spam/SpamPickupTest.php

   1 <?php
   2 // $Id: SpamPickupTest.php,v 1.7 2008/12/31 15:44:14 henoheno Exp $
   3 // Copyright (C) 2007 heno
   4 //
   5 // Design test case for spam.php (called from runner.php)
   6
   7 if (! defined('SPAM_INI_FILE')) define('SPAM_INI_FILE', 'spam.ini.php');
   8
   9 require_once('spam_pickup.php');
  10 require_once('PHPUnit/PHPUnit.php');
  11
  12 class SpamPickupTest extends PHPUnit_TestCase
  13 {
  14         function setup_string_null()
  15         {
  16                 return array(
  17                         '[NULL]'        => NULL,
  18                         '[TRUE]'        => TRUE,
  19                         '[FALSE]'       => FALSE,
  20                         '[array(foobar)]' => array('foobar'),
  21                         '[]'            => '',
  22                         '[0]'           => 0,
  23                         '[1]'           => 1
  24                 );
  25         }
  26
  27         function testFunc_scheme_normalize()
  28         {
  29                 // Null
  30                 foreach($this->setup_string_null() as $key => $value){
  31                         $this->assertEquals('', scheme_normalize($value), $key);
  32                 }
  33
  34                 // CASE
  35                 $this->assertEquals('http', scheme_normalize('HTTP'));
  36
  37                 // Aliases
  38                 $this->assertEquals('pop3',  scheme_normalize('pop'));
  39                 $this->assertEquals('nntp',  scheme_normalize('news'));
  40                 $this->assertEquals('imap',  scheme_normalize('imap4'));
  41                 $this->assertEquals('nntps', scheme_normalize('snntp'));
  42                 $this->assertEquals('nntps', scheme_normalize('snews'));
  43                 $this->assertEquals('pop3s', scheme_normalize('spop3'));
  44                 $this->assertEquals('pop3s', scheme_normalize('pops'));
  45
  46                 // Abbrevs
  47                 $this->assertEquals('http',  scheme_normalize('ttp'));
  48                 $this->assertEquals('https', scheme_normalize('ttps'));
  49
  50                 // Abbrevs considererd harmless
  51                 $this->assertEquals('', scheme_normalize('ttp',  FALSE));
  52                 $this->assertEquals('', scheme_normalize('ttps', FALSE));
  53         }
  54
  55         function testFunc_host_normalize()
  56         {
  57                 // Invalid: Null
  58                 foreach($this->setup_string_null() as $key => $value){
  59                         $this->assertEquals('', host_normalize($value), $key);
  60                 }
  61
  62                 // Hostname is case-insensitive
  63                 $this->assertEquals('example.org', host_normalize('ExAMPle.ORG'));
  64
  65                 // Cut 'www' with traditional ASCII-based FQDN (destructive)
  66                 $this->assertEquals('example.org', host_normalize('WWW.example.org'));
  67
  68                 // Don't cut 'www' with Non-ASCII-based string such as IDN
  69                 $this->assertEquals("www.example.org\0foobar",
  70                          host_normalize("WWW.example.org\0foobar"));
  71         }
  72
  73         function testFunc_port_normalize()
  74         {
  75                 $scheme = 'dont_care';
  76
  77                 // 1st argument: Null
  78                 $this->assertEquals('', port_normalize(NULL, $scheme));
  79                 $this->assertEquals('', port_normalize(TRUE, $scheme));
  80                 $this->assertEquals('', port_normalize(FALSE, $scheme));
  81                 $this->assertEquals('', port_normalize(array('foobar'), $scheme));
  82                 $this->assertEquals('', port_normalize('',   $scheme));
  83
  84                 // 1st argument: Known port
  85                 $this->assertEquals('',    port_normalize(   -1, $scheme));
  86                 $this->assertEquals(0,     port_normalize(    0, $scheme));
  87                 $this->assertEquals(1,     port_normalize(    1, $scheme));
  88                 $this->assertEquals('',    port_normalize(   21, 'ftp'));
  89                 $this->assertEquals('',    port_normalize(   22, 'ssh'));
  90                 $this->assertEquals('',    port_normalize(   23, 'telnet'));
  91                 $this->assertEquals('',    port_normalize(   25, 'smtp'));
  92                 $this->assertEquals('',    port_normalize(   69, 'tftp'));
  93                 $this->assertEquals('',    port_normalize(   70, 'gopher'));
  94                 $this->assertEquals('',    port_normalize(   79, 'finger'));
  95                 $this->assertEquals('',    port_normalize(   80, 'http'));
  96                 $this->assertEquals('',    port_normalize(  110, 'pop3'));
  97                 $this->assertEquals('',    port_normalize(  115, 'sftp'));
  98                 $this->assertEquals('',    port_normalize(  119, 'nntp'));
  99                 $this->assertEquals('',    port_normalize(  143, 'imap'));
 100                 $this->assertEquals('',    port_normalize(  194, 'irc'));
 101                 $this->assertEquals('',    port_normalize(  210, 'wais'));
 102                 $this->assertEquals('',    port_normalize(  443, 'https'));
 103                 $this->assertEquals('',    port_normalize(  563, 'nntps'));
 104                 $this->assertEquals('',    port_normalize(  873, 'rsync'));
 105                 $this->assertEquals('',    port_normalize(  990, 'ftps'));
 106                 $this->assertEquals('',    port_normalize(  992, 'telnets'));
 107                 $this->assertEquals('',    port_normalize(  993, 'imaps'));
 108                 $this->assertEquals('',    port_normalize(  994, 'ircs'));
 109                 $this->assertEquals('',    port_normalize(  995, 'pop3s'));
 110                 $this->assertEquals('',    port_normalize( 3306, 'mysql'));
 111                 $this->assertEquals(8080,  port_normalize( 8080, $scheme));
 112                 $this->assertEquals(65535, port_normalize(65535, $scheme));
 113                 $this->assertEquals(65536, port_normalize(65536, $scheme)); // Seems not invalid in RFC
 114
 115                 // 1st argument: Invalid type
 116                 $this->assertEquals('1x',  port_normalize('001', $scheme) . 'x');
 117                 $this->assertEquals('',    port_normalize('+0',  $scheme));
 118                 $this->assertEquals('',    port_normalize('0-1', $scheme)); // intval() says '0'
 119                 $this->assertEquals('',    port_normalize('str', $scheme));
 120
 121                 // 2nd and 3rd argument: Null
 122                 $this->assertEquals(80,    port_normalize(80, NULL,  TRUE));
 123                 $this->assertEquals(80,    port_normalize(80, TRUE,  TRUE));
 124                 $this->assertEquals(80,    port_normalize(80, FALSE, TRUE));
 125                 $this->assertEquals(80,    port_normalize(80, array('foobar'), TRUE));
 126                 $this->assertEquals(80,    port_normalize(80, '', TRUE));
 127
 128                 // 2nd and 3rd argument: Do $scheme_normalize
 129                 $this->assertEquals('',    port_normalize(80,  'TTP',  TRUE));
 130                 $this->assertEquals('',    port_normalize(110, 'POP',  TRUE));
 131                 $this->assertEquals(80,    port_normalize(80,  'HTTP', FALSE));
 132         }
 133
 134         function testFunc_path_normalize()
 135         {
 136                 // 1st argument: Null
 137                 foreach($this->setup_string_null() as $key => $value){
 138                         $this->assertEquals('/', path_normalize($value), $key);
 139                 }
 140
 141                 // 1st argument: CASE sensitive
 142                 $this->assertEquals('/ExAMPle', path_normalize('ExAMPle'));
 143                 $this->assertEquals('/#hoge',   path_normalize('#hoge'));
 144                 $this->assertEquals('/a/b/c/d', path_normalize('/a/b/./c////./d'));
 145                 $this->assertEquals('/b/',      path_normalize('/a/../../../b/'));
 146
 147                 // 2nd argument
 148                 $this->assertEquals('\\b\\c\\d\\', path_normalize('\\a\\..\\b\\.\\c\\\\.\\d\\', '\\'));
 149                 $this->assertEquals('str1str3str', path_normalize('str1strstr2str..str3str', 'str'));
 150                 $this->assertEquals('/do/../nothing/', path_normalize('/do/../nothing/', TRUE));
 151                 $this->assertEquals('/do/../nothing/', path_normalize('/do/../nothing/', array('a')));
 152                 $this->assertEquals('',            path_normalize(array('a'), array('b')));
 153         }
 154
 155         function testFunc_query_normalize()
 156         {
 157                 // 1st argument: Null
 158                 foreach($this->setup_string_null() as $key => $value){
 159                         $this->assertEquals('', query_normalize($value), $key);
 160                 }
 161
 162                 $this->assertEquals('a=0dd&b&c&d&f=d', query_normalize('&&&&f=d&b&d&c&a=0dd'));
 163                 $this->assertEquals('eg=foobar',       query_normalize('nothing==&eg=dummy&eg=padding&eg=foobar'));
 164         }
 165
 166         function testFunc_file_normalize()
 167         {
 168                 // 1st argument: Null
 169                 foreach($this->setup_string_null() as $key => $value){
 170                         $this->assertEquals('', file_normalize($value), $key);
 171                 }
 172
 173                 // 1st argument: Cut DirectoryIndexes (Destructive)
 174                 foreach(array(
 175                         'default.htm',
 176                         'default.html',
 177                         'default.asp',
 178                         'default.aspx',
 179 \r                       'index',
 180                         'index.htm',
 181                         'index.html',
 182                         'index.shtml',
 183                         'index.jsp',
 184                         'index.php',
 185                         'index.php',
 186                         'index.php3',
 187                         'index.php4',
 188                         'index.pl',
 189                         'index.py',
 190                         'index.rb',
 191                         'index.cgi',
 192
 193                         // Apache 2.0.59 default 'index.html' variants
 194                         'index.html.ca',
 195                         'index.html.cz.iso8859-2',
 196                         'index.html.de',
 197                         'index.html.dk',
 198                         'index.html.ee',
 199                         'index.html.el',
 200                         'index.html.en',
 201                         'index.html.es',
 202                         'index.html.et',
 203                         'index.html.fr',
 204                         'index.html.he.iso8859-8',
 205                         'index.html.hr.iso8859-2',
 206                         'index.html.it',
 207                         'index.html.ja.iso2022-jp',
 208                         'index.html.ko.euc-kr',
 209                         'index.html.lb.utf8',
 210                         'index.html.nl',
 211                         'index.html.nn',
 212                         'index.html.no',
 213                         'index.html.po.iso8859-2',
 214                         'index.html.pt',
 215                         'index.html.pt-br',
 216                         'index.html.ru.cp866',
 217                         'index.html.ru.cp-1251',
 218                         'index.html.ru.iso-ru',
 219                         'index.html.ru.koi8-r',
 220                         'index.html.ru.utf8',
 221                         'index.html.sv',
 222                         'index.html.var',       // default
 223                         'index.html.zh-cn.gb2312',
 224                         'index.html.zh-tw.big5',
 225
 226                         'index.html.po.iso8859-2',
 227                         'index.html.zh-tw.big5',
 228
 229                         'index.ja.en.de.html',
 230
 231                         // .gz
 232                         'index.html.ca.gz',
 233                         'index.html.en.ja.ca.z',
 234                 ) as $arg){
 235                         $this->assertEquals('', file_normalize($arg));
 236                 }
 237
 238                 //$this->assertEquals('foo/', file_normalize('foo/index.html'));
 239
 240                 //$this->assertEquals('ExAMPle', file_normalize('ExAMPle'));
 241                 //$this->assertEquals('exe.exe', file_normalize('exe.exe'));
 242                 //$this->assertEquals('sample.html', file_normalize('sample.html.en'));
 243                 //$this->assertEquals('sample.html', file_normalize('sample.html.pt-br'));
 244                 //$this->assertEquals('sample.html', file_normalize('sample.html.po.iso8859-2'));
 245                 //$this->assertEquals('sample.html', file_normalize('sample.html.zh-tw.big5'));
 246         }
 247
 248         function testFunc_uri_pickup()
 249         {
 250                 // 1st argument: Null
 251                 foreach($this->setup_string_null() as $key => $value){
 252                         $this->assertEquals(0, count(uri_pickup($value)), $key);
 253                 }
 254
 255                 // 1st argument: Some
 256                 $test_string = <<<EOF
 257                         TTP://wwW.Example.Org#TTP_and_www
 258                         https://nasty.example.org:443/foo/xxx#port443/slash
 259                         sftp://foobar.example.org:80/dfsdfs#ftp_bat_port80
 260                         ftp://cnn.example.com&story=breaking_news@10.0.0.1/top_story.htm
 261                         http://192.168.1.4:443#IPv4
 262 EOF;
 263                 $results = uri_pickup_normalize(uri_pickup($test_string));
 264                 $this->assertEquals(5, count($results));
 265
 266                 // ttp://wwW.Example.Org:80#TTP_and_www
 267                 $this->assertEquals('http',           $results[0]['scheme']);
 268                 $this->assertEquals('',               $results[0]['userinfo']);
 269                 $this->assertEquals('example.org',    $results[0]['host']);
 270                 $this->assertEquals('',               $results[0]['port']);
 271                 $this->assertEquals('/',              $results[0]['path']);
 272                 $this->assertEquals('',               $results[0]['file']);
 273                 $this->assertEquals('',               $results[0]['query']);
 274                 $this->assertEquals('ttp_and_www',    $results[0]['fragment']);
 275
 276                 // https://nasty.example.org:443/foo/xxx#port443/slash
 277                 $this->assertEquals('https',          $results[1]['scheme']);
 278                 $this->assertEquals('',               $results[1]['userinfo']);
 279                 $this->assertEquals('nasty.example.org', $results[1]['host']);
 280                 $this->assertEquals('',               $results[1]['port']);
 281                 $this->assertEquals('/foo/',          $results[1]['path']);
 282                 $this->assertEquals('xxx',            $results[1]['file']);
 283                 $this->assertEquals('',               $results[1]['query']);
 284                 $this->assertEquals('port443',        $results[1]['fragment']);
 285
 286                 // sftp://foobar.example.org:80/dfsdfs#sftp_bat_port80
 287                 $this->assertEquals('sftp',           $results[2]['scheme']);
 288                 $this->assertEquals('',               $results[2]['userinfo']);
 289                 $this->assertEquals('foobar.example.org', $results[2]['host']);
 290                 $this->assertEquals('80',             $results[2]['port']);
 291                 $this->assertEquals('/',              $results[2]['path']);
 292                 $this->assertEquals('dfsdfs',         $results[2]['file']);
 293                 $this->assertEquals('',               $results[2]['query']);
 294                 $this->assertEquals('ftp_bat_port80', $results[2]['fragment']);
 295
 296                 // ftp://cnn.example.com&story=breaking_news@10.0.0.1/top_story.htm
 297                 $this->assertEquals('ftp',            $results[3]['scheme']);
 298                 $this->assertEquals('cnn.example.com&story=breaking_news', $results[3]['userinfo']);
 299                 $this->assertEquals('10.0.0.1',       $results[3]['host']);
 300                 $this->assertEquals('',               $results[3]['port']);
 301                 $this->assertEquals('/',              $results[3]['path']);
 302                 $this->assertEquals('top_story.htm',  $results[3]['file']);
 303                 $this->assertEquals('',               $results[3]['query']);
 304                 $this->assertEquals('',               $results[3]['fragment']);
 305
 306
 307                 // Specific tests ----
 308
 309                 // Divider: Back-slash
 310                 $test_string = ' http:\\backslash.org\fobar.html ';
 311                 $results = uri_pickup_normalize(uri_pickup($test_string));
 312                 $this->assertEquals('backslash.org',  $results[0]['host']);
 313
 314                 // Divider: percent-encoded
 315                 //$test_string = ' http%3A%2F%5Cpercent-encoded.org%5Cfobar.html ';
 316                 //$results = uri_pickup_normalize(uri_pickup($test_string));
 317                 //$this->assertEquals('percent-encoded.org',  $results[0]['host']);
 318
 319                 // Host: Without path
 320                 $test_string = ' http://nopathstring.com ';
 321                 $results = uri_pickup($test_string);
 322                 $this->assertEquals('', $results[0]['path']);
 323                 $this->assertEquals('', $results[0]['file']);
 324                 $results[0]['path'] = '/';
 325                 $this->assertEquals('', $results[0]['file'], '[Seems referense trouble]');
 326                 //
 327                 $results = uri_pickup($test_string);
 328                 $results = uri_pickup_normalize($results);
 329                 $this->assertEquals('/',$results[0]['path']);
 330                 $this->assertEquals('', $results[0]['file']);
 331
 332                 // Host: Underscore
 333                 $test_string = ' http://under_score.org/fobar.html ';
 334                 $results = uri_pickup_normalize(uri_pickup($test_string));
 335                 $this->assertEquals('under_score.org',$results[0]['host']);     // Not 'under'
 336
 337                 // Host: IPv4
 338                 $test_string = ' http://192.168.0.1/fobar.html ';
 339                 $results = uri_pickup_normalize(uri_pickup($test_string));
 340                 $this->assertEquals('192.168.0.1',    $results[0]['host']);
 341
 342                 // Host: Starts
 343                 $test_string = ' http://_sss/foo.html ';
 344                 $results = uri_pickup_normalize(uri_pickup($test_string));
 345                 $this->assertEquals('_sss',           $results[0]['host']);
 346                 $this->assertEquals('foo.html',       $results[0]['file']);
 347
 348                 // Host: Ends
 349                 $test_string = ' http://sss_/foo.html ';
 350                 $results = uri_pickup_normalize(uri_pickup($test_string));
 351                 $this->assertEquals('sss_',           $results[0]['host']);
 352                 $this->assertEquals('foo.html',       $results[0]['file']);
 353
 354
 355                 // Specific tests ---- Fails
 356
 357                 // Divider: Colon only (Too sensitive to capture)
 358                 $test_string = ' http:colon.org ';
 359                 $results = uri_pickup_normalize(uri_pickup($test_string));
 360                 $this->assertEquals(0, count($results));
 361
 362                 // Host: Too short
 363                 $test_string = ' http://s/foo.html http://ss/foo.html ';
 364                 $results = uri_pickup_normalize(uri_pickup($test_string));
 365                 $this->assertEquals(0, count($results));
 366
 367                 $test_string = ' http://sss/foo.html ';
 368                 $results = uri_pickup_normalize(uri_pickup($test_string));
 369                 $this->assertEquals('sss',            $results[0]['host']);
 370                 $this->assertEquals('foo.html',       $results[0]['file']);
 371
 372                 // uri_pickup_normalize_pathtofile()
 373                 $test_string = ' http://example.com/path/to/directory-accidentally-not-ended-with-slash ';
 374                 $results = uri_pickup_normalize_pathtofile(uri_pickup($test_string));
 375                 $this->assertEquals('/path/to/directory-accidentally-not-ended-with-slash',
 376                         $results[0]['pathtofile']);
 377                 $this->assertEquals(FALSE, isset($results[0]['path']));
 378                 $this->assertEquals(FALSE, isset($results[0]['file']));
 379         }
 380
 381         function testFunc_spam_uri_pickup()
 382         {
 383                 // Divider: percent-encoded
 384                 $test_string = ' http://victim.example.org/http%3A%2F%5Cnasty.example.org ';
 385                 $results = spam_uri_pickup($test_string);
 386                 $this->assertEquals('victim.example.org', $results[0]['host']);
 387                 $this->assertEquals('nasty.example.org',  $results[1]['host']);
 388
 389                 // Domain exposure (site:)
 390                 $test_string = ' http://search.example.org/?q=%20site:nasty.example.org ';
 391                 $results = spam_uri_pickup($test_string);
 392                 $this->assertEquals('nasty.example.org', $results[0]['host']);
 393                 $this->assertEquals('search.example.org',  $results[1]['host']);
 394
 395                 // Domain exposure (%20site:)
 396                 $test_string = ' http://search2.example.org/?q=%20site:nasty2.example.org ';
 397                 $results = spam_uri_pickup($test_string);
 398                 $this->assertEquals('nasty2.example.org', $results[0]['host']);
 399                 $this->assertEquals('search2.example.org',  $results[1]['host']);
 400         }
 401 }
 402
 403 ?>