case-folding.rb: modularize

* enc/unicode/case-folding.rb (CaseFolding): modularize, and add
  --output-file option.

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@46267 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

ChangeLog

@@ -1,4 +1,7 @@
-Sat May 31 08:49:52 2014  Nobuyoshi Nakada  <nobu@ruby-lang.org>
+Sat May 31 08:54:58 2014  Nobuyoshi Nakada  <nobu@ruby-lang.org>
+
+	* enc/unicode/case-folding.rb (CaseFolding): modularize, and add
+	  --output-file option.
 
 	* enc/unicode/case-folding.rb: script to convert CaseFolding.txt,
 	  tranlated from CaseFolding.py.

enc/unicode/case-folding.rb

@@ -4,35 +4,44 @@
 #   $ wget http://www.unicode.org/Public/UNIDATA/CaseFolding.txt
 #   $ ruby CaseFolding.rb CaseFolding.txt > ../enc/unicode/casefold.h
 
+class CaseFolding
+  module Util
+    module_function
 
     def hex_seq(v)
       v.map {|i| "0x%04x" % i}.join(", ")
     end
 
-def print_table(table, data)
+    def print_table_1(dest, data)
-    print("static const #{table}[] = {\n")
       for k, v in data.sort
-      if Array === k and k.length > 1
+        sk = (Array === k and k.length > 1) ? "{#{hex_seq(k)}}" : ("0x%04x" % k)
-        sk = "{#{hex_seq(k)}}"
+        dest.print("  {#{sk}, {#{v.length}, {#{hex_seq(v)}}}},\n")
-      else
-        sk = "0x%04x" % k
       end
-      print("  {#{sk}, {#{v.length}, {#{hex_seq(v)}}}},\n")
-    end
-    print("};\n\n")
     end
 
-def print_case_folding_data(filename)
+    def print_table(dest, type, data)
+      data.each do |n, d|
+        dest.print("static const #{type}_Type #{n}[] = {\n")
+        print_table_1(dest, d)
+        dest.print("};\n\n")
+      end
+    end
+  end
+
+  include Util
+
+  attr_reader :fold, :fold_locale, :unfold, :unfold_locale
+
+  def load(filename)
     pattern = /([0-9A-F]{4,6}); ([CFT]); ([0-9A-F]{4,6})(?: ([0-9A-F]{4,6}))?(?: ([0-9A-F]{4,6}))?;/
 
-  fold = {}
+    @fold = fold = {}
-  unfold = [{}, {}, {}]
+    @unfold = unfold = [{}, {}, {}]
     turkic = []
 
     IO.foreach(filename) do |line|
       next unless res = pattern.match(line)
       ch_from = res[1].to_i(16)
-    ch_to = []
 
       if res[2] == 'T'
         # Turkic case folding
@@ -41,10 +50,9 @@ def print_case_folding_data(filename)
       end
 
       # store folding data
-    (3..6).each do |i|
+      ch_to = res[3..6].inject([]) do |a, i|
-      if res[i]
+        break a unless i
-        ch_to << res[i].to_i(16)
+        a << i.to_i(16)
-      end
       end
       fold[ch_from] = ch_to
 
@@ -54,54 +62,80 @@ def print_case_folding_data(filename)
     end
 
     # move locale dependent data to (un)fold_locale
-  fold_locale = {}
+    @fold_locale = fold_locale = {}
-  unfold_locale = [{}, {}]
+    @unfold_locale = unfold_locale = [{}, {}]
     for ch_from in turkic
       key = fold[ch_from]
       i = key.length - 1
       unfold_locale[i][i == 0 ? key[0] : key] = unfold[i].delete(key)
       fold_locale[ch_from] = fold.delete(ch_from)
     end
+    self
+  end
 
+  def display(dest)
     # print the header
-  print("/* DO NOT EDIT THIS FILE. */\n")
+    dest.print("/* DO NOT EDIT THIS FILE. */\n")
-  print("/* Generated by tool/CaseFolding.py */\n\n")
+    dest.print("/* Generated by enc/unicode/case-folding.rb */\n\n")
 
     # print folding data
 
-  # CaseFold
+    # CaseFold + CaseFold_Locale
-  print_table("CaseFold_11_Type CaseFold", fold)
+    name = "CaseFold_11"
-
+    print_table(dest, name, "CaseFold"=>fold)
-  # CaseFold_Locale
+    print_table(dest, name, "CaseFold_Locale"=>fold_locale)
-  print_table("CaseFold_11_Type CaseFold_Locale", fold_locale)
 
     # print unfolding data
 
-  # CaseUnfold_11
+    # CaseUnfold_11 + CaseUnfold_11_Locale
-  print_table("CaseUnfold_11_Type CaseUnfold_11", unfold[0])
+    name = "CaseUnfold_11"
+    print_table(dest, name, name=>unfold[0])
+    print_table(dest, name, "#{name}_Locale"=>unfold_locale[0])
 
-  # CaseUnfold_11_Locale
+    # CaseUnfold_12 + CaseUnfold_12_Locale
-  print_table("CaseUnfold_11_Type CaseUnfold_11_Locale", unfold_locale[0])
+    name = "CaseUnfold_12"
-
+    print_table(dest, name, name=>unfold[1])
-  # CaseUnfold_12
+    print_table(dest, name, "#{name}_Locale"=>unfold_locale[1])
-  print_table("CaseUnfold_12_Type CaseUnfold_12", unfold[1])
-
-  # CaseUnfold_12_Locale
-  print_table("CaseUnfold_12_Type CaseUnfold_12_Locale", unfold_locale[1])
 
     # CaseUnfold_13
-  print_table("CaseUnfold_13_Type CaseUnfold_13", unfold[2])
+    name = "CaseUnfold_13"
+    print_table(dest, name, name=>unfold[2])
 
     # table sizes
     fold_table_size = fold.size + fold_locale.size
-  printf("#define FOLD_TABLE_SIZE\t\t%d\n", (fold_table_size * 1.2))
+    dest.printf("#define FOLD_TABLE_SIZE\t\t%d\n", (fold_table_size * 1.2))
     unfold1_table_size = unfold[0].size + unfold_locale[0].size
-  printf("#define UNFOLD1_TABLE_SIZE\t%d\n", (unfold1_table_size * 1.2))
+    dest.printf("#define UNFOLD1_TABLE_SIZE\t%d\n", (unfold1_table_size * 1.2))
     unfold2_table_size = unfold[1].size + unfold_locale[1].size
-  printf("#define UNFOLD2_TABLE_SIZE\t%d\n", (unfold2_table_size * 1.5))
+    dest.printf("#define UNFOLD2_TABLE_SIZE\t%d\n", (unfold2_table_size * 1.5))
     unfold3_table_size = unfold[2].size
-  printf("#define UNFOLD3_TABLE_SIZE\t%d\n", (unfold3_table_size * 1.7))
+    dest.printf("#define UNFOLD3_TABLE_SIZE\t%d\n", (unfold3_table_size * 1.7))
   end
 
+  def self.load(*args)
+    new.load(*args)
+  end
+end
+
+if $0 == __FILE__
+  require 'optparse'
+  dest = nil
+  fold_1 = false
+  ARGV.options do |opt|
+    opt.banner << " [INPUT]"
+    opt.on("--output-file=FILE", "-o", "output to the FILE instead of STDOUT") {|output|
+      dest = (output unless output == '-')
+    }
+    opt.parse!
+    abort(opt.to_s) if ARGV.size > 1
+  end
   filename = ARGV[0] || 'CaseFolding.txt'
-print_case_folding_data(filename)
+  data = CaseFolding.load(filename)
+  if dest
+    open(dest, "wb") do |f|
+      data.display(f)
+    end
+  else
+    data.display(STDOUT)
+  end
+end

enc/unicode/casefold.h

@@ -1,5 +1,5 @@
 /* DO NOT EDIT THIS FILE. */
-/* Generated by tool/CaseFolding.py */
+/* Generated by enc/unicode/case-folding.rb */
 
 static const CaseFold_11_Type CaseFold[] = {
   {0x0041, {1, {0x0061}}},