word_filter 0.0.8 → 0.0.9

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: a81dac029f338386b201365215420886d300afcb
-  data.tar.gz: e108e5ef8d32656eb688d5e3690e1af0fbedbea4
+  metadata.gz: 06e6a40537fb1cfad2ee83e8a598c8826bd40fda
+  data.tar.gz: 14efc8f39dc0a84df54e2195799ea65214134d32
 SHA512:
-  metadata.gz: 3a909da71e2df9ddf2acf718510c7f223712fdbc130b4129f867f4b3840eeb928992b06ed1dc8584837624204f859e72d6f1e70fd1b0b928466274db63d88340
-  data.tar.gz: 1eeeaeba7351963169e980e3bd8aa269191ff6a37bced95ef28aa01f3a5a14b7aba24eab78e6ce115adb9e48c02b4e2b706ed9b8ea1da203ba7bcb1cb8994b94
+  metadata.gz: 921118f7c180ba70f95b4df31297b14351f1777a0dcfca5edaa4d2b7e77be84045880f0edc5b303c3db3059a32e07f098f6eba1d79ee607bbeb5a0fe7224ac62
+  data.tar.gz: 6d79bc70cb2d79ea499faa05533dfeccabbecafde561db16b4045085b0828d54b2ab902164a0bff8678bb519acc5e0b1921ea90b7d9dabea28fec60d4eb28df7

data/lib/word_filter/version.rb

@@ -1,3 +1,3 @@
 module WordFilter
-  VERSION = "0.0.8"
+  VERSION = "0.0.9"
 end

data/lib/word_filter.rb

@@ -1,4 +1,183 @@
 require "word_filter/version"
 
 module WordFilter
+  class Filter
+
+    @@emailRegex = /[a-zA-Z0-9._%+-]+@[a-z0-9.-]+\\.[a-zA-Z]{2,4}/
+    @@alphaNumericDigit = /(zero|one|two|three|four|five|six|seven|eight|nine|ten|eleven|twelve|thirteen|fourteen|fifteen|sixteen|seventeen|eighteen|nineteen|twenty|\d)/; 
+    @@digitsRegex = Regexp.new("\b(\s*" + @@alphaNumericDigit.source + ")+\b")
+    @@streetNameRegex = Regexp.new("\b(\s*" + @@alphaNumericDigit.source + ")+\s([a-z\d]+\.?\s*){1,5}\b(avenue|ave|street|st|court|ct|circle|boulevard|blvd|lane|ln|trail|tr|loop|lp|route|rt|drive|dr|road|rd|terrace|tr|way|wy|highway|hiway|hw)\b")
+    @@phoneNumber = Regexp.new("((" + @@alphaNumericDigit.source + ")\W*?){3}((" + @@alphaNumericDigit.source + ")\W*?){4}\b")
+    @@urlRegex = /(?:http|https):\/\/[a-z0-9]+(?:[\-\.]{1}[a-z0-9]+)*\.[a-z]{2,5}(?:(?::[0-9]{1,5})?\/[^\s]*)?/
+    
+   NONE = 0;
+   REPEATED_VOWELS = 1;
+   SWAPPABLE_VOWELS = 2;
+   SWAPPABLE_AND_REPEATED_VOWELS = 3;
+   SWAPPABLE_AND_REPEATED_VOWELS_INCLUDING_NONE = 4;
+
+
+
+    def initialize()
+        @filterLevel = NONE
+    end
+
+    attr_accessor :filterLevel
+
+    def filterInit(dictionaryFile, badwordslist)
+        @goodWords = loadDictionary(dictionaryFile)
+        #The original java class requiere other three word's list
+        @datingWordsRegex = /dating/
+        @deviantWordsRegex = /deviant/
+        @badWordsRegex = loadBadwords(badwordslist)
+        
+        vowels = /([aeiou])/
+        @vowelSwappedAndRepeatedRegex = Regexp.new(@badWordsRegex.source.gsub(vowels, "[aeiou]+"))
+        @vowelSwappedAndRepeatedRegexIncludingEmpty = Regexp.new(@badWordsRegex.source.gsub(vowels, "[aeiou]*"))
+	    @vowelRepeatedRegex = Regexp.new(@badWordsRegex.source.gsub(vowels, "\\1+"))
+	    @vowelSwappedRegex = Regexp.new(@badWordsRegex.source.gsub(vowels, "[aeiou]"))
+        
+    end
+
+
+    def loadDictionary(path)
+        words = []
+        File.open(path, "r").each_line do |line|
+            splitted = line.split(" ")
+            splitted.each do |w|
+                words << w
+            end
+        end
+        return words
+    end
+
+    def loadBadwords(path)
+      words = File.read(path).gsub("\r", '').split("\n")
+      regex = words.join('|')
+      regex = '(' + regex + ')'
+      regex = Regexp.new(regex)
+      return regex
+    end
+ 
+    def filterString(input)
+
+#        Output:
+#		-1: An exception occured while trying to check the string, do not post
+#		0: string is safe to post
+#		1: string contains an email address
+#		2: string contains a URL
+#		3: string contains a street address
+#		4: string contains a phone number
+#		5: string contains a dating word
+#		6: string contains a deviant word
+#		9: string contains any other bad word
+
+        input = input.strip.downcase
+        workingCopy = input
+        
+        if input == ""
+            return 0
+        end
+        
+        if @@emailRegex.match(input)
+            return 1
+        end
+        
+        if @@urlRegex.match(input)
+            return 2
+        end
+        
+        if @@streetNameRegex.match(input)
+            return 3
+        end
+        
+        if @@phoneNumber.match(input)
+            return 4
+        end
+        
+        workingCopy.gsub("\s+", " ")
+        workingCopy.gsub!(/["',.;:?-]/, " ")
+        workingCopy.gsub!(/!+\s/, " ")
+        workingCopy.gsub!(/!+\z/, " ")
+        workingCopy.gsub!(/\br\su/, " ")
+        
+        cleanVersion = stripGoodWords(workingCopy)
+        
+        if cleanVersion == nil or cleanVersion.length == 0
+            return 0
+        end
+        
+        if @datingWordsRegex.match(cleanVersion)
+            return 5
+        end
+        
+        if @deviantWordsRegex.match(cleanVersion)
+            return 6
+        end
+       
+        if @badWordsRegex.match(cleanVersion)
+            return 7
+        end
+
+        #let's try various combinations of bad word tricks
+        currentVersion = cleanVersion
+        
+        #compress the string then check it again
+        if @badWordsRegex.match(currentVersion.gsub("[ \t\n\f\r]", ""))
+            return 9
+        end
+        
+        #zap special characters and check it again
+        if @badWordsRegex.match(currentVersion.gsub("[^a-z]", ""))
+            return 9
+        end
+        
+        #replace certain special characters with their letter equivalents
+        #NOTE: This one maps vertical non-letter chars (!1|) to i
+        specialCharsReplaced_i = currentVersion.tr("@683!1|0$+","abbeiiiost")
+        if @badWordsRegex.match(specialCharsReplaced_i)
+            return 9
+        end
+        
+        #replace certain special characters with their letter equivalents
+        #NOTE: This one maps vertical non-letter chars (!1|) to l
+        specialCharsReplaced_l = currentVersion.tr("@683!1|0$+","abbelllost")
+        if @badWordsRegex.match(specialCharsReplaced_l)
+            return 9
+        end
+        
+        case @filterLevel
+        when NONE
+            return 0
+        when REPEATED_VOWELS
+            if @vowelRepeatedRegex.match(specialCharsReplaced_i) or @vowelRepeatedRegex.match(specialCharsReplaced_l)
+                return 9
+            end
+        when SWAPPABLE_VOWELS
+            if @vowelSwappedRegex.match(specialCharsReplaced_i) or @vowelSwappedRegex.match(specialCharsReplaced_l)
+                return 9
+            end
+        when SWAPPABLE_AND_REPEATED_VOWELS
+            if @vowelSwappedAndRepeatedRegex.match(specialCharsReplaced_i) or @vowelSwappedAndRepeatedRegex.match(specialCharsReplaced_l)
+                return 9
+            end
+        when SWAPPABLE_AND_REPEATED_VOWELS_INCLUDING_NONE
+            if @vowelSwappedAndRepeatedRegexIncludingEmpty.match(specialCharsReplaced_i) or @vowelSwappedAndRepeatedRegexIncludingEmpty.match(specialCharsReplaced_l)
+                return 9
+            end
+        end
+
+    end
+
+    def stripGoodWords(input)
+        result = []
+        input = input.split(" ")
+        input.each do |w|
+            if not @goodWords.include? w
+                result << w
+            end
+        end
+        return result.join(" ")
+    end
+  end
 end

data/word_filter.gemspec

@@ -6,7 +6,7 @@ require 'word_filter/version'
 Gem::Specification.new do |spec|
   spec.name          = "word_filter"
   spec.version       = WordFilter::VERSION
-  spec.authors       = ["Huascar Oña"]
+  spec.authors       = ["Huascar Ona"]
   spec.email         = ["huascarking@hotmail.com"]
   spec.description   = %q{A bad word filter for the input text.}
   spec.summary       = %q{A word filter gem}

metadata

@@ -1,10 +1,10 @@
 --- !ruby/object:Gem::Specification
 name: word_filter
 version: !ruby/object:Gem::Version
-  version: 0.0.8
+  version: 0.0.9
 platform: ruby
 authors:
-- Huascar Oña
+- Huascar Ona
 autorequire: 
 bindir: bin
 cert_chain: []