minori: dep/utf8proc/data/data_generator.rb annotate

annotate dep/utf8proc/data/data_generator.rb @ 355:83e6ba09338f

gui/pages/seasons: fix function disambiguation on win32

author	Paper <paper@paper.us.eu.org>
date	Sun, 14 Jul 2024 23:46:18 -0400
parents	1faa72660932
children

rev	line source
343 1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	1 #!/usr/bin/env ruby
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	2
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	3 # This file was used to generate the 'unicode_data.c' file by parsing the
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	4 # Unicode data file 'UnicodeData.txt' of the Unicode Character Database.
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	5 # It is included for informational purposes only and not intended for
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	6 # production use.
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	7
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	8
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	9 # Copyright (c) 2018 Steven G. Johnson, Tony Kelman, Keno Fischer,
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	10 # Benito van der Zander, Michaël Meyer, and other contributors.
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	11 # Copyright (c) 2009 Public Software Group e. V., Berlin, Germany
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	12 #
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	13 # Permission is hereby granted, free of charge, to any person obtaining a
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	14 # copy of this software and associated documentation files (the "Software"),
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	15 # to deal in the Software without restriction, including without limitation
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	16 # the rights to use, copy, modify, merge, publish, distribute, sublicense,
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	17 # and/or sell copies of the Software, and to permit persons to whom the
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	18 # Software is furnished to do so, subject to the following conditions:
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	19 #
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	20 # The above copyright notice and this permission notice shall be included in
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	21 # all copies or substantial portions of the Software.
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	22 #
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	23 # THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	24 # IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	25 # FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	26 # AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	27 # LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	28 # FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	29 # DEALINGS IN THE SOFTWARE.
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	30
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	31
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	32 # This file contains derived data from a modified version of the
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	33 # Unicode data files. The following license applies to that data:
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	34 #
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	35 # COPYRIGHT AND PERMISSION NOTICE
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	36 #
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	37 # Copyright (c) 1991-2007 Unicode, Inc. All rights reserved. Distributed
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	38 # under the Terms of Use in http://www.unicode.org/copyright.html.
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	39 #
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	40 # Permission is hereby granted, free of charge, to any person obtaining a
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	41 # copy of the Unicode data files and any associated documentation (the "Data
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	42 # Files") or Unicode software and any associated documentation (the
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	43 # "Software") to deal in the Data Files or Software without restriction,
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	44 # including without limitation the rights to use, copy, modify, merge,
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	45 # publish, distribute, and/or sell copies of the Data Files or Software, and
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	46 # to permit persons to whom the Data Files or Software are furnished to do
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	47 # so, provided that (a) the above copyright notice(s) and this permission
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	48 # notice appear with all copies of the Data Files or Software, (b) both the
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	49 # above copyright notice(s) and this permission notice appear in associated
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	50 # documentation, and (c) there is clear notice in each modified Data File or
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	51 # in the Software as well as in the documentation associated with the Data
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	52 # File(s) or Software that the data or software has been modified.
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	53 #
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	54 # THE DATA FILES AND SOFTWARE ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	55 # KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	56 # MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT OF
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	57 # THIRD PARTY RIGHTS. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR HOLDERS
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	58 # INCLUDED IN THIS NOTICE BE LIABLE FOR ANY CLAIM, OR ANY SPECIAL INDIRECT OR
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	59 # CONSEQUENTIAL DAMAGES, OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	60 # USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	61 # TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	62 # PERFORMANCE OF THE DATA FILES OR SOFTWARE.
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	63 #
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	64 # Except as contained in this notice, the name of a copyright holder shall
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	65 # not be used in advertising or otherwise to promote the sale, use or other
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	66 # dealings in these Data Files or Software without prior written
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	67 # authorization of the copyright holder.
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	68
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	69
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	70 $ignorable_list = File.read("DerivedCoreProperties.txt", :encoding => 'utf-8')[/# Derived Property: Default_Ignorable_Code_Point.*?# Total code points:/m]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	71 $ignorable = []
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	72 $ignorable_list.each_line do \|entry\|
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	73 if entry =~ /^([0-9A-F]+)\.\.([0-9A-F]+)/
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	74 $1.hex.upto($2.hex) { \|e2\| $ignorable << e2 }
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	75 elsif entry =~ /^[0-9A-F]+/
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	76 $ignorable << $&.hex
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	77 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	78 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	79
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	80 $uppercase_list = File.read("DerivedCoreProperties.txt", :encoding => 'utf-8')[/# Derived Property: Uppercase.*?# Total code points:/m]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	81 $uppercase = []
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	82 $uppercase_list.each_line do \|entry\|
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	83 if entry =~ /^([0-9A-F]+)\.\.([0-9A-F]+)/
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	84 $1.hex.upto($2.hex) { \|e2\| $uppercase << e2 }
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	85 elsif entry =~ /^[0-9A-F]+/
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	86 $uppercase << $&.hex
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	87 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	88 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	89
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	90 $lowercase_list = File.read("DerivedCoreProperties.txt", :encoding => 'utf-8')[/# Derived Property: Lowercase.*?# Total code points:/m]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	91 $lowercase = []
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	92 $lowercase_list.each_line do \|entry\|
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	93 if entry =~ /^([0-9A-F]+)\.\.([0-9A-F]+)/
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	94 $1.hex.upto($2.hex) { \|e2\| $lowercase << e2 }
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	95 elsif entry =~ /^[0-9A-F]+/
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	96 $lowercase << $&.hex
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	97 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	98 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	99
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	100 $icb_linker_list = File.read("DerivedCoreProperties.txt", :encoding => 'utf-8')[/# Indic_Conjunct_Break=Linker.*?# Total code points:/m]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	101 $icb = Hash.new("UTF8PROC_INDIC_CONJUNCT_BREAK_NONE")
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	102 $icb_linker_list.each_line do \|entry\|
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	103 if entry =~ /^([0-9A-F]+)\.\.([0-9A-F]+)/
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	104 $1.hex.upto($2.hex) { \|e2\| $icb[e2] = "UTF8PROC_INDIC_CONJUNCT_BREAK_LINKER" }
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	105 elsif entry =~ /^[0-9A-F]+/
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	106 $icb[$&.hex] = "UTF8PROC_INDIC_CONJUNCT_BREAK_LINKER"
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	107 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	108 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	109 $icb_consonant_list = File.read("DerivedCoreProperties.txt", :encoding => 'utf-8')[/# Indic_Conjunct_Break=Consonant.*?# Total code points:/m]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	110 $icb_consonant_list.each_line do \|entry\|
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	111 if entry =~ /^([0-9A-F]+)\.\.([0-9A-F]+)/
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	112 $1.hex.upto($2.hex) { \|e2\| $icb[e2] = "UTF8PROC_INDIC_CONJUNCT_BREAK_CONSONANT" }
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	113 elsif entry =~ /^[0-9A-F]+/
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	114 $icb[$&.hex] = "UTF8PROC_INDIC_CONJUNCT_BREAK_CONSONANT"
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	115 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	116 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	117 $icb_extend_list = File.read("DerivedCoreProperties.txt", :encoding => 'utf-8')[/# Indic_Conjunct_Break=Extend.*?# Total code points:/m]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	118 $icb_extend_list.each_line do \|entry\|
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	119 if entry =~ /^([0-9A-F]+)\.\.([0-9A-F]+)/
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	120 $1.hex.upto($2.hex) { \|e2\| $icb[e2] = "UTF8PROC_INDIC_CONJUNCT_BREAK_EXTEND" }
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	121 elsif entry =~ /^[0-9A-F]+/
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	122 $icb[$&.hex] = "UTF8PROC_INDIC_CONJUNCT_BREAK_EXTEND"
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	123 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	124 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	125
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	126 $grapheme_boundclass_list = File.read("GraphemeBreakProperty.txt", :encoding => 'utf-8')
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	127 $grapheme_boundclass = Hash.new("UTF8PROC_BOUNDCLASS_OTHER")
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	128 $grapheme_boundclass_list.each_line do \|entry\|
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	129 if entry =~ /^([0-9A-F]+)\.\.([0-9A-F]+)\s;\s([A-Za-z_]+)/
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	130 $1.hex.upto($2.hex) { \|e2\| $grapheme_boundclass[e2] = "UTF8PROC_BOUNDCLASS_" + $3.upcase }
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	131 elsif entry =~ /^([0-9A-F]+)\s;\s([A-Za-z_]+)/
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	132 $grapheme_boundclass[$1.hex] = "UTF8PROC_BOUNDCLASS_" + $2.upcase
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	133 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	134 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	135
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	136 $emoji_data_list = File.read("emoji-data.txt", :encoding => 'utf-8')
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	137 $emoji_data_list.each_line do \|entry\|
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	138 if entry =~ /^([0-9A-F]+)\.\.([0-9A-F]+)\s;\sExtended_Pictographic\W/
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	139 $1.hex.upto($2.hex) { \|e2\| $grapheme_boundclass[e2] = "UTF8PROC_BOUNDCLASS_EXTENDED_PICTOGRAPHIC" }
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	140 elsif entry =~ /^([0-9A-F]+)\s;\sExtended_Pictographic\W/
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	141 $grapheme_boundclass[$1.hex] = "UTF8PROC_BOUNDCLASS_EXTENDED_PICTOGRAPHIC"
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	142 elsif entry =~ /^([0-9A-F]+)\.\.([0-9A-F]+)\s;\sEmoji_Modifier\W/
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	143 $1.hex.upto($2.hex) { \|e2\| $grapheme_boundclass[e2] = "UTF8PROC_BOUNDCLASS_EXTEND" }
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	144 elsif entry =~ /^([0-9A-F]+)\s;\sEmoji_Modifier\W/
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	145 $grapheme_boundclass[$1.hex] = "UTF8PROC_BOUNDCLASS_EXTEND"
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	146 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	147 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	148
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	149 $charwidth_list = File.read("CharWidths.txt", :encoding => 'utf-8')
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	150 $charwidth = Hash.new(0)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	151 $charwidth_list.each_line do \|entry\|
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	152 if entry =~ /^([0-9A-F]+)\.\.([0-9A-F]+)\s;\s([0-9]+)/
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	153 $1.hex.upto($2.hex) { \|e2\| $charwidth[e2] = $3.to_i }
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	154 elsif entry =~ /^([0-9A-F]+)\s;\s([0-9]+)/
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	155 $charwidth[$1.hex] = $2.to_i
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	156 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	157 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	158
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	159 $exclusions = File.read("CompositionExclusions.txt", :encoding => 'utf-8')[/# $1$ Script Specifics.*?# Total code points:/m]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	160 $exclusions = $exclusions.chomp.split("\n").collect { \|e\| e.hex }
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	161
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	162 $excl_version = File.read("CompositionExclusions.txt", :encoding => 'utf-8')[/# $2$ Post Composition Version precomposed characters.*?# Total code points:/m]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	163 $excl_version = $excl_version.chomp.split("\n").collect { \|e\| e.hex }
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	164
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	165 $case_folding_string = File.read("CaseFolding.txt", :encoding => 'utf-8')
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	166 $case_folding = {}
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	167 $case_folding_string.chomp.split("\n").each do \|line\|
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	168 next unless line =~ /([0-9A-F]+); [CF]; ([0-9A-F ]+);/i
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	169 $case_folding[$1.hex] = $2.split(" ").collect { \|e\| e.hex }
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	170 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	171
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	172 $int_array = []
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	173 $int_array_indicies = {}
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	174
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	175 def str2c(string, prefix)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	176 return "0" if string.nil?
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	177 return "UTF8PROC_#{prefix}_#{string.upcase}"
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	178 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	179 def pushary(array)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	180 idx = $int_array_indicies[array]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	181 unless idx
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	182 $int_array_indicies[array] = $int_array.length
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	183 idx = $int_array.length
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	184 array.each { \|entry\| $int_array << entry }
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	185 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	186 return idx
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	187 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	188 def cpary2utf16encoded(array)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	189 return array.flat_map { \|cp\|
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	190 if (cp <= 0xFFFF)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	191 raise "utf-16 code: #{cp}" if cp & 0b1111100000000000 == 0b1101100000000000
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	192 cp
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	193 else
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	194 temp = cp - 0x10000
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	195 [(temp >> 10) \| 0b1101100000000000, (temp & 0b0000001111111111) \| 0b1101110000000000]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	196 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	197 }
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	198 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	199 def cpary2c(array)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	200 return "UINT16_MAX" if array.nil? \|\| array.length == 0
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	201 lencode = array.length - 1 #no sequence has len 0, so we encode len 1 as 0, len 2 as 1, ...
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	202 array = cpary2utf16encoded(array)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	203 if lencode >= 3 #we have only 2 bits for the length
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	204 array = [lencode] + array
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	205 lencode = 3
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	206 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	207 idx = pushary(array)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	208 raise "Array index out of bound" if idx > 0x3FFF
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	209 return "#{idx \| (lencode << 14)}"
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	210 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	211 def singlecpmap(cp)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	212 return "UINT16_MAX" if cp == nil
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	213 idx = pushary(cpary2utf16encoded([cp]))
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	214 raise "Array index out of bound" if idx > 0xFFFF
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	215 return "#{idx}"
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	216 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	217
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	218 class UnicodeChar
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	219 attr_accessor :code, :name, :category, :combining_class, :bidi_class,
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	220 :decomp_type, :decomp_mapping,
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	221 :bidi_mirrored,
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	222 :uppercase_mapping, :lowercase_mapping, :titlecase_mapping,
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	223 #caches:
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	224 :c_entry_index, :c_decomp_mapping, :c_case_folding
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	225 def initialize(line)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	226 raise "Could not parse input." unless line =~ /^
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	227 ([0-9A-F]+); # code
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	228 ([^;]+); # name
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	229 ([A-Z]+); # general category
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	230 ([0-9]+); # canonical combining class
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	231 ([A-Z]+); # bidi class
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	232 (<([A-Z]*)>)? # decomposition type
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	233 ((\ ?[0-9A-F]+)*); # decompomposition mapping
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	234 ([0-9]*); # decimal digit
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	235 ([0-9]*); # digit
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	236 ([^;]*); # numeric
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	237 ([YN]*); # bidi mirrored
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	238 ([^;]*); # unicode 1.0 name
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	239 ([^;]*); # iso comment
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	240 ([0-9A-F]*); # simple uppercase mapping
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	241 ([0-9A-F]*); # simple lowercase mapping
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	242 ([0-9A-F]*)$/ix # simple titlecase mapping
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	243 @code = $1.hex
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	244 @name = $2
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	245 @category = $3
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	246 @combining_class = Integer($4)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	247 @bidi_class = $5
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	248 @decomp_type = $7
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	249 @decomp_mapping = ($8=='') ? nil :
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	250 $8.split.collect { \|element\| element.hex }
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	251 @bidi_mirrored = ($13=='Y') ? true : false
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	252 # issue #130: use nonstandard uppercase ß -> ẞ
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	253 # issue #195: if character is uppercase but has no lowercase mapping,
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	254 # then make lowercase mapping = itself (vice versa for lowercase)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	255 @uppercase_mapping = ($16=='') ? (code==0x00df ? 0x1e9e : ($17=='' && $lowercase.include?(code) ? code : nil)) : $16.hex
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	256 @lowercase_mapping = ($17=='') ? ($16=='' && $uppercase.include?(code) ? code : nil) : $17.hex
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	257 @titlecase_mapping = ($18=='') ? (code==0x00df ? 0x1e9e : nil) : $18.hex
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	258 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	259 def case_folding
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	260 $case_folding[code]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	261 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	262 def c_entry(comb_indicies)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	263 " " <<
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	264 "{#{str2c category, 'CATEGORY'}, #{combining_class}, " <<
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	265 "#{str2c bidi_class, 'BIDI_CLASS'}, " <<
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	266 "#{str2c decomp_type, 'DECOMP_TYPE'}, " <<
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	267 "#{c_decomp_mapping}, " <<
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	268 "#{c_case_folding}, " <<
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	269 "#{singlecpmap uppercase_mapping }, " <<
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	270 "#{singlecpmap lowercase_mapping }, " <<
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	271 "#{singlecpmap titlecase_mapping }, " <<
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	272 "#{comb_indicies[code] ? comb_indicies[code]: 'UINT16_MAX'}, " <<
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	273 "#{bidi_mirrored}, " <<
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	274 "#{$exclusions.include?(code) or $excl_version.include?(code)}, " <<
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	275 "#{$ignorable.include?(code)}, " <<
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	276 "#{%W[Zl Zp Cc Cf].include?(category) and not [0x200C, 0x200D].include?(category)}, " <<
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	277 "#{$charwidth[code]}, 0, " <<
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	278 "#{$grapheme_boundclass[code]}, " <<
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	279 "#{$icb[code]}},\n"
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	280 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	281 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	282
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	283 chars = []
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	284 char_hash = {}
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	285
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	286 while gets
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	287 if $_ =~ /^([0-9A-F]+);<[^;>,]+, First>;/i
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	288 first = $1.hex
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	289 gets
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	290 char = UnicodeChar.new($_)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	291 raise "No last character of sequence found." unless
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	292 $_ =~ /^([0-9A-F]+);<([^;>,]+), Last>;/i
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	293 last = $1.hex
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	294 name = "<#{$2}>"
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	295 for i in first..last
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	296 char_clone = char.clone
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	297 char_clone.code = i
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	298 char_clone.name = name
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	299 char_hash[char_clone.code] = char_clone
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	300 chars << char_clone
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	301 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	302 else
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	303 char = UnicodeChar.new($_)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	304 char_hash[char.code] = char
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	305 chars << char
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	306 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	307 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	308
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	309 comb1st_indicies = {}
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	310 comb2nd_indicies = {}
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	311 comb2nd_indicies_sorted_keys = []
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	312 comb2nd_indicies_nonbasic = {}
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	313 comb_array = []
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	314
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	315 chars.each do \|char\|
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	316 if !char.nil? and char.decomp_type.nil? and char.decomp_mapping and
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	317 char.decomp_mapping.length == 2 and !char_hash[char.decomp_mapping[0]].nil? and
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	318 char_hash[char.decomp_mapping[0]].combining_class == 0 and
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	319 not $exclusions.include?(char.code)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	320
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	321 dm0 = char.decomp_mapping[0]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	322 dm1 = char.decomp_mapping[1]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	323 unless comb1st_indicies[dm0]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	324 comb1st_indicies[dm0] = comb1st_indicies.keys.length
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	325 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	326 unless comb2nd_indicies[dm1]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	327 comb2nd_indicies_sorted_keys << dm1
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	328 comb2nd_indicies[dm1] = comb2nd_indicies.keys.length
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	329 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	330 comb_array[comb1st_indicies[dm0]] \|\|= []
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	331 raise "Duplicate canonical mapping: #{char.code} #{dm0} #{dm1}" if comb_array[comb1st_indicies[dm0]][comb2nd_indicies[dm1]]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	332 comb_array[comb1st_indicies[dm0]][comb2nd_indicies[dm1]] = char.code
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	333
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	334 comb2nd_indicies_nonbasic[dm1] = true if char.code > 0xFFFF
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	335 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	336 char.c_decomp_mapping = cpary2c(char.decomp_mapping)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	337 char.c_case_folding = cpary2c(char.case_folding)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	338 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	339
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	340 comb_indicies = {}
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	341 cumoffset = 0
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	342 comb1st_indicies_lastoffsets = []
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	343 comb1st_indicies_firstoffsets = []
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	344 comb1st_indicies.each do \|dm0, index\|
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	345 first = nil
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	346 last = nil
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	347 offset = 0
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	348 comb2nd_indicies_sorted_keys.each_with_index do \|dm1, b\|
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	349 if comb_array[index][b]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	350 first = offset unless first
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	351 last = offset
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	352 last += 1 if comb2nd_indicies_nonbasic[dm1]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	353 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	354 offset += 1
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	355 offset += 1 if comb2nd_indicies_nonbasic[dm1]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	356 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	357 comb1st_indicies_firstoffsets[index] = first
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	358 comb1st_indicies_lastoffsets[index] = last
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	359 raise "double index" if comb_indicies[dm0]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	360 comb_indicies[dm0] = cumoffset
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	361 cumoffset += last - first + 1 + 2
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	362 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	363
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	364 offset = 0
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	365 comb2nd_indicies_sorted_keys.each do \|dm1\|
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	366 raise "double index" if comb_indicies[dm1]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	367 comb_indicies[dm1] = 0x8000 \| (comb2nd_indicies[dm1] + offset)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	368 raise "too large comb index" if comb2nd_indicies[dm1] + offset > 0x4000
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	369 if comb2nd_indicies_nonbasic[dm1]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	370 comb_indicies[dm1] = comb_indicies[dm1] \| 0x4000
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	371 offset += 1
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	372 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	373 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	374
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	375 properties_indicies = {}
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	376 properties = []
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	377 chars.each do \|char\|
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	378 c_entry = char.c_entry(comb_indicies)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	379 char.c_entry_index = properties_indicies[c_entry]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	380 unless char.c_entry_index
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	381 properties_indicies[c_entry] = properties.length
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	382 char.c_entry_index = properties.length
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	383 properties << c_entry
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	384 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	385 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	386
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	387 stage1 = []
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	388 stage2 = []
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	389 for code in 0...0x110000
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	390 next unless code % 0x100 == 0
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	391 stage2_entry = []
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	392 for code2 in code...(code+0x100)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	393 if char_hash[code2]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	394 stage2_entry << (char_hash[code2].c_entry_index + 1)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	395 else
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	396 stage2_entry << 0
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	397 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	398 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	399 old_index = stage2.index(stage2_entry)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	400 if old_index
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	401 stage1 << (old_index * 0x100)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	402 else
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	403 stage1 << (stage2.length * 0x100)
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	404 stage2 << stage2_entry
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	405 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	406 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	407
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	408 $stdout << "static const utf8proc_uint16_t utf8proc_sequences[] = {\n "
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	409 i = 0
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	410 $int_array.each do \|entry\|
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	411 i += 1
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	412 if i == 8
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	413 i = 0
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	414 $stdout << "\n "
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	415 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	416 $stdout << entry << ", "
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	417 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	418 $stdout << "};\n\n"
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	419
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	420 $stdout << "static const utf8proc_uint16_t utf8proc_stage1table[] = {\n "
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	421 i = 0
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	422 stage1.each do \|entry\|
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	423 i += 1
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	424 if i == 8
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	425 i = 0
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	426 $stdout << "\n "
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	427 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	428 $stdout << entry << ", "
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	429 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	430 $stdout << "};\n\n"
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	431
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	432 $stdout << "static const utf8proc_uint16_t utf8proc_stage2table[] = {\n "
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	433 i = 0
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	434 stage2.flatten.each do \|entry\|
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	435 i += 1
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	436 if i == 8
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	437 i = 0
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	438 $stdout << "\n "
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	439 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	440 $stdout << entry << ", "
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	441 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	442 $stdout << "};\n\n"
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	443
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	444 $stdout << "static const utf8proc_property_t utf8proc_properties[] = {\n"
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	445 $stdout << " {0, 0, 0, 0, UINT16_MAX, UINT16_MAX, UINT16_MAX, UINT16_MAX, UINT16_MAX, UINT16_MAX, false,false,false,false, 1, 0, UTF8PROC_BOUNDCLASS_OTHER, UTF8PROC_INDIC_CONJUNCT_BREAK_NONE},\n"
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	446 properties.each { \|line\|
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	447 $stdout << line
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	448 }
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	449 $stdout << "};\n\n"
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	450
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	451
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	452
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	453 $stdout << "static const utf8proc_uint16_t utf8proc_combinations[] = {\n "
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	454 i = 0
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	455 comb1st_indicies.keys.each_index do \|a\|
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	456 offset = 0
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	457 $stdout << comb1st_indicies_firstoffsets[a] << ", " << comb1st_indicies_lastoffsets[a] << ", "
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	458 comb2nd_indicies_sorted_keys.each_with_index do \|dm1, b\|
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	459 break if offset > comb1st_indicies_lastoffsets[a]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	460 if offset >= comb1st_indicies_firstoffsets[a]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	461 i += 1
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	462 if i == 8
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	463 i = 0
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	464 $stdout << "\n "
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	465 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	466 v = comb_array[a][b] ? comb_array[a][b] : 0
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	467 $stdout << (( v & 0xFFFF0000 ) >> 16) << ", " if comb2nd_indicies_nonbasic[dm1]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	468 $stdout << (v & 0xFFFF) << ", "
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	469 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	470 offset += 1
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	471 offset += 1 if comb2nd_indicies_nonbasic[dm1]
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	472 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	473 $stdout << "\n"
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	474 end
1faa72660932 : transfer back to cmake from autotools Paper <paper@paper.us.eu.org>* parents: diff changeset	475 $stdout << "};\n\n"

Mercurial > minori

annotate dep/utf8proc/data/data_generator.rb @ 355:83e6ba09338f