unisec 0.0.1 → 0.0.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 1496d192c32a345de077d1643d041f15f960095f4931165a19b1250a52c5897e
-  data.tar.gz: 553ce1f9fa9d21895d31c144dc15fc73483a7301ececa1046138eec9b51a8707
+  metadata.gz: ce2d8764170d8ac6976db2dd115a0d31c232b1c933de4d9b55b47cc7d8af9f1e
+  data.tar.gz: 29279259ec14a9ff8d1752d5554c601259999d10fcd447795b2bdcbd6be984af
 SHA512:
-  metadata.gz: ae33d34f6bdf6ae0c5a3dd97ffdb3beaf9b2b7a3d3d48697502b272a8db4fc17a6f91bec5596b12c83d7cc609f0b3f7c9a80b7da09d5584537388868364810a5
-  data.tar.gz: c440d1868a5a97a8d6a126c66541d5627ea039086f85d90347dcb52910c32ebc9d1c97f28ce3e0fef62b209f0c982e24414edce98a32406bbc1f802f42bd8ef4
+  metadata.gz: 1e23f9ec4a267c592d4caad37fd452c0ee9498ed4d06578868acc69022ebf55cd5614ef18b0099272de040be7eb85f34d9e1f6f0665112407189835e879ccd3a
+  data.tar.gz: 42a10be34f6f3ee20288d5896971acabdd6082fd9c739e9edb7173c2246f44f609a59d677977d4f50dbaafcbd3603b30174bf0115fe718c5381048754b665607

data/lib/unisec/cli/cli.rb

@@ -4,6 +4,8 @@ require 'unisec/cli/surrogates'
 require 'unisec/cli/hexdump'
 require 'unisec/cli/properties'
 require 'unisec/cli/confusables'
+require 'unisec/cli/versions'
+require 'unisec/cli/size'
 
 module Unisec
   # Module used to create the CLI for the executable
@@ -22,6 +24,8 @@ module Unisec
       register 'properties char', Properties::Char
       register 'confusables list', Confusables::List
       register 'confusables randomize', Confusables::Randomize
+      register 'versions', Versions
+      register 'size', Size
     end
   end
 end

data/lib/unisec/cli/size.rb

@@ -0,0 +1,38 @@
+# frozen_string_literal: true
+
+require 'dry/cli'
+require 'unisec'
+
+module Unisec
+  module CLI
+    module Commands
+      # CLI command `unisec size` for the class {Unisec::Size} from the lib.
+      #
+      # Example:
+      #
+      # ```plaintext
+      # $ unisec size 🧑🏼‍🔬
+      # Code point(s):   4
+      # Grapheme(s):     1
+      # UTF-8 byte(s):   15
+      # UTF-16 byte(s):  14
+      # UTF-32 byte(s):  16
+      # UTF-8 unit(s):   15
+      # UTF-16 unit(s):  7
+      # UTF-32 unit(s):  4
+      # ```
+      class Size < Dry::CLI::Command
+        desc 'All kinf of size information about a Unicode string'
+
+        argument :input, required: true,
+                         desc: 'String input'
+
+        # All kinf of size information about a Unicode string.
+        # @param input [String] Input sting we want to know the size of
+        def call(input: nil, **)
+          puts Unisec::Size.new(input).display
+        end
+      end
+    end
+  end
+end

data/lib/unisec/cli/versions.rb

@@ -0,0 +1,38 @@
+# frozen_string_literal: true
+
+require 'dry/cli'
+require 'unisec'
+
+module Unisec
+  module CLI
+    module Commands
+      # CLI command `unisec versions` for the class {Unisec::Versions} from the lib.
+      #
+      # Example:
+      #
+      # ```plaintext
+      # $ unisec versions
+      # Unicode:
+      # Unicode (Ruby)                    15.0.0
+      # Unicode (twitter_cldr gem)        14.0.0
+      # Unicode (unicode-confusable gem)  15.0.0
+      # ICU (twitter_cldr gem)            70.1
+      # CLDR (twitter_cldr gem)           40
+      # Unicode emoji (Ruby)              15.0
+      #
+      # Gems:
+      # unisec                            0.0.1
+      # twitter_cldr gem                  6.11.5
+      # unicode-confusable gem            1.9.0
+      # ```
+      class Versions < Dry::CLI::Command
+        desc 'Version of anything related to Unicode as used in unisec'
+
+        # Version of anything related to Unicode as used in unisec.
+        def call(**)
+          puts Unisec::Versions.display
+        end
+      end
+    end
+  end
+end

data/lib/unisec/size.rb

@@ -0,0 +1,171 @@
+# frozen_string_literal: true
+
+require 'paint'
+
+module Unisec
+  # All kinf of size information about a Unicode string
+  class Size
+    # Number of code points
+    # @return [Integer] number of code points
+    # @example
+    #   us = Unisec::Size.new('👩‍❤️‍👩')
+    #   us.code_points_size # => 6
+    attr_reader :code_points_size
+
+    # Number of graphemes
+    # @return [Integer] number of graphemes
+    # @example
+    #   us = Unisec::Size.new('👩‍❤️‍👩')
+    #   us.grapheme_size # => 1
+    attr_reader :grapheme_size
+
+    # UTF-8 size in bytes
+    # @return [Integer] UTF-8 size in bytes
+    # @example
+    #   us = Unisec::Size.new('👩‍❤️‍👩')
+    #   us.utf8_bytesize # => 20
+    attr_reader :utf8_bytesize
+
+    # UTF-16 size in bytes
+    # @return [Integer] UTF-16 size in bytes
+    # @example
+    #   us = Unisec::Size.new('👩‍❤️‍👩')
+    #   us.utf16_bytesize # => 16
+    attr_reader :utf16_bytesize
+
+    # UTF-32 size in bytes
+    # @return [Integer] UTF-32 size in bytes
+    # @example
+    #   us = Unisec::Size.new('👩‍❤️‍👩')
+    #   us.utf32_bytesize # => 24
+    attr_reader :utf32_bytesize
+
+    # Number of UTF-8 units
+    # @return [Integer] number of UTF-8 units
+    # @example
+    #   us = Unisec::Size.new('👩‍❤️‍👩')
+    #   us.utf8_unitsize # => 20
+    attr_reader :utf8_unitsize
+
+    # Number of UTF-16 units
+    # @return [Integer] number of UTF-16 units
+    # @example
+    #   us = Unisec::Size.new('👩‍❤️‍👩')
+    #   us.utf16_unitsize # => 8
+    attr_reader :utf16_unitsize
+
+    # Number of UTF-32 units
+    # @return [Integer] number of UTF-32 units
+    # @example
+    #   us = Unisec::Size.new('👩‍❤️‍👩')
+    #   us.utf32_unitsize # => 6
+    attr_reader :utf32_unitsize
+
+    def initialize(str)
+      @code_points_size = Size.code_points_size(str)
+      @grapheme_size = Size.grapheme_size(str)
+      @utf8_bytesize = Size.utf8_bytesize(str)
+      @utf16_bytesize = Size.utf16_bytesize(str)
+      @utf32_bytesize = Size.utf32_bytesize(str)
+      @utf8_unitsize = Size.utf8_unitsize(str)
+      @utf16_unitsize = Size.utf16_unitsize(str)
+      @utf32_unitsize = Size.utf32_unitsize(str)
+    end
+
+    # Number of code points
+    # @param str [String] Input sting we want to know the size of
+    # @return [Integer] number of code points
+    # @example
+    #   Unisec::Size.code_points_size('👩‍❤️‍👩') # => 6
+    def self.code_points_size(str)
+      str.size
+    end
+
+    # Number of graphemes
+    # @param str [String] Input sting we want to know the size of
+    # @return [Integer] number of graphemes
+    # @example
+    #   Unisec::Size.grapheme_size('👩‍❤️‍👩') # => 1
+    def self.grapheme_size(str)
+      str.grapheme_clusters.size
+    end
+
+    # UTF-8 size in bytes
+    # @param str [String] Input sting we want to know the size of
+    # @return [Integer] UTF-8 size in bytes
+    # @example
+    #   Unisec::Size.utf8_bytesize('👩‍❤️‍👩') # => 20
+    def self.utf8_bytesize(str)
+      str.bytesize
+    end
+
+    # UTF-16 size in bytes
+    # @param str [String] Input sting we want to know the size of
+    # @return [Integer] UTF-16 size in bytes
+    # @example
+    #   Unisec::Size.utf16_bytesize('👩‍❤️‍👩') # => 16
+    def self.utf16_bytesize(str)
+      str.encode('UTF-16BE').bytesize
+    end
+
+    # UTF-32 size in bytes
+    # @param str [String] Input sting we want to know the size of
+    # @return [Integer] UTF-32 size in bytes
+    # @example
+    #   Unisec::Size.utf32_bytesize('👩‍❤️‍👩') # => 24
+    def self.utf32_bytesize(str)
+      str.encode('UTF-32BE').bytesize
+    end
+
+    # Number of UTF-8 units
+    # @param str [String] Input sting we want to know the size of
+    # @return [Integer] number of UTF-8 units
+    # @example
+    #   Unisec::Size.utf8_unitsize('👩‍❤️‍👩') # => 20
+    def self.utf8_unitsize(str)
+      utf8_bytesize(str)
+    end
+
+    # Number of UTF-16 units
+    # @param str [String] Input sting we want to know the size of
+    # @return [Integer] number of UTF-16 units
+    # @example
+    #   Unisec::Size.utf16_unitsize('👩‍❤️‍👩') # => 8
+    def self.utf16_unitsize(str)
+      utf16_bytesize(str) / 2
+    end
+
+    # Number of UTF-32 units
+    # @param str [String] Input sting we want to know the size of
+    # @return [Integer] number of UTF-32 units
+    # @example
+    #   Unisec::Size.utf32_unitsize('👩‍❤️‍👩') # => 6
+    def self.utf32_unitsize(str)
+      utf32_bytesize(str) / 4
+    end
+
+    # Display a CLI-friendly output summurizing the size information about a Unicode string.
+    # @example
+    #   Unisec::Size.new('👩‍❤️‍👨').display
+    #   # =>
+    #   # Code point(s):   6
+    #   # Grapheme(s):     1
+    #   # UTF-8 byte(s):   20
+    #   # UTF-16 byte(s):  16
+    #   # UTF-32 byte(s):  24
+    #   # UTF-8 unit(s):   20
+    #   # UTF-16 unit(s):  8
+    #   # UTF-32 unit(s):  6
+    def display
+      display = ->(key, value) { puts Paint[key, :red, :bold].ljust(27) + " #{value}" }
+      display.call('Code point(s):', @code_points_size)
+      display.call('Grapheme(s):', @grapheme_size)
+      display.call('UTF-8 byte(s):', @utf8_bytesize)
+      display.call('UTF-16 byte(s):', @utf16_bytesize)
+      display.call('UTF-32 byte(s):', @utf32_bytesize)
+      display.call('UTF-8 unit(s):', @utf8_unitsize)
+      display.call('UTF-16 unit(s):', @utf16_unitsize)
+      display.call('UTF-32 unit(s):', @utf32_unitsize)
+    end
+  end
+end

data/lib/unisec/version.rb

@@ -2,5 +2,5 @@
 
 module Unisec
   # Version of unisec library and app
-  VERSION = '0.0.1'
+  VERSION = '0.0.2'
 end

data/lib/unisec/versions.rb

@@ -0,0 +1,85 @@
+# frozen_string_literal: true
+
+require 'twitter_cldr'
+require 'unicode/confusable'
+require 'paint'
+
+module Unisec
+  # Version information related to Unicode used in Unisec
+  class Versions
+    # Version and label of anything related to Unicode used in Unisec
+    # @return [Hash] versions of each component
+    # @example
+    #   Unisec::Versions.versions
+    #   # =>
+    #   # {:unisec=>{:version=>"0.0.1", :label=>"unisec"},
+    #   #  … }
+    def self.versions # rubocop:disable Metrics/MethodLength
+      {
+        unisec: {
+          version: Unisec::VERSION,
+          label: 'unisec'
+        },
+        ruby_unicode: {
+          version: RbConfig::CONFIG['UNICODE_VERSION'],
+          label: 'Unicode (Ruby)'
+        },
+        ruby_unicode_emoji: {
+          version: RbConfig::CONFIG['UNICODE_EMOJI_VERSION'],
+          label: 'Unicode emoji (Ruby)'
+        },
+        twittercldr_cldr: {
+          version: TwitterCldr::Versions::CLDR_VERSION,
+          label: 'CLDR (twitter_cldr gem)'
+        },
+        twittercldr_icu: {
+          version: TwitterCldr::Versions::ICU_VERSION,
+          label: 'ICU (twitter_cldr gem)'
+        },
+        twittercldr_unicode: {
+          version: TwitterCldr::Versions::UNICODE_VERSION,
+          label: 'Unicode (twitter_cldr gem)'
+        },
+        twittercldr: {
+          version: TwitterCldr::VERSION,
+          label: 'twitter_cldr gem'
+        },
+        unicodeconfusable: {
+          version: Unicode::Confusable::VERSION,
+          label: 'unicode-confusable gem'
+        },
+        unicodeconfusable_unicode: {
+          version: Unicode::Confusable::UNICODE_VERSION,
+          label: 'Unicode (unicode-confusable gem)'
+        }
+      }
+    end
+
+    # Display a CLI-friendly output of the version of anything related to Unicode used in unisec
+    # @example
+    #   Unisec::Versions.display
+    #   # =>
+    #   # Unicode:
+    #   # Unicode (Ruby)                    15.0.0
+    #   # …
+    #   #
+    #   # Gems:
+    #   # unisec                            0.0.1
+    #   # …
+    def self.display # rubocop:disable Metrics/AbcSize
+      data = versions
+      display = ->(node) { puts Paint[data[node][:label], :red, :bold].ljust(44) + " #{data[node][:version]}" }
+      puts Paint['Unicode:', :underline]
+      display.call(:ruby_unicode)
+      display.call(:twittercldr_unicode)
+      display.call(:unicodeconfusable_unicode)
+      display.call(:twittercldr_icu)
+      display.call(:twittercldr_cldr)
+      display.call(:ruby_unicode_emoji)
+      puts Paint["\nGems:", :underline]
+      display.call(:unisec)
+      display.call(:twittercldr)
+      display.call(:unicodeconfusable)
+    end
+  end
+end

data/lib/unisec.rb

@@ -6,3 +6,5 @@ require 'unisec/surrogates'
 require 'unisec/hexdump'
 require 'unisec/properties'
 require 'unisec/confusables'
+require 'unisec/versions'
+require 'unisec/size'

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: unisec
 version: !ruby/object:Gem::Version
-  version: 0.0.1
+  version: 0.0.2
 platform: ruby
 authors:
 - Alexandre ZANNI
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2023-07-21 00:00:00.000000000 Z
+date: 2023-08-18 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: ctf-party
@@ -100,13 +100,17 @@ files:
 - lib/unisec/cli/confusables.rb
 - lib/unisec/cli/hexdump.rb
 - lib/unisec/cli/properties.rb
+- lib/unisec/cli/size.rb
 - lib/unisec/cli/surrogates.rb
+- lib/unisec/cli/versions.rb
 - lib/unisec/confusables.rb
 - lib/unisec/hexdump.rb
 - lib/unisec/properties.rb
+- lib/unisec/size.rb
 - lib/unisec/surrogates.rb
 - lib/unisec/utils.rb
 - lib/unisec/version.rb
+- lib/unisec/versions.rb
 homepage: https://github.com/Acceis/unisec
 licenses:
 - MIT