scripts/punct.pl

   1 #!/usr/bin/perl
   2
   3 while (<>) {
   4     $text = $_;
   5
   6     # entity
   7     $text =~ s/&\#([0-9]*);/ \#\1 /g;
   8     # foreign proper name
   9     while ($text =~ s/([A-ZÐ])\. ([A-ZÐ])/\1.\2/g) {}
  10     # double colon
  11     $chars = "a-záàäãÕâ¤¥¦ç§åå¡¢ÆÇ£éèë¨©êª«¬®íìïî¸óòöõ÷ô¯°±²µ½¾¶·ÞþúùüûøßÑ×ØæñýÏÖÛÜðA-ZÁÀÄÃ€Â„…†\x06‡ÅÅ�‚\x02\x05ƒÉÈËˆ‰ÊŠ‹Œ�ŽÍÌ›Î˜ÓÒ™ šÔ��‘’“´•–—³”ÚÙœ�ž¿º»¼ÿ¹ÝŸ\x14\x19\x1eÐ";
  12     $text =~ s/([$chars])([!~():,.;!?-])([^$chars])/\1\n\3/g;
  13     $text =~ s/([^$chars])([!~():,.;!?-])([$chars])/\1\n\3/g;
  14     $text =~ s/^([!~():,.;-])([$chars])/\n\2/g;
  15     # dot
  16     # $text =~ m/\. ([a-zðA-ZÐ])/ . \1/g;
  17     print $text;
  18 }
  19
  20 # Local Variables:
  21 # coding: viscii
  22 # End: