n.c.

[chise/ruby.git] / chise / parser.rb

diff --git a/chise/parser.rb b/chise/parser.rb
index 5a6cf7e..1570f20 100755 (executable)
--- a/chise/parser.rb
+++ b/chise/parser.rb
@@ -1,9 +1,10 @@
 # Copyright (C) 2002-2004 Kouichirou Eto, All rights reserved.
 
 require "chise/chisedb"
+require "chise/utf8"
 
 module CHISE
-  module EntityReference
+  module EntityReferenceModule
     PART    = "&([-+0-9A-Za-z#]+);"
     ALL     = '\A'+PART+'\Z'
     PART_RE = Regexp.new(PART)
@@ -13,7 +14,7 @@ module CHISE
     def is_er?(s)      (ALL_RE =~ s)  != nil;  end
 
     # the order is important.  The primary charset should be selectable.
-    CODESYS_TABLE = [
+    CCS_TABLE = [
       %w( =jis-x0208-1990      J90- 4 X),
       %w( =jis-x0208-1983      J83- 4 X),
       %w( =jis-x0208-1978      J78- 4 X),
@@ -24,6 +25,7 @@ module CHISE
       %w( =jis-x0213-2-2000    JX2- 4 X),
       %w( =jis-x0212           JSP- 4 X),
       %w( =big5-cdp            CDP- 4 X),
+      %w( =big5                        B-   4 X),
       %w( =cns11643-1          C1-  4 X),
       %w( =cns11643-2          C2-  4 X),
       %w( =cns11643-3          C3-  4 X),
@@ -36,12 +38,28 @@ module CHISE
       %w( =cbeta               CB   5 d),
       %w( =gt                  GT-  5 d),
       %w( =gt-k                        GT-K 5 d),
+      %w( =hanziku-1           HZK01- 4 X),
+      %w( =hanziku-2           HZK02- 4 X),
+      %w( =hanziku-3           HZK03- 4 X),
+      %w( =hanziku-4           HZK04- 4 X),
+      %w( =hanziku-5           HZK05- 4 X),
+      %w( =hanziku-6           HZK06- 4 X),
+      %w( =hanziku-7           HZK07- 4 X),
+      %w( =hanziku-8           HZK08- 4 X),
+      %w( =hanziku-9           HZK09- 4 X),
+      %w( =hanziku-10          HZK10- 4 X),
+      %w( =hanziku-11          HZK11- 4 X),
+      %w( =hanziku-12          HZK12- 4 X),
+      %w( =ruimoku-v6          RUI6-  4 X),
+      %w( =jef-china3          JC3-   4 X),
     ]
-    PRIVATE_USE_AREA = 0xe000
   end
 
   class CharacterParser
-    include EntityReference
+    include EntityReferenceModule
+    include UTF8Value
+
+    PRIVATE_USE_AREA = 0xe000
 
     def parse(c) # parse a value and return a number (MCS)
       raise "c is nil" if c.nil?
@@ -49,8 +67,9 @@ module CHISE
       if c.kind_of?(String)
        if /\A\?/ =~ c
          c = c.sub(/\A\?/, "") # remove "?" in the head
-         u4 = c.u8tou32 # translate from UTF-8 to UTF-32
-         return u4.u32to_i # translate UTF-32 to UCS number
+         #u4 = c.u8tou32 # translate from UTF-8 to UTF-32
+         #return u4.u32to_i # translate UTF-32 to UCS number
+         return u8toi(c)
        end
 
        return parse_er(c) if is_er?(c) # ER?
@@ -88,7 +107,7 @@ module CHISE
        s = s.sub(/\AI-/, "")
       end
 
-      CODESYS_TABLE.each {|codesys, er_prefix, keta, numtype|
+      CCS_TABLE.each {|ccs, er_prefix, keta, numtype|
        if numtype == "d"
          nre = '\d'
        elsif numtype == "X"
@@ -99,6 +118,7 @@ module CHISE
 
        re = "\\A#{er_prefix}(#{nre}{#{keta},#{keta}})\\Z"
        next unless Regexp.new(re) =~ s
+       #qp s
 
        codestr = $1
        if numtype == "d"
@@ -107,8 +127,8 @@ module CHISE
          code = codestr.hex
        end
 
-       u8 = get_ccs(codesys, code)
-#      qp s, u8
+       u8 = get_ccs(ccs, code)
+       #qp ccs, s, u8
        next if u8.nil?
 
        num = parse(u8)
@@ -128,7 +148,7 @@ module CHISE
   end
 
   class EntityReferenceParser
-    include EntityReference
+    include EntityReferenceModule
 
     def de_er(s) # replace EntityReference with corresponding character.
       return s unless PART_RE =~ s # don't use contain_er? to get $1
@@ -143,15 +163,15 @@ module CHISE
   end
 
   class EntityReferenceEncoder
-    include EntityReference
+    include EntityReferenceModule
 
     def to_er(char)
       cid = char.char_id
       return "&#x%04x;" % cid if cid <=  0xffff
       return "&#x%05x;" % cid if cid <= 0xfffff
 
-      CODESYS_TABLE.each {|codesys, er_prefix, keta, numtype|
-       code = char[codesys]
+      CCS_TABLE.each {|ccs, er_prefix, keta, numtype|
+       code = char[ccs]
        next if code.nil?
        return "&#{er_prefix}%0#{keta}#{numtype};" % code
       }
@@ -159,7 +179,7 @@ module CHISE
       "&MCS-%08X;" % cid # the last answer
     end
 
-    def to_er_by_ccs(cid, codesys) # not yet
+    def to_er_by_ccs(cid, ccs) # not yet
     end
 
   end