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 strict;
  10
  11 my $file = join /$\//, <>;
  12
  13 $file =~ s/&#(\d+);/chr $1/eg;
  14 $file =~ s/&#x([0-9a-fA-F]+);/chr hex $1/eg;
  15
  16 while (<DATA>) {
  17         chomp;
  18         my ($number, $entity) = split / +/;
  19         $file =~ s/&$entity;/chr $number/eg;
  20 }
  21 print $file;
  22
  23 __DATA__
  24 34   quot
  25 38   amp
  26 60   lt
  27 62   gt
  28 160  nbsp
  29 161  iexcl
  30 162  cent
  31 163  pound
  32 164  curren
  33 165  yen
  34 166  brvbar
  35 167  sect
  36 168  uml
  37 169  copy
  38 170  ordf
  39 171  laquo
  40 172  not
  41 173  shy
  42 174  reg
  43 175  macr
  44 176  deg
  45 177  plusmn
  46 178  sup2
  47 179  sup3
  48 180  acute
  49 181  micro
  50 182  para
  51 183  middot
  52 184  cedil
  53 185  sup1
  54 186  ordm
  55 187  raquo
  56 188  frac14
  57 189  frac12
  58 190  frac34
  59 191  iquest
  60 192  Agrave
  61 193  Aacute
  62 194  Acirc
  63 195  Atilde
  64 196  Auml
  65 197  Aring
  66 198  AElig
  67 199  Ccedil
  68 200  Egrave
  69 201  Eacute
  70 202  Ecirc
  71 203  Euml
  72 204  Igrave
  73 205  Iacute
  74 206  Icirc
  75 207  Iuml
  76 208  ETH
  77 209  Ntilde
  78 210  Ograve
  79 211  Oacute
  80 212  Ocirc
  81 213  Otilde
  82 214  Ouml
  83 215  times
  84 216  Oslash
  85 217  Ugrave
  86 218  Uacute
  87 219  Ucirc
  88 220  Uuml
  89 221  Yacute
  90 222  THORN
  91 223  szlig
  92 224  agrave
  93 225  aacute
  94 226  acirc
  95 227  atilde
  96 228  auml
  97 229  aring
  98 230  aelig
  99 231  ccedil
 100 232  egrave
 101 233  eacute
 102 234  ecirc
 103 235  euml
 104 236  igrave
 105 237  iacute
 106 238  icirc
 107 239  iuml
 108 240  eth
 109 241  ntilde
 110 242  ograve
 111 243  oacute
 112 244  ocirc
 113 245  otilde
 114 246  ouml
 115 247  divide
 116 248  oslash
 117 249  ugrave
 118 250  uacute
 119 251  ucirc
 120 252  uuml
 121 253  yacute
 122 254  thorn
 123 255  yuml
 124 338  OElig
 125 339  oelig
 126 352  Scaron
 127 353  scaron
 128 376  Yuml
 129 402  fnof
 130 710  circ
 131 732  tilde
 132 913  Alpha
 133 914  Beta
 134 915  Gamma
 135 916  Delta
 136 917  Epsilon
 137 918  Zeta
 138 919  Eta
 139 920  Theta
 140 921  Iota
 141 922  Kappa
 142 923  Lambda
 143 924  Mu
 144 925  Nu
 145 926  Xi
 146 927  Omicron
 147 928  Pi
 148 929  Rho
 149 931  Sigma
 150 932  Tau
 151 933  Upsilon
 152 934  Phi
 153 935  Chi
 154 936  Psi
 155 937  Omega
 156 945  alpha
 157 946  beta
 158 947  gamma
 159 948  delta
 160 949  epsilon
 161 950  zeta
 162 951  eta
 163 952  theta
 164 953  iota
 165 954  kappa
 166 955  lambda
 167 956  mu
 168 957  nu
 169 958  xi
 170 959  omicron
 171 960  pi
 172 961  rho
 173 962  sigmaf
 174 963  sigma
 175 964  tau
 176 965  upsilon
 177 966  phi
 178 967  chi
 179 968  psi
 180 969  omega
 181 977  thetasym
 182 978  upsih
 183 982  piv
 184 8194 ensp
 185 8195 emsp
 186 8201 thinsp
 187 8204 zwnj
 188 8205 zwj
 189 8206 lrm
 190 8207 rlm
 191 8211 ndash
 192 8212 mdash
 193 8216 lsquo
 194 8217 rsquo
 195 8218 sbquo
 196 8220 ldquo
 197 8221 rdquo
 198 8222 bdquo
 199 8224 dagger
 200 8225 Dagger
 201 8226 bull
 202 8230 hellip
 203 8240 permil
 204 8242 prime
 205 8243 Prime
 206 8249 lsaquo
 207 8250 rsaquo
 208 8254 oline
 209 8260 frasl
 210 8364 euro
 211 8465 image
 212 8472 weierp
 213 8476 real
 214 8482 trade
 215 8501 alefsym
 216 8592 larr
 217 8593 uarr
 218 8594 rarr
 219 8595 darr
 220 8596 harr
 221 8629 crarr
 222 8656 lArr
 223 8657 uArr
 224 8658 rArr
 225 8659 dArr
 226 8660 hArr
 227 8704 forall
 228 8706 part
 229 8707 exist
 230 8709 empty
 231 8711 nabla
 232 8712 isin
 233 8713 notin
 234 8715 ni
 235 8719 prod
 236 8721 sum
 237 8722 minus
 238 8727 lowast
 239 8730 radic
 240 8733 prop
 241 8734 infin
 242 8736 ang
 243 8743 and
 244 8744 or
 245 8745 cap
 246 8746 cup
 247 8747 int
 248 8756 there4
 249 8764 sim
 250 8773 cong
 251 8776 asymp
 252 8800 ne
 253 8801 equiv
 254 8804 le
 255 8805 ge
 256 8834 sub
 257 8835 sup
 258 8836 nsub
 259 8838 sube
 260 8839 supe
 261 8853 oplus
 262 8855 otimes
 263 8869 perp
 264 8901 sdot
 265 8968 lceil
 266 8969 rceil
 267 8970 lfloor
 268 8971 rfloor
 269 9001 lang
 270 9002 rang
 271 9674 loz
 272 9824 spades
 273 9827 clubs
 274 9829 hearts
 275 9830 diams