perl/lib/Thrasher/Nodeprep.pm

   1 package Thrasher::Nodeprep;
   2 use strict;
   3 use warnings;
   4
   5 # This module uses the Unicode::Stringprep module, combined with the
   6 # specification in RFC3920, to define "nodeprep" and "stringprep"
   7 # functions that can be used for converting strings of unknown safety
   8 # into JIDs.
   9
  10 use Unicode::Stringprep;
  11 use Unicode::Stringprep::Mapping;
  12 use Unicode::Stringprep::Prohibited;
  13
  14 use base 'Exporter';
  15
  16 use Data::Dumper;
  17
  18 our @EXPORT_OK = qw(nodeprep stringprep);
  19 our %EXPORT_TAGS = (all => \@EXPORT_OK);
  20
  21 *nodeprep = Unicode::Stringprep->new
  22     (3.2,
  23
  24      [
  25       # In accordance with RFC3920 section A.3
  26       @Unicode::Stringprep::Mapping::B1,
  27       @Unicode::Stringprep::Mapping::B2,
  28
  29       # The standard specifies these as banned in A.5, but I'm going to map
  30       # them to blank strings instead, so they are eliminated instead
  31       # of causing "die"s.
  32       (map { defined($_) ? ($_, '') : () }
  33         @Unicode::Stringprep::Prohibited::C11,
  34         @Unicode::Stringprep::Prohibited::C12,
  35         @Unicode::Stringprep::Prohibited::C21,
  36         @Unicode::Stringprep::Prohibited::C22,
  37         @Unicode::Stringprep::Prohibited::C3,
  38         @Unicode::Stringprep::Prohibited::C4,
  39         @Unicode::Stringprep::Prohibited::C5,
  40         @Unicode::Stringprep::Prohibited::C6,
  41         @Unicode::Stringprep::Prohibited::C7,
  42         @Unicode::Stringprep::Prohibited::C8,
  43         @Unicode::Stringprep::Prohibited::C9,
  44
  45        (
  46         0x22, # double-quote
  47         0x26, # ampersand
  48         0x27, # apostrophe
  49         0x2f, # backslash,
  50         0x3a, # colon
  51         0x3c, # less than
  52         0x3e, # greater than
  53        )),
  54
  55       ord('@') => '%'  # I don't really like this but it's traditional
  56      ],
  57
  58      # Section A.4 specifies normalization form KC
  59      'KC',
  60
  61      # Prohibited characters are replaced with blanks
  62      [
  63      ],
  64
  65      # Section A.6 specifies handling bi-directional strings.
  66      1
  67     );
  68
  69 1;
  70
  71 *resource_prep = Unicode::Stringprep->new
  72     (3.2,
  73
  74      # Section B.3
  75      [
  76       @Unicode::Stringprep::Mapping::B1,
  77       (map { defined($_) ? ($_, '') : () }
  78        @Unicode::Stringprep::Prohibited::C12,
  79        @Unicode::Stringprep::Prohibited::C21,
  80        @Unicode::Stringprep::Prohibited::C22,
  81        @Unicode::Stringprep::Prohibited::C3,
  82        @Unicode::Stringprep::Prohibited::C4,
  83        @Unicode::Stringprep::Prohibited::C5,
  84        @Unicode::Stringprep::Prohibited::C6,
  85        @Unicode::Stringprep::Prohibited::C7,
  86        @Unicode::Stringprep::Prohibited::C8,
  87        @Unicode::Stringprep::Prohibited::C9
  88        )
  89       ],
  90
  91      # Section B.4
  92      'KC',
  93
  94      [
  95      ],
  96
  97      # Section B.6
  98      1);
  99
 100
 101
 102
 103 1;