regexp_parser 0.2.1 → 0.3.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: ccb0d0c3dacb9f2285a8c2458dd9e8a207a4c836
-  data.tar.gz: 27f379340236393b559aa622bd2763d3d27ec3a1
+  metadata.gz: 682c13ce2716430eea274098ce40b3e842b47eba
+  data.tar.gz: 15faf04abe034eb801292781bb2d1503f0b1df0d
 SHA512:
-  metadata.gz: 29fe46ace91ba3b6b106423fa212707e98ccc7f1559094f14bc9a017f4ec28e2ab9deae1e000a71c8329d65791445e73f004eda77cbd7293865ae518df1b894d
-  data.tar.gz: 0f6f3530c2de9e60782ed914deb594137ce37735b21a50f305f1548df06896d6618b95cd69e25de9dcc0a58b03d8a08306355e67c47478e4ef09cb278caf815c
+  metadata.gz: 5991a90cc872e72b4361692deb7d5370c70f5baafeed5afe3e90db4a95947bdaf7e072f77c6e2ca07eb839c79bcebd9022312bcbe3f8ccfc59b84b445d889c1c
+  data.tar.gz: 46d4cebfd6c904002da41db9cfc7747d42775c6e55d9e2853a4fca1a18652ef368e381057dd8a83fa59908c68f2f3c45e6bee19b0a3b8fa282036cfcdef90443

data/ChangeLog

@@ -1,3 +1,15 @@
+Sun Aug 6 2015 Ammar Ali <ammarabuali@gmail.com>
+
+	* Added UnicodeBlocks support to the parser.
+
+Mon Aug 3 2015 Garen Torikian <gjtorikian@gmail.com>
+
+	* Added UnicodeBlocks support to the scanner.
+
+Sat Apr 18 14:38:12 2015 Ammar Ali <ammarabuali@gmail.com>
+
+	* Updated ruby versions for latest releases.
+
 Wed Dec 3 05:21:27 2014 Ammar Ali <ammarabuali@gmail.com>
 
 	* Added expand_members method to CharacterSet, returns traditional

data/README.md

@@ -355,12 +355,7 @@ _Note that not all of these are available in all versions of Ruby_
 |   _**General Categories**_  | `\p{Lu}`, `\P{Cs}`                                      | ✓ |
 |   _**Scripts**_             | `\p{Arabic}`, `\P{Hiragana}`                            | ✓ |
 |   _**Simple**_              | `\p{Dash}`, `\p{Extender}`                              | ✓ |
-
-
-<br/>
-##### Missing Features
-
-  - Unicode blocks, e.g. \p{InArrows}, \p{InArmenian}. _(h/t @gjtorikian for pointing it out)_
+| &emsp;&nbsp;_**Blocks**_             | `\p{InArmenian}`, `\P{InKhmer}`                            | &#x2713; |
 
 ##### Inapplicable Features
 

data/lib/regexp_parser/expression/classes/property.rb

@@ -103,6 +103,7 @@ module Regexp::Expression
     class Age     < UnicodeProperty::Base; end
     class Derived < UnicodeProperty::Base; end
     class Script  < UnicodeProperty::Base; end
+    class Block   < UnicodeProperty::Base; end
   end
 
 end # module Regexp::Expression

data/lib/regexp_parser/parser.rb

@@ -270,6 +270,9 @@ module Regexp::Parser
     when *Regexp::Syntax::Token::UnicodeProperty::Script
       @node << Script.new(token)
 
+    when *Regexp::Syntax::Token::UnicodeProperty::UnicodeBlock
+      @node << Block.new(token)
+
     else
       raise UnknownTokenError.new('UnicodeProperty', token)
     end

data/lib/regexp_parser/scanner.rb

@@ -346,9 +346,9 @@ self._re_scanner_indicies = [
 	40, 40, 40, 40, 40, 40, 40, 40, 
 	40, 40, 40, 40, 40, 40, 40, 40, 
 	41, 40, 40, 40, 40, 40, 40, 40, 
-	40, 40, 40, 40, 40, 40, 40, 40, 
-	40, 40, 40, 40, 40, 40, 40, 40, 
-	40, 40, 40, 40, 40, 40, 40, 40, 
+	40, 40, 40, 40, 40, 41, 40, 40, 
+	41, 41, 41, 41, 41, 41, 41, 41, 
+	41, 41, 40, 40, 40, 40, 40, 40, 
 	40, 42, 41, 41, 41, 41, 41, 41, 
 	41, 41, 41, 41, 41, 41, 41, 41, 
 	41, 41, 41, 41, 41, 41, 41, 41, 
@@ -361,8 +361,8 @@ self._re_scanner_indicies = [
 	40, 40, 40, 40, 40, 40, 40, 40, 
 	40, 40, 40, 41, 40, 40, 40, 40, 
 	40, 40, 40, 40, 40, 40, 40, 40, 
-	40, 40, 40, 40, 40, 40, 40, 40, 
-	40, 40, 40, 40, 40, 40, 40, 40, 
+	41, 40, 40, 41, 41, 41, 41, 41, 
+	41, 41, 41, 41, 41, 40, 40, 40, 
 	40, 40, 40, 40, 41, 41, 41, 41, 
 	41, 41, 41, 41, 41, 41, 41, 41, 
 	41, 41, 41, 41, 41, 41, 41, 41, 
@@ -375,9 +375,9 @@ self._re_scanner_indicies = [
 	40, 40, 40, 40, 40, 40, 40, 40, 
 	40, 40, 40, 40, 40, 40, 40, 40, 
 	40, 41, 40, 40, 40, 40, 40, 40, 
-	40, 40, 40, 40, 40, 40, 40, 40, 
-	40, 40, 40, 40, 40, 40, 40, 40, 
-	40, 40, 40, 40, 40, 40, 40, 40, 
+	40, 40, 40, 40, 40, 40, 41, 40, 
+	40, 41, 41, 41, 41, 41, 41, 41, 
+	41, 41, 41, 40, 40, 40, 40, 40, 
 	40, 40, 41, 41, 41, 41, 41, 41, 
 	45, 41, 41, 41, 41, 41, 41, 41, 
 	41, 41, 41, 41, 41, 41, 41, 41, 
@@ -390,9 +390,9 @@ self._re_scanner_indicies = [
 	40, 40, 40, 40, 40, 40, 40, 40, 
 	40, 40, 40, 40, 40, 40, 40, 41, 
 	40, 40, 40, 40, 40, 40, 40, 40, 
-	40, 40, 40, 40, 40, 40, 40, 40, 
-	40, 40, 40, 40, 40, 40, 40, 40, 
-	40, 40, 40, 40, 40, 40, 40, 40, 
+	40, 40, 40, 40, 41, 40, 40, 41, 
+	41, 41, 41, 41, 41, 41, 41, 41, 
+	41, 40, 40, 40, 40, 40, 40, 40, 
 	41, 41, 41, 41, 46, 41, 41, 41, 
 	41, 41, 41, 41, 41, 41, 41, 41, 
 	41, 41, 41, 41, 41, 41, 41, 41, 
@@ -405,8 +405,8 @@ self._re_scanner_indicies = [
 	40, 40, 40, 40, 40, 40, 40, 40, 
 	40, 40, 40, 40, 40, 41, 40, 40, 
 	40, 40, 40, 40, 40, 40, 40, 40, 
-	40, 40, 40, 40, 40, 40, 40, 40, 
-	40, 40, 40, 40, 40, 40, 40, 40, 
+	40, 40, 41, 40, 40, 41, 41, 41, 
+	41, 41, 41, 41, 41, 41, 41, 40, 
 	40, 40, 47, 40, 40, 40, 41, 41, 
 	41, 41, 41, 41, 41, 41, 41, 41, 
 	41, 41, 41, 41, 41, 41, 41, 41, 
@@ -424,9 +424,9 @@ self._re_scanner_indicies = [
 	40, 40, 40, 40, 40, 40, 40, 40, 
 	40, 40, 40, 40, 40, 40, 40, 41, 
 	40, 40, 40, 40, 40, 40, 40, 40, 
-	40, 40, 40, 40, 40, 40, 40, 40, 
-	40, 40, 40, 40, 40, 40, 40, 40, 
-	40, 40, 40, 40, 40, 40, 40, 40, 
+	40, 40, 40, 40, 41, 40, 40, 41, 
+	41, 41, 41, 41, 41, 41, 41, 41, 
+	41, 40, 40, 40, 40, 40, 40, 40, 
 	42, 41, 41, 41, 41, 41, 41, 41, 
 	41, 41, 41, 41, 41, 41, 41, 41, 
 	41, 41, 41, 41, 41, 41, 41, 41, 
@@ -2205,7 +2205,7 @@ te = p+1
         self.emit(type, :script_tagalog,                  text, ts-1, te)
       when 'thaa', 'thaana'
         self.emit(type, :script_thaana,                   text, ts-1, te)
-      when 'thai' 
+      when 'thai'
         self.emit(type, :script_thai,                     text, ts-1, te)
       when 'tibt', 'tibetan'
         self.emit(type, :script_tibetan,                  text, ts-1, te)
@@ -2230,6 +2230,220 @@ te = p+1
       when 'zzzz', 'unknown'
         self.emit(type, :script_unknown,                  text, ts-1, te)
 
+      # Unicode blocks
+      when 'inalphabeticpresentationforms'
+        self.emit(type, :block_inalphabetic_presentation_forms,           text, ts-1, te)
+      when 'inalphabeticpresentationforms'
+        self.emit(type, :block_inalphabetic_presentation_forms,           text, ts-1, te)
+      when 'inarabicpresentationforms-a'
+        self.emit(type, :block_inarabic_presentation_forms_a,             text, ts-1, te)
+      when 'inarabicpresentationforms-b'
+        self.emit(type, :block_inarabic_presentation_forms_b,             text, ts-1, te)
+      when 'inarabic'
+        self.emit(type, :block_inarabic,                                  text, ts-1, te)
+      when 'inarmenian'
+        self.emit(type, :block_inarmenian,                                text, ts-1, te)
+      when 'inarrows'
+        self.emit(type, :block_inarrows,                                  text, ts-1, te)
+      when 'inbasiclatin'
+        self.emit(type, :block_inbasic_latin,                             text, ts-1, te)
+      when 'inbengali'
+        self.emit(type, :block_inbengali,                                 text, ts-1, te)
+      when 'inblockelements'
+        self.emit(type, :block_inblock_elements,                          text, ts-1, te)
+      when 'inbopomofoextended'
+        self.emit(type, :block_inbopomofo_extended,                       text, ts-1, te)
+      when 'inbopomofo'
+        self.emit(type, :block_inbopomofo,                                text, ts-1, te)
+      when 'inboxdrawing'
+        self.emit(type, :block_inbox_drawing,                             text, ts-1, te)
+      when 'inbraillepatterns'
+        self.emit(type, :block_inbraille_patterns,                        text, ts-1, te)
+      when 'inbuhid'
+        self.emit(type, :block_inbuhid,                                   text, ts-1, te)
+      when 'incjkcompatibilityforms'
+        self.emit(type, :block_incjk_compatibility_forms,                 text, ts-1, te)
+      when 'incjkcompatibilityideographs'
+        self.emit(type, :block_incjk_compatibility_ideographs,            text, ts-1, te)
+      when 'incjkcompatibility'
+        self.emit(type, :block_incjk_compatibility,                       text, ts-1, te)
+      when 'incjkradicalssupplement'
+        self.emit(type, :block_incjk_radicals_supplement,                 text, ts-1, te)
+      when 'incjksymbolsandpunctuation'
+        self.emit(type, :block_incjk_symbols_and_punctuation,             text, ts-1, te)
+      when 'incjkunifiedideographsextensiona'
+        self.emit(type, :block_incjk_unified_ideographs_extension_a,      text, ts-1, te)
+      when 'incjkunifiedideographs'
+        self.emit(type, :block_incjk_unified_ideographs,                  text, ts-1, te)
+      when 'incherokee'
+        self.emit(type, :block_incherokee,                                text, ts-1, te)
+      when 'incombiningdiacriticalmarksforsymbols'
+        self.emit(type, :block_incombining_diacritical_marks_for_symbols, text, ts-1, te)
+      when 'incombiningdiacriticalmarks'
+        self.emit(type, :block_incombining_diacritical_marks,             text, ts-1, te)
+      when 'incombininghalfmarks'
+        self.emit(type, :block_incombining_half_marks,                    text, ts-1, te)
+      when 'incontrolpictures'
+        self.emit(type, :block_incontrol_pictures,                        text, ts-1, te)
+      when 'incurrencysymbols'
+        self.emit(type, :block_incurrency_symbols,                        text, ts-1, te)
+      when 'incyrillicsupplementary'
+        self.emit(type, :block_incyrillic_supplementary,                  text, ts-1, te)
+      when 'incyrillic'
+        self.emit(type, :block_incyrillic,                                text, ts-1, te)
+      when 'indevanagari'
+        self.emit(type, :block_indevanagari,                              text, ts-1, te)
+      when 'indingbats'
+        self.emit(type, :block_indingbats,                                text, ts-1, te)
+      when 'inenclosedalphanumerics'
+        self.emit(type, :block_inenclosed_alphanumerics,                  text, ts-1, te)
+      when 'inenclosedcjklettersandmonths'
+        self.emit(type, :block_inenclosed_cjk_letters_and_months,         text, ts-1, te)
+      when 'inethiopic'
+        self.emit(type, :block_inethiopic,                                text, ts-1, te)
+      when 'ingeneralpunctuation'
+        self.emit(type, :block_ingeneral_punctuation,                     text, ts-1, te)
+      when 'ingeometricshapes'
+        self.emit(type, :block_ingeometric_shapes,                        text, ts-1, te)
+      when 'ingeorgian'
+        self.emit(type, :block_ingeorgian,                                text, ts-1, te)
+      when 'ingreekextended'
+        self.emit(type, :block_ingreek_extended,                          text, ts-1, te)
+      when 'ingreekandcoptic'
+        self.emit(type, :block_ingreek_and_coptic,                        text, ts-1, te)
+      when 'ingujarati'
+        self.emit(type, :block_ingujarati,                                text, ts-1, te)
+      when 'ingurmukhi'
+        self.emit(type, :block_ingurmukhi,                                text, ts-1, te)
+      when 'inhalfwidthandfullwidthforms'
+        self.emit(type, :block_inhalfwidth_and_fullwidth_forms,           text, ts-1, te)
+      when 'inhangulcompatibilityjamo'
+        self.emit(type, :block_inhangul_compatibility_jamo,               text, ts-1, te)
+      when 'inhanguljamo'
+        self.emit(type, :block_inhangul_jamo,                             text, ts-1, te)
+      when 'inhangulsyllables'
+        self.emit(type, :block_inhangul_syllables,                        text, ts-1, te)
+      when 'inhanunoo'
+        self.emit(type, :block_inhanunoo,                                 text, ts-1, te)
+      when 'inhebrew'
+        self.emit(type, :block_inhebrew,                                  text, ts-1, te)
+      when 'inhighprivateusesurrogates'
+        self.emit(type, :block_inhigh_private_use_surrogates,             text, ts-1, te)
+      when 'inhighsurrogates'
+        self.emit(type, :block_inhigh_surrogates,                         text, ts-1, te)
+      when 'inhiragana'
+        self.emit(type, :block_inhiragana,                                text, ts-1, te)
+      when 'inipaextensions'
+        self.emit(type, :block_inipa_extensions,                          text, ts-1, te)
+      when 'inideographicdescriptioncharacters'
+        self.emit(type, :block_inideographic_description_characters,      text, ts-1, te)
+      when 'inkanbun'
+        self.emit(type, :block_inkanbun,                                  text, ts-1, te)
+      when 'inkangxiradicals'
+        self.emit(type, :block_inkangxi_radicals,                         text, ts-1, te)
+      when 'inkannada'
+        self.emit(type, :block_inkannada,                                 text, ts-1, te)
+      when 'inkatakanaphoneticextensions'
+        self.emit(type, :block_inkatakana_phonetic_extensions,            text, ts-1, te)
+      when 'inkatakana'
+        self.emit(type, :block_inkatakana,                                text, ts-1, te)
+      when 'inkhmersymbols'
+        self.emit(type, :block_inkhmer_symbols,                           text, ts-1, te)
+      when 'inkhmer'
+        self.emit(type, :block_inkhmer,                                   text, ts-1, te)
+      when 'inlao'
+        self.emit(type, :block_inlao,                                     text, ts-1, te)
+      when 'inlatin-1supplement'
+        self.emit(type, :block_inlatin_1_supplement,                      text, ts-1, te)
+      when 'inlatinextended-a'
+        self.emit(type, :block_inlatin_extended_a,                        text, ts-1, te)
+      when 'inlatinextended-b'
+        self.emit(type, :block_inlatin_extended_b,                        text, ts-1, te)
+      when 'inlatinextendedadditional'
+        self.emit(type, :block_inlatin_extended_additional,               text, ts-1, te)
+      when 'inletterlikesymbols'
+        self.emit(type, :block_inletterlike_symbols,                      text, ts-1, te)
+      when 'inlimbu'
+        self.emit(type, :block_inlimbu,                                   text, ts-1, te)
+      when 'inlowsurrogates'
+        self.emit(type, :block_inlow_surrogates,                          text, ts-1, te)
+      when 'inmalayalam'
+        self.emit(type, :block_inmalayalam,                               text, ts-1, te)
+      when 'inmathematicaloperators'
+        self.emit(type, :block_inmathematical_operators,                  text, ts-1, te)
+      when 'inmiscellaneousmathematicalsymbols-a'
+        self.emit(type, :block_inmiscellaneous_mathematical_symbols_a,    text, ts-1, te)
+      when 'inmiscellaneousmathematicalsymbols-b'
+        self.emit(type, :block_inmiscellaneous_mathematical_symbols_b,    text, ts-1, te)
+      when 'inmiscellaneoussymbolsandarrows'
+        self.emit(type, :block_inmiscellaneous_symbols_and_arrows,        text, ts-1, te)
+      when 'inmiscellaneoussymbols'
+        self.emit(type, :block_inmiscellaneous_symbols,                   text, ts-1, te)
+      when 'inmiscellaneoustechnical'
+        self.emit(type, :block_inmiscellaneous_technical,                 text, ts-1, te)
+      when 'inmongolian'
+        self.emit(type, :block_inmongolian,                               text, ts-1, te)
+      when 'inmyanmar'
+        self.emit(type, :block_inmyanmar,                                 text, ts-1, te)
+      when 'innumberforms'
+        self.emit(type, :block_innumber_forms,                            text, ts-1, te)
+      when 'inogham'
+        self.emit(type, :block_inogham,                                   text, ts-1, te)
+      when 'inopticalcharacterrecognition'
+        self.emit(type, :block_inoptical_character_recognition,           text, ts-1, te)
+      when 'inoriya'
+        self.emit(type, :block_inoriya,                                   text, ts-1, te)
+      when 'inphoneticextensions'
+        self.emit(type, :block_inphonetic_extensions,                     text, ts-1, te)
+      when 'inprivateusearea'
+        self.emit(type, :block_inprivate_use_area,                        text, ts-1, te)
+      when 'inrunic'
+        self.emit(type, :block_inrunic,                                   text, ts-1, te)
+      when 'insinhala'
+        self.emit(type, :block_insinhala,                                 text, ts-1, te)
+      when 'insmallformvariants'
+        self.emit(type, :block_insmall_form_variants,                     text, ts-1, te)
+      when 'inspacingmodifierletters'
+        self.emit(type, :block_inspacing_modifier_letters,                text, ts-1, te)
+      when 'inspecials'
+        self.emit(type, :block_inspecials,                                text, ts-1, te)
+      when 'insuperscriptsandsubscripts'
+        self.emit(type, :block_insuperscripts_and_subscripts,             text, ts-1, te)
+      when 'insupplementalarrows-a'
+        self.emit(type, :block_insupplemental_arrows_a,                   text, ts-1, te)
+      when 'insupplementalarrows-b'
+        self.emit(type, :block_insupplemental_arrows_b,                   text, ts-1, te)
+      when 'insupplementalmathematicaloperators'
+        self.emit(type, :block_insupplemental_mathematical_operators,     text, ts-1, te)
+      when 'insyriac'
+        self.emit(type, :block_insyriac,                                  text, ts-1, te)
+      when 'intagalog'
+        self.emit(type, :block_intagalog,                                 text, ts-1, te)
+      when 'intagbanwa'
+        self.emit(type, :block_intagbanwa,                                text, ts-1, te)
+      when 'intaile'
+        self.emit(type, :block_intai_le,                                  text, ts-1, te)
+      when 'intamil'
+        self.emit(type, :block_intamil,                                   text, ts-1, te)
+      when 'intelugu'
+        self.emit(type, :block_intelugu,                                  text, ts-1, te)
+      when 'inthaana'
+        self.emit(type, :block_inthaana,                                  text, ts-1, te)
+      when 'inthai'
+        self.emit(type, :block_inthai,                                    text, ts-1, te)
+      when 'intibetan'
+        self.emit(type, :block_intibetan,                                 text, ts-1, te)
+      when 'inunifiedcanadianaboriginalsyllabics'
+        self.emit(type, :block_inunified_canadian_aboriginal_syllabics,   text, ts-1, te)
+      when 'invariationselectors'
+        self.emit(type, :block_invariation_selectors,                     text, ts-1, te)
+      when 'inyiradicals'
+        self.emit(type, :block_inyi_radicals,                             text, ts-1, te)
+      when 'inyisyllables'
+        self.emit(type, :block_inyi_syllables,                            text, ts-1, te)
+      when 'inyijinghexagramsymbols'
+        self.emit(type, :block_inyijing_hexagram_symbols,                 text, ts-1, te)
+
       else
         # Should this really be an error? Or would emitting
         # an :unknown for the property be better?
@@ -3791,7 +4005,7 @@ te = p+1
 # line 764 "/Users/ammar/src/ruby/projects/regexp_parser/lib/regexp_parser/scanner/scanner.rl"
 		begin
 act = 59;		end
-# line 3795 "/Users/ammar/src/ruby/projects/regexp_parser/lib/regexp_parser/scanner.rb"
+# line 4009 "/Users/ammar/src/ruby/projects/regexp_parser/lib/regexp_parser/scanner.rb"
 	end
 	end
 	end
@@ -3809,7 +4023,7 @@ ts = nil;		end
 		begin
 act = 0
 		end
-# line 3813 "/Users/ammar/src/ruby/projects/regexp_parser/lib/regexp_parser/scanner.rb"
+# line 4027 "/Users/ammar/src/ruby/projects/regexp_parser/lib/regexp_parser/scanner.rb"
 	end
 
 	if cs == 0
@@ -3843,7 +4057,7 @@ act = 0
     text = ts ? copy(data, ts-1..-1) : data.pack('c*')
     raise PrematureEndError.new( text )
   		end
-# line 3847 "/Users/ammar/src/ruby/projects/regexp_parser/lib/regexp_parser/scanner.rb"
+# line 4061 "/Users/ammar/src/ruby/projects/regexp_parser/lib/regexp_parser/scanner.rb"
 	  end
 	end
 

data/lib/regexp_parser/scanner/property.rl

@@ -33,7 +33,7 @@
                           'id_start'i       | 'id_continue'i      |
                           'xid_start'i      | 'xid_continue'i     |
                           'grapheme_base'i  | 'grapheme_extend'i  |
-                          'default_ignorable_code_point'i; 
+                          'default_ignorable_code_point'i;
 
   property_age          = 'age=1.1'i | 'age=2.0'i | 'age=2.1'i |
                           'age=3.0'i | 'age=3.1'i | 'age=3.2'i |
@@ -42,7 +42,7 @@
                           'age=6.1'i | 'age=6.2'i | 'age=6.3'i |
                           'age=7.0'i;
 
-  property_script       = (alpha | space | '_')+; # everything else
+  property_script       = (alnum | space | '_' | '-')+; # everything else
 
   property_sequence     = property_char . '{' . '^'? (
                             property_name | general_category |
@@ -553,7 +553,7 @@
         self.emit(type, :script_tagalog,                  text, ts-1, te)
       when 'thaa', 'thaana'
         self.emit(type, :script_thaana,                   text, ts-1, te)
-      when 'thai' 
+      when 'thai'
         self.emit(type, :script_thai,                     text, ts-1, te)
       when 'tibt', 'tibetan'
         self.emit(type, :script_tibetan,                  text, ts-1, te)
@@ -578,6 +578,220 @@
       when 'zzzz', 'unknown'
         self.emit(type, :script_unknown,                  text, ts-1, te)
 
+      # Unicode blocks
+      when 'inalphabeticpresentationforms'
+        self.emit(type, :block_inalphabetic_presentation_forms,           text, ts-1, te)
+      when 'inalphabeticpresentationforms'
+        self.emit(type, :block_inalphabetic_presentation_forms,           text, ts-1, te)
+      when 'inarabicpresentationforms-a'
+        self.emit(type, :block_inarabic_presentation_forms_a,             text, ts-1, te)
+      when 'inarabicpresentationforms-b'
+        self.emit(type, :block_inarabic_presentation_forms_b,             text, ts-1, te)
+      when 'inarabic'
+        self.emit(type, :block_inarabic,                                  text, ts-1, te)
+      when 'inarmenian'
+        self.emit(type, :block_inarmenian,                                text, ts-1, te)
+      when 'inarrows'
+        self.emit(type, :block_inarrows,                                  text, ts-1, te)
+      when 'inbasiclatin'
+        self.emit(type, :block_inbasic_latin,                             text, ts-1, te)
+      when 'inbengali'
+        self.emit(type, :block_inbengali,                                 text, ts-1, te)
+      when 'inblockelements'
+        self.emit(type, :block_inblock_elements,                          text, ts-1, te)
+      when 'inbopomofoextended'
+        self.emit(type, :block_inbopomofo_extended,                       text, ts-1, te)
+      when 'inbopomofo'
+        self.emit(type, :block_inbopomofo,                                text, ts-1, te)
+      when 'inboxdrawing'
+        self.emit(type, :block_inbox_drawing,                             text, ts-1, te)
+      when 'inbraillepatterns'
+        self.emit(type, :block_inbraille_patterns,                        text, ts-1, te)
+      when 'inbuhid'
+        self.emit(type, :block_inbuhid,                                   text, ts-1, te)
+      when 'incjkcompatibilityforms'
+        self.emit(type, :block_incjk_compatibility_forms,                 text, ts-1, te)
+      when 'incjkcompatibilityideographs'
+        self.emit(type, :block_incjk_compatibility_ideographs,            text, ts-1, te)
+      when 'incjkcompatibility'
+        self.emit(type, :block_incjk_compatibility,                       text, ts-1, te)
+      when 'incjkradicalssupplement'
+        self.emit(type, :block_incjk_radicals_supplement,                 text, ts-1, te)
+      when 'incjksymbolsandpunctuation'
+        self.emit(type, :block_incjk_symbols_and_punctuation,             text, ts-1, te)
+      when 'incjkunifiedideographsextensiona'
+        self.emit(type, :block_incjk_unified_ideographs_extension_a,      text, ts-1, te)
+      when 'incjkunifiedideographs'
+        self.emit(type, :block_incjk_unified_ideographs,                  text, ts-1, te)
+      when 'incherokee'
+        self.emit(type, :block_incherokee,                                text, ts-1, te)
+      when 'incombiningdiacriticalmarksforsymbols'
+        self.emit(type, :block_incombining_diacritical_marks_for_symbols, text, ts-1, te)
+      when 'incombiningdiacriticalmarks'
+        self.emit(type, :block_incombining_diacritical_marks,             text, ts-1, te)
+      when 'incombininghalfmarks'
+        self.emit(type, :block_incombining_half_marks,                    text, ts-1, te)
+      when 'incontrolpictures'
+        self.emit(type, :block_incontrol_pictures,                        text, ts-1, te)
+      when 'incurrencysymbols'
+        self.emit(type, :block_incurrency_symbols,                        text, ts-1, te)
+      when 'incyrillicsupplementary'
+        self.emit(type, :block_incyrillic_supplementary,                  text, ts-1, te)
+      when 'incyrillic'
+        self.emit(type, :block_incyrillic,                                text, ts-1, te)
+      when 'indevanagari'
+        self.emit(type, :block_indevanagari,                              text, ts-1, te)
+      when 'indingbats'
+        self.emit(type, :block_indingbats,                                text, ts-1, te)
+      when 'inenclosedalphanumerics'
+        self.emit(type, :block_inenclosed_alphanumerics,                  text, ts-1, te)
+      when 'inenclosedcjklettersandmonths'
+        self.emit(type, :block_inenclosed_cjk_letters_and_months,         text, ts-1, te)
+      when 'inethiopic'
+        self.emit(type, :block_inethiopic,                                text, ts-1, te)
+      when 'ingeneralpunctuation'
+        self.emit(type, :block_ingeneral_punctuation,                     text, ts-1, te)
+      when 'ingeometricshapes'
+        self.emit(type, :block_ingeometric_shapes,                        text, ts-1, te)
+      when 'ingeorgian'
+        self.emit(type, :block_ingeorgian,                                text, ts-1, te)
+      when 'ingreekextended'
+        self.emit(type, :block_ingreek_extended,                          text, ts-1, te)
+      when 'ingreekandcoptic'
+        self.emit(type, :block_ingreek_and_coptic,                        text, ts-1, te)
+      when 'ingujarati'
+        self.emit(type, :block_ingujarati,                                text, ts-1, te)
+      when 'ingurmukhi'
+        self.emit(type, :block_ingurmukhi,                                text, ts-1, te)
+      when 'inhalfwidthandfullwidthforms'
+        self.emit(type, :block_inhalfwidth_and_fullwidth_forms,           text, ts-1, te)
+      when 'inhangulcompatibilityjamo'
+        self.emit(type, :block_inhangul_compatibility_jamo,               text, ts-1, te)
+      when 'inhanguljamo'
+        self.emit(type, :block_inhangul_jamo,                             text, ts-1, te)
+      when 'inhangulsyllables'
+        self.emit(type, :block_inhangul_syllables,                        text, ts-1, te)
+      when 'inhanunoo'
+        self.emit(type, :block_inhanunoo,                                 text, ts-1, te)
+      when 'inhebrew'
+        self.emit(type, :block_inhebrew,                                  text, ts-1, te)
+      when 'inhighprivateusesurrogates'
+        self.emit(type, :block_inhigh_private_use_surrogates,             text, ts-1, te)
+      when 'inhighsurrogates'
+        self.emit(type, :block_inhigh_surrogates,                         text, ts-1, te)
+      when 'inhiragana'
+        self.emit(type, :block_inhiragana,                                text, ts-1, te)
+      when 'inipaextensions'
+        self.emit(type, :block_inipa_extensions,                          text, ts-1, te)
+      when 'inideographicdescriptioncharacters'
+        self.emit(type, :block_inideographic_description_characters,      text, ts-1, te)
+      when 'inkanbun'
+        self.emit(type, :block_inkanbun,                                  text, ts-1, te)
+      when 'inkangxiradicals'
+        self.emit(type, :block_inkangxi_radicals,                         text, ts-1, te)
+      when 'inkannada'
+        self.emit(type, :block_inkannada,                                 text, ts-1, te)
+      when 'inkatakanaphoneticextensions'
+        self.emit(type, :block_inkatakana_phonetic_extensions,            text, ts-1, te)
+      when 'inkatakana'
+        self.emit(type, :block_inkatakana,                                text, ts-1, te)
+      when 'inkhmersymbols'
+        self.emit(type, :block_inkhmer_symbols,                           text, ts-1, te)
+      when 'inkhmer'
+        self.emit(type, :block_inkhmer,                                   text, ts-1, te)
+      when 'inlao'
+        self.emit(type, :block_inlao,                                     text, ts-1, te)
+      when 'inlatin-1supplement'
+        self.emit(type, :block_inlatin_1_supplement,                      text, ts-1, te)
+      when 'inlatinextended-a'
+        self.emit(type, :block_inlatin_extended_a,                        text, ts-1, te)
+      when 'inlatinextended-b'
+        self.emit(type, :block_inlatin_extended_b,                        text, ts-1, te)
+      when 'inlatinextendedadditional'
+        self.emit(type, :block_inlatin_extended_additional,               text, ts-1, te)
+      when 'inletterlikesymbols'
+        self.emit(type, :block_inletterlike_symbols,                      text, ts-1, te)
+      when 'inlimbu'
+        self.emit(type, :block_inlimbu,                                   text, ts-1, te)
+      when 'inlowsurrogates'
+        self.emit(type, :block_inlow_surrogates,                          text, ts-1, te)
+      when 'inmalayalam'
+        self.emit(type, :block_inmalayalam,                               text, ts-1, te)
+      when 'inmathematicaloperators'
+        self.emit(type, :block_inmathematical_operators,                  text, ts-1, te)
+      when 'inmiscellaneousmathematicalsymbols-a'
+        self.emit(type, :block_inmiscellaneous_mathematical_symbols_a,    text, ts-1, te)
+      when 'inmiscellaneousmathematicalsymbols-b'
+        self.emit(type, :block_inmiscellaneous_mathematical_symbols_b,    text, ts-1, te)
+      when 'inmiscellaneoussymbolsandarrows'
+        self.emit(type, :block_inmiscellaneous_symbols_and_arrows,        text, ts-1, te)
+      when 'inmiscellaneoussymbols'
+        self.emit(type, :block_inmiscellaneous_symbols,                   text, ts-1, te)
+      when 'inmiscellaneoustechnical'
+        self.emit(type, :block_inmiscellaneous_technical,                 text, ts-1, te)
+      when 'inmongolian'
+        self.emit(type, :block_inmongolian,                               text, ts-1, te)
+      when 'inmyanmar'
+        self.emit(type, :block_inmyanmar,                                 text, ts-1, te)
+      when 'innumberforms'
+        self.emit(type, :block_innumber_forms,                            text, ts-1, te)
+      when 'inogham'
+        self.emit(type, :block_inogham,                                   text, ts-1, te)
+      when 'inopticalcharacterrecognition'
+        self.emit(type, :block_inoptical_character_recognition,           text, ts-1, te)
+      when 'inoriya'
+        self.emit(type, :block_inoriya,                                   text, ts-1, te)
+      when 'inphoneticextensions'
+        self.emit(type, :block_inphonetic_extensions,                     text, ts-1, te)
+      when 'inprivateusearea'
+        self.emit(type, :block_inprivate_use_area,                        text, ts-1, te)
+      when 'inrunic'
+        self.emit(type, :block_inrunic,                                   text, ts-1, te)
+      when 'insinhala'
+        self.emit(type, :block_insinhala,                                 text, ts-1, te)
+      when 'insmallformvariants'
+        self.emit(type, :block_insmall_form_variants,                     text, ts-1, te)
+      when 'inspacingmodifierletters'
+        self.emit(type, :block_inspacing_modifier_letters,                text, ts-1, te)
+      when 'inspecials'
+        self.emit(type, :block_inspecials,                                text, ts-1, te)
+      when 'insuperscriptsandsubscripts'
+        self.emit(type, :block_insuperscripts_and_subscripts,             text, ts-1, te)
+      when 'insupplementalarrows-a'
+        self.emit(type, :block_insupplemental_arrows_a,                   text, ts-1, te)
+      when 'insupplementalarrows-b'
+        self.emit(type, :block_insupplemental_arrows_b,                   text, ts-1, te)
+      when 'insupplementalmathematicaloperators'
+        self.emit(type, :block_insupplemental_mathematical_operators,     text, ts-1, te)
+      when 'insyriac'
+        self.emit(type, :block_insyriac,                                  text, ts-1, te)
+      when 'intagalog'
+        self.emit(type, :block_intagalog,                                 text, ts-1, te)
+      when 'intagbanwa'
+        self.emit(type, :block_intagbanwa,                                text, ts-1, te)
+      when 'intaile'
+        self.emit(type, :block_intai_le,                                  text, ts-1, te)
+      when 'intamil'
+        self.emit(type, :block_intamil,                                   text, ts-1, te)
+      when 'intelugu'
+        self.emit(type, :block_intelugu,                                  text, ts-1, te)
+      when 'inthaana'
+        self.emit(type, :block_inthaana,                                  text, ts-1, te)
+      when 'inthai'
+        self.emit(type, :block_inthai,                                    text, ts-1, te)
+      when 'intibetan'
+        self.emit(type, :block_intibetan,                                 text, ts-1, te)
+      when 'inunifiedcanadianaboriginalsyllabics'
+        self.emit(type, :block_inunified_canadian_aboriginal_syllabics,   text, ts-1, te)
+      when 'invariationselectors'
+        self.emit(type, :block_invariation_selectors,                     text, ts-1, te)
+      when 'inyiradicals'
+        self.emit(type, :block_inyi_radicals,                             text, ts-1, te)
+      when 'inyisyllables'
+        self.emit(type, :block_inyi_syllables,                            text, ts-1, te)
+      when 'inyijinghexagramsymbols'
+        self.emit(type, :block_inyijing_hexagram_symbols,                 text, ts-1, te)
+
       else
         # Should this really be an error? Or would emitting
         # an :unknown for the property be better?

data/lib/regexp_parser/syntax/tokens/unicode_property.rb

@@ -225,7 +225,115 @@ module Regexp::Syntax
         :script_warang_citi
       ]
 
-      V190 = CharType + POSIX + Category::All + Derived + Script
+      UnicodeBlock = [
+        :block_inalphabetic_presentation_forms,
+        :block_inarabic_presentation_forms_a,
+        :block_inarabic_presentation_forms_b,
+        :block_inarabic,
+        :block_inarmenian,
+        :block_inarrows,
+        :block_inbasic_latin,
+        :block_inbengali,
+        :block_inblock_elements,
+        :block_inbopomofo_extended,
+        :block_inbopomofo,
+        :block_inbox_drawing,
+        :block_inbraille_patterns,
+        :block_inbuhid,
+        :block_incjk_compatibility_forms,
+        :block_incjk_compatibility_ideographs,
+        :block_incjk_compatibility,
+        :block_incjk_radicals_supplement,
+        :block_incjk_symbols_and_punctuation,
+        :block_incjk_unified_ideographs_extension_a,
+        :block_incjk_unified_ideographs,
+        :block_incherokee,
+        :block_incombining_diacritical_marks_for_symbols,
+        :block_incombining_diacritical_marks,
+        :block_incombining_half_marks,
+        :block_incontrol_pictures,
+        :block_incurrency_symbols,
+        :block_incyrillic_supplementary,
+        :block_incyrillic,
+        :block_indevanagari,
+        :block_indingbats,
+        :block_inenclosed_alphanumerics,
+        :block_inenclosed_cjk_letters_and_months,
+        :block_inethiopic,
+        :block_ingeneral_punctuation,
+        :block_ingeometric_shapes,
+        :block_ingeorgian,
+        :block_ingreek_extended,
+        :block_ingreek_and_coptic,
+        :block_ingujarati,
+        :block_ingurmukhi,
+        :block_inhalfwidth_and_fullwidth_forms,
+        :block_inhangul_compatibility_jamo,
+        :block_inhangul_jamo,
+        :block_inhangul_syllables,
+        :block_inhanunoo,
+        :block_inhebrew,
+        :block_inhigh_private_use_surrogates,
+        :block_inhigh_surrogates,
+        :block_inhiragana,
+        :block_inipa_extensions,
+        :block_inideographic_description_characters,
+        :block_inkanbun,
+        :block_inkangxi_radicals,
+        :block_inkannada,
+        :block_inkatakana_phonetic_extensions,
+        :block_inkatakana,
+        :block_inkhmer_symbols,
+        :block_inkhmer,
+        :block_inlao,
+        :block_inlatin_1_supplement,
+        :block_inlatin_extended_a,
+        :block_inlatin_extended_b,
+        :block_inlatin_extended_additional,
+        :block_inletterlike_symbols,
+        :block_inlimbu,
+        :block_inlow_surrogates,
+        :block_inmalayalam,
+        :block_inmathematical_operators,
+        :block_inmiscellaneous_mathematical_symbols_a,
+        :block_inmiscellaneous_mathematical_symbols_b,
+        :block_inmiscellaneous_symbols_and_arrows,
+        :block_inmiscellaneous_symbols,
+        :block_inmiscellaneous_technical,
+        :block_inmongolian,
+        :block_inmyanmar,
+        :block_innumber_forms,
+        :block_inogham,
+        :block_inoptical_character_recognition,
+        :block_inoriya,
+        :block_inphonetic_extensions,
+        :block_inprivate_use_area,
+        :block_inrunic,
+        :block_insinhala,
+        :block_insmall_form_variants,
+        :block_inspacing_modifier_letters,
+        :block_inspecials,
+        :block_insuperscripts_and_subscripts,
+        :block_insupplemental_arrows_a,
+        :block_insupplemental_arrows_b,
+        :block_insupplemental_mathematical_operators,
+        :block_insyriac,
+        :block_intagalog,
+        :block_intagbanwa,
+        :block_intai_le,
+        :block_intamil,
+        :block_intelugu,
+        :block_inthaana,
+        :block_inthai,
+        :block_intibetan,
+        :block_inunified_canadian_aboriginal_syllabics,
+        :block_invariation_selectors,
+        :block_inyi_radicals,
+        :block_inyi_syllables,
+        :block_inyijing_hexagram_symbols,
+      ]
+
+      V190 = CharType + POSIX + Category::All + Derived + Script + UnicodeBlock
       V193 = Age_V193 + Script_6_0
 
       V200 = Age_V200

data/lib/regexp_parser/version.rb

@@ -1,5 +1,5 @@
 class Regexp
   module Parser
-    VERSION = '0.2.1'
+    VERSION = '0.3.0'
   end
 end

data/test/parser/test_properties.rb

@@ -337,6 +337,13 @@ class ParserProperties < Test::Unit::TestCase
            "Expected Script property, but got #{t.expressions[1].class.name}")
   end
 
+  def test_parse_property_block
+    t = RP.parse 'ab\p{InArmenian}cd', 'ruby/1.9'
+
+    assert( t.expressions[1].is_a?(UnicodeProperty::Block),
+           "Expected Block property, but got #{t.expressions[1].class.name}")
+  end
+
   def test_parse_property_following_literal
     t = RP.parse 'ab\p{Lu}cd', 'ruby/1.9'
 

data/test/scanner/test_all.rb

@@ -2,7 +2,7 @@ require File.expand_path("../../helpers", __FILE__)
 
 %w{
   anchors errors escapes free_space groups literals
-  meta properties quantifiers scripts sets types
+  meta properties quantifiers scripts sets types unicode_blocks
 }.each do|tc|
   require File.expand_path("../test_#{tc}", __FILE__)
 end

data/test/scanner/test_unicode_blocks.rb

@@ -0,0 +1,130 @@
+require File.expand_path("../../helpers", __FILE__)
+
+class ScannerUnicodeBlocks < Test::Unit::TestCase
+
+  tests = {
+    'InAlphabetic_Presentation_Forms'           => :block_inalphabetic_presentation_forms,
+    'InArabic_Presentation_Forms-A'             => :block_inarabic_presentation_forms_a,
+    'InArabic_Presentation_Forms-B'             => :block_inarabic_presentation_forms_b,
+    'InArabic'                                  => :block_inarabic,
+    'InArmenian'                                => :block_inarmenian,
+    'InArrows'                                  => :block_inarrows,
+    'InBasic_Latin'                             => :block_inbasic_latin,
+    'InBengali'                                 => :block_inbengali,
+    'InBlock_Elements'                          => :block_inblock_elements,
+    'InBopomofo_Extended'                       => :block_inbopomofo_extended,
+    'InBopomofo'                                => :block_inbopomofo,
+    'InBox_Drawing'                             => :block_inbox_drawing,
+    'InBraille_Patterns'                        => :block_inbraille_patterns,
+    'InBuhid'                                   => :block_inbuhid,
+    'InCJK_Compatibility_Forms'                 => :block_incjk_compatibility_forms,
+    'InCJK_Compatibility_Ideographs'            => :block_incjk_compatibility_ideographs,
+    'InCJK_Compatibility'                       => :block_incjk_compatibility,
+    'InCJK_Radicals_Supplement'                 => :block_incjk_radicals_supplement,
+    'InCJK_Symbols_and_Punctuation'             => :block_incjk_symbols_and_punctuation,
+    'InCJK_Unified_Ideographs_Extension_A'      => :block_incjk_unified_ideographs_extension_a,
+    'InCJK_Unified_Ideographs'                  => :block_incjk_unified_ideographs,
+    'InCherokee'                                => :block_incherokee,
+    'InCombining_Diacritical_Marks_for_Symbols' => :block_incombining_diacritical_marks_for_symbols,
+    'InCombining_Diacritical_Marks'             => :block_incombining_diacritical_marks,
+    'InCombining_Half_Marks'                    => :block_incombining_half_marks,
+    'InControl_Pictures'                        => :block_incontrol_pictures,
+    'InCurrency_Symbols'                        => :block_incurrency_symbols,
+    'InCyrillic_Supplementary'                  => :block_incyrillic_supplementary,
+    'InCyrillic'                                => :block_incyrillic,
+    'InDevanagari'                              => :block_indevanagari,
+    'InDingbats'                                => :block_indingbats,
+    'InEnclosed_Alphanumerics'                  => :block_inenclosed_alphanumerics,
+    'InEnclosed_CJK_Letters_and_Months'         => :block_inenclosed_cjk_letters_and_months,
+    'InEthiopic'                                => :block_inethiopic,
+    'InGeneral_Punctuation'                     => :block_ingeneral_punctuation,
+    'InGeometric_Shapes'                        => :block_ingeometric_shapes,
+    'InGeorgian'                                => :block_ingeorgian,
+    'InGreek_Extended'                          => :block_ingreek_extended,
+    'InGreek_and_Coptic'                        => :block_ingreek_and_coptic,
+    'InGujarati'                                => :block_ingujarati,
+    'InGurmukhi'                                => :block_ingurmukhi,
+    'InHalfwidth_and_Fullwidth_Forms'           => :block_inhalfwidth_and_fullwidth_forms,
+    'InHangul_Compatibility_Jamo'               => :block_inhangul_compatibility_jamo,
+    'InHangul_Jamo'                             => :block_inhangul_jamo,
+    'InHangul_Syllables'                        => :block_inhangul_syllables,
+    'InHanunoo'                                 => :block_inhanunoo,
+    'InHebrew'                                  => :block_inhebrew,
+    'InHigh_Private_Use_Surrogates'             => :block_inhigh_private_use_surrogates,
+    'InHigh_Surrogates'                         => :block_inhigh_surrogates,
+    'InHiragana'                                => :block_inhiragana,
+    'InIPA_Extensions'                          => :block_inipa_extensions,
+    'InIdeographic_Description_Characters'      => :block_inideographic_description_characters,
+    'InKanbun'                                  => :block_inkanbun,
+    'InKangxi_Radicals'                         => :block_inkangxi_radicals,
+    'InKannada'                                 => :block_inkannada,
+    'InKatakana_Phonetic_Extensions'            => :block_inkatakana_phonetic_extensions,
+    'InKatakana'                                => :block_inkatakana,
+    'InKhmer_Symbols'                           => :block_inkhmer_symbols,
+    'InKhmer'                                   => :block_inkhmer,
+    'InLao'                                     => :block_inlao,
+    'InLatin-1_Supplement'                      => :block_inlatin_1_supplement,
+    'InLatin_Extended-A'                        => :block_inlatin_extended_a,
+    'InLatin_Extended-B'                        => :block_inlatin_extended_b,
+    'InLatin_Extended_Additional'               => :block_inlatin_extended_additional,
+    'InLetterlike_Symbols'                      => :block_inletterlike_symbols,
+    'InLimbu'                                   => :block_inlimbu,
+    'InLow_Surrogates'                          => :block_inlow_surrogates,
+    'InMalayalam'                               => :block_inmalayalam,
+    'InMathematical_Operators'                  => :block_inmathematical_operators,
+    'InMiscellaneous_Mathematical_Symbols-A'    => :block_inmiscellaneous_mathematical_symbols_a,
+    'InMiscellaneous_Mathematical_Symbols-B'    => :block_inmiscellaneous_mathematical_symbols_b,
+    'InMiscellaneous_Symbols_and_Arrows'        => :block_inmiscellaneous_symbols_and_arrows,
+    'InMiscellaneous_Symbols'                   => :block_inmiscellaneous_symbols,
+    'InMiscellaneous_Technical'                 => :block_inmiscellaneous_technical,
+    'InMongolian'                               => :block_inmongolian,
+    'InMyanmar'                                 => :block_inmyanmar,
+    'InNumber_Forms'                            => :block_innumber_forms,
+    'InOgham'                                   => :block_inogham,
+    'InOptical_Character_Recognition'           => :block_inoptical_character_recognition,
+    'InOriya'                                   => :block_inoriya,
+    'InPhonetic_Extensions'                     => :block_inphonetic_extensions,
+    'InPrivate_Use_Area'                        => :block_inprivate_use_area,
+    'InRunic'                                   => :block_inrunic,
+    'InSinhala'                                 => :block_insinhala,
+    'InSmall_Form_Variants'                     => :block_insmall_form_variants,
+    'InSpacing_Modifier_Letters'                => :block_inspacing_modifier_letters,
+    'InSpecials'                                => :block_inspecials,
+    'InSuperscripts_and_Subscripts'             => :block_insuperscripts_and_subscripts,
+    'InSupplemental_Arrows-A'                   => :block_insupplemental_arrows_a,
+    'InSupplemental_Arrows-B'                   => :block_insupplemental_arrows_b,
+    'InSupplemental_Mathematical_Operators'     => :block_insupplemental_mathematical_operators,
+    'InSyriac'                                  => :block_insyriac,
+    'InTagalog'                                 => :block_intagalog,
+    'InTagbanwa'                                => :block_intagbanwa,
+    'InTai_Le'                                  => :block_intai_le,
+    'InTamil'                                   => :block_intamil,
+    'InTelugu'                                  => :block_intelugu,
+    'InThaana'                                  => :block_inthaana,
+    'InThai'                                    => :block_inthai,
+    'InTibetan'                                 => :block_intibetan,
+    'InUnified_Canadian_Aboriginal_Syllabics'   => :block_inunified_canadian_aboriginal_syllabics,
+    'InVariation_Selectors'                     => :block_invariation_selectors,
+    'InYi_Radicals'                             => :block_inyi_radicals,
+    'InYi_Syllables'                            => :block_inyi_syllables,
+    'InYijing_Hexagram_Symbols'                 => :block_inyijing_hexagram_symbols
+  }
+
+  count = 0
+  tests.each do |property, test|
+    define_method "test_scan_property_#{test}_#{count+=1}" do
+      token = RS.scan("a\\p{#{property}}c")[1]
+
+      assert_equal( :property,  token[0] )
+      assert_equal( test,       token[1] )
+    end
+
+    define_method "test_scan_nonproperty_#{test}_#{count+=1}" do
+      token = RS.scan("a\\P{#{property}}c")[1]
+
+      assert_equal( :nonproperty,   token[0] )
+      assert_equal( test,           token[1] )
+    end
+  end
+
+end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: regexp_parser
 version: !ruby/object:Gem::Version
-  version: 0.2.1
+  version: 0.3.0
 platform: ruby
 authors:
 - Ammar Ali
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-04-18 00:00:00.000000000 Z
+date: 2015-08-09 00:00:00.000000000 Z
 dependencies: []
 description: A library for tokenizing, lexing, and parsing Ruby regular expressions.
 email:
@@ -133,6 +133,7 @@ files:
 - test/scanner/test_scripts.rb
 - test/scanner/test_sets.rb
 - test/scanner/test_types.rb
+- test/scanner/test_unicode_blocks.rb
 - test/syntax/ruby/test_1.8.rb
 - test/syntax/ruby/test_1.9.1.rb
 - test/syntax/ruby/test_1.9.3.rb
@@ -220,6 +221,7 @@ test_files:
 - test/scanner/test_scripts.rb
 - test/scanner/test_sets.rb
 - test/scanner/test_types.rb
+- test/scanner/test_unicode_blocks.rb
 - test/syntax/ruby/test_1.8.rb
 - test/syntax/ruby/test_1.9.1.rb
 - test/syntax/ruby/test_1.9.3.rb