confidential_info_redactor_lite 1.0.8 → 1.0.9
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
checksums.yaml
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
---
|
|
2
2
|
SHA1:
|
|
3
|
-
metadata.gz:
|
|
4
|
-
data.tar.gz:
|
|
3
|
+
metadata.gz: 7df96ec7fe0c910eb8ab77e6cc54bf018d142ded
|
|
4
|
+
data.tar.gz: 3951ed10c6e222bb24e1725a0944cd65807e02e9
|
|
5
5
|
SHA512:
|
|
6
|
-
metadata.gz:
|
|
7
|
-
data.tar.gz:
|
|
6
|
+
metadata.gz: 9cbff4361d65830508c0dc36174485f97df3e8e6451cf0f7ada707752e50720e1690ea14aa99f7e4ca49914974916f3b0c4ec1a47a4d22f25fc96f2e1f9aed79
|
|
7
|
+
data.tar.gz: e794763d38415090f1b3af1c0bd497543d5e65dc3651942e969155034554a38389068989b501d3f9d97f8aa8346a3a262c656eb86521e790424b53e44d6bc190
|
|
@@ -5,7 +5,7 @@ module ConfidentialInfoRedactorLite
|
|
|
5
5
|
# This class redacts various tokens from a text
|
|
6
6
|
class Redactor
|
|
7
7
|
# Rubular: http://rubular.com/r/OI2wQZ0KSl
|
|
8
|
-
NUMBER_REGEX = /(?<=\A|\A\()[^(]?\d+((,|\.|\/)*\d)*(\D?\s|\s|[[:cntrl:]]|[[:space:]]|\.?\s|\.$|$)|(?<=[[:cntrl:]]|[[:space:]]|\s|\s\(|\s'|\s‘)[^('‘]?\d+((,|\.|\/)*\d)*\"*(?=(\D?\s|\s|[[:cntrl:]]|[[:space:]]|\.?\s|\.$|$))|(?<=\s)\d+(nd|th|st)|(?<=\s)\d+\/\d+\"*(?=\s)|(?<=\()\S{1}\d+(?=\))|(?<=\s{1})\S{1}\d+\z|^\d+$|(?<=\A|\A\(|\s|[[:cntrl:]]|[[:space:]]|\s\()[^(]?\d+((,|\.|\/)*\d)*\D{2}(?=($|\s+))
|
|
8
|
+
NUMBER_REGEX = /(?<=\A|\A\()[^(]?\d+((,|\.|\/)*\d)*(\D?\s|\s|[[:cntrl:]]|[[:space:]]|\.?\s|\.$|$)|(?<=[[:cntrl:]]|[[:space:]]|\s|\s\(|\s'|\s‘)[^('‘]?\d+((,|\.|\/)*\d)*\"*(?=(\D?\s|\s|[[:cntrl:]]|[[:space:]]|\.?\s|\.$|$))|(?<=\s)\d+(nd|th|st)|(?<=\s)\d+\/\d+\"*(?=\s)|(?<=\()\S{1}\d+(?=\))|(?<=\s{1})\S{1}\d+\z|^\d+$|(?<=\A|\A\(|\s|[[:cntrl:]]|[[:space:]]|\s\()[^(]?\d+((,|\.|\/)*\d)*\D{2}(?=($|\s+))|\d+/
|
|
9
9
|
# Rubular: http://rubular.com/r/mxcj2G0Jfa
|
|
10
10
|
EMAIL_REGEX = /(?<=\A|\s|\()[\w+\-.]+@[a-z\d\-]+(\.[a-z]+)*\.[a-z]+(?=\z|\s|\.|\))/i
|
|
11
11
|
|
|
@@ -117,13 +117,17 @@ module ConfidentialInfoRedactorLite
|
|
|
117
117
|
original_sentence_array = txt.split(' ')
|
|
118
118
|
redacted_sentence_array = redacted_text.split(' ')
|
|
119
119
|
diff = original_sentence_array - redacted_sentence_array
|
|
120
|
-
|
|
121
|
-
|
|
122
|
-
|
|
123
|
-
|
|
124
|
-
|
|
125
|
-
|
|
126
|
-
|
|
120
|
+
if original_sentence_array.length.eql?(1)
|
|
121
|
+
final_number_tokens = diff
|
|
122
|
+
else
|
|
123
|
+
final_number_tokens = diff.map { |token| token[-1].eql?('.') ? token[0...-1] : token }
|
|
124
|
+
.map { |token| token[-1].eql?(')') ? token[0...-1] : token }
|
|
125
|
+
.map { |token| token[-1].eql?("'") ? token[0...-1] : token }
|
|
126
|
+
.map { |token| token[-1].eql?('’') ? token[0...-1] : token }
|
|
127
|
+
.map { |token| token[0].eql?('(') ? token[1..token.length] : token }
|
|
128
|
+
.map { |token| token[0].eql?("'") ? token[1..token.length] : token }
|
|
129
|
+
.map { |token| token[0].eql?("‘") ? token[1..token.length] : token }
|
|
130
|
+
end
|
|
127
131
|
end
|
|
128
132
|
[redacted_text.gsub(/(?<=[^\>]|\A)#{Regexp.escape(number_text)}/, "<span class='confidentialNumber'>#{number_text}</span>"), final_number_tokens]
|
|
129
133
|
end
|
|
@@ -157,6 +157,31 @@ RSpec.describe ConfidentialInfoRedactorLite::Redactor do
|
|
|
157
157
|
text = "Page 4"
|
|
158
158
|
expect(described_class.new(language: 'en', dow: en_dow, dow_abbr: en_dow_abbr, months: en_months, months_abbr: en_month_abbr).numbers_html(text)[1]).to eq(["4"])
|
|
159
159
|
end
|
|
160
|
+
|
|
161
|
+
it 'redacts numbers from a text #014' do
|
|
162
|
+
text = "88966-5.0-ENG"
|
|
163
|
+
expect(described_class.new(language: 'en', dow: en_dow, dow_abbr: en_dow_abbr, months: en_months, months_abbr: en_month_abbr).numbers_html(text)[1]).to eq(["88966-5.0-ENG"])
|
|
164
|
+
end
|
|
165
|
+
|
|
166
|
+
it 'redacts numbers from a text #015' do
|
|
167
|
+
text = "85dB(A)"
|
|
168
|
+
expect(described_class.new(language: 'en', dow: en_dow, dow_abbr: en_dow_abbr, months: en_months, months_abbr: en_month_abbr).numbers_html(text)[1]).to eq(["85dB(A)"])
|
|
169
|
+
end
|
|
170
|
+
|
|
171
|
+
it 'redacts numbers from a text #016' do
|
|
172
|
+
text = "Machine Standard operating conditions are between 2.3, and +40°C."
|
|
173
|
+
expect(described_class.new(language: 'en', dow: en_dow, dow_abbr: en_dow_abbr, months: en_months, months_abbr: en_month_abbr).numbers_html(text)[1]).to eq(["2.3,", "+40°C"])
|
|
174
|
+
end
|
|
175
|
+
|
|
176
|
+
it 'redacts numbers from a text #017' do
|
|
177
|
+
text = "(inH2O). CP2.0 RM6 +40°C RM6-Anlage RM6C (see Fig.6)."
|
|
178
|
+
expect(described_class.new(language: 'en', dow: en_dow, dow_abbr: en_dow_abbr, months: en_months, months_abbr: en_month_abbr).numbers_html(text)[1]).to eq(["inH2O", "CP2.0", "RM6", "+40°C", "RM6-Anlage", "RM6C", "Fig.6"])
|
|
179
|
+
end
|
|
180
|
+
|
|
181
|
+
it 'redacts numbers from a text #018' do
|
|
182
|
+
text = "CP3.0"
|
|
183
|
+
expect(described_class.new(language: 'en', dow: en_dow, dow_abbr: en_dow_abbr, months: en_months, months_abbr: en_month_abbr).numbers_html(text)[1]).to eq(["CP3.0"])
|
|
184
|
+
end
|
|
160
185
|
end
|
|
161
186
|
|
|
162
187
|
describe '#numbers_html' do
|
|
@@ -324,7 +349,7 @@ RSpec.describe ConfidentialInfoRedactorLite::Redactor do
|
|
|
324
349
|
Don’t forget to use your imagination and creativity!
|
|
325
350
|
EOF
|
|
326
351
|
tokens = ConfidentialInfoRedactorLite::Extractor.new(corpus: corpus).extract(text)
|
|
327
|
-
expect(described_class.new(language: 'en', tokens: tokens, dow: en_dow, dow_abbr: en_dow_abbr, months: en_months, months_abbr: en_month_abbr).redact(text)).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
|
|
352
|
+
expect(described_class.new(language: 'en', tokens: tokens, dow: en_dow, dow_abbr: en_dow_abbr, months: en_months, months_abbr: en_month_abbr).redact(text)).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 <redacted number> -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!")
|
|
328
353
|
end
|
|
329
354
|
|
|
330
355
|
it 'redacts all confidential information from a text #003' do
|