stringex 1.5.1 → 2.0.0

data/test/redcloth_to_html_test.rb

@@ -0,0 +1,37 @@
+require 'test_helper'
+require 'stringex'
+
+begin
+  require 'rubygems'
+  require 'RedCloth'
+rescue LoadError
+  puts
+  puts ">> Could not load RedCloth. String#to_html was not tested."
+  puts ">> Please gem install RedCloth if you'd like to use this functionality."
+  puts
+end
+
+class RedclothToHTMLTest < Test::Unit::TestCase
+  if defined?(RedCloth)
+    def test_to_html
+      {
+        "h1. A Solution" => "<h1>A Solution</h1>",
+        "I hated wrapping textilize around a string.\n\nIt always felt dirty." =>
+          "<p>I hated wrapping textilize around a string.</p>\n<p>It always felt dirty.</p>",
+        "I think _this_ is awesome" => "<p>I think <em>this</em> is awesome</p>",
+        "Um... _*really*_, man" => "<p>Um&#8230; <em><strong>really</strong></em>, man</p>"
+      }.each do |plain, html|
+        assert_equal html, plain.to_html
+      end
+    end
+
+    def test_to_html_lite
+      {
+        "I have no pee on me" => "I have no pee on me",
+        "But I _do_ get Textile!" => "But I <em>do</em> get Textile!"
+      }.each do |plain, html|
+        assert_equal html, plain.to_html(:lite)
+      end
+    end
+  end
+end

data/test/string_extensions_test.rb

@@ -1,39 +1,15 @@
 # encoding: UTF-8
-$KCODE = "U"
 
-require "test/unit"
+require "test_helper"
+require "stringex"
 
-$: << File.join(File.expand_path(File.dirname(__FILE__)), '../lib')
-require File.join(File.expand_path(File.dirname(__FILE__)), "../init.rb")
+if RUBY_VERSION.to_f < 1.9
+  $KCODE = "U"
+end
 
 class StringExtensionsTest < Test::Unit::TestCase
-  def test_to_html
-    require "rubygems"
-    require "RedCloth"
-    {
-      "h1. A Solution" => "<h1>A Solution</h1>",
-      "I hated wrapping textilize around a string.\n\nIt always felt dirty." =>
-        "<p>I hated wrapping textilize around a string.</p>\n<p>It always felt dirty.</p>",
-      "I think _this_ is awesome" => "<p>I think <em>this</em> is awesome</p>",
-      "Um... _*really*_, man" => "<p>Um&#8230; <em><strong>really</strong></em>, man</p>"
-    }.each do |plain, html|
-      assert_equal html, plain.to_html
-    end
-  rescue LoadError
-    puts "\n>> Could not load RedCloth. String#to_html was not tested.\n>> Please gem install RedCloth if you'd like to use this functionality."
-  end
-
-  def test_to_html_lite
-    require "rubygems"
-    require "RedCloth"
-    {
-      "I have no pee on me" => "I have no pee on me",
-      "But I _do_ get Textile!" => "But I <em>do</em> get Textile!"
-    }.each do |plain, html|
-      assert_equal html, plain.to_html(:lite)
-    end
-  rescue LoadError
-    puts "\n>> Could not load RedCloth. String#to_html (with :lite argument) was not tested.\n>> Please gem install RedCloth if you'd like to use this functionality."
+  def setup
+    Stringex::Localization.reset!
   end
 
   def test_to_url
@@ -77,7 +53,7 @@ class StringExtensionsTest < Test::Unit::TestCase
       "21'17ʼ51" =>
         "21-17-51",
       "ITCZ÷1 (21°17ʼ51.78”N / 89°35ʼ28.18”O / 26-04-08 / 09:00 am)" =>
-        "itcz-divide-1-21-degrees-17-51-dot-78-n-slash-89-degrees-35-28-dot-18-o-slash-26-04-08-slash-09-00-am",
+        "itcz-divided-by-1-21-degrees-17-51-dot-78-n-slash-89-degrees-35-28-dot-18-o-slash-26-04-08-slash-09-00-am",
       "／" =>
         "slash",
       "私はガラスを食べられます。それは私を傷つけません。" =>
@@ -93,7 +69,9 @@ class StringExtensionsTest < Test::Unit::TestCase
       "AVアンプ、ホームシアターシステム、スピーカーシステム等" =>
         "avanpu-homusiatasisutemu-supikasisutemudeng",
       "У лукоморья дуб зеленый" =>
-        "u-lukomoria-dub-zielienyi"
+        "u-lukomoria-dub-zielienyi",
+      "Here are some {braces}" =>
+        "here-are-some-braces"
     }.each do |html, plain|
       assert_equal plain, html.to_url
     end
@@ -103,6 +81,18 @@ class StringExtensionsTest < Test::Unit::TestCase
     assert_equal "So Fucking Special", "So Fucking Special".to_url(:exclude => "So Fucking Special")
   end
 
+  def test_to_url_without_forcing_downcase
+    assert_equal "I-have-CAPS", "I have CAPS".to_url(:force_downcase => false)
+  end
+
+  def test_to_url_with_limit
+    assert_equal "I am much too long".to_url(:limit => 13), "i-am-much-too"
+  end
+
+  def test_to_url_with_alternate_whitespace_replacement
+    assert_equal "under_scores", "Under Scores".to_url(:replace_whitespace_with => "_")
+  end
+
   def test_remove_formatting
     {
       "<p>This has 100% too much    <em>formatting</em></p>" =>
@@ -128,7 +118,7 @@ class StringExtensionsTest < Test::Unit::TestCase
     end
   end
 
-  def test_convert_accented_entities
+  def test_convert_accented_html_entities
     {
       "&aring;"  => "a",
       "&egrave;" => "e",
@@ -138,60 +128,41 @@ class StringExtensionsTest < Test::Unit::TestCase
       "&Ntilde;" => "N",
       "&ccedil;" => "c"
     }.each do |entitied, plain|
-      assert_equal plain, entitied.convert_accented_entities
+      assert_equal plain, entitied.convert_accented_html_entities
     end
   end
 
-  def test_convert_vulgar_fractions
+  def test_localized_vulgar_fractions_conversion
+    Stringex::Localization.backend = :internal
+    Stringex::Localization.store_translations :de, :vulgar_fractions, {
+      :one_fourth => "en fjerdedel",
+      :half => "en halv"
+    }
+    Stringex::Localization.locale = :de
+
     {
-      "&frac14;" => "one fourth",
-      "¼" => "one fourth",
-      "&#188;" => "one fourth",
-      "&frac12;" => "half",
-      "½" => "half",
-      "&#189;" => "half",
-      "&frac34;" => "three fourths",
-      "¾" => "three fourths",
-      "&#190;" => "three fourths",
-      "⅓" => "one third",
-      "&#8531;" => "one third",
-      "⅔" => "two thirds",
-      "&#8532;" => "two thirds",
-      "⅕" => "one fifth",
-      "&#8533;" => "one fifth",
-      "⅖" => "two fifths",
-      "&#8534;" => "two fifths",
-      "⅗" => "three fifths",
-      "&#8535;" => "three fifths",
-      "⅘" => "four fifths",
-      "&#8536;" => "four fifths",
-      "⅙" => "one sixth",
-      "&#8537;" => "one sixth",
-      "⅚" => "five sixths",
-      "&#8538;" => "five sixths",
-      "⅛" => "one eighth",
-      "&#8539;" => "one eighth",
-      "⅜" => "three eighths",
-      "&#8540;" => "three eighths",
-      "⅝" => "five eighths",
-      "&#8541;" => "five eighths",
-      "⅞" => "seven eighths",
-      "&#8542;" => "seven eighths"
+      "&frac14;" => "en fjerdedel",
+      "½" => "en halv"
     }.each do |entitied, plain|
       assert_equal plain, entitied.convert_vulgar_fractions
     end
   end
 
-  def test_convert_misc_entities
+  def test_localized_html_entities_conversion
+    Stringex::Localization.backend = :internal
+    Stringex::Localization.store_translations :de, :html_entities, {
+      :amp => "und",
+      :ellipsis => " prik prik prik",
+      :frac14 => "en fjerdedel"
+    }
+    Stringex::Localization.locale = :de
+
     {
-      "America&#8482;" => "America(tm)",
-      "Tea &amp; Sympathy" => "Tea and Sympathy",
-      "To be continued&#8230;" => "To be continued...",
-      "Foo&nbsp;Bar" => "Foo Bar",
-      "100&#163;" => "100 pound",
-      "35&deg;" => "35 degrees"
+      "Tea &amp; Sympathy" => "Tea und Sympathy",
+      "To be continued&#8230;" => "To be continued prik prik prik",
+      "Det var til &frac14; af prisen" => "Det var til en fjerdedel af prisen"
     }.each do |entitied, plain|
-      assert_equal plain, entitied.convert_misc_entities
+      assert_equal plain, entitied.convert_miscellaneous_html_entities
     end
   end
 
@@ -205,18 +176,19 @@ class StringExtensionsTest < Test::Unit::TestCase
     end
   end
 
-  def test_convert_misc_characters
+  def test_localized_character_conversions
+    Stringex::Localization.backend = :internal
+    Stringex::Localization.store_translations :de, :characters, {
+      "and" => "und",
+      :percent => "procent"
+    }
+    Stringex::Localization.locale = :de
+
     {
-      "Foo & bar make foobar" => "Foo and bar make foobar",
-      "Breakdown #9" => "Breakdown number 9",
-      "foo@bar.com" => "foo at bar dot com",
-      "100% of yr love" => "100 percent of yr love",
-      "Kisses are $3.25 each" => "Kisses are 3 dollars 25 cents each",
-      "That CD is £3.25 plus tax" => "That CD is 3 pounds 25 pence plus tax",
-      "This CD is ¥1000 instead" => "This CD is 1000 yen instead",
-      "Food+Drink" => "Food plus Drink"
+      "ich & dich" => "ich und dich",
+      "det var 100% godt" => "det var 100 procent godt"
     }.each do |misc, plain|
-      assert_equal plain, misc.convert_misc_characters
+      assert_equal plain, misc.convert_miscellaneous_characters
     end
   end
 
@@ -242,8 +214,4 @@ class StringExtensionsTest < Test::Unit::TestCase
 
     assert_equal "now-with-hyphens", "----now---------with-hyphens--------".collapse("-")
   end
-
-  def test_to_url_limit
-    assert_equal "I am much too long".to_url(:limit => 13), "i-am-much-too"
-  end
 end

data/test/test_helper.rb

@@ -0,0 +1,2 @@
+require 'test/unit'
+$: << File.join(File.expand_path(File.dirname(__FILE__)), '../lib')

data/test/unicode_point_suite/basic_greek_test.rb

@@ -0,0 +1,113 @@
+# encoding: UTF-8
+
+require "test/unit"
+require "stringex"
+require File.join(File.expand_path(File.dirname(__FILE__)), "codepoint_test_helper.rb")
+include CodepointTestHelper
+
+class BasicGreekTest < Test::Unit::TestCase
+  # This test suite is just regression test and debugging
+  # to better transliterate the Basic Latin Unicode codepoints
+  #
+  # http://unicode.org/charts/
+  # http://unicode.org/charts/PDF/U0000.pdf
+
+  # NOTE: I can't figure out how to test control characters.
+  # Get weird results trying to pack them to unicode.
+
+  def test_greek_letters
+    {
+      "Α" => "A",
+      "Ά" => "A",
+      "Β" => "B",
+      "Γ" => "G",
+      "Δ" => "D",
+      "Ε" => "E",
+      "Έ" => "E",
+      "Ζ" => "Z",
+      "Η" => "H",
+      "Θ" => "Th",
+      "Ι" => "I",
+      "Ί" => "I",
+      "Ϊ" => "I",
+      "Κ" => "K",
+      "Λ" => "L",
+      "Μ" => "M",
+      "Ν" => "N",
+      "Ξ" => "Ks",
+      "Ο" => "O",
+      "Ό" => "O",
+      "Π" => "P",
+      "Ρ" => "R",
+      "Σ" => "S",
+      "Τ" => "T",
+      "Υ" => "Y",
+      "Ϋ" => "Y",
+      "Ύ" => "Y",
+      "Φ" => "Ph",
+      "Χ" => "Ks",
+      "Ψ" => "Ps",
+      "Ω" => "O",
+      "Ώ" => "O",
+      "α" => "a",
+      "ά" => "a",
+      "β" => "b",
+      "γ" => "g",
+      "δ" => "d",
+      "ε" => "e",
+      "έ" => "e",
+      "ζ" => "z",
+      "η" => "i",
+      "ή" => "i",
+      "θ" => "th",
+      "ι" => "i",
+      "ί" => "i",
+      "ϊ" => "i",
+      "ΐ" => "i",
+      "κ" => "k",
+      "λ" => "l",
+      "μ" => "m",
+      "ν" => "n",
+      "ξ" => "ks",
+      "ο" => "o",
+      "ό" => "o",
+      "π" => "p",
+      "ρ" => "r",
+      "σ" => "s",
+      "ς" => "s",
+      "τ" => "t",
+      "υ" => "u",
+      "ύ" => "u",
+      "ϋ" => "u",
+      "ΰ" => "u",
+      "φ" => "ph",
+      "χ" => "x",
+      "ψ" => "ps",
+      "ω" => "o",
+      "ώ" => "o"
+    }.each do |letter, ascii|
+      assert_equal letter.to_ascii, ascii
+    end
+  end
+
+
+  def test_greek_words
+    {
+      "Καλημέρα" => "Kalimera",
+      "Προϊόν" => "Proion",
+      "Θαλασσογραφία" => "Thalassographia",
+      "Να μας πάρεις μακριά" => "Na mas pareis makria",
+      "να μας πας στα πέρα μέρη" => "na mas pas sta pera meri",
+      "φύσα θάλασσα πλατιά" => "phusa thalassa platia",
+      "φύσα αγέρι φύσα αγέρι" => "phusa ageri phusa ageri",
+      "Αν θέλεις να λέγεσαι άνθρωπος" => "An theleis na legesai anthropos",
+      "δεν θα πάψεις ούτε στιγμή ν΄αγωνίζεσαι για την ειρήνη και" => "den tha papseis oute stigmi nagonizesai gia tin eirini kai",
+      "για το δίκαιο." => "gia to dikaio.",
+      "Θα βγείς στους δρόμους, θα φωνάξεις, τα χείλια σου θα" => "Tha bgeis stous dromous, tha phonakseis, ta xeilia sou tha",
+      "ματώσουν απ΄τις φωνές" => "matosoun aptis phones",
+      "το πρόσωπό σου θα ματώσει από τις σφαίρες – μα ούτε βήμα πίσω." => "to prosopo sou tha matosei apo tis sphaires -- ma oute bima piso.",
+    }.each do |letter, ascii|
+      assert_equal letter.to_ascii, ascii
+    end
+  end
+end

data/test/unicode_point_suite/basic_latin_test.rb

@@ -0,0 +1,142 @@
+# encoding: UTF-8
+
+require "test/unit"
+require "stringex"
+require File.join(File.expand_path(File.dirname(__FILE__)), "codepoint_test_helper.rb")
+include CodepointTestHelper
+
+class BasicLatinTest < Test::Unit::TestCase
+  # This test suite is just regression test and debugging
+  # to better transliterate the Basic Latin Unicode codepoints
+  #
+  # http://unicode.org/charts/
+  # http://unicode.org/charts/PDF/U0000.pdf
+
+  # NOTE: I can't figure out how to test control characters.
+  # Get weird results trying to pack them to unicode.
+
+  def test_spaces
+    assert_equal_encoded " ", %w{0020 00a0}
+    assert_equal_encoded "",  %w{200b 2060}
+  end
+
+  def test_exclamation_marks
+    assert_equal_encoded "!", %w{0021 2762}
+    assert_equal_encoded "!!", "203c"
+    assert_equal_encoded "", "00a1"
+    assert_equal_encoded "?!", "203d"
+  end
+
+  def test_quotation_marks
+    assert_equal_encoded "\"", %w{0022 02ba 2033 3003}
+  end
+
+  def test_apostrophes
+    assert_equal_encoded "'", %w{0027 02b9 02bc 02c8 2032}
+  end
+
+  def test_asterisks
+    assert_equal_encoded "*", %w{002a 066d 204e 2217 26b9 2731}
+  end
+
+  def test_commas
+    assert_equal_encoded ",", %w{002c 060c}
+  end
+
+  def test_periods
+    assert_equal_encoded ".", %w{002e 06d4}
+  end
+
+  def test_hyphens
+    assert_equal_encoded "-", %w{002d 2010 2011 2012 2212}
+  end
+
+  def test_endash
+    assert_equal_encoded "--", %w{2013 2015}
+  end
+
+  def test_emdash
+    assert_equal_encoded "---", %w{2014}
+  end
+
+  def test_dotleader
+    assert_equal_encoded "..", %w{2025}
+  end
+
+  def test_ellipsis
+    assert_equal_encoded "...", %w{2026}
+  end
+
+  def test_slashes
+    assert_equal_encoded "/", %w{002f 2044 2215}
+    assert_equal_encoded "\\", %w{005c 2216}
+  end
+
+  def test_colons
+    assert_equal_encoded ":", %w{003a 2236}
+  end
+
+  def test_semicolons
+    assert_equal_encoded ";", %w{003b 061b}
+  end
+
+  def test_less_thans
+    assert_equal_encoded "<", %w{003c 2039 2329 27e8 3008}
+  end
+
+  def test_equals
+    assert_equal_encoded "=", "003d"
+  end
+
+  def test_greater_thans
+    assert_equal_encoded ">", %w{003e 203a 232a 27e9 3009}
+  end
+
+  def test_question_marks
+    assert_equal_encoded "?", %w{003f 061f}
+    assert_equal_encoded "", "00bf"
+    assert_equal_encoded "?!", %w{203d 2048}
+    assert_equal_encoded "!?", "2049"
+  end
+
+  def test_circumflexes
+    assert_equal_encoded "^", %w{005e 2038 2303}
+  end
+
+  def test_underscores
+    assert_equal_encoded "_", %w{005f 02cd 2017}
+  end
+
+  def test_grave_accents
+    assert_equal_encoded "'", %w{0060 02cb 2035}
+  end
+
+  def test_bars
+    assert_equal_encoded "|", %w{007c 2223 2758}
+  end
+
+  def test_tildes
+    assert_equal_encoded "~", %w{007e 02dc 2053 223c ff5e}
+  end
+
+  def test_related_letters
+    {
+      "B" => "212c",
+      "C" => %w{2102 212d},
+      "E" => %w{2107 2130},
+      "F" => "2131",
+      "H" => %w{210b 210c 210d},
+      "I" => %w{0130 0406 04c0 2110 2111 2160},
+      "K" => "212a",
+      "L" => "2112",
+      "M" => "2133",
+      "N" => "2115",
+      "P" => "2119",
+      "Q" => "211a",
+      "R" => %w{211b 211c 211d},
+      "Z" => %w{2124 2128}
+    }.each do |expected, encode_mes|
+      assert_equal_encoded expected, encode_mes
+    end
+  end
+end