Jiahao@15: #!/usr/bin/env ruby
jbe@7: 
jbe@7: #  This file was used to generate the 'unicode_data.c' file by parsing the
jbe@7: #  Unicode data file 'UnicodeData.txt' of the Unicode Character Database.
jbe@7: #  It is included for informational purposes only and not intended for
jbe@7: #  production use.
jbe@7: 
jbe@7: 
jbe@10: #  Copyright (c) 2009 Public Software Group e. V., Berlin, Germany
jbe@7: #
jbe@7: #  Permission is hereby granted, free of charge, to any person obtaining a
jbe@7: #  copy of this software and associated documentation files (the "Software"),
jbe@7: #  to deal in the Software without restriction, including without limitation
jbe@7: #  the rights to use, copy, modify, merge, publish, distribute, sublicense,
jbe@7: #  and/or sell copies of the Software, and to permit persons to whom the
jbe@7: #  Software is furnished to do so, subject to the following conditions:
jbe@7: #
jbe@7: #  The above copyright notice and this permission notice shall be included in
jbe@7: #  all copies or substantial portions of the Software.
jbe@7: #
jbe@7: #  THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
jbe@7: #  IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
jbe@7: #  FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
jbe@7: #  AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
jbe@7: #  LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
jbe@7: #  FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
jbe@7: #  DEALINGS IN THE SOFTWARE.
jbe@7: 
jbe@7: 
jbe@7: #  This file contains derived data from a modified version of the
jbe@7: #  Unicode data files. The following license applies to that data:
jbe@7: #
jbe@7: #  COPYRIGHT AND PERMISSION NOTICE
jbe@7: #
jbe@7: #  Copyright (c) 1991-2007 Unicode, Inc. All rights reserved. Distributed
jbe@7: #  under the Terms of Use in http://www.unicode.org/copyright.html.
jbe@7: #
jbe@7: #  Permission is hereby granted, free of charge, to any person obtaining a
jbe@7: #  copy of the Unicode data files and any associated documentation (the "Data
jbe@7: #  Files") or Unicode software and any associated documentation (the
jbe@7: #  "Software") to deal in the Data Files or Software without restriction,
jbe@7: #  including without limitation the rights to use, copy, modify, merge,
jbe@7: #  publish, distribute, and/or sell copies of the Data Files or Software, and
jbe@7: #  to permit persons to whom the Data Files or Software are furnished to do
jbe@7: #  so, provided that (a) the above copyright notice(s) and this permission
jbe@7: #  notice appear with all copies of the Data Files or Software, (b) both the
jbe@7: #  above copyright notice(s) and this permission notice appear in associated
jbe@7: #  documentation, and (c) there is clear notice in each modified Data File or
jbe@7: #  in the Software as well as in the documentation associated with the Data
jbe@7: #  File(s) or Software that the data or software has been modified.
jbe@7: #
jbe@7: #  THE DATA FILES AND SOFTWARE ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY
jbe@7: #  KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
jbe@7: #  MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT OF
jbe@7: #  THIRD PARTY RIGHTS. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR HOLDERS
jbe@7: #  INCLUDED IN THIS NOTICE BE LIABLE FOR ANY CLAIM, OR ANY SPECIAL INDIRECT OR
jbe@7: #  CONSEQUENTIAL DAMAGES, OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF
jbe@7: #  USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
jbe@7: #  TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
jbe@7: #  PERFORMANCE OF THE DATA FILES OR SOFTWARE.
jbe@7: #
jbe@7: #  Except as contained in this notice, the name of a copyright holder shall
jbe@7: #  not be used in advertising or otherwise to promote the sale, use or other
jbe@7: #  dealings in these Data Files or Software without prior written
jbe@7: #  authorization of the copyright holder.
jbe@7: 
jbe@7: 
Jiahao@15: $ignorable_list = File.read("DerivedCoreProperties.txt")[/# Derived Property: Default_Ignorable_Code_Point.*?# Total code points:/m]
jbe@7: $ignorable = []
Jiahao@15: $ignorable_list.each_line do |entry|
jbe@8:   if entry =~ /^([0-9A-F]+)\.\.([0-9A-F]+)/
jbe@7:     $1.hex.upto($2.hex) { |e2| $ignorable << e2 }
jbe@7:   elsif entry =~ /^[0-9A-F]+/
jbe@7:     $ignorable << $&.hex
jbe@7:   end
jbe@7: end
jbe@7: 
Jiahao@15: $grapheme_extend_list = File.read("DerivedCoreProperties.txt")[/# Derived Property: Grapheme_Extend.*?# Total code points:/m]
jbe@7: $grapheme_extend = []
Jiahao@15: $grapheme_extend_list.each_line do |entry|
jbe@8:   if entry =~ /^([0-9A-F]+)\.\.([0-9A-F]+)/
jbe@7:     $1.hex.upto($2.hex) { |e2| $grapheme_extend << e2 }
jbe@7:   elsif entry =~ /^[0-9A-F]+/
jbe@7:     $grapheme_extend << $&.hex
jbe@7:   end
jbe@7: end
jbe@7: 
Jiahao@15: $exclusions = File.read("CompositionExclusions.txt")[/# \(1\) Script Specifics.*?# Total code points:/m]
jbe@7: $exclusions = $exclusions.chomp.split("\n").collect { |e| e.hex }
jbe@7: 
Jiahao@15: $excl_version = File.read("CompositionExclusions.txt")[/# \(2\) Post Composition Version precomposed characters.*?# Total code points:/m]
jbe@7: $excl_version = $excl_version.chomp.split("\n").collect { |e| e.hex }
jbe@7: 
Jiahao@15: $case_folding_string = File.open("CaseFolding.txt").read
jbe@7: 
jbe@7: $case_folding = {}
jbe@7: $case_folding_string.chomp.split("\n").each do |line|
jbe@7:   next unless line =~ /([0-9A-F]+); [CFS]; ([0-9A-F ]+);/i
jbe@7:   $case_folding[$1.hex] = $2.split(" ").collect { |e| e.hex }
jbe@7: end
jbe@7: 
jbe@7: $int_array = []
jbe@7: $int_array_indicies = {}
jbe@7: 
jbe@7: def str2c(string, prefix)
jbe@7:   return "0" if string.nil?
jbe@7:   return "UTF8PROC_#{prefix}_#{string.upcase}"
jbe@7: end
jbe@7: def ary2c(array)
jbe@7:   return "NULL" if array.nil?
jbe@7:   unless $int_array_indicies[array]
jbe@7:     $int_array_indicies[array] = $int_array.length
jbe@7:     array.each { |entry| $int_array << entry }
jbe@7:     $int_array << -1
jbe@7:   end
jbe@7:   return "utf8proc_sequences + #{$int_array_indicies[array]}"
jbe@7: end
jbe@7: 
jbe@7: class UnicodeChar
jbe@7:   attr_accessor :code, :name, :category, :combining_class, :bidi_class,
jbe@7:                 :decomp_type, :decomp_mapping,
jbe@7:                 :bidi_mirrored,
jbe@7:                 :uppercase_mapping, :lowercase_mapping, :titlecase_mapping
jbe@7:   def initialize(line)
jbe@7:     raise "Could not parse input." unless line =~ /^
jbe@7:       ([0-9A-F]+);        # code
jbe@7:       ([^;]+);            # name
jbe@7:       ([A-Z]+);           # general category
jbe@7:       ([0-9]+);           # canonical combining class
jbe@7:       ([A-Z]+);           # bidi class
jbe@7:       (<([A-Z]*)>)?       # decomposition type
jbe@7:       ((\ ?[0-9A-F]+)*);  # decompomposition mapping
jbe@7:       ([0-9]*);           # decimal digit
jbe@7:       ([0-9]*);           # digit
jbe@7:       ([^;]*);            # numeric
jbe@7:       ([YN]*);            # bidi mirrored
jbe@7:       ([^;]*);            # unicode 1.0 name
jbe@7:       ([^;]*);            # iso comment
jbe@7:       ([0-9A-F]*);        # simple uppercase mapping
jbe@7:       ([0-9A-F]*);        # simple lowercase mapping
jbe@7:       ([0-9A-F]*)$/ix     # simple titlecase mapping
jbe@7:     @code              = $1.hex
jbe@7:     @name              = $2
jbe@7:     @category          = $3
jbe@7:     @combining_class   = Integer($4)
jbe@7:     @bidi_class        = $5
jbe@7:     @decomp_type       = $7
jbe@7:     @decomp_mapping    = ($8=='') ? nil :
jbe@7:                          $8.split.collect { |element| element.hex }
jbe@7:     @bidi_mirrored     = ($13=='Y') ? true : false
jbe@7:     @uppercase_mapping = ($16=='') ? nil : $16.hex
jbe@7:     @lowercase_mapping = ($17=='') ? nil : $17.hex
jbe@7:     @titlecase_mapping = ($18=='') ? nil : $18.hex
jbe@7:   end
jbe@7:   def case_folding
jbe@7:     $case_folding[code]
jbe@7:   end
jbe@7:   def c_entry(comb1_indicies, comb2_indicies)
jbe@7:     "  " <<
jbe@7:     "{#{str2c category, 'CATEGORY'}, #{combining_class}, " <<
jbe@7:     "#{str2c bidi_class, 'BIDI_CLASS'}, " <<
jbe@7:     "#{str2c decomp_type, 'DECOMP_TYPE'}, " <<
jbe@7:     "#{ary2c decomp_mapping}, " <<
jbe@7:     "#{bidi_mirrored}, " <<
jbe@7:     "#{uppercase_mapping or -1}, " <<
jbe@7:     "#{lowercase_mapping or -1}, " <<
jbe@7:     "#{titlecase_mapping or -1}, " <<
jbe@7:     "#{comb1_indicies[code] ?
jbe@7:        (comb1_indicies[code]*comb2_indicies.keys.length) : -1
jbe@7:       }, #{comb2_indicies[code] or -1}, " <<
jbe@7:     "#{$exclusions.include?(code) or $excl_version.include?(code)}, " <<
jbe@7:     "#{$ignorable.include?(code)}, " <<
jbe@7:     "#{%W[Zl Zp Cc Cf].include?(category) and not [0x200C, 0x200D].include?(category)}, " <<
jbe@7:     "#{$grapheme_extend.include?(code)}, " <<
jbe@7:     "#{ary2c case_folding}},\n"
jbe@7:   end
jbe@7: end
jbe@7: 
jbe@7: chars = []
jbe@7: char_hash = {}
jbe@7: 
jbe@7: while gets
jbe@7:   if $_ =~ /^([0-9A-F]+);<[^;>,]+, First>;/i
jbe@7:     first = $1.hex
jbe@7:     gets
jbe@7:     char = UnicodeChar.new($_)
jbe@7:     raise "No last character of sequence found." unless
jbe@7:       $_ =~ /^([0-9A-F]+);<([^;>,]+), Last>;/i
jbe@7:     last = $1.hex
jbe@7:     name = "<#{$2}>"
jbe@7:     for i in first..last
jbe@7:       char_clone = char.clone
jbe@7:       char_clone.code = i
jbe@7:       char_clone.name = name
jbe@7:       char_hash[char_clone.code] = char_clone
jbe@7:       chars << char_clone
jbe@7:     end
jbe@7:   else
jbe@7:     char = UnicodeChar.new($_)
jbe@7:     char_hash[char.code] = char
jbe@7:     chars << char
jbe@7:   end
jbe@7: end
jbe@7: 
jbe@7: comb1st_indicies = {}
jbe@7: comb2nd_indicies = {}
jbe@7: comb_array = []
jbe@7: 
jbe@7: chars.each do |char|
Jiahao@15:   if !char.nil? and char.decomp_type.nil? and char.decomp_mapping and
Jiahao@15:       char.decomp_mapping.length == 2 and !char_hash[char.decomp_mapping[0]].nil? and
jbe@7:       char_hash[char.decomp_mapping[0]].combining_class == 0 and
jbe@7:       not $exclusions.include?(char.code)
jbe@7:     unless comb1st_indicies[char.decomp_mapping[0]]
jbe@7:       comb1st_indicies[char.decomp_mapping[0]] = comb1st_indicies.keys.length
jbe@7:     end
jbe@7:     unless comb2nd_indicies[char.decomp_mapping[1]]
jbe@7:       comb2nd_indicies[char.decomp_mapping[1]] = comb2nd_indicies.keys.length
jbe@7:     end
jbe@7:     comb_array[comb1st_indicies[char.decomp_mapping[0]]] ||= []
jbe@7:     raise "Duplicate canonical mapping" if
jbe@7:       comb_array[comb1st_indicies[char.decomp_mapping[0]]][
jbe@7:       comb2nd_indicies[char.decomp_mapping[1]]]
jbe@7:     comb_array[comb1st_indicies[char.decomp_mapping[0]]][
jbe@7:       comb2nd_indicies[char.decomp_mapping[1]]] = char.code
jbe@7:   end
jbe@7: end
jbe@7: 
jbe@7: properties_indicies = {}
jbe@7: properties = []
jbe@7: chars.each do |char|
jbe@7:   c_entry = char.c_entry(comb1st_indicies, comb2nd_indicies)
jbe@7:   unless properties_indicies[c_entry]
jbe@7:     properties_indicies[c_entry] = properties.length
jbe@7:     properties << c_entry
jbe@7:   end
jbe@7: end
jbe@7: 
jbe@7: stage1 = []
jbe@7: stage2 = []
jbe@7: for code in 0...0x110000
jbe@7:   next unless code % 0x100 == 0
jbe@7:   stage2_entry = []
jbe@7:   for code2 in code...(code+0x100)
jbe@7:     if char_hash[code2]
jbe@7:       stage2_entry << (properties_indicies[char_hash[code2].c_entry(
jbe@7:         comb1st_indicies, comb2nd_indicies)] + 1)
jbe@7:     else
jbe@7:       stage2_entry << 0
jbe@7:     end
jbe@7:   end
jbe@7:   old_index = stage2.index(stage2_entry)
jbe@7:   if old_index
jbe@7:     stage1 << (old_index * 0x100)
jbe@7:   else
jbe@7:     stage1 << (stage2.length * 0x100)
jbe@7:     stage2 << stage2_entry
jbe@7:   end
jbe@7: end
jbe@7: 
jbe@7: $stdout << "const int32_t utf8proc_sequences[] = {\n  "
jbe@7: i = 0
jbe@7: $int_array.each do |entry|
jbe@7:   i += 1
jbe@7:   if i == 8
jbe@7:     i = 0
jbe@7:     $stdout << "\n  "
jbe@7:   end
jbe@7:   $stdout << entry << ", "
jbe@7: end
jbe@7: $stdout << "};\n\n"
jbe@7: 
jbe@7: $stdout << "const uint16_t utf8proc_stage1table[] = {\n  "
jbe@7: i = 0
jbe@7: stage1.each do |entry|
jbe@7:   i += 1
jbe@7:   if i == 8
jbe@7:     i = 0
jbe@7:     $stdout << "\n  "
jbe@7:   end
jbe@7:   $stdout << entry << ", "
jbe@7: end
jbe@7: $stdout << "};\n\n"
jbe@7: 
jbe@7: $stdout << "const uint16_t utf8proc_stage2table[] = {\n  "
jbe@7: i = 0
jbe@7: stage2.flatten.each do |entry|
jbe@7:   i += 1
jbe@7:   if i == 8
jbe@7:     i = 0
jbe@7:     $stdout << "\n  "
jbe@7:   end
jbe@7:   $stdout << entry << ", "
jbe@7: end
jbe@7: $stdout << "};\n\n"
jbe@7: 
jbe@7: $stdout << "const utf8proc_property_t utf8proc_properties[] = {\n"
jbe@7: $stdout << "  {0, 0, 0, 0, NULL, false, -1, -1, -1, -1, -1, false},\n"
jbe@7: properties.each { |line|
jbe@7:   $stdout << line
jbe@7: }
jbe@7: $stdout << "};\n\n"
jbe@7: 
jbe@7: $stdout << "const int32_t utf8proc_combinations[] = {\n  "
jbe@7: i = 0
jbe@7: comb1st_indicies.keys.each_index do |a|
jbe@7:   comb2nd_indicies.keys.each_index do |b|
jbe@7:     i += 1
jbe@7:     if i == 8
jbe@7:       i = 0
jbe@7:       $stdout << "\n  "
jbe@7:     end
jbe@7:     $stdout << ( comb_array[a][b] or -1 ) << ", "
jbe@7:   end
jbe@7: end
jbe@7: $stdout << "};\n\n"
jbe@7: