maintenance/entities2literals.pl

   1 #!/usr/bin/evn perl
   2 # Takes STDIN and converts Converts hexadecimal, decimal and named HTML
   3 # entities to their respective literals.
   4 #
   5 # Usage: perl entities2literals.pl < file_to_convert [> outfile]
   6 # Reference: http://www.w3.org/TR/REC-html40/sgml/entities.html
   7 # Copyright 2005 Ævar Arnfjörð Bjarmason <avarab@gmail.com> No rights reserved
   8
   9 use encoding 'utf8';
  10 use strict;
  11
  12 my $file = join /$\//, <>;
  13
  14 $file =~ s/&#(\d+);/chr $1/eg;
  15 $file =~ s/&#x([0-9a-fA-F]+);/chr hex $1/eg;
  16
  17 while (<DATA>) {
  18         chomp;
  19         my ($number, $entity) = split / +/;
  20         $file =~ s/&$entity;/chr $number/eg;
  21 }
  22 print $file;
  23
  24 __DATA__
  25 34   quot
  26 38   amp
  27 60   lt
  28 62   gt
  29 160  nbsp
  30 161  iexcl
  31 162  cent
  32 163  pound
  33 164  curren
  34 165  yen
  35 166  brvbar
  36 167  sect
  37 168  uml
  38 169  copy
  39 170  ordf
  40 171  laquo
  41 172  not
  42 173  shy
  43 174  reg
  44 175  macr
  45 176  deg
  46 177  plusmn
  47 178  sup2
  48 179  sup3
  49 180  acute
  50 181  micro
  51 182  para
  52 183  middot
  53 184  cedil
  54 185  sup1
  55 186  ordm
  56 187  raquo
  57 188  frac14
  58 189  frac12
  59 190  frac34
  60 191  iquest
  61 192  Agrave
  62 193  Aacute
  63 194  Acirc
  64 195  Atilde
  65 196  Auml
  66 197  Aring
  67 198  AElig
  68 199  Ccedil
  69 200  Egrave
  70 201  Eacute
  71 202  Ecirc
  72 203  Euml
  73 204  Igrave
  74 205  Iacute
  75 206  Icirc
  76 207  Iuml
  77 208  ETH
  78 209  Ntilde
  79 210  Ograve
  80 211  Oacute
  81 212  Ocirc
  82 213  Otilde
  83 214  Ouml
  84 215  times
  85 216  Oslash
  86 217  Ugrave
  87 218  Uacute
  88 219  Ucirc
  89 220  Uuml
  90 221  Yacute
  91 222  THORN
  92 223  szlig
  93 224  agrave
  94 225  aacute
  95 226  acirc
  96 227  atilde
  97 228  auml
  98 229  aring
  99 230  aelig
 100 231  ccedil
 101 232  egrave
 102 233  eacute
 103 234  ecirc
 104 235  euml
 105 236  igrave
 106 237  iacute
 107 238  icirc
 108 239  iuml
 109 240  eth
 110 241  ntilde
 111 242  ograve
 112 243  oacute
 113 244  ocirc
 114 245  otilde
 115 246  ouml
 116 247  divide
 117 248  oslash
 118 249  ugrave
 119 250  uacute
 120 251  ucirc
 121 252  uuml
 122 253  yacute
 123 254  thorn
 124 255  yuml
 125 338  OElig
 126 339  oelig
 127 352  Scaron
 128 353  scaron
 129 376  Yuml
 130 402  fnof
 131 710  circ
 132 732  tilde
 133 913  Alpha
 134 914  Beta
 135 915  Gamma
 136 916  Delta
 137 917  Epsilon
 138 918  Zeta
 139 919  Eta
 140 920  Theta
 141 921  Iota
 142 922  Kappa
 143 923  Lambda
 144 924  Mu
 145 925  Nu
 146 926  Xi
 147 927  Omicron
 148 928  Pi
 149 929  Rho
 150 931  Sigma
 151 932  Tau
 152 933  Upsilon
 153 934  Phi
 154 935  Chi
 155 936  Psi
 156 937  Omega
 157 945  alpha
 158 946  beta
 159 947  gamma
 160 948  delta
 161 949  epsilon
 162 950  zeta
 163 951  eta
 164 952  theta
 165 953  iota
 166 954  kappa
 167 955  lambda
 168 956  mu
 169 957  nu
 170 958  xi
 171 959  omicron
 172 960  pi
 173 961  rho
 174 962  sigmaf
 175 963  sigma
 176 964  tau
 177 965  upsilon
 178 966  phi
 179 967  chi
 180 968  psi
 181 969  omega
 182 977  thetasym
 183 978  upsih
 184 982  piv
 185 8194 ensp
 186 8195 emsp
 187 8201 thinsp
 188 8204 zwnj
 189 8205 zwj
 190 8206 lrm
 191 8207 rlm
 192 8211 ndash
 193 8212 mdash
 194 8216 lsquo
 195 8217 rsquo
 196 8218 sbquo
 197 8220 ldquo
 198 8221 rdquo
 199 8222 bdquo
 200 8224 dagger
 201 8225 Dagger
 202 8226 bull
 203 8230 hellip
 204 8240 permil
 205 8242 prime
 206 8243 Prime
 207 8249 lsaquo
 208 8250 rsaquo
 209 8254 oline
 210 8260 frasl
 211 8364 euro
 212 8465 image
 213 8472 weierp
 214 8476 real
 215 8482 trade
 216 8501 alefsym
 217 8592 larr
 218 8593 uarr
 219 8594 rarr
 220 8595 darr
 221 8596 harr
 222 8629 crarr
 223 8656 lArr
 224 8657 uArr
 225 8658 rArr
 226 8659 dArr
 227 8660 hArr
 228 8704 forall
 229 8706 part
 230 8707 exist
 231 8709 empty
 232 8711 nabla
 233 8712 isin
 234 8713 notin
 235 8715 ni
 236 8719 prod
 237 8721 sum
 238 8722 minus
 239 8727 lowast
 240 8730 radic
 241 8733 prop
 242 8734 infin
 243 8736 ang
 244 8743 and
 245 8744 or
 246 8745 cap
 247 8746 cup
 248 8747 int
 249 8756 there4
 250 8764 sim
 251 8773 cong
 252 8776 asymp
 253 8800 ne
 254 8801 equiv
 255 8804 le
 256 8805 ge
 257 8834 sub
 258 8835 sup
 259 8836 nsub
 260 8838 sube
 261 8839 supe
 262 8853 oplus
 263 8855 otimes
 264 8869 perp
 265 8901 sdot
 266 8968 lceil
 267 8969 rceil
 268 8970 lfloor
 269 8971 rfloor
 270 9001 lang
 271 9002 rang
 272 9674 loz
 273 9824 spades
 274 9827 clubs
 275 9829 hearts
 276 9830 diams