wikipedia-vandalism_detection 0.1.0-java
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.gitignore +19 -0
- data/.rspec +2 -0
- data/.rubocop.yml +35 -0
- data/.travis.yml +6 -0
- data/Gemfile +4 -0
- data/LICENSE.txt +4 -0
- data/README.md +288 -0
- data/Rakefile +11 -0
- data/config/wikipedia-vandalism-detection.yml.example +103 -0
- data/lib/java/SMOTE.jar +0 -0
- data/lib/java/balancedRandomForest.jar +0 -0
- data/lib/java/diffutils-1.3.0.jar +0 -0
- data/lib/java/oneClassClassifier.jar +0 -0
- data/lib/java/realAdaBoost.jar +0 -0
- data/lib/java/swc-engine-1.1.0-jar-with-dependencies.jar +0 -0
- data/lib/java/sweble-wikitext-extractor.jar +0 -0
- data/lib/weka/classifiers/meta/one_class_classifier.rb +21 -0
- data/lib/weka/classifiers/meta/real_ada_boost.rb +15 -0
- data/lib/weka/classifiers/trees/balanced_random_forest.rb +16 -0
- data/lib/weka/filters/supervised/instance/smote.rb +15 -0
- data/lib/wikipedia/vandalism_detection/algorithms/kullback_leibler_divergence.rb +103 -0
- data/lib/wikipedia/vandalism_detection/algorithms.rb +12 -0
- data/lib/wikipedia/vandalism_detection/classifier.rb +202 -0
- data/lib/wikipedia/vandalism_detection/configuration.rb +350 -0
- data/lib/wikipedia/vandalism_detection/diff.rb +36 -0
- data/lib/wikipedia/vandalism_detection/edit.rb +81 -0
- data/lib/wikipedia/vandalism_detection/evaluator.rb +640 -0
- data/lib/wikipedia/vandalism_detection/exceptions.rb +47 -0
- data/lib/wikipedia/vandalism_detection/feature_calculator.rb +94 -0
- data/lib/wikipedia/vandalism_detection/features/all_wordlists_frequency.rb +22 -0
- data/lib/wikipedia/vandalism_detection/features/all_wordlists_impact.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/anonymity.rb +17 -0
- data/lib/wikipedia/vandalism_detection/features/anonymity_previous.rb +32 -0
- data/lib/wikipedia/vandalism_detection/features/article_size.rb +17 -0
- data/lib/wikipedia/vandalism_detection/features/bad_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/bad_impact.rb +20 -0
- data/lib/wikipedia/vandalism_detection/features/base.rb +61 -0
- data/lib/wikipedia/vandalism_detection/features/biased_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/biased_impact.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/blanking.rb +26 -0
- data/lib/wikipedia/vandalism_detection/features/character_diversity.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/character_sequence.rb +19 -0
- data/lib/wikipedia/vandalism_detection/features/comment_bad_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/comment_biased_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/comment_length.rb +16 -0
- data/lib/wikipedia/vandalism_detection/features/comment_markup_frequency.rb +30 -0
- data/lib/wikipedia/vandalism_detection/features/comment_pronoun_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/comment_sex_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/comment_vulgarism_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/compressibility.rb +31 -0
- data/lib/wikipedia/vandalism_detection/features/contains_base.rb +16 -0
- data/lib/wikipedia/vandalism_detection/features/copyedit.rb +17 -0
- data/lib/wikipedia/vandalism_detection/features/digit_ratio.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/edits_per_user.rb +72 -0
- data/lib/wikipedia/vandalism_detection/features/emoticons_frequency.rb +27 -0
- data/lib/wikipedia/vandalism_detection/features/emoticons_impact.rb +29 -0
- data/lib/wikipedia/vandalism_detection/features/frequency_base.rb +18 -0
- data/lib/wikipedia/vandalism_detection/features/impact_base.rb +25 -0
- data/lib/wikipedia/vandalism_detection/features/inserted_character_distribution.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/inserted_external_links.rb +19 -0
- data/lib/wikipedia/vandalism_detection/features/inserted_internal_links.rb +18 -0
- data/lib/wikipedia/vandalism_detection/features/inserted_size.rb +19 -0
- data/lib/wikipedia/vandalism_detection/features/inserted_words.rb +17 -0
- data/lib/wikipedia/vandalism_detection/features/longest_word.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/markup_frequency.rb +29 -0
- data/lib/wikipedia/vandalism_detection/features/markup_impact.rb +30 -0
- data/lib/wikipedia/vandalism_detection/features/non_alphanumeric_ratio.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/personal_life.rb +17 -0
- data/lib/wikipedia/vandalism_detection/features/pronoun_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/pronoun_impact.rb +20 -0
- data/lib/wikipedia/vandalism_detection/features/removed_all_wordlists_frequency.rb +22 -0
- data/lib/wikipedia/vandalism_detection/features/removed_bad_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/removed_biased_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/removed_character_distribution.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/removed_emoticons_frequency.rb +28 -0
- data/lib/wikipedia/vandalism_detection/features/removed_markup_frequency.rb +30 -0
- data/lib/wikipedia/vandalism_detection/features/removed_pronoun_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/removed_sex_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/removed_size.rb +19 -0
- data/lib/wikipedia/vandalism_detection/features/removed_vulgarism_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/removed_words.rb +17 -0
- data/lib/wikipedia/vandalism_detection/features/replacement_similarity.rb +20 -0
- data/lib/wikipedia/vandalism_detection/features/reverted.rb +16 -0
- data/lib/wikipedia/vandalism_detection/features/revisions_character_distribution.rb +25 -0
- data/lib/wikipedia/vandalism_detection/features/same_editor.rb +32 -0
- data/lib/wikipedia/vandalism_detection/features/sex_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/sex_impact.rb +20 -0
- data/lib/wikipedia/vandalism_detection/features/size_increment.rb +20 -0
- data/lib/wikipedia/vandalism_detection/features/size_ratio.rb +28 -0
- data/lib/wikipedia/vandalism_detection/features/term_frequency.rb +31 -0
- data/lib/wikipedia/vandalism_detection/features/time_interval.rb +38 -0
- data/lib/wikipedia/vandalism_detection/features/time_of_day.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/upper_case_ratio.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/upper_case_words_ratio.rb +33 -0
- data/lib/wikipedia/vandalism_detection/features/upper_to_lower_case_ratio.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/vulgarism_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/vulgarism_impact.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/weekday.rb +19 -0
- data/lib/wikipedia/vandalism_detection/features/words_increment.rb +20 -0
- data/lib/wikipedia/vandalism_detection/features.rb +66 -0
- data/lib/wikipedia/vandalism_detection/instances.rb +121 -0
- data/lib/wikipedia/vandalism_detection/page.rb +101 -0
- data/lib/wikipedia/vandalism_detection/page_parser.rb +63 -0
- data/lib/wikipedia/vandalism_detection/revision.rb +75 -0
- data/lib/wikipedia/vandalism_detection/revision_parser.rb +67 -0
- data/lib/wikipedia/vandalism_detection/test_dataset.rb +374 -0
- data/lib/wikipedia/vandalism_detection/text.rb +23 -0
- data/lib/wikipedia/vandalism_detection/training_dataset.rb +282 -0
- data/lib/wikipedia/vandalism_detection/version.rb +5 -0
- data/lib/wikipedia/vandalism_detection/wikitext_extractor.rb +80 -0
- data/lib/wikipedia/vandalism_detection/word_lists/bad.rb +11 -0
- data/lib/wikipedia/vandalism_detection/word_lists/biased.rb +20 -0
- data/lib/wikipedia/vandalism_detection/word_lists/emoticons.rb +26 -0
- data/lib/wikipedia/vandalism_detection/word_lists/markup.rb +19 -0
- data/lib/wikipedia/vandalism_detection/word_lists/pronouns.rb +11 -0
- data/lib/wikipedia/vandalism_detection/word_lists/sex.rb +10 -0
- data/lib/wikipedia/vandalism_detection/word_lists/vulgarism.rb +96 -0
- data/lib/wikipedia/vandalism_detection/word_lists.rb +17 -0
- data/lib/wikipedia/vandalism_detection.rb +29 -0
- data/lib/wikipedia.rb +41 -0
- data/spec/factories/edit.rb +19 -0
- data/spec/factories/page.rb +12 -0
- data/spec/factories/revision.rb +51 -0
- data/spec/resources/config/wikipedia-vandalism-detection.yml +35 -0
- data/spec/resources/corpora/test/edits.csv +8 -0
- data/spec/resources/corpora/test/ground-truth.txt +3 -0
- data/spec/resources/corpora/test/revisions/part-1/326471754.txt +199 -0
- data/spec/resources/corpora/test/revisions/part-1/326873205.txt +137 -0
- data/spec/resources/corpora/test/revisions/part-1/328774035.txt +162 -0
- data/spec/resources/corpora/test/revisions/part-2/307084144.txt +137 -0
- data/spec/resources/corpora/test/revisions/part-2/326978767.txt +199 -0
- data/spec/resources/corpora/test/revisions/part-2/328774110.txt +162 -0
- data/spec/resources/corpora/training/annotations.csv +7 -0
- data/spec/resources/corpora/training/edits.csv +7 -0
- data/spec/resources/corpora/training/revisions/part-1/326471754.txt +199 -0
- data/spec/resources/corpora/training/revisions/part-1/326873205.txt +137 -0
- data/spec/resources/corpora/training/revisions/part-1/328774035.txt +162 -0
- data/spec/resources/corpora/training/revisions/part-2/307084144.txt +137 -0
- data/spec/resources/corpora/training/revisions/part-2/326978767.txt +199 -0
- data/spec/resources/corpora/training/revisions/part-2/328774110.txt +162 -0
- data/spec/resources/page_with_redirects.xml +85 -0
- data/spec/resources/redirect_page.xml +59 -0
- data/spec/resources/revision_simplified.xml +13 -0
- data/spec/resources/sample_revision.txt +137 -0
- data/spec/resources/sample_revision_clean_text.txt +1 -0
- data/spec/resources/sample_revision_plain_text.txt +183 -0
- data/spec/resources/vandalism_on_wikipedia.xml +234 -0
- data/spec/resources/vandalism_on_wikipedia_simplified.xml +119 -0
- data/spec/resources/wikipedia_tokens.txt +30 -0
- data/spec/spec_helper.rb +38 -0
- data/spec/support/macros/file_reading.rb +6 -0
- data/spec/support/macros/test_configuration.rb +81 -0
- data/spec/vandalism_detection/algorithms/kullback_leibler_divergence_spec.rb +34 -0
- data/spec/vandalism_detection/classifier_spec.rb +330 -0
- data/spec/vandalism_detection/configuration_spec.rb +601 -0
- data/spec/vandalism_detection/diff_spec.rb +40 -0
- data/spec/vandalism_detection/edit_spec.rb +122 -0
- data/spec/vandalism_detection/evaluator_spec.rb +711 -0
- data/spec/vandalism_detection/feature_calculator_spec.rb +135 -0
- data/spec/vandalism_detection/features/all_wordlists_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/all_wordlists_impact_spec.rb +52 -0
- data/spec/vandalism_detection/features/anonymity_previous_spec.rb +67 -0
- data/spec/vandalism_detection/features/anonymity_spec.rb +17 -0
- data/spec/vandalism_detection/features/article_size_spec.rb +29 -0
- data/spec/vandalism_detection/features/bad_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/bad_impact_spec.rb +53 -0
- data/spec/vandalism_detection/features/base_spec.rb +41 -0
- data/spec/vandalism_detection/features/biased_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/biased_impact_spec.rb +52 -0
- data/spec/vandalism_detection/features/blanking_spec.rb +35 -0
- data/spec/vandalism_detection/features/character_diversity_spec.rb +30 -0
- data/spec/vandalism_detection/features/character_sequence_spec.rb +31 -0
- data/spec/vandalism_detection/features/comment_bad_frequency_spec.rb +28 -0
- data/spec/vandalism_detection/features/comment_biased_frequency_spec.rb +28 -0
- data/spec/vandalism_detection/features/comment_length_spec.rb +21 -0
- data/spec/vandalism_detection/features/comment_markup_frequency_spec.rb +28 -0
- data/spec/vandalism_detection/features/comment_pronoun_frequency_spec.rb +28 -0
- data/spec/vandalism_detection/features/comment_sex_frequency_spec.rb +28 -0
- data/spec/vandalism_detection/features/comment_vulgarism_frequency_spec.rb +28 -0
- data/spec/vandalism_detection/features/compressibility_spec.rb +38 -0
- data/spec/vandalism_detection/features/contains_base_spec.rb +27 -0
- data/spec/vandalism_detection/features/copyedit_spec.rb +30 -0
- data/spec/vandalism_detection/features/digit_ratio_spec.rb +30 -0
- data/spec/vandalism_detection/features/edits_per_user_spec.rb +51 -0
- data/spec/vandalism_detection/features/emoticons_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/emoticons_impact_spec.rb +45 -0
- data/spec/vandalism_detection/features/frequency_base_spec.rb +20 -0
- data/spec/vandalism_detection/features/impact_base_spec.rb +36 -0
- data/spec/vandalism_detection/features/inserted_character_distribution_spec.rb +40 -0
- data/spec/vandalism_detection/features/inserted_external_links_spec.rb +29 -0
- data/spec/vandalism_detection/features/inserted_internal_links_spec.rb +29 -0
- data/spec/vandalism_detection/features/inserted_size_spec.rb +29 -0
- data/spec/vandalism_detection/features/inserted_words_spec.rb +29 -0
- data/spec/vandalism_detection/features/longest_word_spec.rb +29 -0
- data/spec/vandalism_detection/features/markup_frequency_spec.rb +31 -0
- data/spec/vandalism_detection/features/markup_impact_spec.rb +53 -0
- data/spec/vandalism_detection/features/non_alphanumeric_ratio_spec.rb +30 -0
- data/spec/vandalism_detection/features/personal_life_spec.rb +22 -0
- data/spec/vandalism_detection/features/pronoun_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/pronoun_impact_spec.rb +53 -0
- data/spec/vandalism_detection/features/removed_all_wordlists_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/removed_bad_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/removed_biased_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/removed_character_distribution_spec.rb +40 -0
- data/spec/vandalism_detection/features/removed_emoticons_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/removed_markup_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/removed_pronoun_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/removed_sex_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/removed_size_spec.rb +29 -0
- data/spec/vandalism_detection/features/removed_vulgarism_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/removed_words_spec.rb +29 -0
- data/spec/vandalism_detection/features/replacement_similarity_spec.rb +37 -0
- data/spec/vandalism_detection/features/reverted_spec.rb +24 -0
- data/spec/vandalism_detection/features/revisions_character_distribution_spec.rb +40 -0
- data/spec/vandalism_detection/features/same_editor_spec.rb +71 -0
- data/spec/vandalism_detection/features/sex_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/sex_impact_spec.rb +53 -0
- data/spec/vandalism_detection/features/size_increment_spec.rb +29 -0
- data/spec/vandalism_detection/features/size_ratio_spec.rb +48 -0
- data/spec/vandalism_detection/features/term_frequency_spec.rb +32 -0
- data/spec/vandalism_detection/features/time_interval_spec.rb +56 -0
- data/spec/vandalism_detection/features/time_of_day_spec.rb +16 -0
- data/spec/vandalism_detection/features/upper_case_ratio_spec.rb +30 -0
- data/spec/vandalism_detection/features/upper_case_words_ratio_spec.rb +33 -0
- data/spec/vandalism_detection/features/upper_to_lower_case_ratio_spec.rb +30 -0
- data/spec/vandalism_detection/features/vulgarism_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/vulgarism_impact_spec.rb +52 -0
- data/spec/vandalism_detection/features/weekday_spec.rb +16 -0
- data/spec/vandalism_detection/features/words_increment_spec.rb +29 -0
- data/spec/vandalism_detection/instances_spec.rb +146 -0
- data/spec/vandalism_detection/page_parser_spec.rb +190 -0
- data/spec/vandalism_detection/page_spec.rb +134 -0
- data/spec/vandalism_detection/revision_parser_spec.rb +53 -0
- data/spec/vandalism_detection/revision_spec.rb +148 -0
- data/spec/vandalism_detection/test_dataset_spec.rb +227 -0
- data/spec/vandalism_detection/text_spec.rb +29 -0
- data/spec/vandalism_detection/training_dataset_spec.rb +266 -0
- data/spec/vandalism_detection/wikitext_extractor_spec.rb +97 -0
- data/spec/weka/classifiers/meta/one_class_classifier_spec.rb +82 -0
- data/spec/weka/classifiers/meta/real_ada_boost_spec.rb +37 -0
- data/spec/weka/classifiers/trees/balanced_random_forest_spec.rb +37 -0
- data/spec/weka/filters/supervised/instance/smote_spec.rb +5 -0
- data/wikipedia-vandalism_detection.gemspec +37 -0
- metadata +550 -0
@@ -0,0 +1,135 @@
|
|
1
|
+
require 'spec_helper'
|
2
|
+
|
3
|
+
describe Wikipedia::VandalismDetection::FeatureCalculator do
|
4
|
+
let(:edit) { build(:edit) }
|
5
|
+
|
6
|
+
it 'raises NoFeaturesConfiguredError when no features are configured' do
|
7
|
+
config = Wikipedia::VandalismDetection::Configuration.send(:new)
|
8
|
+
config.instance_variable_set(:@features, nil)
|
9
|
+
|
10
|
+
use_configuration(config)
|
11
|
+
|
12
|
+
expect { Wikipedia::VandalismDetection::FeatureCalculator.new }
|
13
|
+
.to raise_error Wikipedia::VandalismDetection::FeaturesNotConfiguredError
|
14
|
+
end
|
15
|
+
|
16
|
+
before do
|
17
|
+
use_test_configuration
|
18
|
+
@calculator = Wikipedia::VandalismDetection::FeatureCalculator.new
|
19
|
+
end
|
20
|
+
|
21
|
+
describe '#calculate_features_for' do
|
22
|
+
it { is_expected.to respond_to :calculate_features_for }
|
23
|
+
|
24
|
+
it 'takes an edit as parameter' do
|
25
|
+
expect { @calculator.calculate_features_for(edit) }
|
26
|
+
.not_to raise_error ArgumentError
|
27
|
+
end
|
28
|
+
|
29
|
+
it 'raises an error if called with wrong parameter type' do
|
30
|
+
revision = build(:empty_revision)
|
31
|
+
expect { @calculator.calculate_features_for(revision) }
|
32
|
+
.to raise_error ArgumentError
|
33
|
+
end
|
34
|
+
|
35
|
+
it 'returns an array' do
|
36
|
+
expect(@calculator.calculate_features_for(edit)).to be_an Array
|
37
|
+
end
|
38
|
+
|
39
|
+
it 'returns the computed numeric feature values' do
|
40
|
+
feature_values = @calculator.calculate_features_for(edit)
|
41
|
+
expect(feature_values.all? { |value| value.is_a?(Numeric) }).to be true
|
42
|
+
end
|
43
|
+
|
44
|
+
it 'returns the right number of feature values' do
|
45
|
+
count = @calculator.used_features.count
|
46
|
+
expect(@calculator.calculate_features_for(edit).count).to eq count
|
47
|
+
end
|
48
|
+
|
49
|
+
it 'uses the cleaned up text if revision contains a #REDIRECT' do
|
50
|
+
redirect_text = Text.new('#REDIRECT [[Redirect page]]')
|
51
|
+
old_revision_redirect = build(:old_revision, text: redirect_text)
|
52
|
+
new_revision_redirect = build(:new_revision, text: redirect_text)
|
53
|
+
old_revision = build(:old_revision)
|
54
|
+
new_revision = build(:new_revision)
|
55
|
+
|
56
|
+
edit_a = Wikipedia::VandalismDetection::Edit.new(old_revision_redirect, new_revision)
|
57
|
+
edit_b = Wikipedia::VandalismDetection::Edit.new(old_revision, new_revision_redirect)
|
58
|
+
|
59
|
+
config = Wikipedia::VandalismDetection.config
|
60
|
+
count = config.features.count
|
61
|
+
|
62
|
+
expect(@calculator.calculate_features_for(edit_a).count).to eq count
|
63
|
+
expect(@calculator.calculate_features_for(edit_b).count).to eq count
|
64
|
+
end
|
65
|
+
|
66
|
+
it 'includes -1 for not extractable texts in either revision' do
|
67
|
+
config = Wikipedia::VandalismDetection::Configuration.instance
|
68
|
+
config.instance_variable_set(:@features, ['all wordlists impact'])
|
69
|
+
|
70
|
+
use_configuration(config)
|
71
|
+
|
72
|
+
unparsable_wiki_text = Text.new("[[Image:img.jpg|\n{|\n|-\n|||| |}")
|
73
|
+
|
74
|
+
old_revision_unparsable = build(:old_revision, text: unparsable_wiki_text)
|
75
|
+
new_revision_unparsable = build(:new_revision, text: unparsable_wiki_text)
|
76
|
+
|
77
|
+
old_revision = build(:old_revision)
|
78
|
+
new_revision = build(:new_revision)
|
79
|
+
|
80
|
+
edit_a = Wikipedia::VandalismDetection::Edit.new(old_revision_unparsable, new_revision)
|
81
|
+
edit_b = Wikipedia::VandalismDetection::Edit.new(old_revision, new_revision_unparsable)
|
82
|
+
|
83
|
+
expect(subject.calculate_features_for(edit_a)).to include Features::MISSING_VALUE
|
84
|
+
expect(subject.calculate_features_for(edit_b)).to include Features::MISSING_VALUE
|
85
|
+
end
|
86
|
+
end
|
87
|
+
|
88
|
+
describe '#claculate_feature_for' do
|
89
|
+
let(:feature_name) { 'anonymity' }
|
90
|
+
let(:random_number) { rand(1000) }
|
91
|
+
let(:empty_revision) { build(:empty_revision) }
|
92
|
+
|
93
|
+
before do
|
94
|
+
allow_any_instance_of(Features::Anonymity)
|
95
|
+
.to receive(:calculate)
|
96
|
+
.and_return(random_number)
|
97
|
+
end
|
98
|
+
|
99
|
+
it { is_expected.to respond_to :calculate_feature_for }
|
100
|
+
|
101
|
+
it 'takes an edit and feature name as parameter' do
|
102
|
+
expect { @calculator.calculate_feature_for(edit, feature_name) }
|
103
|
+
.not_to raise_error ArgumentError
|
104
|
+
end
|
105
|
+
|
106
|
+
it 'raises an error if called with wrong parameter type edit' do
|
107
|
+
expect { @calculator.calculate_feature_for(empty_revision, feature_name) }
|
108
|
+
.to raise_error ArgumentError
|
109
|
+
end
|
110
|
+
|
111
|
+
it 'raises an error if called with wrong parameter type feature name' do
|
112
|
+
expect { @calculator.calculate_feature_for(edit, empty_revision) }
|
113
|
+
.to raise_error ArgumentError
|
114
|
+
end
|
115
|
+
|
116
|
+
it 'returns a Numeric' do
|
117
|
+
expect(@calculator.calculate_feature_for(edit, feature_name))
|
118
|
+
.to be_a Numeric
|
119
|
+
end
|
120
|
+
|
121
|
+
it 'returns the value calculated by the feature class' do
|
122
|
+
expect(@calculator.calculate_feature_for(edit, feature_name))
|
123
|
+
.to eq random_number
|
124
|
+
end
|
125
|
+
end
|
126
|
+
|
127
|
+
describe '#used_features' do
|
128
|
+
it { is_expected.to respond_to :used_features }
|
129
|
+
|
130
|
+
it 'returns an array of the features defined in the config feature.yml' do
|
131
|
+
features = Wikipedia::VandalismDetection.config.features
|
132
|
+
expect(@calculator.used_features).to match_array features
|
133
|
+
end
|
134
|
+
end
|
135
|
+
end
|
@@ -0,0 +1,30 @@
|
|
1
|
+
require 'spec_helper'
|
2
|
+
|
3
|
+
describe Wikipedia::VandalismDetection::Features::AllWordlistsFrequency do
|
4
|
+
it { is_expected.to be_a Features::FrequencyBase }
|
5
|
+
|
6
|
+
describe '#calculate' do
|
7
|
+
it 'returns the inserted number of all lists words over all inserted' do
|
8
|
+
# inserted: total 7 words, 1 vulgarism, 1 biased, 1 pronouns = 3 bad
|
9
|
+
old_text = Text.new('Your old shit. ')
|
10
|
+
new_text = Text.new('Your old shit. Fuck you great, and all the others.')
|
11
|
+
|
12
|
+
old_rev = build(:old_revision, text: old_text)
|
13
|
+
new_rev = build(:new_revision, text: new_text)
|
14
|
+
edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
|
15
|
+
|
16
|
+
expect(subject.calculate(edit)).to eq 3.0 / 7.0
|
17
|
+
end
|
18
|
+
|
19
|
+
it 'returns 0.0 on empty clean inserted text' do
|
20
|
+
old_text = Text.new('Your old shit. ')
|
21
|
+
new_text = Text.new('Your old shit. {{speedy deletion}}')
|
22
|
+
|
23
|
+
old_rev = build(:old_revision, text: old_text)
|
24
|
+
new_rev = build(:new_revision, text: new_text)
|
25
|
+
edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
|
26
|
+
|
27
|
+
expect(subject.calculate(edit)).to eq 0.0
|
28
|
+
end
|
29
|
+
end
|
30
|
+
end
|
@@ -0,0 +1,52 @@
|
|
1
|
+
require 'spec_helper'
|
2
|
+
|
3
|
+
describe Wikipedia::VandalismDetection::Features::AllWordlistsImpact do
|
4
|
+
it { is_expected.to be_a Features::ImpactBase }
|
5
|
+
|
6
|
+
describe '#calculate' do
|
7
|
+
it 'returns the impact of all wordlists words of the new revision text' do
|
8
|
+
# 1 vulgarism, 2 pronouns, 0 biased = 3 bad
|
9
|
+
old_text = Text.new('Fuck you, you and all the others')
|
10
|
+
# 3 vulgarism, 3 pronouns, 1 biased = 7 bad
|
11
|
+
new_text = Text.new('Fuck you great, fuck you, you all others sluts')
|
12
|
+
|
13
|
+
old_rev = build(:old_revision, text: old_text)
|
14
|
+
new_rev = build(:new_revision, text: new_text)
|
15
|
+
edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
|
16
|
+
|
17
|
+
expect(subject.calculate(edit)).to eq 3.0 / (3.0 + 7.0)
|
18
|
+
end
|
19
|
+
|
20
|
+
it 'returns 0.5 if both text revisions have no terms' do
|
21
|
+
text = Text.new('{{speedy deletion}}')
|
22
|
+
|
23
|
+
old_rev = build(:old_revision, text: text)
|
24
|
+
new_rev = build(:new_revision, text: text)
|
25
|
+
edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
|
26
|
+
|
27
|
+
expect(subject.calculate(edit)).to eq 0.5
|
28
|
+
end
|
29
|
+
|
30
|
+
it 'returns 0.0 for an emtpy clean text in the old revision' do
|
31
|
+
old_text = Text.new('{{speedy deletion}}')
|
32
|
+
new_text = Text.new('fuck')
|
33
|
+
|
34
|
+
old_rev = build(:old_revision, text: old_text)
|
35
|
+
new_rev = build(:new_revision, text: new_text)
|
36
|
+
edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
|
37
|
+
|
38
|
+
expect(subject.calculate(edit)).to eq 0.0
|
39
|
+
end
|
40
|
+
|
41
|
+
it 'returns 1.0 for an emtpy clean text in the new revision' do
|
42
|
+
old_text = Text.new('fuck')
|
43
|
+
new_text = Text.new('{{speedy deletion}}')
|
44
|
+
|
45
|
+
old_rev = build(:old_revision, text: old_text)
|
46
|
+
new_rev = build(:new_revision, text: new_text)
|
47
|
+
edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
|
48
|
+
|
49
|
+
expect(subject.calculate(edit)).to eq 1.0
|
50
|
+
end
|
51
|
+
end
|
52
|
+
end
|
@@ -0,0 +1,67 @@
|
|
1
|
+
require 'spec_helper'
|
2
|
+
|
3
|
+
describe Wikipedia::VandalismDetection::Features::AnonymityPrevious do
|
4
|
+
it { is_expected.to be_a Features::Base }
|
5
|
+
|
6
|
+
describe '#calculate' do
|
7
|
+
context 'both contributors are given' do
|
8
|
+
it 'return 1.0 in case of an registered previous editor' do
|
9
|
+
old_rev = build(:old_revision, contributor: 'Peter')
|
10
|
+
new_rev = build(:new_revision)
|
11
|
+
edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
|
12
|
+
|
13
|
+
expect(subject.calculate(edit)).to eq 1
|
14
|
+
end
|
15
|
+
|
16
|
+
it 'returns 0.0 in case of an anonymous previous editor' do
|
17
|
+
old_rev = build(:old_revision, contributor: '137.163.16.199')
|
18
|
+
new_rev = build(:new_revision)
|
19
|
+
edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
|
20
|
+
|
21
|
+
expect(subject.calculate(edit)).to eq 0
|
22
|
+
end
|
23
|
+
end
|
24
|
+
|
25
|
+
context 'previous contributor not given' do
|
26
|
+
context 'for a registered previous editor' do
|
27
|
+
it 'requests the user from Wikipedia API and returns 1' do
|
28
|
+
old_rev = build(:old_revision, id: '324557983', contributor: nil)
|
29
|
+
new_rev = build(
|
30
|
+
:new_revision,
|
31
|
+
id: '329962649',
|
32
|
+
parent_id: '324557983',
|
33
|
+
contributor: 'Tomaxer'
|
34
|
+
)
|
35
|
+
|
36
|
+
edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
|
37
|
+
|
38
|
+
expect(subject.calculate(edit)).to eq 1
|
39
|
+
end
|
40
|
+
end
|
41
|
+
|
42
|
+
context 'for an anonymous previous editor' do # also same editor!
|
43
|
+
it 'requests the user from Wikipedia API and returns 0' do
|
44
|
+
old_rev = build(:old_revision, id: '328774110', contributor: nil)
|
45
|
+
new_rev = build(:new_revision, id: '328774035', parent_id: '328774110')
|
46
|
+
edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
|
47
|
+
|
48
|
+
expect(subject.calculate(edit)).to eq 0
|
49
|
+
end
|
50
|
+
end
|
51
|
+
|
52
|
+
context 'if old reivision is not available anymore' do
|
53
|
+
it 'returns missing' do
|
54
|
+
# to get api call, see:
|
55
|
+
# https://en.wikipedia.org/w/api.php?action=query&prop=revisions&rvprop=timestamp&revids=325218985
|
56
|
+
# <rev revid="325218985"/>
|
57
|
+
|
58
|
+
old_rev = build(:old_revision, id: '325218985', contributor: nil)
|
59
|
+
new_rev = build(:new_revision, id: '326980599', parent_id: '325218985')
|
60
|
+
edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
|
61
|
+
|
62
|
+
expect(subject.calculate(edit)).to eq Features::MISSING_VALUE
|
63
|
+
end
|
64
|
+
end
|
65
|
+
end
|
66
|
+
end
|
67
|
+
end
|
@@ -0,0 +1,17 @@
|
|
1
|
+
require 'spec_helper'
|
2
|
+
|
3
|
+
describe Wikipedia::VandalismDetection::Features::Anonymity do
|
4
|
+
it { is_expected.to be_a Features::Base }
|
5
|
+
|
6
|
+
describe '#calculate' do
|
7
|
+
it 'return 1.0 in case of an registered editor' do
|
8
|
+
edit = build :edit, new_revision: build(:registered_revision)
|
9
|
+
expect(subject.calculate(edit)).to eq 1
|
10
|
+
end
|
11
|
+
|
12
|
+
it 'returns 0.0 in case of an anonymous editor' do
|
13
|
+
edit = build :edit, new_revision: build(:anonymous_revision)
|
14
|
+
expect(subject.calculate(edit)).to eq 0
|
15
|
+
end
|
16
|
+
end
|
17
|
+
end
|
@@ -0,0 +1,29 @@
|
|
1
|
+
require 'spec_helper'
|
2
|
+
|
3
|
+
describe Wikipedia::VandalismDetection::Features::ArticleSize do
|
4
|
+
it { is_expected.to be_a Features::Base }
|
5
|
+
|
6
|
+
describe '#calculate' do
|
7
|
+
it 'returns the size of the edit’s new revisions' do
|
8
|
+
old_rev_text = Text.new('123')
|
9
|
+
new_rev_text = Text.new('123 456789') # size 10 (with spaces)
|
10
|
+
|
11
|
+
old_rev = build(:old_revision, text: old_rev_text)
|
12
|
+
new_rev = build(:new_revision, text: new_rev_text)
|
13
|
+
edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
|
14
|
+
|
15
|
+
expect(subject.calculate(edit)).to eq 10
|
16
|
+
end
|
17
|
+
|
18
|
+
it "returns 0 if the edit's new revisions is empty" do
|
19
|
+
old_rev_text = Text.new('123')
|
20
|
+
new_rev_text = Text.new # size 0
|
21
|
+
|
22
|
+
old_rev = build(:old_revision, text: old_rev_text)
|
23
|
+
new_rev = build(:new_revision, text: new_rev_text)
|
24
|
+
edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
|
25
|
+
|
26
|
+
expect(subject.calculate(edit)).to eq 0
|
27
|
+
end
|
28
|
+
end
|
29
|
+
end
|
@@ -0,0 +1,30 @@
|
|
1
|
+
require 'spec_helper'
|
2
|
+
|
3
|
+
describe Wikipedia::VandalismDetection::Features::BadFrequency do
|
4
|
+
it { is_expected.to be_a Features::FrequencyBase }
|
5
|
+
|
6
|
+
describe '#calculate' do
|
7
|
+
it 'returns the number of vulgarism words relative to all words count' do
|
8
|
+
# total 6 words, 3 bad
|
9
|
+
old_text = Text.new('Old whatever.')
|
10
|
+
new_text = Text.new('Old whatever. New ugly contents. Hi, gotta fun.')
|
11
|
+
|
12
|
+
old_rev = build(:old_revision, text: old_text)
|
13
|
+
new_rev = build(:new_revision, text: new_text)
|
14
|
+
edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
|
15
|
+
|
16
|
+
expect(subject.calculate(edit)).to eq 3.0 / 6.0
|
17
|
+
end
|
18
|
+
|
19
|
+
it 'returns 0.0 for an emtpy clean text in the new revision' do
|
20
|
+
old_text = Text.new('Old guy.')
|
21
|
+
new_text = Text.new('Old guy. {{speedy deletion}}')
|
22
|
+
|
23
|
+
old_rev = build(:old_revision, text: old_text)
|
24
|
+
new_rev = build(:new_revision, text: new_text)
|
25
|
+
edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
|
26
|
+
|
27
|
+
expect(subject.calculate(edit)).to eq 0.0
|
28
|
+
end
|
29
|
+
end
|
30
|
+
end
|
@@ -0,0 +1,53 @@
|
|
1
|
+
require 'spec_helper'
|
2
|
+
|
3
|
+
describe Wikipedia::VandalismDetection::Features::BadImpact do
|
4
|
+
it { is_expected.to be_a Features::ImpactBase }
|
5
|
+
|
6
|
+
describe '#calculate' do
|
7
|
+
it 'returns the impact of bad words of the edit’s new revision text' do
|
8
|
+
# 3 bad words
|
9
|
+
old_text = Text.new('Hi, old text 666, dont know')
|
10
|
+
|
11
|
+
# 4 bad words
|
12
|
+
new_text = Text.new('Hi, new text dosent, whatever, guy')
|
13
|
+
|
14
|
+
old_rev = build(:old_revision, text: old_text)
|
15
|
+
new_rev = build(:new_revision, text: new_text)
|
16
|
+
edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
|
17
|
+
|
18
|
+
expect(subject.calculate(edit)).to eq 3.0 / (3.0 + 4.0)
|
19
|
+
end
|
20
|
+
|
21
|
+
it 'returns 0.5 on both no terms in text revisions' do
|
22
|
+
text = Text.new('{speedy deletion}}')
|
23
|
+
|
24
|
+
old_rev = build(:old_revision, text: text)
|
25
|
+
new_rev = build(:new_revision, text: text)
|
26
|
+
edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
|
27
|
+
|
28
|
+
expect(subject.calculate(edit)).to eq 0.5
|
29
|
+
end
|
30
|
+
|
31
|
+
it 'returns 0.0 on emtpy clean text of old revision' do
|
32
|
+
old_text = Text.new('{{speedy deletion}}')
|
33
|
+
new_text = Text.new('Guy')
|
34
|
+
|
35
|
+
old_rev = build(:old_revision, text: old_text)
|
36
|
+
new_rev = build(:new_revision, text: new_text)
|
37
|
+
edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
|
38
|
+
|
39
|
+
expect(subject.calculate(edit)).to eq 0.0
|
40
|
+
end
|
41
|
+
|
42
|
+
it 'returns 1.0 on emtpy clean text of new revision' do
|
43
|
+
old_text = Text.new('Guy')
|
44
|
+
new_text = Text.new('{{speedy deletion}}')
|
45
|
+
|
46
|
+
old_rev = build(:old_revision, text: old_text)
|
47
|
+
new_rev = build(:new_revision, text: new_text)
|
48
|
+
edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
|
49
|
+
|
50
|
+
expect(subject.calculate(edit)).to eq 1.0
|
51
|
+
end
|
52
|
+
end
|
53
|
+
end
|
@@ -0,0 +1,41 @@
|
|
1
|
+
require 'spec_helper'
|
2
|
+
|
3
|
+
describe Wikipedia::VandalismDetection::Features::Base do
|
4
|
+
describe '#count' do
|
5
|
+
let(:text) { 'I, you: i will help You' }
|
6
|
+
|
7
|
+
it { is_expected.to respond_to(:count).with(2).arguments }
|
8
|
+
|
9
|
+
it 'raises an error if option :in is not defined' do
|
10
|
+
expect { subject.count(%i[i you], from: text) }
|
11
|
+
.to raise_error ArgumentError
|
12
|
+
end
|
13
|
+
|
14
|
+
it 'raises an error if terms is not an Array or String' do
|
15
|
+
expect { subject.count({ term: 'You' }, in: text) }
|
16
|
+
.to raise_error ArgumentError
|
17
|
+
end
|
18
|
+
|
19
|
+
it 'returns the total number of terms found for the given terms array' do
|
20
|
+
terms = %i[i you]
|
21
|
+
expect(subject.count(terms, in: text)).to eq 4
|
22
|
+
end
|
23
|
+
|
24
|
+
it 'returns the number of terms found for the given single term' do
|
25
|
+
expect(subject.count('You', in: text)).to eq 2
|
26
|
+
end
|
27
|
+
end
|
28
|
+
|
29
|
+
describe '#calculate' do
|
30
|
+
it { is_expected.to respond_to :calculate }
|
31
|
+
|
32
|
+
it 'takes an Wikipedia::Edit as argument' do
|
33
|
+
edit = build(:edit)
|
34
|
+
expect { subject.calculate(edit) }.not_to raise_error ArgumentError
|
35
|
+
end
|
36
|
+
|
37
|
+
it 'raises an ArgumentError if argument is no Wikipedia::Edit' do
|
38
|
+
expect { subject.calculate('string') }.to raise_error ArgumentError
|
39
|
+
end
|
40
|
+
end
|
41
|
+
end
|
@@ -0,0 +1,30 @@
|
|
1
|
+
require 'spec_helper'
|
2
|
+
|
3
|
+
describe Wikipedia::VandalismDetection::Features::BiasedFrequency do
|
4
|
+
it { is_expected.to be_a Features::FrequencyBase }
|
5
|
+
|
6
|
+
describe '#calculate' do
|
7
|
+
it 'returns the inserted number of biased words over all inserted words' do
|
8
|
+
# inserted: total 7 words, 3 biased
|
9
|
+
old_text = Text.new('Great old.')
|
10
|
+
new_text = Text.new('Great old. This is so great, really a classic.')
|
11
|
+
|
12
|
+
old_rev = build(:old_revision, text: old_text)
|
13
|
+
new_rev = build(:new_revision, text: new_text)
|
14
|
+
edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
|
15
|
+
|
16
|
+
expect(subject.calculate(edit)).to eq 3.0 / 7.0
|
17
|
+
end
|
18
|
+
|
19
|
+
it 'returns 0.0 on emtpy clean inserted text' do
|
20
|
+
old_text = Text.new('Great old.')
|
21
|
+
new_text = Text.new('Great old. {{speedy deletion}}')
|
22
|
+
|
23
|
+
old_rev = build(:old_revision, text: old_text)
|
24
|
+
new_rev = build(:new_revision, text: new_text)
|
25
|
+
edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
|
26
|
+
|
27
|
+
expect(subject.calculate(edit)).to eq 0.0
|
28
|
+
end
|
29
|
+
end
|
30
|
+
end
|
@@ -0,0 +1,52 @@
|
|
1
|
+
require 'spec_helper'
|
2
|
+
|
3
|
+
describe Wikipedia::VandalismDetection::Features::BiasedImpact do
|
4
|
+
it { is_expected.to be_a Features::ImpactBase }
|
5
|
+
|
6
|
+
describe '#calculate' do
|
7
|
+
it 'returns the impact of biased words of the new revision text' do
|
8
|
+
# 1 vulgarism
|
9
|
+
old_text = Text.new('this is classic!')
|
10
|
+
# 3 vulgarism
|
11
|
+
new_text = Text.new('This is classic, legendary and amazing!')
|
12
|
+
|
13
|
+
old_rev = build(:old_revision, text: old_text)
|
14
|
+
new_rev = build(:new_revision, text: new_text)
|
15
|
+
edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
|
16
|
+
|
17
|
+
expect(subject.calculate(edit)).to eq 1.0 / (1.0 + 3.0)
|
18
|
+
end
|
19
|
+
|
20
|
+
it 'returns 0.5 on both no terms in text revisions' do
|
21
|
+
text = Text.new('{{speedy deletion}}')
|
22
|
+
|
23
|
+
old_rev = build(:old_revision, text: text)
|
24
|
+
new_rev = build(:new_revision, text: text)
|
25
|
+
edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
|
26
|
+
|
27
|
+
expect(subject.calculate(edit)).to eq 0.5
|
28
|
+
end
|
29
|
+
|
30
|
+
it 'returns 0.0 on emtpy clean text of old revision' do
|
31
|
+
old_text = Text.new('{{speedy deletion}}')
|
32
|
+
new_text = Text.new('great')
|
33
|
+
|
34
|
+
old_rev = build(:old_revision, text: old_text)
|
35
|
+
new_rev = build(:new_revision, text: new_text)
|
36
|
+
edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
|
37
|
+
|
38
|
+
expect(subject.calculate(edit)).to eq 0.0
|
39
|
+
end
|
40
|
+
|
41
|
+
it 'returns 1.0 on emtpy clean text of new revision' do
|
42
|
+
old_text = Text.new('great')
|
43
|
+
new_text = Text.new('{{speedy deletion}}')
|
44
|
+
|
45
|
+
old_rev = build(:old_revision, text: old_text)
|
46
|
+
new_rev = build(:new_revision, text: new_text)
|
47
|
+
edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
|
48
|
+
|
49
|
+
expect(subject.calculate(edit)).to eq 1.0
|
50
|
+
end
|
51
|
+
end
|
52
|
+
end
|
@@ -0,0 +1,35 @@
|
|
1
|
+
require 'spec_helper'
|
2
|
+
|
3
|
+
describe Wikipedia::VandalismDetection::Features::Blanking do
|
4
|
+
let(:blank_text) { 'a' * (Features::Blanking::BLANKING_THRESHOLD - 1) }
|
5
|
+
|
6
|
+
it { is_expected.to be_a Features::Base }
|
7
|
+
|
8
|
+
describe '#calculate' do
|
9
|
+
it 'returns 1.0 in case of full blanking the new revision' do
|
10
|
+
# full blanking means size < BLANKING_THRESHOLD.
|
11
|
+
old_rev = build(:old_revision, text: "#{blank_text} additional text")
|
12
|
+
new_rev = build(:new_revision, text: blank_text)
|
13
|
+
edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
|
14
|
+
|
15
|
+
expect(subject.calculate(edit)).to eq 1.0
|
16
|
+
end
|
17
|
+
|
18
|
+
it 'returns 0.0 in case of not full blanking the new revision' do
|
19
|
+
# not full blanking means size > BLANKING_THRESHOLD.
|
20
|
+
old_rev = build(:old_revision, text: "#{blank_text} additional text")
|
21
|
+
new_rev = build(:new_revision, text: "#{blank_text}a")
|
22
|
+
edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
|
23
|
+
|
24
|
+
expect(subject.calculate(edit)).to eq 0.0
|
25
|
+
end
|
26
|
+
|
27
|
+
it 'returns 0.0 if old revision is <= new revision' do
|
28
|
+
old_rev = build(:old_revision, text: blank_text)
|
29
|
+
new_rev = build(:new_revision, text: blank_text.next!)
|
30
|
+
edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
|
31
|
+
|
32
|
+
expect(subject.calculate(edit)).to eq 0.0
|
33
|
+
end
|
34
|
+
end
|
35
|
+
end
|
@@ -0,0 +1,30 @@
|
|
1
|
+
require 'spec_helper'
|
2
|
+
|
3
|
+
describe Wikipedia::VandalismDetection::Features::CharacterDiversity do
|
4
|
+
it { is_expected.to be_a Features::Base }
|
5
|
+
|
6
|
+
describe '#calculate' do
|
7
|
+
it 'returns the character diversity of the new inserted text' do
|
8
|
+
old_text = Text.new('text')
|
9
|
+
# 9 unique characters of total 14
|
10
|
+
new_text = Text.new('text [[aa ab cdeefg]]')
|
11
|
+
|
12
|
+
old_rev = build(:old_revision, text: old_text)
|
13
|
+
new_rev = build(:new_revision, text: new_text)
|
14
|
+
edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
|
15
|
+
|
16
|
+
expect(subject.calculate(edit)).to eq 14**(1.0 / 9)
|
17
|
+
end
|
18
|
+
|
19
|
+
it 'returns 0.0 if no text inserted' do
|
20
|
+
old_text = Text.new('deletion text')
|
21
|
+
new_text = Text.new('text')
|
22
|
+
|
23
|
+
old_rev = build(:old_revision, text: old_text)
|
24
|
+
new_rev = build(:new_revision, text: new_text)
|
25
|
+
edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
|
26
|
+
|
27
|
+
expect(subject.calculate(edit)).to eq 0.0
|
28
|
+
end
|
29
|
+
end
|
30
|
+
end
|
@@ -0,0 +1,31 @@
|
|
1
|
+
require 'spec_helper'
|
2
|
+
|
3
|
+
describe Wikipedia::VandalismDetection::Features::CharacterSequence do
|
4
|
+
it { is_expected.to be_a Features::Base }
|
5
|
+
|
6
|
+
describe '#calculate' do
|
7
|
+
it 'returns the number of the new revision’s longest character sequence' do
|
8
|
+
old_text = Text.new('a 666666')
|
9
|
+
new_text = Text.new("a 666666 4444ccc eefffff gggg g ''fffaffff''")
|
10
|
+
|
11
|
+
old_rev = build(:old_revision, text: old_text)
|
12
|
+
new_rev = build(:new_revision, text: new_text)
|
13
|
+
|
14
|
+
edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
|
15
|
+
|
16
|
+
expect(subject.calculate(edit)).to eq 5
|
17
|
+
end
|
18
|
+
|
19
|
+
it 'returns 0 if no text was inserted' do
|
20
|
+
old_text = Text.new('a 666666 4444ccc eeeefffff gggg g')
|
21
|
+
new_text = Text.new('a 666666 ')
|
22
|
+
|
23
|
+
old_rev = build(:old_revision, text: old_text)
|
24
|
+
new_rev = build(:new_revision, text: new_text)
|
25
|
+
|
26
|
+
edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
|
27
|
+
|
28
|
+
expect(subject.calculate(edit)).to eq 0
|
29
|
+
end
|
30
|
+
end
|
31
|
+
end
|
@@ -0,0 +1,28 @@
|
|
1
|
+
require 'spec_helper'
|
2
|
+
|
3
|
+
describe Wikipedia::VandalismDetection::Features::CommentBadFrequency do
|
4
|
+
it { is_expected.to be_a Features::FrequencyBase }
|
5
|
+
|
6
|
+
describe '#calculate' do
|
7
|
+
it 'returns the number of bad words in comment over all words' do
|
8
|
+
# total 11 words, 7 bad words
|
9
|
+
comment = Text.new('666 was 666 if 666 was 666 and guy are 666')
|
10
|
+
|
11
|
+
old_rev = build(:old_revision)
|
12
|
+
new_rev = build(:new_revision, comment: comment)
|
13
|
+
edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
|
14
|
+
|
15
|
+
expect(subject.calculate(edit)).to eq 6.0 / 11.0
|
16
|
+
end
|
17
|
+
|
18
|
+
it 'returns 0.0 on emtpy clean text comment' do
|
19
|
+
comment = Text.new('{{speedy deletion}}')
|
20
|
+
|
21
|
+
old_rev = build(:old_revision)
|
22
|
+
new_rev = build(:new_revision, comment: comment)
|
23
|
+
edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
|
24
|
+
|
25
|
+
expect(subject.calculate(edit)).to eq 0.0
|
26
|
+
end
|
27
|
+
end
|
28
|
+
end
|