confidential_info_redactor_lite 0.0.19 → 0.0.20

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: ecfaf02e5c062c12167515680d1ce6e4a0f73d62
-  data.tar.gz: 069f0d297228e1fa76ab940776b43398a0ca130e
+  metadata.gz: 536892192827f07c05d351d34ba72add5dc8b572
+  data.tar.gz: dd8caa21ae31b337c9d366e4b0dc792ee699d646
 SHA512:
-  metadata.gz: fe912ff29029746481b880e81090fb8298f3b0581270bb954f903afd8a9c6ff7d0e6da0269af0eb2a8ad444562811d68d0faee43eadd329e3dedd59eb6330400
-  data.tar.gz: 4dd204b8f0d34e0d13beaa1795806c4a39c406d01ee258367a7c877cf144e8ea7c3866c65b4910dc6a13b4352a669b11c61fb9df9813fe64de5546a89b97005c
+  metadata.gz: 8cf63069cb1f56353515eb0c0bb4d425c2d15749e119f1cb6477b3c4a456cd5742a52729be1a79fa71669d0e11665fa90aa7a3545f4c9ce0ba0529ce680ffac5
+  data.tar.gz: b22085b7813a3f3261711b6c5da379d5582fb6ae4d0b3eba0b2cfdb324abd3ee7c9a8218d61af5cf3e6a9e2bc2cd4993b9778be910731eccdf3febad6fd41f00

data/lib/confidential_info_redactor_lite/extractor.rb

@@ -14,6 +14,7 @@ module ConfidentialInfoRedactorLite
       extracted_terms = []
       PragmaticSegmenter::Segmenter.new(text: text, language: language).segment.each do |segment|
         initial_extracted_terms = segment.gsub(EXTRACT_REGEX).map { |match| match unless corpus.include?(match.downcase.gsub(/[\?\.\)\(\!\\\/\"\:\;]/, '').gsub(/”/,'').gsub(/\'$/, '')) }.compact
+        next if initial_extracted_terms.length.eql?(segment.split(' ').length)
         initial_extracted_terms.each do |ngram|
           ngram.split(/[\?\)\(\!\\\/\"\:\;\,]/).each do |t|
             next if !(t !~ /.*\d+.*/)

data/lib/confidential_info_redactor_lite/version.rb

@@ -1,3 +1,3 @@
 module ConfidentialInfoRedactorLite
-  VERSION = "0.0.19"
+  VERSION = "0.0.20"
 end

data/spec/confidential_info_redactor_lite/extractor_spec.rb

@@ -139,6 +139,11 @@ RSpec.describe ConfidentialInfoRedactorLite::Extractor do
         text = 'Corrigendum to Council Regulation (EC) No 85/2009 of 19 January 2009 amending Regulation (EC) No 1083/2006 laying down general provisions on the European Regional Development Fund, the European Social Fund and the Cohesion Fund concerning certain provisions relating to financial management'
         expect(described_class.new(text: text, corpus: corpus, language: 'en').extract).to eq(["Corrigendum", "Council Regulation", "No", "January", "Regulation", "European Regional Development Fund", "European Social Fund", "Cohesion Fund"])
       end
+
+      it 'extracts the proper nouns from a text #017' do
+        text = 'John'
+        expect(described_class.new(text: text, corpus: corpus, language: 'en').extract).to eq([])
+      end
     end
 
     context 'German (de)' do

data/spec/confidential_info_redactor_lite/redactor_spec.rb

@@ -198,7 +198,7 @@ RSpec.describe ConfidentialInfoRedactorLite::Redactor do
         Don’t forget to use your imagination and creativity!
       EOF
       tokens = ConfidentialInfoRedactorLite::Extractor.new(text: text, corpus: corpus).extract
-      expect(described_class.new(text: text, language: 'en', tokens: tokens, dow: en_dow, dow_abbr: en_dow_abbr, months: en_months, months_abbr: en_month_abbr).redact).to eq("        <redacted>\n\n        <redacted> is hosting the <redacted number> <redacted>.  So get out your putter and your camera and see if you have what it takes.  Are you a <redacted>?\n\n        <redacted>: <redacted number>) <redacted> of <redacted number> professional miniature golfers, each from a different country. (<redacted number> points; <redacted number> bonus points if the professional miniature golfers are also from <redacted number> different continents) <redacted number>) <redacted> of yourself next to each obstacle in our list of the Top <redacted number> <redacted>. (<redacted number> points; <redacted number> bonus points for each obstacle that exactly matches the one pictured in the article) <redacted number>) <redacted> your own full-size miniature golf hole. (<redacted number> points; up to <redacted number> bonus points available depending on the craftsmanship, playability, creativity and fun factor of your hole) <redacted number>) <redacted> of yourself making a hole-in-one on two consecutive miniature golf holes. <redacted> video must be one continuous shot with no editing. (<redacted number> points) <redacted number>) <redacted> of yourself with the <redacted> mascot. (<redacted number> points; <redacted number> bonus points if you are wearing a <redacted> t-shirt) <redacted number>) <redacted> of yourself with the completed <redacted> wobblehead. (<redacted number> points; <redacted number> bonus points if the picture is taken at a miniature golf course) <redacted number>) <redacted> of a completed scorecard from a round of miniature golf. <redacted> round of golf must have taken place after the start of this scavenger hunt. (<redacted number> points) <redacted number>) <redacted> of completed scorecards from <redacted number> different miniature golf courses. <redacted> round of golf must have taken place after the start of this scavenger hunt. (<redacted number> points) <redacted number>) <redacted> an entry to the <redacted number> <redacted>. (<redacted number> points; <redacted number> bonus points if your entry gets more than <redacted number> votes) <redacted number>) <redacted> from the <redacted> app showing a 9-hole score below par. (<redacted number> points) <redacted number>) <redacted> from the <redacted> app showing that you have successfully unlocked all of the holes in the game. (<redacted number> points) <redacted number>) <redacted> of the <redacted> wobblehead at a <redacted>. (<redacted number> points) <redacted number>) <redacted> and submit the <redacted> ‘Practice <redacted>’ and ‘Final <redacted>’ for any one of the <redacted> math or physics lessons. (<redacted number> points; <redacted number> bonus points if you complete two lessons) <redacted number>) <redacted> of yourself with at least <redacted number> different colored miniature golf balls. (<redacted number> points; <redacted number> bonus points for each additional color {limit of <redacted number> bonus points}) <redacted number>) <redacted> of yourself with a famous golfer or miniature golfer. (<redacted number> points; <redacted number> bonus points if the golfer is on the <redacted> tour <redacted> you are wearing a <redacted> t-shirt in the picture) <redacted number>) <redacted> of yourself making a hole-in-one on a miniature golf hole with a loop-de-loop obstacle. (<redacted number> points) <redacted number>) <redacted> of yourself successfully making a trick miniature golf shot. (<redacted number> points; up to <redacted number> bonus points available depending on the difficulty and complexity of the trick shot)\n\n\n        <redacted>: <redacted number> <redacted> <redacted>\n\n        <redacted>\n        (<redacted number>  <redacted number> <redacted> - <redacted>)\n\n        <redacted> team will judge the scavenger hunt and all decisions will be final. <redacted> is sponsoring it. <redacted> scavenger hunt is open to anyone and everyone.  <redacted> scavenger hunt ends on <redacted date>.\n\n        <redacted> enter the scavenger hunt, send an email to info <redacted> putterking <redacted> com with the subject line: \"<redacted>\".  In the email please include links to the pictures and videos you are submitting.  You can utilize free photo and video hosting sites such as <redacted>, <redacted>, <redacted>, <redacted>, etc. for your submissions.\n\n        <redacted> entering the <redacted>, you allow <redacted> to use or link to any of the pictures or videos you submit for advertisements and promotions.\n\n        Don’t forget to use your imagination and creativity!\n")
+      expect(described_class.new(text: text, language: 'en', tokens: tokens, dow: en_dow, dow_abbr: en_dow_abbr, months: en_months, months_abbr: en_month_abbr).redact).to eq("        <redacted>\n\n        <redacted> is hosting the <redacted number> <redacted>.  So get out your putter and your camera and see if you have what it takes.  Are you a <redacted>?\n\n        <redacted>: <redacted number>) <redacted> of <redacted number> professional miniature golfers, each from a different country. (<redacted number> points; <redacted number> bonus points if the professional miniature golfers are also from <redacted number> different continents) <redacted number>) <redacted> of yourself next to each obstacle in our list of the Top <redacted number> <redacted>. (<redacted number> points; <redacted number> bonus points for each obstacle that exactly matches the one pictured in the article) <redacted number>) <redacted> your own full-size miniature golf hole. (<redacted number> points; up to <redacted number> bonus points available depending on the craftsmanship, playability, creativity and fun factor of your hole) <redacted number>) <redacted> of yourself making a hole-in-one on two consecutive miniature golf holes. <redacted> video must be one continuous shot with no editing. (<redacted number> points) <redacted number>) <redacted> of yourself with the <redacted> mascot. (<redacted number> points; <redacted number> bonus points if you are wearing a <redacted> t-shirt) <redacted number>) <redacted> of yourself with the completed <redacted> wobblehead. (<redacted number> points; <redacted number> bonus points if the picture is taken at a miniature golf course) <redacted number>) <redacted> of a completed scorecard from a round of miniature golf. <redacted> round of golf must have taken place after the start of this scavenger hunt. (<redacted number> points) <redacted number>) <redacted> of completed scorecards from <redacted number> different miniature golf courses. <redacted> round of golf must have taken place after the start of this scavenger hunt. (<redacted number> points) <redacted number>) <redacted> an entry to the <redacted number> <redacted>. (<redacted number> points; <redacted number> bonus points if your entry gets more than <redacted number> votes) <redacted number>) <redacted> from the <redacted> app showing a 9-hole score below par. (<redacted number> points) <redacted number>) <redacted> from the <redacted> app showing that you have successfully unlocked all of the holes in the game. (<redacted number> points) <redacted number>) <redacted> of the <redacted> wobblehead at a <redacted>. (<redacted number> points) <redacted number>) <redacted> and submit the <redacted> ‘Practice <redacted>’ and ‘Final <redacted>’ for any one of the <redacted> math or physics lessons. (<redacted number> points; <redacted number> bonus points if you complete two lessons) <redacted number>) <redacted> of yourself with at least <redacted number> different colored miniature golf balls. (<redacted number> points; <redacted number> bonus points for each additional color {limit of <redacted number> bonus points}) <redacted number>) <redacted> of yourself with a famous golfer or miniature golfer. (<redacted number> points; <redacted number> bonus points if the golfer is on the <redacted> tour <redacted> you are wearing a <redacted> t-shirt in the picture) <redacted number>) <redacted> of yourself making a hole-in-one on a miniature golf hole with a loop-de-loop obstacle. (<redacted number> points) <redacted number>) <redacted> of yourself successfully making a trick miniature golf shot. (<redacted number> points; up to <redacted number> bonus points available depending on the difficulty and complexity of the trick shot)\n\n\n        Prizes: <redacted number> <redacted> <redacted>\n\n        <redacted>\n        (<redacted number>  <redacted number> <redacted> - <redacted>)\n\n        <redacted> team will judge the scavenger hunt and all decisions will be final. <redacted> is sponsoring it. <redacted> scavenger hunt is open to anyone and everyone.  <redacted> scavenger hunt ends on <redacted date>.\n\n        <redacted> enter the scavenger hunt, send an email to info <redacted> putterking <redacted> com with the subject line: \"<redacted>\".  In the email please include links to the pictures and videos you are submitting.  You can utilize free photo and video hosting sites such as <redacted>, <redacted>, <redacted>, <redacted>, etc. for your submissions.\n\n        <redacted> entering the <redacted>, you allow <redacted> to use or link to any of the pictures or videos you submit for advertisements and promotions.\n\n        Don’t forget to use your imagination and creativity!\n")
     end
 
     it 'redacts all confidential information from a text #003' do

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: confidential_info_redactor_lite
 version: !ruby/object:Gem::Version
-  version: 0.0.19
+  version: 0.0.20
 platform: ruby
 authors:
 - Kevin S. Dias