element.rb in regextest-0.1.6

- old
+ new

@@ -6,14 +6,18 @@
 require 'regextest/common'
 
 class Regextest::Back::Element
   include Regextest::Common
   def initialize(param)
-    # puts "Element param:#{param[:cmd]} data:#{param[:data].size}"
-    @command = param[:cmd]
+    # puts "Element param:#{param[:cmd]} data:#{param[:ranges].size}"
     @param = param
-    @candidates = param[:data] if @command == :CMD_SELECT
+    @command = param[:cmd]
+    @charset = param[:charset]
+    if @command == :CMD_SELECT
+      @candidates = param[:ranges].inject([]){|result, range| result += range.to_a}
+    end
+    # @candidates = param[:data] if @command == :CMD_SELECT
   end
   
   attr_reader :param, :command, :candidates
   
   # random fix
@@ -31,11 +35,12 @@
   # size of candidates
   def size
     if(@candidates)
       @candidates.size
     else
-      raise "internal error: candidates not found at size-method"
+      # raise "internal error: candidates not found at size-method"
+      0
     end
   end
   
   # [] of candidates
   def [](num)
@@ -80,17 +85,18 @@
     case @command
     when :CMD_SELECT
       if(@candidates)
         @candidates.inspect
       else
-        @param[:data].inspect
+        @param[:ranges].inspect
       end
     when :CMD_LOOK_BEHIND, :CMD_LOOK_AHEAD, :CMD_NOT_LOOK_BEHIND, :CMD_NOT_LOOK_AHEAD
       @param.inspect
     when :CMD_ANC_LINE_BEGIN, :CMD_ANC_LINE_END, :CMD_ANC_WORD_BOUND, :CMD_ANC_WORD_UNBOUND,
          :CMD_ANC_STRING_BEGIN, :CMD_ANC_STRING_END, :CMD_ANC_STRING_END2, :CMD_ANC_MATCH_START,
-         :CMD_ANC_LOOK_BEHIND2, :CMD_ANC_RELUCTANT_BEGIN, :CMD_ANC_RELUCTANT_END
+         :CMD_ANC_LOOK_BEHIND2, :CMD_ANC_RELUCTANT_BEGIN, :CMD_ANC_RELUCTANT_END,
+         :CMD_ANC_POSSESSIVE_BEGIN, :CMD_ANC_POSSESSIVE_END
       @param.inspect
     else
       raise "inner error, invalid command #{@command}"
     end
   end
@@ -105,38 +111,55 @@
     @candidates = [0xa]
   end
   
   # Is word-elements only?
   def word_elements?
-    letters = @candidates.map{|elem| [elem].pack("U*")}
-    letters.join("").match(/^\p{Word}+$/)
+    letters = @candidates.map{|elem| [elem].pack("U*")}.join("")
+    if @charset == "u" || @charset == "d"
+      letters.match(/^\p{Word}+$/)
+    else
+      letters.match(/^\w+$/)
+    end
   end
   
   # is non-word-elements only?
   def non_word_elements?
-    letters = @candidates.map{|elem| [elem].pack("U*")}
-    letters.join("").match(/^\p{^Word}+$/)
+    letters = @candidates.map{|elem| [elem].pack("U*")}.join("")
+    if @charset == "u" || @charset == "d"
+      letters.match(/^\p{^Word}+$/)
+    else
+      letters.match(/^\W+$/)
+    end
   end
 
   # set word-elements
   def set_word_elements
-    @candidates.select!{|elem| [elem].pack("U*").match(/^\w$/)}
+    if @charset == "u" || @charset == "d"
+      @candidates.select!{|elem| [elem].pack("U*").match(/^\p{Word}$/)}
+    else
+      @candidates.select!{|elem| [elem].pack("U*").match(/^\w$/)}
+    end
   end
   
   # set non_word-elements
   def set_non_word_elements
-    @candidates.select!{|elem| [elem].pack("U*").match(/^\W$/)}
+    if @charset == "u" || @charset == "d"
+      @candidates.select!{|elem| [elem].pack("U*").match(/^\p{^Word}$/)}
+      #@candidates.select!{|elem| [elem].pack("U*").match(/^[[:^word:]]$/)}
+    else
+      @candidates.select!{|elem| [elem].pack("U*").match(/^[[:^word:]]$/)}
+    end
   end
   
   # checks empty
   def empty?
     @candidates.size == 0
   end
   
   # factory method to generate any char element
   def self.any_char
     # BUG: must consider other character set!
-    Regextest::Back::Element.new({cmd: :CMD_SELECT, data:  (0x20..0x7e).to_a})
+    Regextest::Back::Element.new({cmd: :CMD_SELECT, ranges:  [0x20..0x7e]})
   end
   
   # factory method to generate any char element
   def reverse
     @candidates = ((0x20..0x7e).to_a) - @candidates