u 0.5.0

data/ext/encoding/character/utf-8/data/generate-unicode-data.rb

@@ -0,0 +1,1070 @@
+#! /usr/bin/ruby -w
+=begin
+  :contents: Generate Unicode table headers.
+  :arch-tag: 98c7456d-c7d9-4b40-9971-409428593ad5
+
+  Copyright (C) 2004 Nikolai Weibull <source@pcppopper.org>
+
+  This program is free software; you can redistribute it and/or modify
+  it under the terms of the GNU General Public License as published by
+  the Free Software Foundation; either version 2 of the License, or
+  (at your option) any later version.
+  
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+  
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+=end
+
+
+
+def error(fmt, *args)
+  $stderr.printf("%s: %s\n", File.basename($0), sprintf(fmt, *args))
+  exit(1)
+end
+
+class File
+  def self.process(path)
+    begin
+      File.open(path) do |file|
+	file.each_line do |line|
+	  next if line =~ /^(#|\s*$)/
+	  yield line
+	end
+      end
+    rescue IOError => e
+      error("I/O error while processing input:\n" +
+	    "    file: %s\n" +
+	    "    error: %s\n", path, e.message)
+    end
+  end
+end
+
+class String
+  def escape
+    self.unpack('H*')[0].gsub(/(.{2})/, '\\x\1')
+  end
+
+  def width
+    self.gsub(/\t/, ' ' * 8).length
+  end
+end
+
+class Array
+  def verify_size(wanted, path, index)
+    if !(wanted === self.size)
+      error("entry doesn't contain the required %s fields:\n" +
+	    "    file: %s\n" +
+	    "    entry: %s\n" +
+	    "    field count: %d\n",
+	    wanted.to_s,
+	    path,
+	    (self.size > index) ? self[index] : 'N/A',
+	    self.size)
+    end
+  end
+
+  def verify_field(index, code, path, raw_code, type, ccase)
+    if self[index].to_i(16) != code
+      error("entry has type %s but UCD_%s(%s) != %s:\n" +
+	    "    file: %s\n" +
+	    "    entry: %s\n",
+	    type, ccase, raw_code, raw_code, path, raw_code)
+    end
+  end
+end
+
+class Hash
+  def enumerate_ordered
+    n = 0
+    self.keys.sort.each do |code|
+      if self[code] == 1
+	self.delete(code)
+	next
+      end
+      self[code] = n
+      n += 1
+    end
+    n
+  end
+end
+
+# XXX: this is too memory consuming to keep like this.  We need to split it up
+# like the perl script does in hashes and arrays.  Argh!
+class UnicodeCodepoint
+  def initialize(code)
+    @code = code
+    @type = @value = @lower = @upper = @cclass = @compat = nil
+    @compositions = @decompositions = @break_props = nil
+  end
+
+  attr_accessor :code
+  attr_accessor :type, :value, :lower, :upper, :cclass, :compat
+  attr_accessor :compositions, :decompositions, :break_props
+end
+
+# XXX: cleanup
+class CollectedData
+  def initialize(dir = '.', indent = "\t")
+    @dir = dir
+    @indent = indent 
+    @cps = []
+
+    @excludes = nil
+
+    @pages_before_e0000 = 0
+    @last = 0x10ffff
+
+    @type = []
+    @value = []
+    @title_to_lower = {}
+    @title_to_upper = {}
+    @cclass = []
+    @decompose_compat = []
+    @compositions = {}
+    @decompositions = []
+
+    @break_props = []
+
+    @special_case_offsets = []
+    @special_cases = []
+
+    @casefold = []
+    @casefold_longest = -1
+
+    @bidimirror = []
+  end
+
+  attr :dir
+  attr :indent
+  attr :cps, true
+  attr :excludes, true
+  attr :pages_before_e0000, true
+  attr :last
+  attr_accessor :type, :value, :title_to_lower, :title_to_upper, :cclass,
+    :decompose_compat, :compositions, :decompositions
+  attr :break_props, true
+  attr :special_case_offsets, true
+  attr :special_cases, true
+  attr :casefold, true
+  attr :casefold_longest, true
+  attr :bidimirror, true
+end
+
+class CompositionExclusions
+  def process(data)
+    data.excludes = Hash.new
+    File.process(File.join(data.dir, 'CompositionExclusions.txt')) do |line|
+      data.excludes[line.chomp.sub(/^\s*(.*?)\s*(#.*)?$/,'\1').to_i(16)] = true
+    end
+  end
+end
+
+class UnicodeData
+  CODE, NAME, CATEGORY, COMBINING_CLASSES, BIDI_CATEGORY,
+    DECOMPOSITION, DECIMAL_VALUE, DIGIT_VALUE, NUMERIC_VALUE, MIRRORED,
+    OLD_NAME, COMMENT, UPPER, LOWER, TITLE = (0..14).to_a
+
+  def process(data)
+    prev_code = -1
+    path = File.join(data.dir, 'UnicodeData.txt')
+    File.process(path) do |line|
+      fields = line.chomp.split(/;/, -1)
+      fields.verify_size(15, path, CODE)
+      code = fields[CODE].to_i(16)
+
+      if code >= 0xe0000 and prev_code < 0xe0000
+	data.pages_before_e0000 = (prev_code >> 8) + 1
+      end
+
+      if code > prev_code + 1
+	process_gap(data,
+		    prev_code + 1,
+		    code - 1,
+		    fields[NAME] =~ /Last>$/ ? fields : new_gap_fields)
+      end
+      process_one(data, code, fields)
+      prev_code = code
+    end
+    process_gap(data, prev_code + 1, 0x10ffff, new_gap_fields)
+  end
+
+private
+
+  def new_gap_fields
+    ['', '', 'Cn', '0', '', '', '', '', '', '', '', '', '', '', '']
+  end
+
+  def process_gap(data, low, hi, fields)
+    low.upto(hi) do |i|
+      fields[CODE] = sprintf('%04x', i)
+      process_one(data, i, fields)
+    end
+  end
+
+  def process_one(data, code, fields)
+#    puts(code.to_s)
+#    data.cps[code] ||= UnicodeCodepoint.new(code)
+    data.type[code] = fields[CATEGORY]
+
+    # TODO: Why not process things like 'Nl'?
+    case data.type[code]
+    when 'Nd'
+      data.value[code] = fields[DECIMAL_VALUE].to_i
+    when 'Ll'
+      data.value[code] = fields[UPPER].to_i(16)
+    when 'Lu'
+      data.value[code] = fields[LOWER].to_i(16)
+    when 'Lt'
+      data.title_to_lower[code] = fields[LOWER].to_i(16)
+      data.title_to_upper[code] = fields[UPPER].to_i(16)
+    end
+
+    data.cclass[code] = fields[COMBINING_CLASSES]
+
+    unless fields[DECOMPOSITION] == ''
+      if fields[DECOMPOSITION] =~ /^\<.*\>\s*(.*)/
+	data.decompose_compat[code] = true
+	fields[DECOMPOSITION] = $1
+      else
+	data.decompose_compat[code] = false
+	unless data.excludes.include?(code)
+	  data.compositions[code] = fields[DECOMPOSITION]
+	end
+      end
+      data.decompositions[code] = fields[DECOMPOSITION]
+    end
+  end
+end
+
+class LineBreak
+  BREAK_CODE, BREAK_PROPERTY = (0..1).to_a
+
+  def process(data)
+    prev_code = -1
+    path = File.join(data.dir, 'LineBreak.txt')
+    File.process(path) do |line|
+      fields = line.chomp.sub(/\s*#.*/, '').split(/;/, -1)
+      fields.verify_size(2, path, BREAK_CODE)
+
+      if fields[BREAK_CODE] =~ /([0-9A-F]{4,6})\.\.([0-9A-F]{4,6})/
+	start_code, end_code = $1.to_i(16), $2.to_i(16)
+      else
+	start_code = end_code = fields[BREAK_CODE].to_i(16)
+      end
+
+      if start_code > prev_code + 1
+	process_gap(data, prev_code + 1, start_code - 1)
+      end
+
+      start_code.upto(end_code) do |i|
+	data.break_props[i] = fields[BREAK_PROPERTY]
+      end
+
+      prev_code = end_code
+    end
+
+    process_gap(data, prev_code + 1, 0x10ffff)
+  end
+
+private
+
+  def process_gap(data, low, hi)
+    low.upto(hi) do |i|
+      data.break_props[i] = (data.type[i] == 'Cn') ? 'XX' : 'AL'
+    end
+  end
+end
+
+class SpecialCasing
+  CASE_CODE, CASE_LOWER, CASE_TITLE, CASE_UPPER, CASE_CONDITION = (0..4).to_a
+
+  def initialize
+    @offset = 0
+  end
+
+  def process(data)
+    path = File.join(data.dir, 'SpecialCasing.txt')
+    File.process(path) do |line|
+      fields = line.chomp.sub(/\s*#.*/, '').split(/\s*;\s*/, -1)
+      fields.verify_size((5..6), path, CASE_CODE)
+      raw_code, code = fields[CASE_CODE], fields[CASE_CODE].to_i(16)
+      unless data.type[code].nil?
+	# We ignore conditional special cases
+	next if fields.size == 6
+
+	case data.type[code]
+	when 'Lu'
+	  fields.verify_field(CASE_UPPER, code, path, raw_code, 'Lu', 'Upper')
+	  add_special_case(data, code, data.value[code],
+			   fields[CASE_LOWER], fields[CASE_TITLE])
+	when 'Lt'
+	  fields.verify_field(CASE_TITLE, code, path, raw_code, 'Lt', 'Title')
+	  add_special_case(data, code, nil,
+			   fields[CASE_LOWER], fields[CASE_UPPER])
+	when 'Ll'
+	  fields.verify_field(CASE_LOWER, code, path, raw_code, 'Ll', 'Lower')
+	  add_special_case(data, code, data.value[code],
+			   fields[CASE_UPPER], fields[CASE_TITLE])
+	else
+	  error("special case for non-alphabetic code point:\n" +
+		"    file: %s\n" +
+		"    type: %s\n" +
+		"    code point/entry: %s\n",
+		path, data.type[code], raw_code)
+	end
+      else
+	error("special case for code point which doesn't have a type:\n" +
+	      "    file: %s\n" +
+	      "    code point/entry: %d\n",
+	      path, code)
+      end
+    end
+  end
+
+private
+
+  def add_special_case(data, code, single, field1, field2)
+    values = [
+      single.nil? ? nil : [single],
+      field1.split(/\s+/).map{ |s| s.to_i(16) },
+      [0],
+      field2.split(/\s+/).map{ |s| s.to_i(16) },
+    ]
+    result = ''
+    values.each{ |value| result += value.pack('U*') unless value.nil? }
+
+    data.special_case_offsets.push(@offset)
+    data.value[code] = 0x1000000 + @offset
+    data.special_cases.push(result.escape)
+    @offset += 1 + result.length
+  end
+end
+
+class CaseFolding
+  FOLDING_CODE, FOLDING_STATUS, FOLDING_MAPPING = (0..2).to_a
+
+  def process(data)
+    path = File.join(data.dir, 'CaseFolding.txt')
+    File.process(path) do |line|
+      fields = line.chomp.sub(/\s*#.*/, '').split(/\s*;\s*/, -1)
+      fields.verify_size(4, path, FOLDING_CODE)
+
+      # skip Simple and Turkic rules
+      next if fields[FOLDING_STATUS] =~ /^[ST]$/
+
+      raw_code, code = fields[FOLDING_CODE], fields[FOLDING_CODE].to_i(16)
+      values = fields[FOLDING_MAPPING].split(/\s+/).map{ |s| s.to_i(16) }
+      if values.size == 1 &&
+	!(!data.value[code].nil? && data.value[code] >= 0x1000000) &&
+	!data.type[code].nil?
+	case data.type[code]
+	when 'Ll'
+	  lower = code
+	when 'Lt'
+	  lower = data.title_to_lower[code]
+	when 'Lu'
+	  lower = data.value[code]
+	else
+	  lower = code
+	end
+	next if lower == values[0]
+      end
+
+      string = values.pack('U*')
+      if string.length + 1 > data.casefold_longest
+	data.casefold_longest = string.length + 1
+      end
+      data.casefold.push([code, string.escape])
+    end
+  end
+end
+
+class BidiMirroring
+  def process(data)
+    path = File.join(data.dir, 'BidiMirroring.txt')
+    File.process(path) do |line|
+      fields = line.chomp.sub(/\s*#.*/, '').split(/\s*;\s*/, -1)
+      fields.verify_size(2, path, 0)
+      data.bidimirror.push([fields[0].to_i(16), fields[1].to_i(16)])
+    end
+  end
+end
+
+class Printer
+  def initialize
+    @index = 0
+  end
+
+  def process(data)
+    @last_char_part1_i = data.pages_before_e0000 * 256 - 1
+    @last_char_part1_x = sprintf('0x%04x', @last_char_part1_i)
+    @last_char_part1_X = sprintf('%04X', @last_char_part1_i)
+    print_tables(data)
+    print_decomp(data)
+    print_composition_table(data)
+    print_line_break(data)
+  end
+
+private
+
+  # Map general category code onto symbolic name.
+  Mappings = {
+    # Normative.
+    'Lu' => 'UNICODE_UPPERCASE_LETTER',
+    'Ll' => 'UNICODE_LOWERCASE_LETTER',
+    'Lt' => 'UNICODE_TITLECASE_LETTER',
+    'Mn' => 'UNICODE_NON_SPACING_MARK',
+    'Mc' => 'UNICODE_COMBINING_MARK',
+    'Me' => 'UNICODE_ENCLOSING_MARK',
+    'Nd' => 'UNICODE_DECIMAL_NUMBER',
+    'Nl' => 'UNICODE_LETTER_NUMBER',
+    'No' => 'UNICODE_OTHER_NUMBER',
+    'Zs' => 'UNICODE_SPACE_SEPARATOR',
+    'Zl' => 'UNICODE_LINE_SEPARATOR',
+    'Zp' => 'UNICODE_PARAGRAPH_SEPARATOR',
+    'Cc' => 'UNICODE_CONTROL',
+    'Cf' => 'UNICODE_FORMAT',
+    'Cs' => 'UNICODE_SURROGATE',
+    'Co' => 'UNICODE_PRIVATE_USE',
+    'Cn' => 'UNICODE_UNASSIGNED',
+
+    # Informative.
+    'Lm' => 'UNICODE_MODIFIER_LETTER',
+    'Lo' => 'UNICODE_OTHER_LETTER',
+    'Pc' => 'UNICODE_CONNECT_PUNCTUATION',
+    'Pd' => 'UNICODE_DASH_PUNCTUATION',
+    'Ps' => 'UNICODE_OPEN_PUNCTUATION',
+    'Pe' => 'UNICODE_CLOSE_PUNCTUATION',
+    'Pi' => 'UNICODE_INITIAL_PUNCTUATION',
+    'Pf' => 'UNICODE_FINAL_PUNCTUATION',
+    'Po' => 'UNICODE_OTHER_PUNCTUATION',
+    'Sm' => 'UNICODE_MATH_SYMBOL',
+    'Sc' => 'UNICODE_CURRENCY_SYMBOL',
+    'Sk' => 'UNICODE_MODIFIER_SYMBOL',
+    'So' => 'UNICODE_OTHER_SYMBOL'
+  }
+
+  BreakMappings = {
+    'BK' => 'UNICODE_BREAK_MANDATORY',
+    'CR' => 'UNICODE_BREAK_CARRIAGE_RETURN',
+    'LF' => 'UNICODE_BREAK_LINE_FEED',
+    'CM' => 'UNICODE_BREAK_COMBINING_MARK',
+    'SG' => 'UNICODE_BREAK_SURROGATE',
+    'ZW' => 'UNICODE_BREAK_ZERO_WIDTH_SPACE',
+    'IN' => 'UNICODE_BREAK_INSEPARABLE',
+    'GL' => 'UNICODE_BREAK_NON_BREAKING_GLUE',
+    'CB' => 'UNICODE_BREAK_CONTINGENT',
+    'SP' => 'UNICODE_BREAK_SPACE',
+    'BA' => 'UNICODE_BREAK_AFTER',
+    'BB' => 'UNICODE_BREAK_BEFORE',
+    'B2' => 'UNICODE_BREAK_BEFORE_AND_AFTER',
+    'HY' => 'UNICODE_BREAK_HYPHEN',
+    'NS' => 'UNICODE_BREAK_NON_STARTER',
+    'OP' => 'UNICODE_BREAK_OPEN_PUNCTUATION',
+    'CL' => 'UNICODE_BREAK_CLOSE_PUNCTUATION',
+    'QU' => 'UNICODE_BREAK_QUOTATION',
+    'EX' => 'UNICODE_BREAK_EXCLAMATION',
+    'ID' => 'UNICODE_BREAK_IDEOGRAPHIC',
+    'NU' => 'UNICODE_BREAK_NUMERIC',
+    'IS' => 'UNICODE_BREAK_INFIX_SEPARATOR',
+    'SY' => 'UNICODE_BREAK_SYMBOL',
+    'AL' => 'UNICODE_BREAK_ALPHABETIC',
+    'PR' => 'UNICODE_BREAK_PREFIX',
+    'PO' => 'UNICODE_BREAK_POSTFIX',
+    'SA' => 'UNICODE_BREAK_COMPLEX_CONTEXT',
+    'AI' => 'UNICODE_BREAK_AMBIGUOUS',
+    'NL' => 'UNICODE_BREAK_NEXT_LINE',
+    'WJ' => 'UNICODE_BREAK_WORD_JOINER',
+    'XX' => 'UNICODE_BREAK_UNKNOWN',
+    'JL' => 'UNICODE_BREAK_HANGUL_L_JAMO',
+    'JV' => "UNICODE_BREAK_HANGUL_V_JAMO",
+    'JT' => "UNICODE_BREAK_HANGUL_T_JAMO",
+    'H2' => "UNICODE_BREAK_HANGUL_LV_SYLLABLE",
+    'H3' => "UNICODE_BREAK_HANGUL_LVT_SYLLABLE"
+  };
+
+  NOT_PRESENT_OFFSET = 65535
+
+  def print_table(data, low, mid, hi, size, header, part1_h, part2_h, &f)
+    @index = 0
+    rows = []
+    print(header)
+    low.step(hi, 256) do |i|
+      rows[i / 256] = print_row(data, i, size){ |i| f.call(i) }
+    end
+    print("\n};\n")
+    print(part1_h)
+    low.step(mid, 256) do |i|
+      printf("%s%s,\n", data.indent, rows[i / 256])
+    end
+    print("};\n")
+    if mid != hi
+      print(part2_h)
+      0xe0000.step(hi, 256) do |i|
+	printf("%s%s,\n", data.indent, rows[i / 256])
+      end
+      print("};\n")
+    end
+  end
+
+  def print_tables(data, outfile = 'character-tables.h')
+    row = []
+    saved_stdout = $stdout
+    File.open(outfile, 'w') do |file|
+      header_h = outfile.upcase.gsub(/[^A-Z0-9]/, '_')
+      $stdout = file
+      print <<EOF
+/* Automatically generated file */
+
+#ifndef #{header_h}
+#define #{header_h}
+
+#define UNICODE_DATA_VERSION "#{UnicodeVersion}"
+
+#define UNICODE_LAST_CHAR #{sprintf('0x%04x', data.last)}
+
+#define UNICODE_MAX_TABLE_INDEX 10000
+
+#define UNICODE_LAST_CHAR_PART1 #{@last_char_part1_x}
+
+#define UNICODE_LAST_PAGE_PART1 #{data.pages_before_e0000 - 1}
+
+#define UNICODE_FIRST_CHAR_PART2 0xe0000
+
+#define UNICODE_SPECIAL_CASE_TABLE_START 0x1000000
+EOF
+      print_table(data, 0, @last_char_part1_i, data.last, 1,
+		  <<EOH, <<EOH1, <<EOH2){ |i| Mappings[data.type[i]] }
+
+
+static const char type_data[][256] = {
+EOH
+
+
+/* U+0000 through U+#{@last_char_part1_X} */
+static const int16_t type_table_part1[#{data.pages_before_e0000}] = {
+EOH1
+
+
+/* U+E0000 through U+#{sprintf('%04X', data.last)} */
+static const int16_t type_table_part2[768] = {
+EOH2
+
+      print_table(data, 0, @last_char_part1_i, data.last, 4,
+		  <<EOH, <<EOH1, <<EOH2) { |i| data.value[i].nil? ? '0x0000' : sprintf('0x%04x', data.value[i]) }
+
+
+static const unichar attr_data[][256] = {
+EOH
+
+
+/* U+0000 through U+#{@last_char_part1_X} */
+static const int16_t attr_table_part1[#{data.pages_before_e0000}] = {
+EOH1
+
+
+/* U+E0000 through U+#{sprintf('%04X', data.last)} */
+static const int16_t attr_table_part2[768] = {
+EOH2
+
+      print <<EOF
+
+
+static const unichar title_table[][3] = {
+EOF
+      data.title_to_lower.keys.sort.each do |code|
+	printf("%s{ 0x%04x, 0x%04x, 0x%04x },\n", data.indent,
+	       code, data.title_to_upper[code], data.title_to_lower[code])
+      end
+      print("};\n")
+
+      print_special_case_table(data)
+      print_case_fold_table(data)
+
+      print <<EOF
+static const struct {
+#{data.indent}unichar ch;
+#{data.indent}unichar mirrored_ch;
+} bidi_mirroring_table[] = {
+EOF
+      data.bidimirror.each do |item|
+	printf("%s{ 0x%04x, 0x%04x },\n", data.indent, item[0], item[1])
+      end
+      print <<EOF
+};
+
+#endif /* #{header_h} */
+EOF
+    end
+    $stdout = saved_stdout
+  end
+
+  def print_row(data, start, type_size)
+    flag = true
+    values = []
+    0.upto(255) do |i|
+      values[i] = yield(start + i)
+      flag = false if values[i] != values[0]
+    end
+    return values[0] + " + UNICODE_MAX_TABLE_INDEX" if flag
+
+    puts(',') if @index != 0
+    printf("%s{ /* page %d, index %d */\n%s",
+	   data.indent, start / 256, @index, data.indent * 2)
+    column = data.indent.width * 2
+    start.upto(start + 255) do |i|
+      text = values[i - start]
+      if text.length + column + 2 > 79
+	printf("\n%s", data.indent * 2)
+	column = data.indent.width * 2
+      end
+
+      printf("%s, ", text)
+      column += text.width + 2
+    end
+
+    print("\n#{data.indent}}")
+    @index += 1
+    return sprintf("%d /* page %d */", @index - 1, start / 256);
+  end
+
+  def print_special_case_table(data)
+    print <<EOF
+
+
+/*
+ * Table of special cases for case conversion; each record contains
+ * First, the best single character mapping to lowercase if Lu,
+ * and to uppercase if Ll, followed by the output mapping for the two cases
+ * other than the case of the codepoint, in the order Ll, Lu, Lt, encoded in
+ * UTF-8, separated and terminated by a NUL character.
+ */
+static const char special_case_table[] = {
+EOF
+    data.special_cases.each_with_index do |sc, i|
+      printf(%Q< "%s\\0" /* offset %d */\n>, sc, data.special_case_offsets[i])
+    end
+    print <<EOF
+};
+
+EOF
+  end
+
+  def print_case_fold_table(data)
+    print <<EOF
+
+/*
+ * Table of casefolding cases that can't be derived by lowercasing.
+ */
+static const struct {
+#{data.indent}uint16_t ch;
+#{data.indent}char data[#{data.casefold_longest}];
+} casefold_table[] = {
+EOF
+    data.casefold.sort_by{ |a| a[0] }.each do |cf|
+      if cf[0] > 0xffff
+	error('casefold_table.ch field too short.' +
+	      '  Upgrade to unichar to fit values beyond 0xffff.')
+      end
+      printf(%Q<%s{ 0x%04x, "%s" },\n>, data.indent, cf[0], cf[1])
+    end
+    print <<EOF
+};
+EOF
+  end
+
+  def print_decomp(data, outfile = 'decompose.h')
+    row = []
+    saved_stdout = $stdout
+    File.open(outfile, 'w') do |file|
+      header_h = outfile.upcase.gsub(/[^A-Z0-9]/, '_')
+      $stdout = file
+      print <<EOF
+/* Automatically generated file */
+
+#ifndef #{header_h}
+#define #{header_h}
+
+
+#define UNICODE_LAST_CHAR #{sprintf('0x%04x', data.last)}
+
+#define UNICODE_MAX_TABLE_INDEX (0x110000 / 256)
+
+#define UNICODE_LAST_CHAR_PART1 #{@last_char_part1_x}
+
+#define UNICODE_LAST_PAGE_PART1 #{data.pages_before_e0000 - 1}
+
+#define UNICODE_FIRST_CHAR_PART2 0xe0000
+
+#define UNICODE_NOT_PRESENT_OFFSET #{NOT_PRESENT_OFFSET}
+EOF
+      print_table(data, 0, @last_char_part1_i, data.last, 1,
+		  <<EOH, <<EOH1, <<EOH2){ |i| data.cclass[i] }
+
+
+static const uint8_t cclass_data[][256] = {
+EOH
+
+
+static const int16_t combining_class_table_part1[#{data.pages_before_e0000}] = {
+EOH1
+
+
+static const int16_t combining_class_table_part2[768] = {
+EOH2
+
+      print <<EOL
+
+
+static const struct {
+#{data.indent}unichar ch;
+#{data.indent}uint16_t canon_offset;
+#{data.indent}uint16_t compat_offset;
+} decomp_table[] = {
+EOL
+      decomp_offsets = {}
+      decomp_string = ''
+      @decomp_string_offset = 0
+      0.upto(data.last) do |i|
+	unless data.decompositions[i].nil?
+	  canon_decomp = data.decompose_compat[i] ?
+	    nil : make_decomp(data, i, false)
+	  compat_decomp = make_decomp(data, i, true)
+	  if not canon_decomp.nil? and compat_decomp == canon_decomp
+	    compat_decomp = nil
+	  end
+	  canon_offset = handle_decomp(canon_decomp, decomp_offsets,
+				       decomp_string)
+	  compat_offset = handle_decomp(compat_decomp, decomp_offsets,
+					decomp_string)
+
+	  if @decomp_string_offset > NOT_PRESENT_OFFSET
+	    error('decomposition string offset beyond not-present-offset,' +
+		  " upgrade value:\n" +
+		  "    offset: %d\n" +
+		  "    max: %d\n",
+		  @decomp_string_offset, NOT_PRESENT_OFFSET)
+	  end
+	  printf("%s{ 0x%04x, %s, %s },\n",
+		 data.indent, i, canon_offset, compat_offset)
+	end
+      end
+      print("\n};")
+
+      print <<EOL
+
+static const char decomp_expansion_string[] = #{decomp_string};
+
+
+#endif /* #{header_h} */
+EOL
+    end
+    $stdout = saved_stdout
+  end
+
+  def expand_decomp(data, code, compat)
+    ary = []
+    data.decompositions[code].split(/ /).each do |item|
+      pos = item.to_i(16)
+      if not data.decompositions[pos].nil? and
+	(compat or not data.decompose_compat[pos])
+	ary.concat(expand_decomp(data, pos, compat))
+      else
+	ary.push(pos)
+      end
+    end
+    ary
+  end
+
+  def make_decomp(data, code, compat)
+    str = ''
+    expand_decomp(data, code, compat).each do |item|
+      str += item.is_a?(Array) ? item.flatten.pack('U') : [item].pack('U')
+    end
+    str
+  end
+
+  def handle_decomp(decomp, decomp_offsets,
+		    decomp_string)
+    offset = 'UNICODE_NOT_PRESENT_OFFSET'
+    unless decomp.nil?
+      if decomp_offsets.member?(decomp)
+	offset = decomp_offsets[decomp]
+      else
+	offset = @decomp_string_offset
+	decomp_offsets[decomp] = offset
+	decomp_string << ("\n  \"" + decomp.escape +
+			  "\\0\" /* offset #{offset} */")
+	@decomp_string_offset += decomp.length + 1
+      end
+    end
+    offset
+  end
+
+  def print_composition_table(data, outfile = 'compose.h')
+    first = Hash.new(0)
+    second = Hash.new(0)
+
+    data.compositions.each do |code, value|
+      values = value.split(/\s+/).map{ |s| s.to_i(16) }
+
+      # skip non-starters and single-character decompositions
+      if data.cclass[values[0]] != '0' or values.size == 1
+	data.compositions.delete(code)
+	next
+      end
+
+      if values.size != 2
+	error("decomposition of entry contains more than two elements:\n" +
+	      "    entry: %d\n" +
+	      "    elements: %d\n",
+	      code, values.size)
+      end
+
+      first[values[0]] += 1
+    end
+    
+    n_first = first.enumerate_ordered
+
+    data.compositions.each do |code, value|
+      values = value.split(/\s+/).map{ |s| s.to_i(16) }
+
+      second[values[1]] += 1 if first.member?(values[0])
+    end
+
+    n_second = second.enumerate_ordered
+
+    first_singletons = []
+    second_singletons = []
+    reverse = {}
+    data.compositions.each do |code, value|
+      values = value.split(/\s+/).map{ |s| s.to_i(16) }
+
+      if first.member?(values[0]) and second.member?(values[1])
+	reverse["#{first[values[0]]}|#{second[values[1]]}"] = code
+      elsif not first.member?(values[0])
+	first_singletons.push([values[0], values[1], code])
+      else
+	second_singletons.push([values[1], values[0], code])
+      end
+    end
+
+    first_singletons = first_singletons.sort_by{ |a| a[0] }
+    second_singletons = second_singletons.sort_by{ |a| a[0] }
+
+    row = []
+    saved_stdout = $stdout
+    File.open(outfile, 'w') do |file|
+      header_h = outfile.upcase.gsub(/[^A-Z0-9]/, '_')
+      $stdout = file
+      values = {}
+      total = first_start = 1
+      last = 0
+
+      first.each do |code, value|
+	values[code] = value + total
+	last = code if code > last
+      end
+      total += n_first
+
+      first_single_start = total
+      first_singletons.each_with_index do |item, i|
+	code = item[0]
+	values[code] = i + total
+	last = code if code > last
+      end
+      total += first_singletons.size
+
+      second_start = total
+      second.each do |code, value|
+	values[code] = value + total
+	last = code if code > last
+      end
+      total += n_second
+
+      second_single_start = total
+      second_singletons.each_with_index do |item, i|
+	code = item[0]
+	values[code] = i + total
+	last = code if code > last
+      end
+
+      print <<EOL
+/* Automatically generated file */
+
+#ifndef #{header_h}
+#define #{header_h}
+
+
+#define COMPOSE_FIRST_START #{first_start}
+#define COMPOSE_FIRST_SINGLE_START #{first_single_start}
+#define COMPOSE_SECOND_START #{second_start}
+#define COMPOSE_SECOND_SINGLE_START #{second_single_start}
+#define COMPOSE_TABLE_LAST #{last / 256}
+EOL
+
+      print_table(data, 0, last, last, 2,
+		  <<EOH, <<EOH1, nil){ |i| values.member?(i) ? values[i].to_s : '0' }
+
+
+static const uint16_t compose_data[][256] = {
+EOH
+
+
+static const int16_t compose_table[COMPOSE_TABLE_LAST + 1] = {
+EOH1
+
+      print <<EOL
+
+
+static const uint16_t compose_first_single[][2] = {
+EOL
+      first_singletons.each_with_index do |item, i|
+	if item[1] > 0xffff or item[2] > 0xffff
+	  error("compose_first_single table field too short." +
+		"  Upgrade to unichar to fit values beyond 0xffff.")
+	end
+	printf("%s{ %#06x, %#06x },\n", data.indent, item[1], item[2])
+      end
+      print("};\n")
+
+      print <<EOL
+
+
+static const uint16_t compose_second_single[][2] = {
+EOL
+      second_singletons.each_with_index do |item, i|
+	if item[1] > 0xffff or item[2] > 0xffff
+	  error("compose_second_single table field too short." +
+		"  Upgrade to unichar to fit values beyond 0xffff.")
+	end
+	printf("%s{ %#06x, %#06x },\n", data.indent, item[1], item[2])
+      end
+      print("};\n")
+
+      print <<EOL
+
+
+static const uint16_t compose_array[#{n_first}][#{n_second}] = {
+EOL
+      0.upto(n_first - 1) do |i|
+	printf("%s{\n%s", data.indent, data.indent * 2)
+	column = data.indent.width * 2
+	0.upto(n_second - 1) do |j|
+	  if column + 8 > 79
+	    printf("\n%s", data.indent * 2)
+	    column = data.indent.width * 2
+	  end
+	  if reverse.member?("#{i}|#{j}")
+	    if reverse["#{i}|#{j}"] > 0xffff
+	      error("compose_array table field too short." +
+		    "  Upgrade to unichar to fit values beyond 0xffff.")
+	    end
+	    printf("0x%04x, ", reverse["#{i}|#{j}"])
+	  else
+	    print("     0, ")
+	  end
+	  column += 8
+	end
+	printf("\n%s},\n", data.indent)
+      end
+      print("};\n")
+
+      print <<EOL
+
+
+#endif /* #{header_h} */
+EOL
+    end
+    $stdout = saved_stdout
+  end
+
+  def print_line_break(data, outfile = 'break.h')
+    row = []
+    saved_stdout = $stdout
+    File.open(outfile, 'w') do |file|
+      header_h = outfile.upcase.gsub(/[^A-Z0-9]/, '_')
+      $stdout = file
+      print <<EOF
+/* Automatically generated file */
+
+#ifndef #{header_h}
+#define #{header_h}
+
+#define UNICODE_DATA_VERSION "#{UnicodeVersion}"
+
+#define UNICODE_LAST_CHAR #{sprintf('0x%04x', data.last)}
+
+#define UNICODE_MAX_TABLE_INDEX 10000
+
+/*
+ * The last code point that should be looked up in break_property_table_part1.
+ */
+#define UNICODE_LAST_CHAR_PART1 #{@last_char_part1_x}
+
+/*
+ * The first code point that should be looked up in break_property_table_part2.
+ */
+#define UNICODE_FIRST_CHAR_PART2 0xe0000
+EOF
+      print_table(data, 0, @last_char_part1_i, data.last, 1,
+		  <<EOH, <<EOH1, <<EOH2){ |i| BreakMappings[data.break_props[i]] }
+
+
+static const int8_t break_property_data[][256] = {
+EOH
+
+
+/* U+0000 through U+#{@last_char_part1_X} */
+static const int16_t break_property_table_part1[#{data.pages_before_e0000}] = {
+EOH1
+
+
+/* U+E0000 through U+#{sprintf('%04X', data.last)} */
+static const int16_t break_property_table_part2[768] = {
+EOH2
+
+      print <<EOF
+
+
+#endif /* #{header_h} */
+EOF
+    end
+    $stdout = saved_stdout
+  end
+end
+
+UnicodeVersion = ARGV[0]
+
+class Runner
+  def main
+    check_for_data_files(ARGV[1])
+    data = CollectedData.new(ARGV[1], "\t")
+    [CompositionExclusions, UnicodeData, LineBreak,
+      SpecialCasing, CaseFolding, BidiMirroring, Printer].each do |klass|
+      klass.new.process(data)
+    end
+  end
+
+private
+  def check_for_data_files(dir)
+    ['UnicodeData.txt', 'LineBreak.txt', 'SpecialCasing.txt', 'CaseFolding.txt',
+      'CompositionExclusions.txt', 'BidiMirroring.txt'].each do |file|
+      path = File.join(dir, file)
+      unless File.readable?(path)
+	error('missing required file: %s', path)
+      end
+    end
+  end
+end
+
+Runner.new.main
+
+
+
+# vim: set sts=2 sw=2: