ultra-clean-kit 0.0.1

data/babosa-2.1.0/lib/babosa/transliterator/macedonian.rb

@@ -0,0 +1,32 @@
+# frozen_string_literal: true
+
+require_relative "cyrillic"
+
+module Babosa
+  module Transliterator
+    class Macedonian < Cyrillic
+      APPROXIMATIONS = {
+        "Ѓ" => "Gj",
+        "Љ" => "Lj",
+        "Њ" => "Nj",
+        "Ќ" => "Kj",
+        "Џ" => "Dzh",
+        "Ж" => "Zh",
+        "Ц" => "C",
+        "Ѕ" => "Z",
+        "Ј" => "J",
+        "Х" => "H",
+        "ѓ" => "gj",
+        "љ" => "lj",
+        "њ" => "nj",
+        "ќ" => "kj",
+        "џ" => "dzh",
+        "ж" => "zh",
+        "ц" => "c",
+        "ѕ" => "z",
+        "ј" => "j",
+        "х" => "h"
+      }.freeze
+    end
+  end
+end

data/babosa-2.1.0/lib/babosa/transliterator/norwegian.rb

@@ -0,0 +1,16 @@
+# frozen_string_literal: true
+
+require_relative "latin"
+
+module Babosa
+  module Transliterator
+    class Norwegian < Latin
+      APPROXIMATIONS = {
+        "ø" => "oe",
+        "å" => "aa",
+        "Ø" => "Oe",
+        "Å" => "Aa"
+      }.freeze
+    end
+  end
+end

data/babosa-2.1.0/lib/babosa/transliterator/romanian.rb

@@ -0,0 +1,16 @@
+# frozen_string_literal: true
+
+require_relative "latin"
+
+module Babosa
+  module Transliterator
+    class Romanian < Latin
+      APPROXIMATIONS = {
+        "ș" => "s",
+        "ț" => "t",
+        "Ș" => "S",
+        "Ț" => "T"
+      }.freeze
+    end
+  end
+end

data/babosa-2.1.0/lib/babosa/transliterator/russian.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+require_relative "cyrillic"
+
+module Babosa
+  module Transliterator
+    class Russian < Cyrillic
+      APPROXIMATIONS = {
+        "Й" => "I",
+        "М" => "M",
+        "Х" => "H",
+        "Ц" => "Ts",
+        "Ш" => "Sh",
+        "Щ" => "Sch",
+        "Ю" => "U",
+        "Я" => "Ya",
+        "й" => "i",
+        "х" => "h",
+        "ц" => "ts",
+        "щ" => "sch",
+        "ю" => "u"
+      }.freeze
+    end
+  end
+end

data/babosa-2.1.0/lib/babosa/transliterator/serbian.rb

@@ -0,0 +1,38 @@
+# frozen_string_literal: true
+
+require_relative "latin"
+
+module Babosa
+  module Transliterator
+    class Serbian < Latin
+      APPROXIMATIONS = Cyrillic.const_get(:APPROXIMATIONS).merge(
+        {
+          "Ð" => "Dj",
+          "Č" => "Ch",
+          "Š" => "Sh",
+          "č" => "ch",
+          "đ" => "dj",
+          "š" => "sh",
+          "Ћ" => "C",
+          "Ц" => "C",
+          "Ч" => "Ch",
+          "Ђ" => "Dj",
+          "Џ" => "Dz",
+          "Х" => "H",
+          "Ј" => "J",
+          "Љ" => "Lj",
+          "Њ" => "Nj",
+          "ц" => "c",
+          "ћ" => "c",
+          "ч" => "ch",
+          "ђ" => "dj",
+          "џ" => "dz",
+          "х" => "h",
+          "ј" => "j",
+          "љ" => "lj",
+          "њ" => "nj"
+        }
+      )
+    end
+  end
+end

data/babosa-2.1.0/lib/babosa/transliterator/spanish.rb

@@ -0,0 +1,11 @@
+# frozen_string_literal: true
+
+require_relative "latin"
+
+module Babosa
+  module Transliterator
+    class Spanish < Latin
+      APPROXIMATIONS = {"ñ" => "ni", "Ñ" => "Ni"}.freeze
+    end
+  end
+end

data/babosa-2.1.0/lib/babosa/transliterator/swedish.rb

@@ -0,0 +1,18 @@
+# frozen_string_literal: true
+
+require_relative "latin"
+
+module Babosa
+  module Transliterator
+    class Swedish < Latin
+      APPROXIMATIONS = {
+        "å" => "aa",
+        "ä" => "ae",
+        "ö" => "oe",
+        "Å" => "Aa",
+        "Ä" => "Ae",
+        "Ö" => "Oe"
+      }.freeze
+    end
+  end
+end

data/babosa-2.1.0/lib/babosa/transliterator/turkish.rb

@@ -0,0 +1,10 @@
+# frozen_string_literal: true
+
+require_relative "latin"
+
+module Babosa
+  module Transliterator
+    class Turkish < Latin
+    end
+  end
+end

data/babosa-2.1.0/lib/babosa/transliterator/ukrainian.rb

@@ -0,0 +1,33 @@
+# frozen_string_literal: true
+
+require_relative "cyrillic"
+
+module Babosa
+  module Transliterator
+    class Ukrainian < Cyrillic
+      APPROXIMATIONS = {
+        "Г" => "H",
+        "г" => "h",
+        "Ґ" => "G",
+        "ґ" => "g",
+        "є" => "ie",
+        "И" => "Y",
+        "и" => "y",
+        "І" => "I",
+        "і" => "i",
+        "ї" => "i",
+        "Й" => "Y",
+        "й" => "i",
+        "Х" => "Kh",
+        "х" => "kh",
+        "Ц" => "Ts",
+        "ц" => "ts",
+        "Щ" => "Shch",
+        "щ" => "shch",
+        "ю" => "iu",
+        "я" => "ia",
+        "'" => ""
+      }.freeze
+    end
+  end
+end

data/babosa-2.1.0/lib/babosa/transliterator/vietnamese.rb

@@ -0,0 +1,146 @@
+# frozen_string_literal: true
+
+require_relative "latin"
+
+module Babosa
+  module Transliterator
+    class Vietnamese < Latin
+      APPROXIMATIONS = {
+        "à" => "a",
+        "á" => "a",
+        "ạ" => "a",
+        "ả" => "a",
+        "ã" => "a",
+        "â" => "a",
+        "ầ" => "a",
+        "ấ" => "a",
+        "ậ" => "a",
+        "ẩ" => "a",
+        "ẫ" => "a",
+        "ă" => "a",
+        "ằ" => "a",
+        "ắ" => "a",
+        "ặ" => "a",
+        "ẳ" => "a",
+        "ẵ" => "a",
+        "À" => "A",
+        "Á" => "A",
+        "Ạ" => "A",
+        "Ả" => "A",
+        "Ã" => "A",
+        "Â" => "A",
+        "Ầ" => "A",
+        "Ấ" => "A",
+        "Ậ" => "A",
+        "Ẩ" => "A",
+        "Ẫ" => "A",
+        "Ă" => "A",
+        "Ằ" => "A",
+        "Ắ" => "A",
+        "Ặ" => "A",
+        "Ẳ" => "A",
+        "Ẵ" => "A",
+        "ì" => "i",
+        "í" => "i",
+        "ị" => "i",
+        "ỉ" => "i",
+        "ĩ" => "i",
+        "Ì" => "I",
+        "Í" => "I",
+        "Ị" => "I",
+        "Ỉ" => "I",
+        "Ĩ" => "I",
+        "ù" => "u",
+        "ú" => "u",
+        "ụ" => "u",
+        "ủ" => "u",
+        "ũ" => "u",
+        "ư" => "u",
+        "ừ" => "u",
+        "ứ" => "u",
+        "ự" => "u",
+        "ử" => "u",
+        "ữ" => "u",
+        "Ù" => "U",
+        "Ú" => "U",
+        "Ụ" => "U",
+        "Ủ" => "U",
+        "Ũ" => "U",
+        "Ư" => "U",
+        "Ừ" => "U",
+        "Ứ" => "U",
+        "Ự" => "U",
+        "Ử" => "U",
+        "Ữ" => "U",
+        "è" => "e",
+        "é" => "e",
+        "ẹ" => "e",
+        "ẻ" => "e",
+        "ẽ" => "e",
+        "ê" => "e",
+        "ề" => "e",
+        "ế" => "e",
+        "ệ" => "e",
+        "ể" => "e",
+        "ễ" => "e",
+        "È" => "E",
+        "É" => "E",
+        "Ẹ" => "E",
+        "Ẻ" => "E",
+        "Ẽ" => "E",
+        "Ê" => "E",
+        "Ề" => "E",
+        "Ế" => "E",
+        "Ệ" => "E",
+        "Ể" => "E",
+        "Ễ" => "E",
+        "ò" => "o",
+        "ó" => "o",
+        "ọ" => "o",
+        "ỏ" => "o",
+        "õ" => "o",
+        "ô" => "o",
+        "ồ" => "o",
+        "ố" => "o",
+        "ộ" => "o",
+        "ổ" => "o",
+        "ỗ" => "o",
+        "ơ" => "o",
+        "ờ" => "o",
+        "ớ" => "o",
+        "ợ" => "o",
+        "ở" => "o",
+        "ỡ" => "o",
+        "Ò" => "O",
+        "Ó" => "O",
+        "Ọ" => "O",
+        "Ỏ" => "O",
+        "Õ" => "O",
+        "Ô" => "O",
+        "Ồ" => "O",
+        "Ố" => "O",
+        "Ộ" => "O",
+        "Ổ" => "O",
+        "Ỗ" => "O",
+        "Ơ" => "O",
+        "Ờ" => "O",
+        "Ớ" => "O",
+        "Ợ" => "O",
+        "Ở" => "O",
+        "Ỡ" => "O",
+        "ỳ" => "y",
+        "ý" => "y",
+        "ỵ" => "y",
+        "ỷ" => "y",
+        "ỹ" => "y",
+        "Ỳ" => "Y",
+        "Ý" => "Y",
+        "Ỵ" => "Y",
+        "Ỷ" => "Y",
+        "Ỹ" => "Y",
+        "đ" => "d",
+        "Đ" => "D"
+      }.freeze
+    end
+  end
+end

data/babosa-2.1.0/lib/babosa/version.rb

@@ -0,0 +1,7 @@
+# frozen_string_literal: true
+
+module Babosa
+  module Version
+    STRING = "2.1.0"
+  end
+end

data/babosa-2.1.0/lib/babosa.rb

@@ -0,0 +1,14 @@
+# frozen_string_literal: true
+
+module Babosa
+end
+
+class String
+  def to_identifier
+    Babosa::Identifier.new self
+  end
+  alias_method :to_slug, :to_identifier
+end
+
+require "babosa/transliterator/base"
+require "babosa/identifier"

data/babosa-2.1.0/spec/identifier_spec.rb

@@ -0,0 +1,161 @@
+# frozen_string_literal: true
+
+require "spec_helper"
+
+describe Babosa::Identifier do
+  it "should respond_to :empty?" do
+    expect("".to_slug).to respond_to(:empty?)
+  end
+
+  %w[approximate_ascii clean downcase word_chars normalize to_ascii upcase with_dashes].each do |method|
+    describe "##{method}" do
+      it "should work with invalid UTF-8 strings" do
+        expect { "\x93abc".to_slug.send method }.not_to raise_exception
+      end
+    end
+  end
+
+  describe "#word_chars" do
+    it "word_chars! should leave only letters, numbers, and spaces" do
+      string = "a*$%^$@!@b$%^&*()*!c 123"
+      expect(string.to_slug.word_chars!).to eq "abc 123"
+    end
+  end
+
+  describe "#transliterate" do
+    it "should transliterate to ascii" do
+      (0xC0..0x17E).to_a.each do |codepoint|
+        ss = [codepoint].pack("U*").to_slug
+        expect(ss.approximate_ascii!).to match(/[\x0-\x7f]/)
+      end
+    end
+
+    it "should transliterate uncomposed utf8" do
+      string = [117, 776].pack("U*") # "ü" as ASCII "u" plus COMBINING DIAERESIS
+      expect(string.to_slug.approximate_ascii).to eql("u")
+    end
+
+    it "should transliterate using multiple transliterators" do
+      string = "свободное režģis"
+      expect(string.to_slug.approximate_ascii(:latin, :russian)).to eql("svobodnoe rezgis")
+    end
+  end
+
+  describe "#downcase" do
+    it "should lowercase strings" do
+      expect("FELIZ AÑO".to_slug.downcase).to eql("feliz año")
+    end
+  end
+
+  describe "#upcase" do
+    it "should uppercase strings" do
+      expect("feliz año".to_slug.upcase).to eql("FELIZ AÑO")
+    end
+  end
+
+  describe "#normalize" do
+    it "should allow passing locale as key for :transliterate" do
+      expect("ö".to_slug.clean.normalize(transliterate: :german)).to eql("oe")
+    end
+
+    it "should replace whitespace with dashes" do
+      expect("a b".to_slug.clean.normalize).to eql("a-b")
+    end
+
+    it "should replace multiple spaces with 1 dash" do
+      expect("a    b".to_slug.clean.normalize).to eql("a-b")
+    end
+
+    it "should replace multiple dashes with 1 dash" do
+      expect("male - female".to_slug.normalize).to eql("male-female")
+    end
+
+    it "should strip trailing space" do
+      expect("ab ".to_slug.normalize).to eql("ab")
+    end
+
+    it "should strip leading space" do
+      expect(" ab".to_slug.normalize).to eql("ab")
+    end
+
+    it "should strip trailing slashes" do
+      expect("ab-".to_slug.normalize).to eql("ab")
+    end
+
+    it "should strip leading slashes" do
+      expect("-ab".to_slug.normalize).to eql("ab")
+    end
+
+    it "should not modify valid name strings" do
+      expect("a-b-c-d".to_slug.normalize).to eql("a-b-c-d")
+    end
+
+    it "should not convert underscores" do
+      expect("hello_world".to_slug.normalize).to eql("hello_world")
+    end
+
+    it "should work with non roman chars" do
+      expect("検 索".to_slug.normalize).to eql("検-索")
+    end
+
+    it "should work with emoji chars" do
+      expect("Emojis on slug 😋".to_slug.normalize).to eql("emojis-on-slug-😋")
+    end
+    
+    context "with to_ascii option" do
+      it "should approximate and strip non ascii" do
+        ss = "カタカナ: katakana is über cool".to_slug
+        expect(ss.normalize(to_ascii: true)).to eql("katakana-is-uber-cool")
+      end
+    end
+  end
+
+  describe "#truncate_bytes" do
+    it "should by byte length" do
+      expect("üa".to_slug.truncate_bytes(2)).to eql("ü")
+      expect("üa".to_slug.truncate_bytes(1)).to eql("")
+      expect("üa".to_slug.truncate_bytes(100)).to eql("üa")
+      expect("üéøá".to_slug.truncate_bytes(3)).to eql("ü")
+    end
+  end
+
+  describe "#truncate" do
+    it "should truncate by char length" do
+      expect("üa".to_slug.truncate(2)).to eql("üa")
+      expect("üa".to_slug.truncate(1)).to eql("ü")
+      expect("üa".to_slug.truncate(100)).to eql("üa")
+    end
+  end
+
+  describe "#with_dashes" do
+    it "should not change byte size when replacing spaces" do
+      expect("".to_slug.with_dashes.bytesize).to eql(0)
+      expect(" ".to_slug.with_dashes.bytesize).to eql(1)
+      expect("-abc-".to_slug.with_dashes.bytesize).to eql(5)
+      expect(" abc ".to_slug.with_dashes.bytesize).to eql(5)
+      expect(" a  bc ".to_slug.with_dashes.bytesize).to eql(7)
+    end
+  end
+
+  describe "#to_ruby_method" do
+    it "should get a string suitable for use as a ruby method" do
+      expect("¿¿¿hello... world???".to_slug.to_ruby_method).to eql("hello_world?")
+      expect("カタカナ: katakana is über cool".to_slug.to_ruby_method).to eql("katakana_is_uber_cool")
+      expect("カタカナ: katakana is über cool!".to_slug.to_ruby_method).to eql("katakana_is_uber_cool!")
+      expect("カタカナ: katakana is über cool".to_slug.to_ruby_method(allow_bangs: false)).to eql("katakana_is_uber_cool")
+      expect("not 2 much 4 ruby".to_slug.to_ruby_method).to eql("not_2_much_4_ruby")
+    end
+
+    it "should optionally remove trailing punctuation" do
+      expect("¿¿¿hello... world???".to_slug.to_ruby_method(allow_bangs: false)).to eql("hello_world")
+    end
+
+    it "should raise an error when it would generate an impossible method name" do
+      expect { "1".to_identifier.to_ruby_method }.to raise_error(Babosa::Identifier::Error)
+    end
+
+    it "should raise Babosa::Error error when the string is nil" do
+      expect { "".to_slug.to_ruby_method }.to raise_error(Babosa::Identifier::Error)
+    end
+  end
+end

data/babosa-2.1.0/spec/spec_helper.rb

@@ -0,0 +1,45 @@
+# frozen_string_literal: true
+
+if ENV["COV"]
+  require "simplecov"
+  SimpleCov.start
+end
+
+require "bundler/setup"
+require "babosa"
+
+shared_examples_for "a latin transliterator" do
+  let(:t) { described_class.instance }
+
+  it "should transliterate latin characters" do
+    string = (0xC0..0x17E).to_a.pack("U*")
+    expect(t.transliterate(string)).to match(/[\x0-\x7f]/)
+  end
+end
+
+shared_examples_for "a cyrillic transliterator" do
+  let(:t) { described_class.instance }
+
+  it "should transliterate cyrillic characters" do
+    string = "Славься, Отечество наше свободное"
+    expect(t.transliterate(string)).to match(/[\x0-\x7f]/)
+  end
+end
+
+shared_examples_for "a greek transliterator" do
+  let(:t) { described_class.instance }
+
+  it "should transliterate greek characters" do
+    string = "Γερμανία"
+    expect(t.transliterate(string)).to match(/[\x0-\x7f]/)
+  end
+end
+
+shared_examples_for "a hindi transliterator" do
+  let(:t) { described_class.instance }
+
+  it "should transliterate hindi characters" do
+    string = "आदित्य तापड़िया"
+    expect(t.transliterate(string)).to match(/[\x0-\x7f]/)
+  end
+end

data/babosa-2.1.0/spec/transliterators/base_spec.rb

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+
+require "spec_helper"
+
+describe Babosa::Transliterator::Base do
+  let(:t) { Babosa::Transliterator::Base.instance }
+
+  it "should transliterate 'smart' quotes" do
+    expect(t.transliterate("’")).to eql("'")
+  end
+
+  it "should transliterate non-breaking spaces" do
+    expect(t.transliterate("\xc2\xa0")).to eql(" ")
+  end
+end

data/babosa-2.1.0/spec/transliterators/bulgarian_spec.rb

@@ -0,0 +1,19 @@
+# frozen_string_literal: true
+
+require "spec_helper"
+
+describe Babosa::Transliterator::Bulgarian do
+  let(:t) { described_class.instance }
+  it_behaves_like "a cyrillic transliterator"
+
+  it "should transliterate Cyrillic characters" do
+    examples = {
+      "Ютия" => "Iutiia",
+      "Чушка" => "Chushka",
+      "кьорав" => "kiorav",
+      "Щъркел" => "Shturkel",
+      "полицай" => "policai"
+    }
+    examples.each { |k, v| expect(t.transliterate(k)).to eql(v) }
+  end
+end

data/babosa-2.1.0/spec/transliterators/danish_spec.rb

@@ -0,0 +1,16 @@
+# frozen_string_literal: true
+
+require "spec_helper"
+
+describe Babosa::Transliterator::Danish do
+  let(:t) { described_class.instance }
+  it_behaves_like "a latin transliterator"
+
+  it "should transliterate various characters" do
+    examples = {
+      "Ærøskøbing" => "Aeroeskoebing",
+      "Årslev" => "Aarslev"
+    }
+    examples.each { |k, v| expect(t.transliterate(k)).to eql(v) }
+  end
+end

data/babosa-2.1.0/spec/transliterators/german_spec.rb

@@ -0,0 +1,16 @@
+# frozen_string_literal: true
+
+require "spec_helper"
+
+describe Babosa::Transliterator::German do
+  let(:t) { described_class.instance }
+  it_behaves_like "a latin transliterator"
+
+  it "should transliterate Eszett" do
+    expect(t.transliterate("ß")).to eql("ss")
+  end
+
+  it "should transliterate vowels with umlauts" do
+    expect(t.transliterate("üöä")).to eql("ueoeae")
+  end
+end

data/babosa-2.1.0/spec/transliterators/greek_spec.rb

@@ -0,0 +1,17 @@
+# frozen_string_literal: true
+
+require "spec_helper"
+
+describe Babosa::Transliterator::Greek do
+  let(:t) { described_class.instance }
+  it_behaves_like "a greek transliterator"
+
+  it "should transliterate various characters" do
+    examples = {
+      "Γερμανία" => "Germania",
+      "Αυστρία" => "Aystria",
+      "Ιταλία" => "Italia"
+    }
+    examples.each { |k, v| expect(t.transliterate(k)).to eql(v) }
+  end
+end