prest/tbin/redom

   1 #!/usr/local/bin/perl -p
   2
   3 # Converts a dom to a regular-expression DOM
   4 s/([\(\)\[\]\{\}\.\+\*\?\|\\\^\$])/\\$1/g;
   5 s/id\d+/id\\d+/g;
   6 s/line="\d+"/line="\\d+"/g;
   7 # Remove bogus A-circumflex
   8 s/\xc2//g;
   9 # Deal with unicode characters
  10 s/([\x00-\x09\x0b-\x1f\x7f-\xff\x{0100}-\x{ffff}])/sprintf '\\\\u%04x', ord $1/ge;
  11 # Handle the "Generated on" date
  12 s/(Generated on: *).*/$1.*/;
  13 # Handle the "Generated by" description
  14 s/^( *Generated by *)/$1.*/;
  15 # Accommodate different date formats
  16 BEGIN {
  17     $time = '(\s+\d+:\d+:\d+(\s+[-+]\d+(\s+\(.+?\))?)?)?';
  18     ($timere = $time) =~ s/\\([()])/\\\\\\$1/g;
  19 }
  20 s!(\d{4})([-/])(\d{2})\2(\d{2})$timere!\\d{4}[-/\]\\d{2}\[-/\]\\d{2}$time!go;
  21 # Handle uppercase vs lowercase unicode
  22 s/(\\u)([\da-f]{4})/$1(?i)$2/ig;
  23 # Deal with revisions that may change
  24 s/(prest release )\d+(\\\.\d+)+/$1\\d+(\\.\\d+)+/;
  25 if ($AFTER_REVISION) {
  26     s/\d+(\\\.\d+)+/\\d+(\\.\\d+)+/;
  27     $AFTER_REVISION = 0;
  28 }
  29 $AFTER_REVISION = /<revision>/;
  30 # Deal with dates that may change
  31 if ($AFTER_DATE) {
  32     s|\d{4}[/-]\d{2}[/-]\d{2}|\\d{4}[/-]\\d{2}[/-]\\d{2}|;
  33     $AFTER_DATE = 0;
  34 }
  35 # Deal with perl eval numbers that may change in error messages
  36 s/\\\((eval )(\d+)\\\).*/\\($1\\d+\\).*/;
  37 $AFTER_DATE = /<date>/;
  38 # Perl trapped error message changed between 5.8.0 and 5.8.6.
  39 s/(: )(.*)( trapped by operation mask)/$1($2|.*)$3/;
  40 # System messages may differ in different locales
  41 s/(\\\[Errno \d+\\\])(.*)/$1($2|.*)/;
  42 s/(\bstates: )(.*)/$1($2|.*)/;
  43 # Handle -d output
  44 s!^.*(/Text/Restructured/Writer/dom\\\.wrt, line )\d+!.*$1\\d+!;
  45 # Handle line numbers in error messages
  46 s!\S+(/\S+\\\.(?:pm|wrt),? line )(\d+)(, <\w+> line \d+)?!.*$1($2|\\d+)(, <\\w+> line \\d+)?!g;
  47