data-anonymization 0.3.0 → 0.5.0

data/lib/strategy/blacklist.rb

@@ -7,7 +7,7 @@ module DataAnon
           database_field_name = record.attributes.select { |k,v| k.downcase == field }.keys[0]
           field_value = record.attributes[database_field_name]
           unless field_value.nil? || is_primary_key?(database_field_name)
-            field = DataAnon::Core::Field.new(database_field_name, field_value, index, record)
+            field = DataAnon::Core::Field.new(database_field_name, field_value, index, record, @name)
             record[database_field_name] = strategy.anonymize(field)
           end
         end

data/lib/strategy/field/anonymize_array.rb

@@ -0,0 +1,28 @@
+module DataAnon
+  module Strategy
+    module Field
+
+
+      class AnonymizeArray
+
+        def self.user_defaults user_defaults
+          @@user_defaults = user_defaults
+        end
+
+        def initialize strategy
+          @strategy = strategy
+        end
+
+        def anonymize field
+          field.value.collect do |v|
+            strategy = @strategy || @@user_defaults[v.class.to_s.downcase.to_sym]
+            strategy.anonymize DataAnon::Core::Field.new(field.name, v, field.row_number, field.ar_record, field.table_name)
+          end
+        end
+
+      end
+
+
+    end
+  end
+end

data/lib/strategy/field/contact/random_address.rb

@@ -2,6 +2,18 @@ module DataAnon
   module Strategy
     module Field
 
+      # Generates address using the [geojson](http://www.geojson.org/geojson-spec.html) format file. The default US/UK file chooses randomly from 300 addresses.
+      # The large data set can be downloaded from [here](http://www.infochimps.com/datasets/simplegeo-places-dump)
+      #
+      #    !!!ruby
+      #    anonymize('Address').using FieldStrategy::RandomAddress.region_US
+      #
+      #    !!!ruby
+      #    anonymize('Address').using FieldStrategy::RandomAddress.region_UK
+      #
+      #    !!!ruby
+      #    # get your own geo_json file and use it
+      #    anonymize('Address').using FieldStrategy::RandomAddress.new('my_geo_json.json')
 
       class RandomAddress  < GeojsonBase
 

data/lib/strategy/field/contact/random_city.rb

@@ -2,6 +2,18 @@ module DataAnon
   module Strategy
     module Field
 
+      # Similar to RandomAddress, generates city using the [geojson](http://www.geojson.org/geojson-spec.html) format file. The default US/UK file chooses randomly from 300 addresses.
+      # The large data set can be downloaded from [here](http://www.infochimps.com/datasets/simplegeo-places-dump)
+      #
+      #    !!!ruby
+      #    anonymize('City').using FieldStrategy::RandomCity.region_US
+      #
+      #    !!!ruby
+      #    anonymize('City').using FieldStrategy::RandomCity.region_UK
+      #
+      #    !!!ruby
+      #    # get your own geo_json file and use it
+      #    anonymize('City').using FieldStrategy::RandomCity.new('my_geo_json.json')
 
       class RandomCity < GeojsonBase
 

data/lib/strategy/field/contact/random_phone_number.rb

@@ -2,6 +2,10 @@ module DataAnon
   module Strategy
     module Field
 
+      # Keeping the format same it changes each digit in the string with random digit.
+      #
+      #    !!!ruby
+      #    anonymize('PhoneNumber').using FieldStrategy::RandomPhoneNumber.new
 
       class RandomPhoneNumber < FormattedStringNumber
 

data/lib/strategy/field/contact/random_province.rb

@@ -2,6 +2,18 @@ module DataAnon
   module Strategy
     module Field
 
+      # Similar to RandomAddress, generates province using the [geojson](http://www.geojson.org/geojson-spec.html) format file. The default US/UK file chooses randomly from 300 addresses.
+      # The large data set can be downloaded from [here](http://www.infochimps.com/datasets/simplegeo-places-dump)
+      #
+      #    !!!ruby
+      #    anonymize('Province').using FieldStrategy::RandomProvince.region_US
+      #
+      #    !!!ruby
+      #    anonymize('Province').using FieldStrategy::RandomProvince.region_UK
+      #
+      #    !!!ruby
+      #    # get your own geo_json file and use it
+      #    anonymize('Province').using FieldStrategy::RandomProvince.new('my_geo_json.json')
 
       class RandomProvince < GeojsonBase
 

data/lib/strategy/field/contact/random_zipcode.rb

@@ -2,6 +2,18 @@ module DataAnon
   module Strategy
     module Field
 
+      # Similar to RandomAddress, generates zipcode using the [geojson](http://www.geojson.org/geojson-spec.html) format file. The default US/UK file chooses randomly from 300 addresses.
+      # The large data set can be downloaded from [here](http://www.infochimps.com/datasets/simplegeo-places-dump)
+      #
+      #    !!!ruby
+      #    anonymize('Address').using FieldStrategy::RandomZipcode.region_US
+      #
+      #    !!!ruby
+      #    anonymize('Address').using FieldStrategy::RandomZipcode.region_UK
+      #
+      #    !!!ruby
+      #    # get your own geo_json file and use it
+      #    anonymize('Address').using FieldStrategy::RandomZipcode.new('my_geo_json.json')
 
       class RandomZipcode < GeojsonBase
 

data/lib/strategy/field/datetime/anonymize_date.rb

@@ -2,6 +2,21 @@ module DataAnon
   module Strategy
     module Field
 
+      # Anonmizes day and month fields within natural range based on true/false input for that field. By defaut both fields are
+      # anonymized
+      #
+      #    !!!ruby
+      #    # anonymizes month and leaves day unchanged
+      #    anonymize('DateOfBirth').using FieldStrategy::AnonymizeDate.new(true,false)
+      #
+      # In addition to customizing which fields you want anonymized, there are some helper methods which allow for quick anonymization
+      #
+      #                                                                                                                                                                                               ```ruby
+      #    # anonymizes only the month field
+      #    anonymize('DateOfBirth').using FieldStrategy::AnonymizeDate.only_month
+      #    # anonymizes only the day field
+      #    anonymize('DateOfBirth').using FieldStrategy::AnonymizeDate.only_day
+
       class AnonymizeDate
 
 

data/lib/strategy/field/datetime/anonymize_datetime.rb

@@ -2,6 +2,25 @@ module DataAnon
   module Strategy
     module Field
 
+      # Anonymizes each field(except year and seconds) within the natural range (e.g. hour between 1-24 and day within the month) based on true/false
+      # input for that field. By default, all fields are anonymized.
+      #
+      #    !!!ruby
+      #    # anonymizes month and hour fields, leaving the day and minute fields untouched
+      #    anonymize('DateOfBirth').using FieldStrategy::AnonymizeDateTime.new(true,false,true,false)
+      #
+      # In addition to customizing which fields you want anonymized, there are some helper methods which allow for quick anonymization
+      #
+      #    !!!ruby
+      #    # anonymizes only the month field
+      #    anonymize('DateOfBirth').using FieldStrategy::AnonymizeDateTime.only_month
+      #    # anonymizes only the day field
+      #    anonymize('DateOfBirth').using FieldStrategy::AnonymizeDateTime.only_day
+      #    # anonymizes only the hour field
+      #    anonymize('DateOfBirth').using FieldStrategy::AnonymizeDateTime.only_hour
+      #    # anonymizes only the minute field
+      #    anonymize('DateOfBirth').using FieldStrategy::AnonymizeDateTime.only_minute
+
       class AnonymizeDateTime < AnonymizeTime
 
         private

data/lib/strategy/field/datetime/anonymize_time.rb

@@ -2,6 +2,25 @@ module DataAnon
   module Strategy
     module Field
 
+      # Anonymizes each field(except year and seconds) within the natural range (e.g. hour between 1-24 and day within the month) based on true/false
+      # input for that field. By default, all fields are anonymized.
+      #
+      #    !!!ruby
+      #    # anonymizes month and hour fields, leaving the day and minute fields untouched
+      #    anonymize('DateOfBirth').using FieldStrategy::AnonymizeTime.new(true,false,true,false)
+      #
+      # In addition to customizing which fields you want anonymized, there are some helper methods which allow for quick anonymization
+      #
+      #    !!!ruby
+      #    # anonymizes only the month field
+      #    anonymize('DateOfBirth').using FieldStrategy::AnonymizeTime.only_month
+      #    # anonymizes only the day field
+      #    anonymize('DateOfBirth').using FieldStrategy::AnonymizeTime.only_day
+      #    # anonymizes only the hour field
+      #    anonymize('DateOfBirth').using FieldStrategy::AnonymizeTime.only_hour
+      #    # anonymizes only the minute field
+      #    anonymize('DateOfBirth').using FieldStrategy::AnonymizeTime.only_minute
+
       class AnonymizeTime
 
         DEFAULT_ANONYMIZATION = true

data/lib/strategy/field/datetime/date_delta.rb

@@ -2,6 +2,16 @@ module DataAnon
   module Strategy
     module Field
 
+      # Shifts date randomly within given delta range. Default shits date within 10 days + or -
+      #
+      #    !!!ruby
+      #    anonymize('DateOfBirth').using FieldStrategy::AnonymizeDate.new
+      #
+      #    !!!ruby
+      #    # shifts date within 25 days
+      #    anonymize('DateOfBirth').using FieldStrategy::DateDelta.new(25)
+      #
+
       class DateDelta
 
         DEFAULT_DAY_DELTA = 10

data/lib/strategy/field/datetime/date_time_delta.rb

@@ -2,6 +2,15 @@ module DataAnon
   module Strategy
     module Field
 
+      # Shifts data randomly within given range. Default shifts date within 10 days + or - and shifts time within 30 minutes.
+      #
+      #    !!!ruby
+      #    anonymize('DateOfBirth').using FieldStrategy::DateTimeDelta.new
+      #
+      #    !!!ruby
+      #    # shifts date within 20 days and time within 50 minutes
+      #    anonymize('DateOfBirth').using FieldStrategy::DateTimeDelta.new(20, 50)
+
       class DateTimeDelta
 
         DEFAULT_DAY_DELTA = 10

data/lib/strategy/field/datetime/time_delta.rb

@@ -2,6 +2,14 @@ module DataAnon
   module Strategy
     module Field
 
+      # Shifts data randomly within given range. Default shifts date within 10 days + or - and shifts time within 30 minutes.
+      #
+      #    !!!ruby
+      #    anonymize('DateOfBirth').using FieldStrategy::TimeDelta.new
+      #
+      #    !!!ruby
+      #    # shifts date within 20 days and time within 50 minutes
+      #    anonymize('DateOfBirth').using FieldStrategy::TimeDelta.new(20, 50)
 
       class TimeDelta < DateTimeDelta
       end

data/lib/strategy/field/default_anon.rb

@@ -12,17 +12,20 @@ module DataAnon
                               :datetime => FieldStrategy::DateTimeDelta.new,
                               :time => FieldStrategy::TimeDelta.new,
                               :date => FieldStrategy::DateDelta.new,
+                              :array => FieldStrategy::AnonymizeArray.new(nil),
                               :trueclass => FieldStrategy::RandomBoolean.new,
+                              :"bson::objectid" => FieldStrategy::Whitelist.new,
                               :falseclass => FieldStrategy::RandomBoolean.new
         }
 
         def initialize user_defaults = {}
           @user_defaults = DEFAULT_STRATEGIES.merge user_defaults
+          FieldStrategy::AnonymizeArray.user_defaults @user_defaults
         end
 
         def anonymize field
           strategy = @user_defaults[field.value.class.to_s.downcase.to_sym]
-          raise "No strategy defined for datatype #{field.value.class}. Use 'default_field_strategies' option in your script. Refer to  http://sunitparekh.github.com/data-anonymization/#default-field-strategies for more details. " unless strategy
+          raise "No strategy defined for datatype #{field.value.class}. Use 'default_field_strategies' option in your script. Refer to  http://sunitparekh.github.com/data-anonymization/#default-field-strategies for more details. #{field.inspect}" unless strategy
           strategy.anonymize field
         end
 

data/lib/strategy/field/email/gmail_template.rb

@@ -1,6 +1,14 @@
 module DataAnon
   module Strategy
     module Field
+
+      # Generates a valid unique gmail address by taking advantage of the gmail + strategy. Takes in a valid gmail username and
+      # generates emails of the form username+<number>@gmail.com
+      #
+      #    !!!ruby
+      #    anonymize('Email').using FieldStrategy::GmailTemplate.new('username')
+      #
+
       class GmailTemplate
 
         def initialize username = 'someusername'

data/lib/strategy/field/email/random_email.rb

@@ -2,6 +2,13 @@ module DataAnon
   module Strategy
     module Field
 
+      # Generates email randomly using the given HOSTNAME and TLD.
+      # By defaults generates hostname randomly along with email id.
+      #
+      #    !!!ruby
+      #    anonymize('Email').using FieldStrategy::RandomEmail.new('thoughtworks','com')
+      #
+
       class RandomEmail
 
         TLDS = ['com','org','net','edu','gov','mil','biz','info']

data/lib/strategy/field/email/random_mailinator_email.rb

@@ -2,6 +2,11 @@ module DataAnon
   module Strategy
     module Field
 
+      # Generates random email using mailinator hostname. e.g. <randomstring>@mailinator.com
+      #
+      #    !!!ruby
+      #    anonymize('Email').using FieldStrategy::RandomMailinatorEmail.new
+
       class RandomMailinatorEmail
 
         def initialize

data/lib/strategy/field/fields.rb

@@ -3,12 +3,16 @@ require 'strategy/field/random_boolean'
 
 require 'strategy/field/anonymous'
 
+#array
+require 'strategy/field/anonymize_array'
+
 # string
 require 'strategy/field/string/lorem_ipsum'
 require 'strategy/field/string/string_template'
 require 'strategy/field/string/random_string'
 require 'strategy/field/string/random_url'
 require 'strategy/field/string/formatted_string_numbers'
+require 'strategy/field/string/random_formatted_string'
 
 require 'strategy/field/string/select_from_file'
 require 'strategy/field/string/select_from_list'

data/lib/strategy/field/name/random_first_name.rb

@@ -2,6 +2,16 @@ module DataAnon
   module Strategy
     module Field
 
+      # Randomly picks up first name from the predefined list in the file. Default [file](https://raw.github.com/sunitparekh/data-anonymization/master/resources/first_names.txt) is part of the gem.
+      # File should contain first name on each line.
+      #
+      #    !!!ruby                                                                                                                                                                                                                                        ```ruby
+      #    anonymize('FirstName').using FieldStrategy::RandomFirstName.new
+      #
+      #    !!!ruby
+      #    anonymize('FirstName').using FieldStrategy::RandomFirstName.new('my_first_names.txt')
+      #
+
       class RandomFirstName < SelectFromFile
 
         def initialize file_path = nil

data/lib/strategy/field/name/random_full_name.rb

@@ -2,6 +2,14 @@ module DataAnon
   module Strategy
     module Field
 
+      # Generates full name using the RandomFirstName and RandomLastName strategies.
+      #
+      #    !!!ruby
+      #    anonymize('FullName').using FieldStrategy::RandomFullName.new
+      #
+      #    !!!ruby
+      #    anonymize('FullName').using FieldStrategy::RandomLastName.new('my_first_names.txt', 'my_last_names.txt')
+
       class RandomFullName
 
         def initialize first_names = nil, last_names = nil
@@ -13,10 +21,10 @@ module DataAnon
 
           name_words = field.value.split(' ')
 
-          anonymized_first_name = @first_name_anonymizer.anonymize(name_words[0])
+          anonymized_first_name = @first_name_anonymizer.anonymize(field)
           anonymized_last_name = ""
           for counter in (1..name_words.size-1)
-            anonymized_last_name = anonymized_last_name + " " + @last_name_anonymizer.anonymize(name_words[counter])
+            anonymized_last_name = anonymized_last_name + " " + @last_name_anonymizer.anonymize(field)
           end
 
           return anonymized_first_name + anonymized_last_name

data/lib/strategy/field/name/random_last_name.rb

@@ -2,6 +2,15 @@ module DataAnon
   module Strategy
     module Field
 
+      # Randomly picks up last name from the predefined list in the file. Default [file](https://raw.github.com/sunitparekh/data-anonymization/master/resources/last_names.txt) is part of the gem.
+      # File should contain last name on each line.
+      #
+      #    !!!ruby
+      #    anonymize('LastName').using FieldStrategy::RandomLastName.new
+      #
+      #    !!!ruby
+      #    anonymize('LastName').using FieldStrategy::RandomLastName.new('my_last_names.txt')
+
       class RandomLastName < SelectFromFile
 
         def initialize file_path = nil

data/lib/strategy/field/name/random_user_name.rb

@@ -2,6 +2,11 @@ module DataAnon
   module Strategy
     module Field
 
+      # Generates random user name of same length as original user name.
+      #
+      #    !!!ruby
+      #    anonymize('Username').using FieldStrategy::RandomUserName.new
+      #
       class RandomUserName
 
         DEFAULT_MIN_LENGTH = 5

data/lib/strategy/field/number/random_big_decimal_delta.rb

@@ -3,6 +3,12 @@ require 'bigdecimal'
 module DataAnon
   module Strategy
     module Field
+
+      # Shifts the current value randomly within given delta + and -. Default is 10.0
+      #
+      #    !!!ruby
+      #    anonymize('points').using FieldStrategy::RandomFloatDelta.new(2.5)
+
       class RandomBigDecimalDelta
 
         def initialize delta = 100.0

data/lib/strategy/field/number/random_float.rb

@@ -2,6 +2,10 @@ module DataAnon
   module Strategy
     module Field
 
+      # Generates random float number between given two numbers. Default range is 0.0 to 100.0
+      #
+      #    !!!ruby
+      #    anonymize('points').using FieldStrategy::RandomFloat.new(3.0,5.0)
 
       class RandomFloat