1
- data/Gemfile.lock +114 -0
- data/Jenkinsfile +20 -0
- data/LittleWeasel.gemspec +31 -18
- data/README.md +408 -42
- data/Rakefile +296 -3
- data/lib/LittleWeasel/block_results.rb +81 -0
- data/lib/LittleWeasel/configure.rb +98 -0
- data/lib/LittleWeasel/dictionary.rb +125 -0
- data/lib/LittleWeasel/dictionary_key.rb +48 -0
- data/lib/LittleWeasel/dictionary_manager.rb +91 -0
- data/lib/LittleWeasel/errors/dictionary_file_already_loaded_error.rb +9 -0
- data/lib/LittleWeasel/errors/dictionary_file_empty_error.rb +8 -0
- data/lib/LittleWeasel/errors/dictionary_file_not_found_error.rb +8 -0
- data/lib/LittleWeasel/errors/dictionary_file_too_large_error.rb +16 -0
- data/lib/LittleWeasel/errors/language_required_error.rb +8 -0
- data/lib/LittleWeasel/errors/must_override_error.rb +8 -0
- data/lib/LittleWeasel/filters/en_us/currency_filter.rb +19 -0
- data/lib/LittleWeasel/filters/en_us/numeric_filter.rb +19 -0
- data/lib/LittleWeasel/filters/en_us/single_character_word_filter.rb +21 -0
- data/lib/LittleWeasel/filters/word_filter.rb +59 -0
- data/lib/LittleWeasel/filters/word_filter_managable.rb +80 -0
- data/lib/LittleWeasel/filters/word_filter_validatable.rb +31 -0
- data/lib/LittleWeasel/filters/word_filterable.rb +19 -0
- data/lib/LittleWeasel/filters/word_filters_validatable.rb +29 -0
- data/lib/LittleWeasel/metadata/dictionary_metadata.rb +145 -0
- data/lib/LittleWeasel/metadata/invalid_words_metadata.rb +134 -0
- data/lib/LittleWeasel/metadata/invalid_words_service_results.rb +45 -0
- data/lib/LittleWeasel/metadata/metadata_observable_validatable.rb +22 -0
- data/lib/LittleWeasel/metadata/metadata_observerable.rb +90 -0
- data/lib/LittleWeasel/metadata/metadatable.rb +134 -0
- data/lib/LittleWeasel/modules/class_name_to_symbol.rb +26 -0
- data/lib/LittleWeasel/modules/configurable.rb +26 -0
- data/lib/LittleWeasel/modules/deep_dup.rb +11 -0
- data/lib/LittleWeasel/modules/dictionary_cache_keys.rb +34 -0
- data/lib/LittleWeasel/modules/dictionary_cache_servicable.rb +26 -0
- data/lib/LittleWeasel/modules/dictionary_cache_validatable.rb +18 -0
- data/lib/LittleWeasel/modules/dictionary_creator_servicable.rb +27 -0
- data/lib/LittleWeasel/modules/dictionary_file_loader.rb +67 -0
- data/lib/LittleWeasel/modules/dictionary_key_validatable.rb +17 -0
- data/lib/LittleWeasel/modules/dictionary_keyable.rb +24 -0
- data/lib/LittleWeasel/modules/dictionary_metadata_servicable.rb +29 -0
- data/lib/LittleWeasel/modules/dictionary_metadata_validatable.rb +15 -0
- data/lib/LittleWeasel/modules/dictionary_source_validatable.rb +15 -0
- data/lib/LittleWeasel/modules/dictionary_sourceable.rb +86 -0
- data/lib/LittleWeasel/modules/dictionary_validatable.rb +18 -0
- data/lib/LittleWeasel/modules/language.rb +24 -0
- data/lib/LittleWeasel/modules/language_validatable.rb +14 -0
- data/lib/LittleWeasel/modules/locale.rb +23 -0
- data/lib/LittleWeasel/modules/order_validatable.rb +16 -0
- data/lib/LittleWeasel/modules/orderable.rb +17 -0
- data/lib/LittleWeasel/modules/region.rb +24 -0
- data/lib/LittleWeasel/modules/region_validatable.rb +14 -0
- data/lib/LittleWeasel/modules/tag_validatable.rb +14 -0
- data/lib/LittleWeasel/modules/taggable.rb +31 -0
- data/lib/LittleWeasel/modules/word_results_validatable.rb +28 -0
- data/lib/LittleWeasel/preprocessors/en_us/capitalize_preprocessor.rb +22 -0
- data/lib/LittleWeasel/preprocessors/preprocessed_word.rb +29 -0
- data/lib/LittleWeasel/preprocessors/preprocessed_word_validatable.rb +56 -0
- data/lib/LittleWeasel/preprocessors/preprocessed_words.rb +59 -0
- data/lib/LittleWeasel/preprocessors/preprocessed_words_validatable.rb +28 -0
- data/lib/LittleWeasel/preprocessors/word_preprocessable.rb +19 -0
- data/lib/LittleWeasel/preprocessors/word_preprocessor.rb +123 -0
- data/lib/LittleWeasel/preprocessors/word_preprocessor_managable.rb +114 -0
- data/lib/LittleWeasel/preprocessors/word_preprocessor_validatable.rb +40 -0
- data/lib/LittleWeasel/preprocessors/word_preprocessors_validatable.rb +24 -0
- data/lib/LittleWeasel/services/dictionary_cache_service.rb +211 -0
- data/lib/LittleWeasel/services/dictionary_creator_service.rb +94 -0
- data/lib/LittleWeasel/services/dictionary_file_loader_service.rb +37 -0
- data/lib/LittleWeasel/services/dictionary_killer_service.rb +35 -0
- data/lib/LittleWeasel/services/dictionary_metadata_service.rb +116 -0
- data/lib/LittleWeasel/services/invalid_words_service.rb +59 -0
- data/lib/LittleWeasel/version.rb +3 -1
- data/lib/LittleWeasel/word_results.rb +146 -0
- data/lib/LittleWeasel.rb +72 -186
- data/spec/factories/dictionary.rb +43 -0
- data/spec/factories/dictionary_cache_service.rb +95 -0
- data/spec/factories/dictionary_creator_service.rb +16 -0
- data/spec/factories/dictionary_file_loader_service.rb +13 -0
- data/spec/factories/dictionary_hash.rb +39 -0
- data/spec/factories/dictionary_key.rb +14 -0
- data/spec/factories/dictionary_killer_service.rb +14 -0
- data/spec/factories/dictionary_manager.rb +10 -0
- data/spec/factories/dictionary_metadata.rb +16 -0
- data/spec/factories/dictionary_metadata_service.rb +16 -0
- data/spec/factories/numeric_filter.rb +12 -0
- data/spec/factories/preprocessed_word.rb +16 -0
- data/spec/factories/preprocessed_words.rb +41 -0
- data/spec/factories/single_character_word_filter.rb +12 -0
- data/spec/factories/word_results.rb +16 -0
- data/spec/lib/LittleWeasel/block_results_spec.rb +248 -0
- data/spec/lib/LittleWeasel/configure_spec.rb +74 -0
- data/spec/lib/LittleWeasel/dictionary_key_spec.rb +118 -0
- data/spec/lib/LittleWeasel/dictionary_manager_spec.rb +166 -0
- data/spec/lib/LittleWeasel/dictionary_spec.rb +289 -0
- data/spec/lib/LittleWeasel/filters/en_us/currency_filter_spec.rb +80 -0
- data/spec/lib/LittleWeasel/filters/en_us/numeric_filter_spec.rb +66 -0
- data/spec/lib/LittleWeasel/filters/en_us/single_character_word_filter_spec.rb +58 -0
- data/spec/lib/LittleWeasel/filters/word_filter_managable_spec.rb +180 -0
- data/spec/lib/LittleWeasel/filters/word_filter_spec.rb +151 -0
- data/spec/lib/LittleWeasel/filters/word_filter_validatable_spec.rb +94 -0
- data/spec/lib/LittleWeasel/filters/word_filters_validatable_spec.rb +48 -0
- data/spec/lib/LittleWeasel/integraton_tests/dictionary_integration_spec.rb +201 -0
- data/spec/lib/LittleWeasel/metadata/dictionary_creator_servicable_spec.rb +54 -0
- data/spec/lib/LittleWeasel/metadata/dictionary_metadata_spec.rb +209 -0
- data/spec/lib/LittleWeasel/metadata/invalid_words_metadata_spec.rb +155 -0
- data/spec/lib/LittleWeasel/metadata/metadata_observerable_spec.rb +31 -0
- data/spec/lib/LittleWeasel/metadata/metadatable_spec.rb +35 -0
- data/spec/lib/LittleWeasel/modules/class_name_to_symbol_spec.rb +21 -0
- data/spec/lib/LittleWeasel/modules/dictionary_file_loader_spec.rb +125 -0
- data/spec/lib/LittleWeasel/modules/dictionary_sourceable_spec.rb +81 -0
- data/spec/lib/LittleWeasel/modules/language_spec.rb +112 -0
- data/spec/lib/LittleWeasel/modules/locale_spec.rb +95 -0
- data/spec/lib/LittleWeasel/modules/region_spec.rb +112 -0
- data/spec/lib/LittleWeasel/preprocessors/en_us/capitalize_preprocessor_spec.rb +34 -0
- data/spec/lib/LittleWeasel/preprocessors/preprocessed_word_spec.rb +105 -0
- data/spec/lib/LittleWeasel/preprocessors/preprocessed_word_validatable_spec.rb +143 -0
- data/spec/lib/LittleWeasel/preprocessors/preprocessed_words_spec.rb +77 -0
- data/spec/lib/LittleWeasel/preprocessors/preprocessed_words_validatable_spec.rb +58 -0
- data/spec/lib/LittleWeasel/preprocessors/word_preprocessor_managable_spec.rb +242 -0
- data/spec/lib/LittleWeasel/preprocessors/word_preprocessor_spec.rb +218 -0
- data/spec/lib/LittleWeasel/preprocessors/word_preprocessor_validatable_spec.rb +109 -0
- data/spec/lib/LittleWeasel/preprocessors/word_preprocessors_validatable_spec.rb +49 -0
- data/spec/lib/LittleWeasel/services/dictionary_cache_service_spec.rb +444 -0
- data/spec/lib/LittleWeasel/services/dictionary_creator_service_spec.rb +119 -0
- data/spec/lib/LittleWeasel/services/dictionary_file_loader_service_spec.rb +71 -0
- data/spec/lib/LittleWeasel/services/dictionary_metadata_service_spec.rb +279 -0
- data/spec/lib/LittleWeasel/word_results_spec.rb +275 -0
- data/spec/lib/LittleWeasel/workflow/workflow_spec.rb +20 -0
- data/spec/spec_helper.rb +117 -6
- data/spec/support/factory_bot.rb +15 -0
- data/spec/support/file_helpers.rb +46 -0
- data/spec/support/files/empty-dictionary.txt +0 -0
- data/{lib/dictionary → spec/support/files/en-US-big.txt} +262156 -31488
- data/spec/support/files/en-US-tagged.txt +26 -0
- data/spec/support/files/en-US.txt +26 -0
- data/spec/support/files/en.txt +26 -0
- data/spec/support/files/es-ES.txt +27 -0
- data/spec/support/files/es.txt +27 -0
- data/spec/support/general_helpers.rb +68 -0
- data/spec/support/shared_contexts.rb +107 -0
- data/spec/support/shared_examples.rb +105 -0
- metadata +418 -70
- data/spec/checker/checker_spec.rb +0 -286
@@ -0,0 +1,95 @@
|
|
1
|
+
# frozen_string_literal: false
|
2
|
+
|
3
|
+
require 'pry'
|
4
|
+
|
5
|
+
FactoryBot.define do
|
6
|
+
factory :dictionary_cache_service, class: LittleWeasel::Services::DictionaryCacheService do
|
7
|
+
dictionary_key { create(:dictionary_key) }
|
8
|
+
dictionary_cache { {} }
|
9
|
+
|
10
|
+
transient do
|
11
|
+
# The dictionary reference created in the cache will point to a MEMORY source.
|
12
|
+
#
|
13
|
+
# Valid values: nil | true | false | <Array of dictionary words>
|
14
|
+
#
|
15
|
+
# If nil or false - No memory source will be added to the dictionary cache.
|
16
|
+
# If true - A memory source will be added to the dictionary cache.
|
17
|
+
# If <An Array of dictionary words> - A memory source will be added to the dictionry cache.
|
18
|
+
# This only makes sense if load == true.
|
19
|
+
dictionary_memory_source {}
|
20
|
+
|
21
|
+
# The dictionary reference created in the cache will point to a FILE source.
|
22
|
+
#
|
23
|
+
# Important: dictionary_file_source will only be used if dictionary_memory_source
|
24
|
+
# is false.
|
25
|
+
#
|
26
|
+
# Valid values: nil | true | false | <Path to dictionary file>
|
27
|
+
#
|
28
|
+
# If nil or false - No file source will be added to the dictionary cache.
|
29
|
+
#
|
30
|
+
# If true - A file source will be added to the dictionry cache.
|
31
|
+
# dictionary_key.key will be used to create the dictionary
|
32
|
+
# file path.
|
33
|
+
# If <Path to dictionary file> - A files source will be added to the dictionry cache.
|
34
|
+
# The file source will point to <Path to dictionary file>.
|
35
|
+
dictionary_file_source {}
|
36
|
+
|
37
|
+
# If load == true - A dictionary object will be created and added to the dictionary cache
|
38
|
+
# depending on the dictionary source (file or memory).
|
39
|
+
load { false }
|
40
|
+
end
|
41
|
+
|
42
|
+
skip_create
|
43
|
+
initialize_with do
|
44
|
+
new(dictionary_key: dictionary_key, dictionary_cache: dictionary_cache)
|
45
|
+
end
|
46
|
+
|
47
|
+
after :create do |dictionary_cache_service, evaluator|
|
48
|
+
dictionary_key = dictionary_cache_service.dictionary_key
|
49
|
+
dictionary_cache = dictionary_cache_service.dictionary_cache
|
50
|
+
|
51
|
+
# Initialize the dictionary cache if the user already passed an
|
52
|
+
# initialized dictionary cache; otherwise, just use what they passed us.
|
53
|
+
dictionary_cache_service.class.init(dictionary_cache: dictionary_cache) \
|
54
|
+
unless dictionary_cache_service.class.count(dictionary_cache: dictionary_cache).positive?
|
55
|
+
|
56
|
+
load = evaluator.load
|
57
|
+
dictionary_memory_source = evaluator.dictionary_memory_source
|
58
|
+
dictionary_file_source = evaluator.dictionary_file_source
|
59
|
+
|
60
|
+
if load
|
61
|
+
unless dictionary_memory_source.present? || dictionary_file_source.present?
|
62
|
+
raise 'Transient attributes dictionary_memory_source or dictionary_file_source ' \
|
63
|
+
"must be present if transient attribute load is true: #{dictionary_reference}"
|
64
|
+
end
|
65
|
+
end
|
66
|
+
|
67
|
+
if dictionary_file_source
|
68
|
+
file_name = if dictionary_file_source == true
|
69
|
+
dictionary_key.key
|
70
|
+
else
|
71
|
+
dictionary_file_source
|
72
|
+
end
|
73
|
+
dictionary_cache_service.add_dictionary_source(dictionary_source: dictionary_path_for(file_name: file_name))
|
74
|
+
elsif dictionary_memory_source
|
75
|
+
dictionary_cache_service.add_dictionary_source(dictionary_source: LittleWeasel::Modules::DictionarSourceable.memory_source)
|
76
|
+
end
|
77
|
+
|
78
|
+
if load
|
79
|
+
dictionary_words = if dictionary_file_source
|
80
|
+
dictionary_file_loader_service = create(:dictionary_file_loader_service, dictionary_key: dictionary_key, dictionary_cache: dictionary_cache)
|
81
|
+
dictionary_file_loader_service.execute
|
82
|
+
else
|
83
|
+
unless dictionary_memory_source.is_a? Array
|
84
|
+
raise 'Transient attribute dictionary_memory_source must be an Array of words ' \
|
85
|
+
"if transient attribute load == true: #{dictionary_memory_source}"
|
86
|
+
end
|
87
|
+
dictionary_memory_source
|
88
|
+
end
|
89
|
+
dictionary_cache_service.dictionary_object = create(:dictionary, dictionary_key: dictionary_key, dictionary_cache: dictionary_cache, dictionary_words: dictionary_words)
|
90
|
+
end
|
91
|
+
|
92
|
+
dictionary_cache_service
|
93
|
+
end
|
94
|
+
end
|
95
|
+
end
|
@@ -0,0 +1,16 @@
|
|
1
|
+
# frozen_string_literal: false
|
2
|
+
|
3
|
+
FactoryBot.define do
|
4
|
+
factory :dictionary_creator_service, class: LittleWeasel::Services::DictionaryCreatorService do
|
5
|
+
dictionary_key { create(:dictionary_key) }
|
6
|
+
dictionary_cache { {} }
|
7
|
+
dictionary_metadata { {} }
|
8
|
+
word_filters {}
|
9
|
+
word_preprocessors {}
|
10
|
+
|
11
|
+
skip_create
|
12
|
+
initialize_with do
|
13
|
+
new(dictionary_key: dictionary_key, dictionary_cache: dictionary_cache, dictionary_metadata: dictionary_metadata, word_filters: word_filters, word_preprocessors: word_preprocessors)
|
14
|
+
end
|
15
|
+
end
|
16
|
+
end
|
@@ -0,0 +1,13 @@
|
|
1
|
+
# frozen_string_literal: false
|
2
|
+
|
3
|
+
FactoryBot.define do
|
4
|
+
factory :dictionary_file_loader_service, class: LittleWeasel::Services::DictionaryFileLoaderService do
|
5
|
+
dictionary_key { create(:dictionary_key) }
|
6
|
+
dictionary_cache { {} }
|
7
|
+
|
8
|
+
skip_create
|
9
|
+
initialize_with do
|
10
|
+
new(dictionary_key: dictionary_key, dictionary_cache: dictionary_cache)
|
11
|
+
end
|
12
|
+
end
|
13
|
+
end
|
@@ -0,0 +1,39 @@
|
|
1
|
+
# frozen_string_literal: false
|
2
|
+
|
3
|
+
FactoryBot.define do
|
4
|
+
factory :dictionary_hash, class: Hash do
|
5
|
+
dictionary_words do
|
6
|
+
%w(apple
|
7
|
+
better
|
8
|
+
cat
|
9
|
+
dog
|
10
|
+
everyone
|
11
|
+
fat
|
12
|
+
game
|
13
|
+
help
|
14
|
+
italic
|
15
|
+
jasmine
|
16
|
+
kelp
|
17
|
+
love
|
18
|
+
man
|
19
|
+
nope
|
20
|
+
octopus
|
21
|
+
popeye
|
22
|
+
queue
|
23
|
+
ruby
|
24
|
+
stop
|
25
|
+
top
|
26
|
+
ultimate
|
27
|
+
very
|
28
|
+
was
|
29
|
+
xylophone
|
30
|
+
yes
|
31
|
+
zebra)
|
32
|
+
end
|
33
|
+
|
34
|
+
skip_create
|
35
|
+
initialize_with do
|
36
|
+
LittleWeasel::Dictionary.to_hash(dictionary_words: dictionary_words)
|
37
|
+
end
|
38
|
+
end
|
39
|
+
end
|
@@ -0,0 +1,14 @@
|
|
1
|
+
# frozen_string_literal: false
|
2
|
+
|
3
|
+
FactoryBot.define do
|
4
|
+
factory :dictionary_key, class: LittleWeasel::DictionaryKey do
|
5
|
+
language { :en }
|
6
|
+
region { :us }
|
7
|
+
tag {}
|
8
|
+
|
9
|
+
skip_create
|
10
|
+
initialize_with do
|
11
|
+
new language: language, region: region, tag: tag
|
12
|
+
end
|
13
|
+
end
|
14
|
+
end
|
@@ -0,0 +1,14 @@
|
|
1
|
+
# frozen_string_literal: false
|
2
|
+
|
3
|
+
FactoryBot.define do
|
4
|
+
factory :dictionary_killer_service, class: LittleWeasel::Services::DictionaryKillerService do
|
5
|
+
dictionary_key { create(:dictionary_key) }
|
6
|
+
dictionary_cache { {} }
|
7
|
+
dictionary_metadata { {} }
|
8
|
+
|
9
|
+
skip_create
|
10
|
+
initialize_with do
|
11
|
+
new(dictionary_key: dictionary_key, dictionary_cache: dictionary_cache, dictionary_metadata: dictionary_metadata)
|
12
|
+
end
|
13
|
+
end
|
14
|
+
end
|
@@ -0,0 +1,16 @@
|
|
1
|
+
# frozen_string_literal: false
|
2
|
+
|
3
|
+
FactoryBot.define do
|
4
|
+
factory :dictionary_metadata, class: LittleWeasel::Metadata::DictionaryMetadata do
|
5
|
+
dictionary_words {}
|
6
|
+
dictionary_key { create(:dictionary_key) }
|
7
|
+
dictionary_cache { {} }
|
8
|
+
dictionary_metadata { {} }
|
9
|
+
|
10
|
+
skip_create
|
11
|
+
initialize_with do
|
12
|
+
dictionary_hash = dictionary_words || create(:dictionary_hash)
|
13
|
+
new dictionary_words: dictionary_hash, dictionary_key: dictionary_key, dictionary_cache: dictionary_cache, dictionary_metadata: dictionary_metadata
|
14
|
+
end
|
15
|
+
end
|
16
|
+
end
|
@@ -0,0 +1,16 @@
|
|
1
|
+
# frozen_string_literal: false
|
2
|
+
|
3
|
+
require 'pry'
|
4
|
+
|
5
|
+
FactoryBot.define do
|
6
|
+
factory :dictionary_metadata_service, class: LittleWeasel::Services::DictionaryMetadataService do
|
7
|
+
dictionary_key { create(:dictionary_key) }
|
8
|
+
dictionary_cache { {} }
|
9
|
+
dictionary_metadata { {} }
|
10
|
+
|
11
|
+
skip_create
|
12
|
+
initialize_with do
|
13
|
+
new(dictionary_key: dictionary_key, dictionary_cache: dictionary_cache, dictionary_metadata: dictionary_metadata)
|
14
|
+
end
|
15
|
+
end
|
16
|
+
end
|
@@ -0,0 +1,16 @@
|
|
1
|
+
# frozen_string_literal: false
|
2
|
+
|
3
|
+
FactoryBot.define do
|
4
|
+
factory :preprocessed_word, class: LittleWeasel::Preprocessors::PreprocessedWord do
|
5
|
+
original_word {}
|
6
|
+
preprocessed {}
|
7
|
+
preprocessed_word {}
|
8
|
+
preprocessor {}
|
9
|
+
preprocessor_order {}
|
10
|
+
|
11
|
+
skip_create
|
12
|
+
initialize_with do
|
13
|
+
new original_word: original_word, preprocessed: preprocessed, preprocessed_word: preprocessed_word, preprocessor: preprocessor, preprocessor_order: preprocessor_order
|
14
|
+
end
|
15
|
+
end
|
16
|
+
end
|
@@ -0,0 +1,41 @@
|
|
1
|
+
# frozen_string_literal: false
|
2
|
+
|
3
|
+
require 'pry'
|
4
|
+
|
5
|
+
FactoryBot.define do
|
6
|
+
factory :preprocessed_words, class: LittleWeasel::Preprocessors::PreprocessedWords do
|
7
|
+
original_word { 'word' }
|
8
|
+
preprocessed_words {}
|
9
|
+
|
10
|
+
transient do
|
11
|
+
with_word_processors { 0 }
|
12
|
+
end
|
13
|
+
|
14
|
+
skip_create
|
15
|
+
initialize_with do
|
16
|
+
new original_word: original_word, preprocessed_words: preprocessed_words
|
17
|
+
end
|
18
|
+
|
19
|
+
after :create do |preprocessed_words, evaluator|
|
20
|
+
with_word_processors = evaluator.with_word_processors
|
21
|
+
|
22
|
+
if with_word_processors.positive?
|
23
|
+
preprocessed_words.preprocessed_words = []
|
24
|
+
preprocessed_word = evaluator.original_word
|
25
|
+
|
26
|
+
with_word_processors.times do |index|
|
27
|
+
preprocessed_word_object = create(:preprocessed_word,
|
28
|
+
original_word: preprocessed_word,
|
29
|
+
preprocessed: true,
|
30
|
+
preprocessed_word: "#{preprocessed_word}-#{index}").tap do |preprocessed_word_object|
|
31
|
+
preprocessed_word_object.preprocessed_word = "#{preprocessed_word}-#{index}"
|
32
|
+
preprocessed_word_object.preprocessor = "preprocesor#{index}"
|
33
|
+
preprocessed_word_object.preprocessor_order = index
|
34
|
+
end
|
35
|
+
preprocessed_words.preprocessed_words << preprocessed_word_object
|
36
|
+
preprocessed_word = preprocessed_word_object.preprocessed_word
|
37
|
+
end
|
38
|
+
end
|
39
|
+
end
|
40
|
+
end
|
41
|
+
end
|
@@ -0,0 +1,12 @@
|
|
1
|
+
# frozen_string_literal: false
|
2
|
+
|
3
|
+
FactoryBot.define do
|
4
|
+
factory :single_character_word_filter, class: LittleWeasel::Filters::EnUs::SingleCharacterWordFilter do
|
5
|
+
filter_on { true }
|
6
|
+
|
7
|
+
skip_create
|
8
|
+
initialize_with do
|
9
|
+
new filter_on: filter_on
|
10
|
+
end
|
11
|
+
end
|
12
|
+
end
|
@@ -0,0 +1,16 @@
|
|
1
|
+
# frozen_string_literal: false
|
2
|
+
|
3
|
+
FactoryBot.define do
|
4
|
+
factory :word_results, class: LittleWeasel::WordResults do
|
5
|
+
original_word {}
|
6
|
+
filters_matched { [] }
|
7
|
+
preprocessed_words {}
|
8
|
+
word_cached { false }
|
9
|
+
word_valid { false }
|
10
|
+
|
11
|
+
skip_create
|
12
|
+
initialize_with do
|
13
|
+
new filters_matched: filters_matched, original_word: original_word, preprocessed_words: preprocessed_words, word_cached: word_cached, word_valid: word_valid
|
14
|
+
end
|
15
|
+
end
|
16
|
+
end
|
@@ -0,0 +1,248 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require 'spec_helper'
|
4
|
+
|
5
|
+
RSpec.describe LittleWeasel::BlockResults do
|
6
|
+
subject { described_class.new original_word_block: original_word_block }
|
7
|
+
|
8
|
+
def ceate_word_results(word:, word_valid: false, word_cached: false, filters_matched: [], preprocessed_words: nil)
|
9
|
+
create(:word_results, original_word: word, word_valid: word_valid, word_cached: word_cached, filters_matched: filters_matched, preprocessed_words: preprocessed_words)
|
10
|
+
end
|
11
|
+
|
12
|
+
let(:original_word_block) { 'Original word block' }
|
13
|
+
|
14
|
+
#.new
|
15
|
+
describe '.new' do
|
16
|
+
it 'instantiates an object' do
|
17
|
+
expect { subject }.to_not raise_error
|
18
|
+
end
|
19
|
+
|
20
|
+
it 'initializes #word_results to an empty Array' do
|
21
|
+
expect(subject.word_results).to eq []
|
22
|
+
end
|
23
|
+
|
24
|
+
it 'initializes #original_word_block to the original_word_block argument passed' do
|
25
|
+
expect(subject.original_word_block).to eq original_word_block
|
26
|
+
end
|
27
|
+
end
|
28
|
+
|
29
|
+
#<<
|
30
|
+
describe '<<' do
|
31
|
+
context 'with an invalid argument' do
|
32
|
+
it 'raises an error' do
|
33
|
+
expect { subject << :not_a_word_results_object }.to raise_error(/Argument word_result is not a WordResults object/)
|
34
|
+
end
|
35
|
+
end
|
36
|
+
|
37
|
+
context 'with a valid argument' do
|
38
|
+
it 'adds the WordResults object to the #word_results Array' do
|
39
|
+
expect { subject << ceate_word_results(word: 'word01') }.to \
|
40
|
+
change { subject.word_results.count }.from(0).to(1)
|
41
|
+
end
|
42
|
+
end
|
43
|
+
end
|
44
|
+
|
45
|
+
#success?
|
46
|
+
describe '#success?' do
|
47
|
+
before do
|
48
|
+
subject << ceate_word_results(word: 'word01', word_valid: true)
|
49
|
+
subject << ceate_word_results(word: 'word02', word_valid: false, filters_matched: [:matched_filter])
|
50
|
+
end
|
51
|
+
|
52
|
+
context 'when all WordResults#successful? return true' do
|
53
|
+
before do
|
54
|
+
subject << ceate_word_results(word: 'word03', word_valid: false, filters_matched: [:matched_filter])
|
55
|
+
subject << ceate_word_results(word: 'word04', word_valid: false, filters_matched: [:matched_filter])
|
56
|
+
end
|
57
|
+
|
58
|
+
it 'returns true' do
|
59
|
+
expect(subject.success?).to be true
|
60
|
+
end
|
61
|
+
end
|
62
|
+
|
63
|
+
context 'when any WordResults#word_valid? objects return false' do
|
64
|
+
before do
|
65
|
+
subject << ceate_word_results(word: 'word03', word_valid: false)
|
66
|
+
end
|
67
|
+
|
68
|
+
it 'returns false' do
|
69
|
+
expect(subject.success?).to be false
|
70
|
+
end
|
71
|
+
end
|
72
|
+
|
73
|
+
context 'when any WordResults#filter_match? objects return false' do
|
74
|
+
before do
|
75
|
+
subject << ceate_word_results(word: 'word03')
|
76
|
+
end
|
77
|
+
|
78
|
+
it 'returns false' do
|
79
|
+
expect(subject.success?).to be false
|
80
|
+
end
|
81
|
+
end
|
82
|
+
end
|
83
|
+
|
84
|
+
#words_valid?
|
85
|
+
describe '#words_valid?' do
|
86
|
+
before do
|
87
|
+
subject << ceate_word_results(word: 'word01', word_valid: true)
|
88
|
+
subject << ceate_word_results(word: 'word02', word_valid: true)
|
89
|
+
end
|
90
|
+
|
91
|
+
context 'when all WordResults#word_valid? objects return true' do
|
92
|
+
it 'returns true' do
|
93
|
+
expect(subject.words_valid?).to be true
|
94
|
+
end
|
95
|
+
end
|
96
|
+
|
97
|
+
context 'when any WordResults#word_valid? objects return false' do
|
98
|
+
before do
|
99
|
+
subject << ceate_word_results(word: 'word03', word_valid: false)
|
100
|
+
end
|
101
|
+
|
102
|
+
it 'returns false' do
|
103
|
+
expect(subject.words_valid?).to be false
|
104
|
+
end
|
105
|
+
end
|
106
|
+
end
|
107
|
+
|
108
|
+
#filters_match?
|
109
|
+
describe '#filters_match?' do
|
110
|
+
before do
|
111
|
+
subject << ceate_word_results(word: 'word01', filters_matched: [:matched_filter])
|
112
|
+
subject << ceate_word_results(word: 'word02', filters_matched: [:matched_filter])
|
113
|
+
end
|
114
|
+
|
115
|
+
context 'when all WordResults#filters_match? objects return true' do
|
116
|
+
it 'returns true' do
|
117
|
+
expect(subject.filters_match?).to be true
|
118
|
+
end
|
119
|
+
end
|
120
|
+
|
121
|
+
context 'when any WordResults#filters_match? objects return false' do
|
122
|
+
before do
|
123
|
+
subject << ceate_word_results(word: 'word03', filters_matched: [])
|
124
|
+
end
|
125
|
+
|
126
|
+
it 'returns false' do
|
127
|
+
expect(subject.filters_match?).to be false
|
128
|
+
end
|
129
|
+
end
|
130
|
+
end
|
131
|
+
|
132
|
+
#preprocessed_words?
|
133
|
+
describe '#preprocessed_words?' do
|
134
|
+
before do
|
135
|
+
subject << ceate_word_results(word: word, preprocessed_words: preprocessed_words)
|
136
|
+
end
|
137
|
+
|
138
|
+
let(:word) { 'word' }
|
139
|
+
let(:preprocessed_words) { create(:preprocessed_words, original_word: word, with_word_processors: 1) }
|
140
|
+
|
141
|
+
context 'when all WordResults#preprocessed_word? objects return true' do
|
142
|
+
it 'returns true' do
|
143
|
+
expect(subject.preprocessed_words?).to be true
|
144
|
+
end
|
145
|
+
end
|
146
|
+
end
|
147
|
+
|
148
|
+
#preprocessed_words_or_original_words
|
149
|
+
describe '#preprocessed_words_or_original_words' do
|
150
|
+
context 'with all preprocessed words' do
|
151
|
+
before do
|
152
|
+
subject << ceate_word_results(word: words[0], preprocessed_words: preprocessed_words[0])
|
153
|
+
subject << ceate_word_results(word: words[1], preprocessed_words: preprocessed_words[1])
|
154
|
+
end
|
155
|
+
|
156
|
+
let(:words) { %w(word1 word2) }
|
157
|
+
let(:preprocessed_words) do
|
158
|
+
[
|
159
|
+
create(:preprocessed_words, original_word: words[0], with_word_processors: 1),
|
160
|
+
create(:preprocessed_words, original_word: words[1], with_word_processors: 2)
|
161
|
+
]
|
162
|
+
end
|
163
|
+
|
164
|
+
it 'returns all the preprocessed words' do
|
165
|
+
expect(subject.preprocessed_words_or_original_words).to eq %w(word1-0 word2-0-1)
|
166
|
+
end
|
167
|
+
end
|
168
|
+
|
169
|
+
context 'with all original words' do
|
170
|
+
before do
|
171
|
+
subject << ceate_word_results(word: words[0])
|
172
|
+
subject << ceate_word_results(word: words[1])
|
173
|
+
end
|
174
|
+
|
175
|
+
let(:words) { %w(word1 word2) }
|
176
|
+
|
177
|
+
it 'returns all the original words' do
|
178
|
+
expect(subject.preprocessed_words_or_original_words).to eq %w(word1 word2)
|
179
|
+
end
|
180
|
+
end
|
181
|
+
|
182
|
+
context 'with mixed original words and preprocessed words' do
|
183
|
+
before do
|
184
|
+
subject << ceate_word_results(word: words[0], word_valid: true)
|
185
|
+
subject << ceate_word_results(word: words[1], word_valid: true, preprocessed_words: preprocessed_words[0])
|
186
|
+
subject << ceate_word_results(word: words[2], word_valid: true)
|
187
|
+
subject << ceate_word_results(word: words[3], word_valid: true, preprocessed_words: preprocessed_words[1])
|
188
|
+
end
|
189
|
+
|
190
|
+
let(:words) { %w(word0 word1 word2 word3) }
|
191
|
+
let(:preprocessed_words) do
|
192
|
+
[
|
193
|
+
create(:preprocessed_words, original_word: words[1], with_word_processors: 1),
|
194
|
+
create(:preprocessed_words, original_word: words[3], with_word_processors: 2)
|
195
|
+
]
|
196
|
+
end
|
197
|
+
|
198
|
+
it 'returns all the original and preprocessed words' do
|
199
|
+
expect(subject.preprocessed_words_or_original_words).to eq %w(word0 word1-0 word2 word3-0-1)
|
200
|
+
end
|
201
|
+
end
|
202
|
+
end
|
203
|
+
|
204
|
+
#words_cached?
|
205
|
+
describe '#words_cached?' do
|
206
|
+
before do
|
207
|
+
subject << ceate_word_results(word: words[0])
|
208
|
+
subject << ceate_word_results(word: words[1])
|
209
|
+
subject << ceate_word_results(word: words[2])
|
210
|
+
subject << ceate_word_results(word: words[3])
|
211
|
+
end
|
212
|
+
|
213
|
+
let(:words) { %w(word1 word2 word3 word4) }
|
214
|
+
|
215
|
+
context 'when all the words are cached' do
|
216
|
+
before do
|
217
|
+
allow_any_instance_of(LittleWeasel::WordResults).to receive(:word_cached?).and_return(true)
|
218
|
+
end
|
219
|
+
|
220
|
+
it 'returns true' do
|
221
|
+
expect(subject.words_cached?).to eq true
|
222
|
+
end
|
223
|
+
end
|
224
|
+
|
225
|
+
context 'when some of the words are cached' do
|
226
|
+
before do
|
227
|
+
allow(subject.word_results[0]).to receive(:word_cached?).and_return(false)
|
228
|
+
allow(subject.word_results[1]).to receive(:word_cached?).and_return(true)
|
229
|
+
allow(subject.word_results[2]).to receive(:word_cached?).and_return(false)
|
230
|
+
allow(subject.word_results[3]).to receive(:word_cached?).and_return(true)
|
231
|
+
end
|
232
|
+
|
233
|
+
it 'returns false' do
|
234
|
+
expect(subject.words_cached?).to eq false
|
235
|
+
end
|
236
|
+
end
|
237
|
+
|
238
|
+
context 'when NONE of the words are cached' do
|
239
|
+
before do
|
240
|
+
allow_any_instance_of(LittleWeasel::WordResults).to receive(:word_cached?).and_return(false)
|
241
|
+
end
|
242
|
+
|
243
|
+
it 'returns false' do
|
244
|
+
expect(subject.words_cached?).to eq false
|
245
|
+
end
|
246
|
+
end
|
247
|
+
end
|
248
|
+
end
|
@@ -0,0 +1,74 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require 'spec_helper'
|
4
|
+
|
5
|
+
RSpec.describe LittleWeasel do
|
6
|
+
subject do
|
7
|
+
described_class.configure { |config| }
|
8
|
+
described_class.configuration
|
9
|
+
end
|
10
|
+
|
11
|
+
context 'default configurable settings' do
|
12
|
+
describe '#max_dictionary_file_megabytes' do
|
13
|
+
it 'set to 5 by default' do
|
14
|
+
expect(subject.max_dictionary_file_megabytes).to eq 5
|
15
|
+
end
|
16
|
+
end
|
17
|
+
|
18
|
+
describe '#max_invalid_words_bytesize' do
|
19
|
+
it 'set to 25_000 by default' do
|
20
|
+
expect(subject.max_invalid_words_bytesize).to eq 25_000
|
21
|
+
end
|
22
|
+
end
|
23
|
+
|
24
|
+
describe '#metadata_observers' do
|
25
|
+
it 'set to an Array with InvalidWordsMetadata by default' do
|
26
|
+
expect(subject.metadata_observers).to eq [
|
27
|
+
LittleWeasel::Metadata::InvalidWordsMetadata
|
28
|
+
]
|
29
|
+
end
|
30
|
+
end
|
31
|
+
|
32
|
+
describe '#word_block_regex' do
|
33
|
+
it 'set to the default regex to split work blocks by default' do
|
34
|
+
expect(subject.word_block_regex).to eq /[[[:word:]]'-]+/
|
35
|
+
end
|
36
|
+
end
|
37
|
+
end
|
38
|
+
|
39
|
+
#.configuration
|
40
|
+
describe '.configuration' do
|
41
|
+
context 'when passing a block' do
|
42
|
+
subject do
|
43
|
+
described_class.configure do |config|
|
44
|
+
config.max_dictionary_file_megabytes = max_dictionary_file_megabytes
|
45
|
+
config.metadata_observers = metadata_observers
|
46
|
+
config.word_block_regex = word_block_regex
|
47
|
+
end
|
48
|
+
described_class.configuration
|
49
|
+
end
|
50
|
+
|
51
|
+
let(:max_dictionary_file_megabytes) { 1_222_333 }
|
52
|
+
let(:metadata_observers) { %i(observer0 observer1) }
|
53
|
+
let(:word_block_regex) { :word_block_regex }
|
54
|
+
|
55
|
+
describe '#max_dictionary_file_megabytes=' do
|
56
|
+
it 'sets the value' do
|
57
|
+
expect(subject.max_dictionary_file_megabytes).to eq max_dictionary_file_megabytes
|
58
|
+
end
|
59
|
+
end
|
60
|
+
|
61
|
+
describe '#metadata_observers=' do
|
62
|
+
it 'sets the value' do
|
63
|
+
expect(subject.metadata_observers).to eq metadata_observers
|
64
|
+
end
|
65
|
+
end
|
66
|
+
|
67
|
+
describe '#word_block_regex=' do
|
68
|
+
it 'sets the value' do
|
69
|
+
expect(subject.word_block_regex).to eq word_block_regex
|
70
|
+
end
|
71
|
+
end
|
72
|
+
end
|
73
|
+
end
|
74
|
+
end
|