"Fossies" - the Fresh Open Source Software Archive

Member "detox-1.4.5/table/unicode.tbl.sample" (15 Aug 2021, 13111 Bytes) of package /linux/privat/detox-1.4.5.tar.gz:


As a special service "Fossies" has tried to format the requested text file into HTML format (style: standard) with prefixed line numbers. Alternatively you can here view or download the uninterpreted source code file.

    1 # This file is part of the Detox package.
    2 #
    3 # Copyright (c) Doug Harple <detox.dharple@gmail.com>
    4 #
    5 # For the full copyright and license information, please view the LICENSE
    6 # file that was distributed with this source code.
    7 
    8 # 
    9 # Special thanks to: http://www.unicode.org/Public/UNIDATA/UnicodeData.txt 
   10 #
   11 # Used for translating Icelandic:
   12 # http://www.bartleby.com/65/th/Thingval.html
   13 # http://www.wordiq.com/definition/Thorn_%28letter%29
   14 # http://www.wordiq.com/definition/Edh
   15 #
   16 # Greenlandic:
   17 # http://std.dkuug.dk/cultreg/registrations/narrative/da_DK,_4.3.html
   18 #
   19 # Sami:
   20 # http://www.wordiq.com/definition/Sami_languages
   21 #
   22 # Germanic:
   23 # http://www.wordiq.com/definition/%DF
   24 # 
   25 
   26 default		_
   27 
   28 start
   29 
   30 #
   31 # Basic Latin - This is a subset of the basic 7-bit ASCII table, with unsafe
   32 # characters removed.
   33 #
   34 
   35 0x0023		#	# NUMBER SIGN
   36 0x0025		%	# PERCENT SIGN
   37 0x0026		_and_	# AMPERSAND
   38 0x002B		+	# PLUS SIGN
   39 0x002C		,	# COMMA
   40 0x002D		-	# HYPHEN-MINUS
   41 0x002E		.	# FULL STOP
   42 
   43 0x0030		0	# DIGIT ZERO
   44 0x0031		1	# DIGIT ONE
   45 0x0032		2	# DIGIT TWO
   46 0x0033		3	# DIGIT THREE
   47 0x0034		4	# DIGIT FOUR
   48 0x0035		5	# DIGIT FIVE
   49 0x0036		6	# DIGIT SIX
   50 0x0037		7	# DIGIT SEVEN
   51 0x0038		8	# DIGIT EIGHT
   52 0x0039		9	# DIGIT NINE
   53 0x003D		=	# EQUALS SIGN
   54 
   55 0x0041		A	# LATIN CAPITAL LETTER A
   56 0x0042		B	# LATIN CAPITAL LETTER B
   57 0x0043		C	# LATIN CAPITAL LETTER C
   58 0x0044		D	# LATIN CAPITAL LETTER D
   59 0x0045		E	# LATIN CAPITAL LETTER E
   60 0x0046		F	# LATIN CAPITAL LETTER F
   61 0x0047		G	# LATIN CAPITAL LETTER G
   62 0x0048		H	# LATIN CAPITAL LETTER H
   63 0x0049		I	# LATIN CAPITAL LETTER I
   64 0x004A		J	# LATIN CAPITAL LETTER J
   65 0x004B		K	# LATIN CAPITAL LETTER K
   66 0x004C		L	# LATIN CAPITAL LETTER L
   67 0x004D		M	# LATIN CAPITAL LETTER M
   68 0x004E		N	# LATIN CAPITAL LETTER N
   69 0x004F		O	# LATIN CAPITAL LETTER O
   70 
   71 0x0050		P	# LATIN CAPITAL LETTER P
   72 0x0051		Q	# LATIN CAPITAL LETTER Q
   73 0x0052		R	# LATIN CAPITAL LETTER R
   74 0x0053		S	# LATIN CAPITAL LETTER S
   75 0x0054		T	# LATIN CAPITAL LETTER T
   76 0x0055		U	# LATIN CAPITAL LETTER U
   77 0x0056		V	# LATIN CAPITAL LETTER V
   78 0x0057		W	# LATIN CAPITAL LETTER W
   79 0x0058		X	# LATIN CAPITAL LETTER X
   80 0x0059		Y	# LATIN CAPITAL LETTER Y
   81 0x005A		Z	# LATIN CAPITAL LETTER Z
   82 0x005E		^	# CIRCUMFLEX ACCENT
   83 
   84 0x0061		a	# LATIN SMALL LETTER A
   85 0x0062		b	# LATIN SMALL LETTER B
   86 0x0063		c	# LATIN SMALL LETTER C
   87 0x0064		d	# LATIN SMALL LETTER D
   88 0x0065		e	# LATIN SMALL LETTER E
   89 0x0066		f	# LATIN SMALL LETTER F
   90 0x0067		g	# LATIN SMALL LETTER G
   91 0x0068		h	# LATIN SMALL LETTER H
   92 0x0069		i	# LATIN SMALL LETTER I
   93 0x006A		j	# LATIN SMALL LETTER J
   94 0x006B		k	# LATIN SMALL LETTER K
   95 0x006C		l	# LATIN SMALL LETTER L
   96 0x006D		m	# LATIN SMALL LETTER M
   97 0x006E		n	# LATIN SMALL LETTER N
   98 0x006F		o	# LATIN SMALL LETTER O
   99 
  100 0x0070		p	# LATIN SMALL LETTER P
  101 0x0071		q	# LATIN SMALL LETTER Q
  102 0x0072		r	# LATIN SMALL LETTER R
  103 0x0073		s	# LATIN SMALL LETTER S
  104 0x0074		t	# LATIN SMALL LETTER T
  105 0x0075		u	# LATIN SMALL LETTER U
  106 0x0076		v	# LATIN SMALL LETTER V
  107 0x0077		w	# LATIN SMALL LETTER W
  108 0x0078		x	# LATIN SMALL LETTER X
  109 0x0079		y	# LATIN SMALL LETTER Y
  110 0x007A		z	# LATIN SMALL LETTER Z
  111 0x007E		~	# TILDE
  112 
  113 #
  114 # CP 1252
  115 #
  116 
  117 0x0085		...	# HORIZONTAL ELLIPSIS
  118 0x0088		^	# MODIFIER LETTER CIRCUMFLEX ACCENT
  119 0x008A		S	# LATIN CAPITAL LETTER S WITH CARON
  120 0x008C		OE	# LATIN CAPITAL LIGATURE OE
  121 
  122 0x0096		-	# EN DASH
  123 0x0097		-	# EM DASH
  124 0x0099		_tm_	# TRADE MARK SIGN
  125 0x009A		s	# LATIN SMALL LETTER S WITH CARON
  126 0x009C		oe	# LATIN SMALL LIGATURE OE
  127 0x009F		Y	# LATIN CAPITAL LETTER Y WITH DIAERESIS
  128 
  129 #
  130 # Latin 1
  131 #
  132 
  133 0x00A2		_cent_
  134 0x00A3		_pound_
  135 0x00A5		_yen_
  136 0x00A9		_copy_
  137 0x00AE		_reg_
  138 
  139 0x00B2		2
  140 0x00B3		3
  141 0x00B5		b
  142 0x00B6		_pp_
  143 0x00B9		1
  144 
  145 0x00C0		A	# LATIN CAPITAL LETTER A WITH GRAVE
  146 0x00C1		A	# LATIN CAPITAL LETTER A WITH ACUTE
  147 0x00C2		A	# LATIN CAPITAL LETTER A WITH CIRCUMFLEX
  148 0x00C3		A	# LATIN CAPITAL LETTER A WITH TILDE
  149 0x00C4		A	# LATIN CAPITAL LETTER A WITH DIAERESIS
  150 0x00C5		A	# LATIN CAPITAL LETTER A WITH RING ABOVE
  151 0x00C6		AE	# LATIN CAPITAL LETTER AE
  152 0x00C7		C	# LATIN CAPITAL LETTER C WITH CEDILLA
  153 0x00C8		E	# LATIN CAPITAL LETTER E WITH GRAVE
  154 0x00C9		E	# LATIN CAPITAL LETTER E WITH ACUTE
  155 0x00CA		E	# LATIN CAPITAL LETTER E WITH CIRCUMFLEX
  156 0x00CB		E	# LATIN CAPITAL LETTER E WITH DIAERESIS
  157 0x00CC		I	# LATIN CAPITAL LETTER I WITH GRAVE
  158 0x00CD		I	# LATIN CAPITAL LETTER I WITH ACUTE
  159 0x00CE		I	# LATIN CAPITAL LETTER I WITH CIRCUMFLEX
  160 0x00CF		I	# LATIN CAPITAL LETTER I WITH DIAERESIS
  161 
  162 0x00D0		TH	# LATIN CAPITAL LETTER ETH (Icelandic)
  163 0x00D1		N	# LATIN CAPITAL LETTER N WITH TILDE
  164 0x00D2		O	# LATIN CAPITAL LETTER O WITH GRAVE
  165 0x00D3		O	# LATIN CAPITAL LETTER O WITH ACUTE
  166 0x00D4		O	# LATIN CAPITAL LETTER O WITH CIRCUMFLEX
  167 0x00D5		O	# LATIN CAPITAL LETTER O WITH TILDE
  168 0x00D6		O	# LATIN CAPITAL LETTER O WITH DIAERESIS
  169 0x00D7		x	# MULTIPLICATION SIGN
  170 0x00D8		O	# LATIN CAPITAL LETTER O WITH STROKE
  171 0x00D9		U	# LATIN CAPITAL LETTER U WITH GRAVE
  172 0x00DA		U	# LATIN CAPITAL LETTER U WITH ACUTE
  173 0x00DB		U	# LATIN CAPITAL LETTER U WITH CIRCUMFLEX
  174 0x00DC		U	# LATIN CAPITAL LETTER U WITH DIAERESIS
  175 0x00DD		Y	# LATIN CAPITAL LETTER Y WITH ACUTE
  176 0x00DE		TH	# LATIN CAPITAL LETTER THORN (Icelandic)
  177 0x00DF		ss	# LATIN SMALL LETTER SHARP S (German)
  178 
  179 0x00E0		a	# LATIN SMALL LETTER A WITH GRAVE
  180 0x00E1		a	# LATIN SMALL LETTER A WITH ACUTE
  181 0x00E2		a	# LATIN SMALL LETTER A WITH CIRCUMFLEX
  182 0x00E3		a	# LATIN SMALL LETTER A WITH TILDE
  183 0x00E4		a	# LATIN SMALL LETTER A WITH DIAERESIS
  184 0x00E5		a	# LATIN SMALL LETTER A WITH RING ABOVE
  185 0x00E6		ae	# LATIN SMALL LETTER AE
  186 0x00E7		c	# LATIN SMALL LETTER C WITH CEDILLA
  187 0x00E8		e	# LATIN SMALL LETTER E WITH GRAVE
  188 0x00E9		e	# LATIN SMALL LETTER E WITH ACUTE
  189 0x00EA		e	# LATIN SMALL LETTER E WITH CIRCUMFLEX
  190 0x00EB		e	# LATIN SMALL LETTER E WITH DIAERESIS
  191 0x00EC		i	# LATIN SMALL LETTER I WITH GRAVE
  192 0x00ED		i	# LATIN SMALL LETTER I WITH ACUTE
  193 0x00EE		i	# LATIN SMALL LETTER I WITH CIRCUMFLEX
  194 0x00EF		i	# LATIN SMALL LETTER I WITH DIAERESIS
  195 
  196 0x00F0		th	# LATIN SMALL LETTER ETH (Icelandic)
  197 0x00F1		n	# LATIN SMALL LETTER N WITH TILDE
  198 0x00F2		o	# LATIN SMALL LETTER O WITH GRAVE
  199 0x00F3		o	# LATIN SMALL LETTER O WITH ACUTE
  200 0x00F4		o	# LATIN SMALL LETTER O WITH CIRCUMFLEX
  201 0x00F5		o	# LATIN SMALL LETTER O WITH TILDE
  202 0x00F6		o	# LATIN SMALL LETTER O WITH DIAERESIS
  203 0x00F8		o	# LATIN SMALL LETTER O WITH STROKE
  204 0x00F9		u	# LATIN SMALL LETTER U WITH GRAVE
  205 0x00FA		u	# LATIN SMALL LETTER U WITH ACUTE
  206 0x00FB		u	# LATIN SMALL LETTER U WITH CIRCUMFLEX
  207 0x00FC		u	# LATIN SMALL LETTER U WITH DIAERESIS
  208 0x00FD		y	# LATIN SMALL LETTER Y WITH ACUTE
  209 0x00FE		th	# LATIN SMALL LETTER THORN (Icelandic)
  210 0x00FF		y	# LATIN SMALL LETTER Y WITH DIAERESIS
  211 
  212 #
  213 # Latin Extended A
  214 #
  215 
  216 0x0100		A	# LATIN CAPITAL LETTER A WITH MACRON
  217 0x0101		a	# LATIN SMALL LETTER A WITH MACRON
  218 0x0102		A	# LATIN CAPITAL LETTER A WITH BREVE
  219 0x0103		a	# LATIN SMALL LETTER A WITH BREVE
  220 0x0104		A	# LATIN CAPITAL LETTER A WITH OGONEK
  221 0x0105		a	# LATIN SMALL LETTER A WITH OGONEK
  222 0x0106		C	# LATIN CAPITAL LETTER C WITH ACUTE
  223 0x0107		c	# LATIN SMALL LETTER C WITH ACUTE
  224 0x0108		C	# LATIN CAPITAL LETTER C WITH CIRCUMFLEX
  225 0x0109		c	# LATIN SMALL LETTER C WITH CIRCUMFLEX
  226 0x010A		C	# LATIN CAPITAL LETTER C WITH DOT ABOVE
  227 0x010B		c	# LATIN SMALL LETTER C WITH DOT ABOVE
  228 0x010C		C	# LATIN CAPITAL LETTER C WITH CARON
  229 0x010D		c	# LATIN SMALL LETTER C WITH CARON
  230 0x010E		D	# LATIN CAPITAL LETTER D WITH CARON
  231 0x010F		d	# LATIN SMALL LETTER D WITH CARON
  232 
  233 0x0110		D	# LATIN CAPITAL LETTER D WITH STROKE
  234 0x0111		d	# LATIN SMALL LETTER D WITH STROKE
  235 0x0112		E	# LATIN CAPITAL LETTER E WITH MACRON
  236 0x0113		e	# LATIN SMALL LETTER E WITH MACRON
  237 0x0114		E	# LATIN CAPITAL LETTER E WITH BREVE
  238 0x0115		e	# LATIN SMALL LETTER E WITH BREVE
  239 0x0116		E	# LATIN CAPITAL LETTER E WITH DOT ABOVE
  240 0x0117		e	# LATIN SMALL LETTER E WITH DOT ABOVE
  241 0x0118		E	# LATIN CAPITAL LETTER E WITH OGONEK
  242 0x0119		e	# LATIN SMALL LETTER E WITH OGONEK
  243 0x011A		E	# LATIN CAPITAL LETTER E WITH CARON
  244 0x011B		e	# LATIN SMALL LETTER E WITH CARON
  245 0x011C		G	# LATIN CAPITAL LETTER G WITH CIRCUMFLEX
  246 0x011D		g	# LATIN SMALL LETTER G WITH CIRCUMFLEX
  247 0x011E		G	# LATIN CAPITAL LETTER G WITH BREVE
  248 0x011F		g	# LATIN SMALL LETTER G WITH BREVE
  249 
  250 0x0120		G	# LATIN CAPITAL LETTER G WITH DOT ABOVE
  251 0x0121		g	# LATIN SMALL LETTER G WITH DOT ABOVE
  252 0x0122		G	# LATIN CAPITAL LETTER G WITH CEDILLA
  253 0x0123		g	# LATIN SMALL LETTER G WITH CEDILLA
  254 0x0124		H	# LATIN CAPITAL LETTER H WITH CIRCUMFLEX
  255 0x0125		h	# LATIN SMALL LETTER H WITH CIRCUMFLEX
  256 0x0126		H	# LATIN CAPITAL LETTER H WITH STROKE
  257 0x0127		h	# LATIN SMALL LETTER H WITH STROKE
  258 0x0128		I	# LATIN CAPITAL LETTER I WITH TILDE
  259 0x0129		i	# LATIN SMALL LETTER I WITH TILDE
  260 0x012A		I	# LATIN CAPITAL LETTER I WITH MACRON
  261 0x012B		i	# LATIN SMALL LETTER I WITH MACRON
  262 0x012C		I	# LATIN CAPITAL LETTER I WITH BREVE
  263 0x012D		i	# LATIN SMALL LETTER I WITH BREVE
  264 0x012E		I	# LATIN CAPITAL LETTER I WITH OGONEK
  265 0x012F		i	# LATIN SMALL LETTER I WITH OGONEK
  266 
  267 0x0130		I	# LATIN CAPITAL LETTER I WITH DOT ABOVE
  268 0x0131		i	# LATIN SMALL LETTER DOTLESS I
  269 0x0132		IJ	# LATIN CAPITAL LIGATURE IJ
  270 0x0133		ij	# LATIN SMALL LIGATURE IJ
  271 0x0134		J	# LATIN CAPITAL LETTER J WITH CIRCUMFLEX
  272 0x0135		j	# LATIN SMALL LETTER J WITH CIRCUMFLEX
  273 0x0136		K	# LATIN CAPITAL LETTER K WITH CEDILLA
  274 0x0137		k	# LATIN SMALL LETTER K WITH CEDILLA
  275 0x0138		q	# LATIN SMALL LETTER KRA (Greenlandic)
  276 0x0139		L	# LATIN CAPITAL LETTER L WITH ACUTE
  277 0x013A		l	# LATIN SMALL LETTER L WITH ACUTE
  278 0x013B		L	# LATIN CAPITAL LETTER L WITH CEDILLA
  279 0x013C		l	# LATIN SMALL LETTER L WITH CEDILLA
  280 0x013D		L	# LATIN CAPITAL LETTER L WITH CARON
  281 0x013E		l	# LATIN SMALL LETTER L WITH CARON
  282 0x013F		L	# LATIN CAPITAL LETTER L WITH MIDDLE DOT
  283 
  284 0x0140		l	# LATIN SMALL LETTER L WITH MIDDLE DOT
  285 0x0141		L	# LATIN CAPITAL LETTER L WITH STROKE
  286 0x0142		l	# LATIN SMALL LETTER L WITH STROKE
  287 0x0143		N	# LATIN CAPITAL LETTER N WITH ACUTE
  288 0x0144		n	# LATIN SMALL LETTER N WITH ACUTE
  289 0x0145		N	# LATIN CAPITAL LETTER N WITH CEDILLA
  290 0x0146		n	# LATIN SMALL LETTER N WITH CEDILLA
  291 0x0147		N	# LATIN CAPITAL LETTER N WITH CARON
  292 0x0148		n	# LATIN SMALL LETTER N WITH CARON
  293 0x0149		n	# LATIN SMALL LETTER N PRECEDED BY APOSTROPHE
  294 0x014A		NG	# LATIN CAPITAL LETTER ENG (Sami)
  295 0x014B		ng	# LATIN SMALL LETTER ENG (Sami)
  296 0x014C		O	# LATIN CAPITAL LETTER O WITH MACRON
  297 0x014D		o	# LATIN SMALL LETTER O WITH MACRON
  298 0x014E		O	# LATIN CAPITAL LETTER O WITH BREVE
  299 0x014F		o	# LATIN SMALL LETTER O WITH BREVE
  300 
  301 0x0150		O	# LATIN CAPITAL LETTER O WITH DOUBLE ACUTE
  302 0x0151		o	# LATIN SMALL LETTER O WITH DOUBLE ACUTE
  303 0x0152		OE	# LATIN CAPITAL LIGATURE OE
  304 0x0153		oe	# LATIN SMALL LIGATURE OE
  305 0x0154		R	# LATIN CAPITAL LETTER R WITH ACUTE
  306 0x0155		r	# LATIN SMALL LETTER R WITH ACUTE
  307 0x0156		R	# LATIN CAPITAL LETTER R WITH CEDILLA
  308 0x0157		r	# LATIN SMALL LETTER R WITH CEDILLA
  309 0x0158		R	# LATIN CAPITAL LETTER R WITH CARON
  310 0x0159		r	# LATIN SMALL LETTER R WITH CARON
  311 0x015A		S	# LATIN CAPITAL LETTER S WITH ACUTE
  312 0x015B		s	# LATIN SMALL LETTER S WITH ACUTE
  313 0x015C		S	# LATIN CAPITAL LETTER S WITH CIRCUMFLEX
  314 0x015D		s	# LATIN SMALL LETTER S WITH CIRCUMFLEX
  315 0x015E		S	# LATIN CAPITAL LETTER S WITH CEDILLA
  316 0x015F		s	# LATIN SMALL LETTER S WITH CEDILLA
  317 
  318 0x0160		S	# LATIN CAPITAL LETTER S WITH CARON
  319 0x0161		s	# LATIN SMALL LETTER S WITH CARON
  320 0x0162		T	# LATIN CAPITAL LETTER T WITH CEDILLA
  321 0x0163		t	# LATIN SMALL LETTER T WITH CEDILLA
  322 0x0164		T	# LATIN CAPITAL LETTER T WITH CARON
  323 0x0165		t	# LATIN SMALL LETTER T WITH CARON
  324 0x0166		T	# LATIN CAPITAL LETTER T WITH STROKE
  325 0x0167		t	# LATIN SMALL LETTER T WITH STROKE
  326 0x0168		U	# LATIN CAPITAL LETTER U WITH TILDE
  327 0x0169		u	# LATIN SMALL LETTER U WITH TILDE
  328 0x016A		U	# LATIN CAPITAL LETTER U WITH MACRON
  329 0x016B		u	# LATIN SMALL LETTER U WITH MACRON
  330 0x016C		U	# LATIN CAPITAL LETTER U WITH BREVE
  331 0x016D		u	# LATIN SMALL LETTER U WITH BREVE
  332 0x016E		U	# LATIN CAPITAL LETTER U WITH RING ABOVE
  333 0x016F		u	# LATIN SMALL LETTER U WITH RING ABOVE
  334 
  335 0x0170		U	# LATIN CAPITAL LETTER U WITH DOUBLE ACUTE
  336 0x0171		u	# LATIN SMALL LETTER U WITH DOUBLE ACUTE
  337 0x0172		U	# LATIN CAPITAL LETTER U WITH OGONEK
  338 0x0173		u	# LATIN SMALL LETTER U WITH OGONEK
  339 0x0174		W	# LATIN CAPITAL LETTER W WITH CIRCUMFLEX
  340 0x0175		w	# LATIN SMALL LETTER W WITH CIRCUMFLEX
  341 0x0176		Y	# LATIN CAPITAL LETTER Y WITH CIRCUMFLEX
  342 0x0177		y	# LATIN SMALL LETTER Y WITH CIRCUMFLEX
  343 0x0178		Y	# LATIN CAPITAL LETTER Y WITH DIAERESIS
  344 0x0179		Z	# LATIN CAPITAL LETTER Z WITH ACUTE
  345 0x017A		z	# LATIN SMALL LETTER Z WITH ACUTE
  346 0x017B		Z	# LATIN CAPITAL LETTER Z WITH DOT ABOVE
  347 0x017C		z	# LATIN SMALL LETTER Z WITH DOT ABOVE
  348 0x017D		Z	# LATIN CAPITAL LETTER Z WITH CARON
  349 0x017E		z	# LATIN SMALL LETTER Z WITH CARON
  350 0x017F		s	# LATIN SMALL LETTER LONG S
  351 
  352 #
  353 # Dunno where
  354 #
  355 
  356 0x02C6		^	# MODIFIER LETTER CIRCUMFLEX ACCENT
  357 
  358 0x2010		-	# HYPHEN
  359 0x2011		-	# NON-BREAKING HYPHEN
  360 0x2012		-	# FIGURE DASH
  361 0x2013		-	# EN DASH
  362 0x2014		-	# EM DASH
  363 0x2015		-	# HORIZONTAL BAR
  364 
  365 0x2026		...	# HORIZONTAL ELLIPSIS
  366 
  367 0x2122		_tm_	# TRADE MARK SIGN
  368 
  369 end
  370 
  371 #
  372 # German specific conversions
  373 #
  374 
  375 start		de
  376 
  377 0x00C4		AE	# LATIN CAPITAL LETTER A WITH UMLAUT
  378 0x00D6		OE	# LATIN CAPITAL LETTER O WITH UMLAUT
  379 0x00DC		UE	# LATIN CAPITAL LETTER U WITH UMLAUT
  380 
  381 0x00E4		ae	# LATIN SMALL LETTER A WITH UMLAUT
  382 0x00F6		oe	# LATIN SMALL LETTER O WITH UMLAUT
  383 0x00FC		ue	# LATIN SMALL LETTER U WITH UMLAUT 
  384 
  385 end
  386