l2e_vocab_blacklist 1.3.1 → 1.4.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 32359d6f8e221d893a5b00b182452508066a8af8
-  data.tar.gz: 37cc97866bb88cf197cc619d63bd6a1a83b5cb88
+  metadata.gz: c0a855f57a636fb8e7d01bc4036ad56b0e35976d
+  data.tar.gz: ab46f63f8dc08c9a81816e1d0c7bb3f2168b8022
 SHA512:
-  metadata.gz: 59abcfa730b9ba973dc356740b38d37f8d82e1f66ebd72902fe85994d756b0580fd48128a343af17e525c80ce2b6c3eba2f3c0844a103fac1c651b14d0d1ef83
-  data.tar.gz: fe35639b3131ff824fda3639979ca129c195775f1710fc9049936cfb5dac58976a494e78e277710bd43e593b65a0427c274a97470cae693316eb9aa9f3f9ccd2
+  metadata.gz: e02de0ce46669ea163efb08cb9c787ddd0c11c9380f902c018344bc4f64815b2bfe2da4cc732831e213ddf00de22de217ae6bf8313976bd134239150c04a46c1
+  data.tar.gz: 131180e147384652932bb0aadf3e3e82e22a78a8e9c1d3d532936978e584d7bf474238c2a7050f38c24c1a8efc1d813f0990458242089a81c7983fb6c9d1be17

data/lib/l2e_vocab_blacklist.rb

@@ -4,12 +4,25 @@ require 'csv'
 class VocabBlacklist
 
 	# Returns true or false, check to see if the 
-	def self.blacklisted?(str,age ="0")
+	def self.blacklisted?(str, age = "0")
 		# Sanitize string
-		str = str.downcase.strip.gsub(CONSIDER_REGEX, '')
+		str = str.downcase.strip
+
+		whitelisted_phrases = self.whitelist_matches(str)
+
 		# Blacklist if any of the words 
-		str.split(" ").each do |word|
-			return true	 if check_full_words_csv(word,age)
+		str.split(/[ -]/).each do |word|
+			word = word.gsub(CONSIDER_REGEX, "")
+
+			if check_full_words_csv(word, age)
+
+				is_whitelisted = whitelisted_phrases.any? do |phrase|
+					phrase.include?(word)
+				end
+				if !is_whitelisted
+					return true
+				end
+			end
 		end
 		# For compound dirty words
 		PHRASES.each do |bad_phrase|
@@ -19,30 +32,46 @@ class VocabBlacklist
 		return GREEDY_WORDS.any? { |s| str.include?(s) }
 	end
 
+	def self.whitelist_matches(text)
+		text = text.downcase.strip.gsub(CONSIDER_REGEX, "")
+		WHITELIST.select do |whitelist_phrase|
+			text.include?(whitelist_phrase)
+		end
+
+	end
 
 	def self.censor(str, age = "0", replace_with = "****")
 		PHRASES.each do |bad_phrase|
-			str.gsub!(/#{bad_phrase}/i, replace_with)
+			str.gsub!(/#{ bad_phrase }/i, replace_with)
 		end
 
+		whitelisted_phrases = self.whitelist_matches(str)
+
 		str.split(/ /).map do |working_word|
+			working_word.split(/-/).map do |sub_working_word|
+				word = sub_working_word.downcase.gsub(CONSIDER_REGEX, "")
 
-			word = working_word.downcase.gsub(CONSIDER_REGEX, '')
+				is_whitelisted = whitelisted_phrases.any? do |phrase|
+					phrase.include?(word)
+				end
 
-			if check_full_words_csv(word,age)
-				working_word.gsub!(/#{word}/i, replace_with)
-			end
+				if !is_whitelisted
+					if check_full_words_csv(word, age)
+						sub_working_word.gsub!(/#{ word }/i, replace_with)
+					end
 
-			if GREEDY_WORDS.any? { |w| word.include?(w) }
-				working_word = replace_with
-			end
-			
-			working_word
+					if GREEDY_WORDS.any? { |w| word.include?(w) }
+						sub_working_word = replace_with
+					end
+				end
+
+				sub_working_word
+			end.join("-")
 		end.join(" ")
 	end
 
 	def self.file_to_normalized_words(file)
-		CSV.parse(File.read(file)).map(&:first).reject { |s| s.to_s.strip.empty? }.map(&:downcase).map { |s| s.gsub(CONSIDER_REGEX, '') }
+		CSV.parse(File.read(file)).map(&:first).reject { |s| s.to_s.strip.empty? }.map(&:downcase).map { |s| s.gsub(CONSIDER_REGEX, "") }
 	end
 
 	def self.words_with_expansions(words)
@@ -55,15 +84,17 @@ class VocabBlacklist
 	PHRASES = file_to_normalized_words("#{BLACKLIST_DIR}/full_words.csv").select { |w| w.split(" ").length > 1 }
 	FULL_WORDS = file_to_normalized_words("#{BLACKLIST_DIR}/full_words.csv").reject { |w| w.split(" ").length > 1 }
 	GREEDY_WORDS = words_with_expansions(file_to_normalized_words("#{BLACKLIST_DIR}/greedy_words.txt")).uniq.freeze
+	WHITELIST = file_to_normalized_words("#{BLACKLIST_DIR}/whitelist.csv").uniq.freeze
 
 	FULL_WORDS_CSV = CSV.parse(File.read("#{BLACKLIST_DIR}/full_words.csv"))
 
 	private
 
-	def self.check_full_words_csv(word,age)
-		FULL_WORDS_CSV.each do |row|
-			return true	 if row[1].to_i >= age.to_i && row[0].downcase == word.downcase
+		def self.check_full_words_csv(word, age)
+			FULL_WORDS_CSV.each do |row|
+				return true	 if row[1].to_i >= age.to_i && row[0].downcase == word.downcase
+			end
+			return false
 		end
-		return false
-	end
+
 end

data/lib/l2e_vocab_blacklist/blacklists/full_words.csv

@@ -7,6 +7,8 @@ Apeshit,15
 Ass,15
 Asshole,10
 B*tch,15
+Badass,15
+Bad-ass,15
 Barbiturates,5
 Bastard,10
 Beaner,15
@@ -52,6 +54,8 @@ Crip,10
 Cum,15
 Cumming,15
 Cumshot,15
+Cumslut,15
+Cum-slut,15
 Cunnilingus,15
 Cunt,15
 Damn,5
@@ -69,7 +73,6 @@ Doggy Style,15
 Doobie,15
 Douche,15
 Drag Queen,5
-Drugs,5
 Dyke,15
 Ejaculate,10
 Ejaculation,10
@@ -121,6 +124,7 @@ Hickey,5
 Hickie,5
 Homosexual,5
 Hooker,10
+Horny,15
 Huff,10
 Huffing,10
 Hungover,10
@@ -132,6 +136,8 @@ Jerk Off,15
 Jerkoff,15
 Jigga,15
 Joint,10
+Kickass,15
+Kick-ass,15
 Kunt,15
 Kyke,15
 Labia,10
@@ -144,7 +150,6 @@ Lubrication,5
 Lust,5
 Lustful,5
 Marijuana,10
-Mary Jane,15
 Masturbate,10
 Masturbating,10
 Masturbation,10
@@ -159,14 +164,13 @@ Moneyshot,15
 Moron,5
 Motherf*cker,15
 Narcotic,5
-Negro,5
-Nicotine,5
 Nigga,15
 Nignog,15
 Nips,15
 Niptip,15
 Nookie,15
 Nooky,15
+Nude,10
 Opiates,5
 Opium,5
 Oral sex,15
@@ -178,7 +182,6 @@ Oxy,15
 Oxycodone,10
 P*ss,15
 Panties,10
-Pee,5
 Penile,5
 Penis,10
 Playboy,10
@@ -235,7 +238,6 @@ Sucker,10
 Suicide,5
 Syphilis,10
 T*t,15
-Tequila,5
 Testes,10
 Testicle,10
 Testicular,10
@@ -252,6 +254,7 @@ Transgendered,5
 Transvestite,10
 Turd,15
 Twerk,15
+Uncensored,10
 Urethra,10
 Urine,5
 Uterus,5

data/lib/l2e_vocab_blacklist/blacklists/whitelist.csv

@@ -0,0 +1 @@
+moby dick,5

metadata

@@ -1,16 +1,16 @@
 --- !ruby/object:Gem::Specification
 name: l2e_vocab_blacklist
 version: !ruby/object:Gem::Version
-  version: 1.3.1
+  version: 1.4.0
 platform: ruby
 authors:
 - Greg Sherrid
-- Learn2Earn
-- Whooo's Reading
+- Gilles Ferone
+- Whooo's Reading by Learn2Earn
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2017-08-13 00:00:00.000000000 Z
+date: 2018-04-02 00:00:00.000000000 Z
 dependencies: []
 description: Detect words and partial-words that shouldn't be used on Learn2Earn Whooo's
   Reading website
@@ -19,10 +19,11 @@ executables: []
 extensions: []
 extra_rdoc_files: []
 files:
+- README.md
 - lib/l2e_vocab_blacklist.rb
-- lib/l2e_vocab_blacklist/blacklists/greedy_words.txt
 - lib/l2e_vocab_blacklist/blacklists/full_words.csv
-- README.md
+- lib/l2e_vocab_blacklist/blacklists/greedy_words.txt
+- lib/l2e_vocab_blacklist/blacklists/whitelist.csv
 homepage: https://github.com/whooosreading/l2e_vocab_blacklist
 licenses:
 - All rights reserved, for now
@@ -33,17 +34,17 @@ require_paths:
 - lib
 required_ruby_version: !ruby/object:Gem::Requirement
   requirements:
-  - - '>='
+  - - ">="
     - !ruby/object:Gem::Version
       version: '0'
 required_rubygems_version: !ruby/object:Gem::Requirement
   requirements:
-  - - '>='
+  - - ">="
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 2.0.3
+rubygems_version: 2.6.11
 signing_key: 
 specification_version: 4
 summary: L2E Vocab Blacklist