tests/phpunit/includes/normal/CleanUpTest.php

   1 <?php
   2 /**
   3  * Tests for UtfNormal::cleanUp() function.
   4  *
   5  * Copyright © 2004 Brion Vibber <brion@pobox.com>
   6  * https://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  */
  25
  26 /**
  27  * Additional tests for UtfNormal::cleanUp() function, inclusion
  28  * regression checks for known problems.
  29  * Requires PHPUnit.
  30  *
  31  * @ingroup UtfNormal
  32  * @group Large
  33  *
  34  * @todo covers tags, will be UtfNormal::cleanUp once the below is resolved
  35  * @todo split me into test methods and providers per the below comment
  36  * @todo Document individual tests
  37  *
  38  * We ignore code coverage for this test suite until they are rewritten
  39  * to use data providers (bug 46561).
  40  * @codeCoverageIgnore
  41  */
  42 class CleanUpTest extends MediaWikiTestCase {
  43         public function testAscii() {
  44                 $text = 'This is plain ASCII text.';
  45                 $this->assertEquals( $text, UtfNormal::cleanUp( $text ) );
  46         }
  47
  48         public function testNull() {
  49                 $text = "a \x00 null";
  50                 $expect = "a \xef\xbf\xbd null";
  51                 $this->assertEquals(
  52                         bin2hex( $expect ),
  53                         bin2hex( UtfNormal::cleanUp( $text ) ) );
  54         }
  55
  56         public function testLatin() {
  57                 $text = "L'\xc3\xa9cole";
  58                 $this->assertEquals( $text, UtfNormal::cleanUp( $text ) );
  59         }
  60
  61         public function testLatinNormal() {
  62                 $text = "L'e\xcc\x81cole";
  63                 $expect = "L'\xc3\xa9cole";
  64                 $this->assertEquals( $expect, UtfNormal::cleanUp( $text ) );
  65         }
  66
  67         /**
  68          * This test is *very* expensive!
  69          */
  70         function XtestAllChars() {
  71                 $rep = UTF8_REPLACEMENT;
  72                 for ( $i = 0x0; $i < UNICODE_MAX; $i++ ) {
  73                         $char = codepointToUtf8( $i );
  74                         $clean = UtfNormal::cleanUp( $char );
  75                         $x = sprintf( "%04X", $i );
  76
  77                         if ( $i % 0x1000 == 0 ) {
  78                                 echo "U+$x\n";
  79                         }
  80
  81                         if ( $i == 0x0009 ||
  82                                 $i == 0x000a ||
  83                                 $i == 0x000d ||
  84                                 ( $i > 0x001f && $i < UNICODE_SURROGATE_FIRST ) ||
  85                                 ( $i > UNICODE_SURROGATE_LAST && $i < 0xfffe ) ||
  86                                 ( $i > 0xffff && $i <= UNICODE_MAX )
  87                         ) {
  88                                 if ( isset( UtfNormal::$utfCanonicalComp[$char] )
  89                                         || isset( UtfNormal::$utfCanonicalDecomp[$char] )
  90                                 ) {
  91                                         $comp = UtfNormal::NFC( $char );
  92                                         $this->assertEquals(
  93                                                 bin2hex( $comp ),
  94                                                 bin2hex( $clean ),
  95                                                 "U+$x should be decomposed" );
  96                                 } else {
  97                                         $this->assertEquals(
  98                                                 bin2hex( $char ),
  99                                                 bin2hex( $clean ),
 100                                                 "U+$x should be intact" );
 101                                 }
 102                         } else {
 103                                 $this->assertEquals( bin2hex( $rep ), bin2hex( $clean ), $x );
 104                         }
 105                 }
 106         }
 107
 108         public static function provideAllBytes() {
 109                 return array(
 110                         array( '', '' ),
 111                         array( 'x', '' ),
 112                         array( '', 'x' ),
 113                         array( 'x', 'x' ),
 114                 );
 115         }
 116
 117         /**
 118          * @dataProvider provideAllBytes
 119          */
 120         function testBytes( $head, $tail ) {
 121                 for ( $i = 0x0; $i < 256; $i++ ) {
 122                         $char = $head . chr( $i ) . $tail;
 123                         $clean = UtfNormal::cleanUp( $char );
 124                         $x = sprintf( "%02X", $i );
 125
 126                         if ( $i == 0x0009 ||
 127                                 $i == 0x000a ||
 128                                 $i == 0x000d ||
 129                                 ( $i > 0x001f && $i < 0x80 )
 130                         ) {
 131                                 $this->assertEquals(
 132                                         bin2hex( $char ),
 133                                         bin2hex( $clean ),
 134                                         "ASCII byte $x should be intact" );
 135                                 if ( $char != $clean ) {
 136                                         return;
 137                                 }
 138                         } else {
 139                                 $norm = $head . UTF8_REPLACEMENT . $tail;
 140                                 $this->assertEquals(
 141                                         bin2hex( $norm ),
 142                                         bin2hex( $clean ),
 143                                         "Forbidden byte $x should be rejected" );
 144                                 if ( $norm != $clean ) {
 145                                         return;
 146                                 }
 147                         }
 148                 }
 149         }
 150
 151         /**
 152          * @dataProvider provideAllBytes
 153          */
 154         function testDoubleBytes( $head, $tail ) {
 155                 for ( $first = 0xc0; $first < 0x100; $first += 2 ) {
 156                         for ( $second = 0x80; $second < 0x100; $second += 2 ) {
 157                                 $char = $head . chr( $first ) . chr( $second ) . $tail;
 158                                 $clean = UtfNormal::cleanUp( $char );
 159                                 $x = sprintf( "%02X,%02X", $first, $second );
 160                                 if ( $first > 0xc1 &&
 161                                         $first < 0xe0 &&
 162                                         $second < 0xc0
 163                                 ) {
 164                                         $norm = UtfNormal::NFC( $char );
 165                                         $this->assertEquals(
 166                                                 bin2hex( $norm ),
 167                                                 bin2hex( $clean ),
 168                                                 "Pair $x should be intact" );
 169                                         if ( $norm != $clean ) {
 170                                                 return;
 171                                         }
 172                                 } elseif ( $first > 0xfd || $second > 0xbf ) {
 173                                         # fe and ff are not legal head bytes -- expect two replacement chars
 174                                         $norm = $head . UTF8_REPLACEMENT . UTF8_REPLACEMENT . $tail;
 175                                         $this->assertEquals(
 176                                                 bin2hex( $norm ),
 177                                                 bin2hex( $clean ),
 178                                                 "Forbidden pair $x should be rejected" );
 179                                         if ( $norm != $clean ) {
 180                                                 return;
 181                                         }
 182                                 } else {
 183                                         $norm = $head . UTF8_REPLACEMENT . $tail;
 184                                         $this->assertEquals(
 185                                                 bin2hex( $norm ),
 186                                                 bin2hex( $clean ),
 187                                                 "Forbidden pair $x should be rejected" );
 188                                         if ( $norm != $clean ) {
 189                                                 return;
 190                                         }
 191                                 }
 192                         }
 193                 }
 194         }
 195
 196         /**
 197          * @dataProvider provideAllBytes
 198          */
 199         function testTripleBytes( $head, $tail ) {
 200                 for ( $first = 0xc0; $first < 0x100; $first += 2 ) {
 201                         for ( $second = 0x80; $second < 0x100; $second += 2 ) {
 202                                 #for( $third = 0x80; $third < 0x100; $third++ ) {
 203                                 for ( $third = 0x80; $third < 0x81; $third++ ) {
 204                                         $char = $head . chr( $first ) . chr( $second ) . chr( $third ) . $tail;
 205                                         $clean = UtfNormal::cleanUp( $char );
 206                                         $x = sprintf( "%02X,%02X,%02X", $first, $second, $third );
 207
 208                                         if ( $first >= 0xe0 &&
 209                                                 $first < 0xf0 &&
 210                                                 $second < 0xc0 &&
 211                                                 $third < 0xc0
 212                                         ) {
 213                                                 if ( $first == 0xe0 && $second < 0xa0 ) {
 214                                                         $this->assertEquals(
 215                                                                 bin2hex( $head . UTF8_REPLACEMENT . $tail ),
 216                                                                 bin2hex( $clean ),
 217                                                                 "Overlong triplet $x should be rejected" );
 218                                                 } elseif ( $first == 0xed &&
 219                                                         ( chr( $first ) . chr( $second ) . chr( $third ) ) >= UTF8_SURROGATE_FIRST
 220                                                 ) {
 221                                                         $this->assertEquals(
 222                                                                 bin2hex( $head . UTF8_REPLACEMENT . $tail ),
 223                                                                 bin2hex( $clean ),
 224                                                                 "Surrogate triplet $x should be rejected" );
 225                                                 } else {
 226                                                         $this->assertEquals(
 227                                                                 bin2hex( UtfNormal::NFC( $char ) ),
 228                                                                 bin2hex( $clean ),
 229                                                                 "Triplet $x should be intact" );
 230                                                 }
 231                                         } elseif ( $first > 0xc1 && $first < 0xe0 && $second < 0xc0 ) {
 232                                                 $this->assertEquals(
 233                                                         bin2hex( UtfNormal::NFC( $head . chr( $first ) .
 234                                                                         chr( $second ) ) . UTF8_REPLACEMENT . $tail ),
 235                                                         bin2hex( $clean ),
 236                                                         "Valid 2-byte $x + broken tail" );
 237                                         } elseif ( $second > 0xc1 && $second < 0xe0 && $third < 0xc0 ) {
 238                                                 $this->assertEquals(
 239                                                         bin2hex( $head . UTF8_REPLACEMENT .
 240                                                                 UtfNormal::NFC( chr( $second ) . chr( $third ) . $tail ) ),
 241                                                         bin2hex( $clean ),
 242                                                         "Broken head + valid 2-byte $x" );
 243                                         } elseif ( ( $first > 0xfd || $second > 0xfd ) &&
 244                                                 ( ( $second > 0xbf && $third > 0xbf ) ||
 245                                                         ( $second < 0xc0 && $third < 0xc0 ) ||
 246                                                         ( $second > 0xfd ) ||
 247                                                         ( $third > 0xfd ) )
 248                                         ) {
 249                                                 # fe and ff are not legal head bytes -- expect three replacement chars
 250                                                 $this->assertEquals(
 251                                                         bin2hex( $head . UTF8_REPLACEMENT . UTF8_REPLACEMENT . UTF8_REPLACEMENT . $tail ),
 252                                                         bin2hex( $clean ),
 253                                                         "Forbidden triplet $x should be rejected" );
 254                                         } elseif ( $first > 0xc2 && $second < 0xc0 && $third < 0xc0 ) {
 255                                                 $this->assertEquals(
 256                                                         bin2hex( $head . UTF8_REPLACEMENT . $tail ),
 257                                                         bin2hex( $clean ),
 258                                                         "Forbidden triplet $x should be rejected" );
 259                                         } else {
 260                                                 $this->assertEquals(
 261                                                         bin2hex( $head . UTF8_REPLACEMENT . UTF8_REPLACEMENT . $tail ),
 262                                                         bin2hex( $clean ),
 263                                                         "Forbidden triplet $x should be rejected" );
 264                                         }
 265                                 }
 266                         }
 267                 }
 268         }
 269
 270         public function testChunkRegression() {
 271                 # Check for regression against a chunking bug
 272                 $text = "\x46\x55\xb8" .
 273                         "\xdc\x96" .
 274                         "\xee" .
 275                         "\xe7" .
 276                         "\x44" .
 277                         "\xaa" .
 278                         "\x2f\x25";
 279                 $expect = "\x46\x55\xef\xbf\xbd" .
 280                         "\xdc\x96" .
 281                         "\xef\xbf\xbd" .
 282                         "\xef\xbf\xbd" .
 283                         "\x44" .
 284                         "\xef\xbf\xbd" .
 285                         "\x2f\x25";
 286
 287                 $this->assertEquals(
 288                         bin2hex( $expect ),
 289                         bin2hex( UtfNormal::cleanUp( $text ) ) );
 290         }
 291
 292         public function testInterposeRegression() {
 293                 $text = "\x4e\x30" .
 294                         "\xb1" . # bad tail
 295                         "\x3a" .
 296                         "\x92" . # bad tail
 297                         "\x62\x3a" .
 298                         "\x84" . # bad tail
 299                         "\x43" .
 300                         "\xc6" . # bad head
 301                         "\x3f" .
 302                         "\x92" . # bad tail
 303                         "\xad" . # bad tail
 304                         "\x7d" .
 305                         "\xd9\x95";
 306
 307                 $expect = "\x4e\x30" .
 308                         "\xef\xbf\xbd" .
 309                         "\x3a" .
 310                         "\xef\xbf\xbd" .
 311                         "\x62\x3a" .
 312                         "\xef\xbf\xbd" .
 313                         "\x43" .
 314                         "\xef\xbf\xbd" .
 315                         "\x3f" .
 316                         "\xef\xbf\xbd" .
 317                         "\xef\xbf\xbd" .
 318                         "\x7d" .
 319                         "\xd9\x95";
 320
 321                 $this->assertEquals(
 322                         bin2hex( $expect ),
 323                         bin2hex( UtfNormal::cleanUp( $text ) ) );
 324         }
 325
 326         public function testOverlongRegression() {
 327                 $text = "\x67" .
 328                         "\x1a" . # forbidden ascii
 329                         "\xea" . # bad head
 330                         "\xc1\xa6" . # overlong sequence
 331                         "\xad" . # bad tail
 332                         "\x1c" . # forbidden ascii
 333                         "\xb0" . # bad tail
 334                         "\x3c" .
 335                         "\x9e"; # bad tail
 336                 $expect = "\x67" .
 337                         "\xef\xbf\xbd" .
 338                         "\xef\xbf\xbd" .
 339                         "\xef\xbf\xbd" .
 340                         "\xef\xbf\xbd" .
 341                         "\xef\xbf\xbd" .
 342                         "\xef\xbf\xbd" .
 343                         "\x3c" .
 344                         "\xef\xbf\xbd";
 345                 $this->assertEquals(
 346                         bin2hex( $expect ),
 347                         bin2hex( UtfNormal::cleanUp( $text ) ) );
 348         }
 349
 350         public function testSurrogateRegression() {
 351                 $text = "\xed\xb4\x96" . # surrogate 0xDD16
 352                         "\x83" . # bad tail
 353                         "\xb4" . # bad tail
 354                         "\xac"; # bad head
 355                 $expect = "\xef\xbf\xbd" .
 356                         "\xef\xbf\xbd" .
 357                         "\xef\xbf\xbd" .
 358                         "\xef\xbf\xbd";
 359                 $this->assertEquals(
 360                         bin2hex( $expect ),
 361                         bin2hex( UtfNormal::cleanUp( $text ) ) );
 362         }
 363
 364         public function testBomRegression() {
 365                 $text = "\xef\xbf\xbe" . # U+FFFE, illegal char
 366                         "\xb2" . # bad tail
 367                         "\xef" . # bad head
 368                         "\x59";
 369                 $expect = "\xef\xbf\xbd" .
 370                         "\xef\xbf\xbd" .
 371                         "\xef\xbf\xbd" .
 372                         "\x59";
 373                 $this->assertEquals(
 374                         bin2hex( $expect ),
 375                         bin2hex( UtfNormal::cleanUp( $text ) ) );
 376         }
 377
 378         public function testForbiddenRegression() {
 379                 $text = "\xef\xbf\xbf"; # U+FFFF, illegal char
 380                 $expect = "\xef\xbf\xbd";
 381                 $this->assertEquals(
 382                         bin2hex( $expect ),
 383                         bin2hex( UtfNormal::cleanUp( $text ) ) );
 384         }
 385
 386         public function testHangulRegression() {
 387                 $text = "\xed\x9c\xaf" . # Hangul char
 388                         "\xe1\x87\x81"; # followed by another final jamo
 389                 $expect = $text; # Should *not* change.
 390                 $this->assertEquals(
 391                         bin2hex( $expect ),
 392                         bin2hex( UtfNormal::cleanUp( $text ) ) );
 393         }
 394 }