t/eml_content_type.t

   1 #!perl -w
   2 # Copyright (C) 2020-2021 all contributors <meta@public-inbox.org>
   3 # Copyright (C) 2004- Simon Cozens, Casey West, Ricardo SIGNES
   4 # This library is free software; you can redistribute it and/or modify
   5 # it under the same terms as Perl itself.
   6 #
   7 # License: GPL-1.0+ or Artistic-1.0-Perl
   8 #  <https://www.gnu.org/licenses/gpl-1.0.txt>
   9 #  <https://dev.perl.org/licenses/artistic.html>
  10 use strict;
  11 use Test::More;
  12 use PublicInbox::EmlContentFoo qw(parse_content_type);
  13
  14 my %ct_tests = (
  15         '' => {
  16                 type       => "text",
  17                 subtype    => "plain",
  18                 attributes => { charset => "us-ascii" }
  19         },
  20
  21         "text/plain" => {
  22                 type => "text",
  23                 subtype => "plain",
  24                 attributes => {}
  25         },
  26         'text/plain; charset=us-ascii' => {
  27                 type       => "text",
  28                 subtype    => "plain",
  29                 attributes => { charset => "us-ascii" }
  30         },
  31         'text/plain; charset="us-ascii"' => {
  32                 type       => "text",
  33                 subtype    => "plain",
  34                 attributes => { charset => "us-ascii" }
  35         },
  36         "text/plain; charset=us-ascii (Plain text)" => {
  37                 type       => "text",
  38                 subtype    => "plain",
  39                 attributes => { charset => "us-ascii" }
  40         },
  41
  42         'text/plain; charset=ISO-8859-1' => {
  43                 type       => "text",
  44                 subtype    => "plain",
  45                 attributes => { charset => "ISO-8859-1" }
  46         },
  47         'text/plain; charset="ISO-8859-1"' => {
  48                 type       => "text",
  49                 subtype    => "plain",
  50                 attributes => { charset => "ISO-8859-1" }
  51         },
  52         'text/plain; charset="ISO-8859-1" (comment)' => {
  53                 type       => "text",
  54                 subtype    => "plain",
  55                 attributes => { charset => "ISO-8859-1" }
  56         },
  57
  58         '(c) text/plain (c); (c) charset=ISO-8859-1 (c)' => {
  59                 type       => "text",
  60                 subtype    => "plain",
  61                 attributes => { charset => "ISO-8859-1" }
  62         },
  63         '(c \( \\\\) (c) text/plain (c) (c) ; (c) (c) charset=utf-8 (c)' => {
  64                 type       => "text",
  65                 subtype    => "plain",
  66                 attributes => { charset => "utf-8" }
  67         },
  68         'text/plain; (c (nested ()c)another c)() charset=ISO-8859-1' => {
  69                 type       => "text",
  70                 subtype    => "plain",
  71                 attributes => { charset => "ISO-8859-1" }
  72         },
  73         'text/plain (c \(!nested ()c\)\)(nested\(c())); charset=utf-8' => {
  74                 type       => "text",
  75                 subtype    => "plain",
  76                 attributes => { charset => "utf-8" }
  77         },
  78
  79         "application/foo" => {
  80                 type       => "application",
  81                 subtype    => "foo",
  82                 attributes => {}
  83         },
  84         "multipart/mixed; boundary=unique-boundary-1" => {
  85                 type       => "multipart",
  86                 subtype    => "mixed",
  87                 attributes => { boundary => "unique-boundary-1" }
  88         },
  89         'message/external-body; access-type=local-file; name="/u/n/m.jpg"' => {
  90                 type       => "message",
  91                 subtype    => "external-body",
  92                 attributes => {
  93                         "access-type" => "local-file",
  94                         "name"        => "/u/n/m.jpg"
  95                 }
  96         },
  97         'multipart/mixed; boundary="----------=_1026452699-10321-0" ' => {
  98                 'type'       => 'multipart',
  99                 'subtype'    => 'mixed',
 100                 'attributes' => {
 101                         'boundary' => '----------=_1026452699-10321-0'
 102                 }
 103         },
 104         'multipart/report; boundary= "=_0=73e476c3-cd5a-5ba3-b910-2="' => {
 105                 'type'       => 'multipart',
 106                 'subtype'    => 'report',
 107                 'attributes' => {
 108                         'boundary' => '=_0=73e476c3-cd5a-5ba3-b910-2='
 109                 }
 110         },
 111         'multipart/report; boundary=' . " \t" . '"=_0=7-c-5-b-2="' => {
 112                 'type'       => 'multipart',
 113                 'subtype'    => 'report',
 114                 'attributes' => {
 115                         'boundary' => '=_0=7-c-5-b-2='
 116                 }
 117         },
 118
 119         'message/external-body; access-type=URL;' .
 120         ' URL*0="ftp://";' .
 121         ' URL*1="example.com/"' => {
 122                 'type'       => 'message',
 123                 'subtype'    => 'external-body',
 124                 'attributes' => {
 125                         'access-type' => 'URL',
 126                         'url' => 'ftp://example.com/'
 127                 }
 128         },
 129         'message/external-body; access-type=URL; URL="ftp://example.com/"' => {
 130                 'type'       => 'message',
 131                 'subtype'    => 'external-body',
 132                 'attributes' => {
 133                         'access-type' => 'URL',
 134                         'url' => 'ftp://example.com/',
 135                 }
 136         },
 137
 138         "application/x-stuff; title*=us-ascii'en-us'This%20is%20f%2Ad" => {
 139                 'type'       => 'application',
 140                 'subtype'    => 'x-stuff',
 141                 'attributes' => {
 142                         'title' => 'This is f*d'
 143                 }
 144         },
 145         "application/x-stuff; title*=us-ascii''This%20is%20f%2Ad" => {
 146                 'type'       => 'application',
 147                 'subtype'    => 'x-stuff',
 148                 'attributes' => {
 149                         'title' => 'This is f*d'
 150                 }
 151         },
 152         "application/x-stuff; title*=''This%20is%20f%2Ad" => {
 153                 'type'       => 'application',
 154                 'subtype'    => 'x-stuff',
 155                 'attributes' => {
 156                         'title' => 'This is f*d'
 157                 }
 158         },
 159         "application/x-stuff; title*='en-us'This%20is%20f%2Ad" => {
 160                 'type'       => 'application',
 161                 'subtype'    => 'x-stuff',
 162                 'attributes' => {
 163                         'title' => 'This is f*d'
 164                 }
 165         },
 166         q(application/x-stuff;) .
 167         q( title*0*=us-ascii'en'This%20is%20even%20more%20;) .
 168         q(title*1*=%2A%2A%2Afun%2A%2A%2A%20; title*2="isn't it!") => {
 169                 'type'       => 'application',
 170                 'subtype'    => 'x-stuff',
 171                 'attributes' => {
 172                         'title' => "This is even more ***fun*** isn't it!"
 173                 }
 174         },
 175         q(application/x-stuff;) .
 176         q( title*0*='en'This%20is%20even%20more%20;) .
 177         q( title*1*=%2A%2A%2Afun%2A%2A%2A%20; title*2="isn't it!") => {
 178                 'type'       => 'application',
 179                 'subtype'    => 'x-stuff',
 180                 'attributes' => {
 181                         'title' => "This is even more ***fun*** isn't it!"
 182                 }
 183         },
 184         q(application/x-stuff;) .
 185         q( title*0*=''This%20is%20even%20more%20;) .
 186         q( title*1*=%2A%2A%2Afun%2A%2A%2A%20; title*2="isn't it!") => {
 187                 'type'       => 'application',
 188                 'subtype'    => 'x-stuff',
 189                 'attributes' => {
 190                         'title' => "This is even more ***fun*** isn't it!"
 191                 }
 192         },
 193         q(application/x-stuff;).
 194         q( title*0*=us-ascii''This%20is%20even%20more%20;).
 195         q( title*1*=%2A%2A%2Afun%2A%2A%2A%20; title*2="isn't it!")
 196           => {
 197                 'type'       => 'application',
 198                 'subtype'    => 'x-stuff',
 199                 'attributes' => {
 200                         'title' => "This is even more ***fun*** isn't it!"
 201                 }
 202         },
 203
 204         'text/plain; attribute="v\"v\\\\v\(v\>\<\)\@\,\;\:\/\]\[\?\=v v";' .
 205         ' charset=us-ascii' => {
 206                 'type'       => 'text',
 207                 'subtype'    => 'plain',
 208                 'attributes' => {
 209                         'attribute' => 'v"v\\v(v><)@,;:/][?=v v',
 210                         'charset' => 'us-ascii',
 211                 },
 212         },
 213
 214         qq(text/plain;\r
 215          charset=us-ascii;\r
 216          attribute="\r value1 \r value2\r\n value3\r\n value4\r\n "\r\n ) => {
 217                 'type'       => 'text',
 218                 'subtype'    => 'plain',
 219                 'attributes' => {
 220                         'attribute' => ' value1  value2 value3 value4 ',
 221                         'charset'   => 'us-ascii',
 222                 },
 223         },
 224 );
 225
 226 my %non_strict_ct_tests = (
 227         "text/plain;" => { type => "text", subtype => "plain", attributes => {} },
 228         "text/plain; " =>
 229           { type => "text", subtype => "plain", attributes => {} },
 230         'image/jpeg;' .
 231         ' x-mac-type="3F3F3F3F";'.
 232         ' x-mac-creator="3F3F3F3F" name="file name.jpg";' => {
 233                 type       => "image",
 234                 subtype    => "jpeg",
 235                 attributes => {
 236                         'x-mac-type'    => "3F3F3F3F",
 237                         'x-mac-creator' => "3F3F3F3F",
 238                         'name'          => "file name.jpg"
 239                 }
 240         },
 241         "text/plain; key=very long value" => {
 242                 type       => "text",
 243                 subtype    => "plain",
 244                 attributes => { key => "very long value" }
 245         },
 246         "text/plain; key=very long value key2=value2" => {
 247                 type    => "text",
 248                 subtype => "plain",
 249                 attributes => { key => "very long value", key2 => "value2" }
 250         },
 251         'multipart/mixed; boundary = "--=_Next_Part_24_Nov_2016_08.09.21"' => {
 252                 type    => "multipart",
 253                 subtype => "mixed",
 254                 attributes => {
 255                         boundary => "--=_Next_Part_24_Nov_2016_08.09.21"
 256                 }
 257         },
 258 );
 259
 260 sub test {
 261         my ($string, $expect, $info) = @_;
 262
 263         local $_;
 264         $info =~ s/\r/\\r/g;
 265         $info =~ s/\n/\\n/g;
 266         is_deeply(parse_content_type($string), $expect, $info);
 267 }
 268
 269 for (sort keys %ct_tests) {
 270         test($_, $ct_tests{$_}, "Can parse C-T <$_>");
 271 }
 272
 273 local $PublicInbox::EmlContentFoo::STRICT_PARAMS = 0;
 274 for (sort keys %ct_tests) {
 275         test($_, $ct_tests{$_}, "Can parse non-strict C-T <$_>");
 276 }
 277 for (sort keys %non_strict_ct_tests) {
 278         test(
 279                 $_,
 280                 $non_strict_ct_tests{$_},
 281                 "Can parse non-strict C-T <$_>"
 282         );
 283 }
 284
 285 done_testing;