Chris@0: <?php
Chris@0: 
Chris@0: namespace PhpParser;
Chris@0: 
Chris@0: use PhpParser\Parser\Tokens;
Chris@0: 
Chris@0: class LexerTest extends \PHPUnit_Framework_TestCase
Chris@0: {
Chris@0:     /* To allow overwriting in parent class */
Chris@0:     protected function getLexer(array $options = array()) {
Chris@0:         return new Lexer($options);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * @dataProvider provideTestError
Chris@0:      */
Chris@0:     public function testError($code, $messages) {
Chris@0:         if (defined('HHVM_VERSION')) {
Chris@0:             $this->markTestSkipped('HHVM does not throw warnings from token_get_all()');
Chris@0:         }
Chris@0: 
Chris@0:         $errorHandler = new ErrorHandler\Collecting();
Chris@0:         $lexer = $this->getLexer(['usedAttributes' => [
Chris@0:             'comments', 'startLine', 'endLine', 'startFilePos', 'endFilePos'
Chris@0:         ]]);
Chris@0:         $lexer->startLexing($code, $errorHandler);
Chris@0:         $errors = $errorHandler->getErrors();
Chris@0: 
Chris@0:         $this->assertSame(count($messages), count($errors));
Chris@0:         for ($i = 0; $i < count($messages); $i++) {
Chris@0:             $this->assertSame($messages[$i], $errors[$i]->getMessageWithColumnInfo($code));
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     public function provideTestError() {
Chris@0:         return array(
Chris@0:             array("<?php /*", array("Unterminated comment from 1:7 to 1:9")),
Chris@0:             array("<?php \1", array("Unexpected character \"\1\" (ASCII 1) from 1:7 to 1:7")),
Chris@0:             array("<?php \0", array("Unexpected null byte from 1:7 to 1:7")),
Chris@0:             // Error with potentially emulated token
Chris@0:             array("<?php ?? \0", array("Unexpected null byte from 1:10 to 1:10")),
Chris@0:             array("<?php\n\0\1 foo /* bar", array(
Chris@0:                 "Unexpected null byte from 2:1 to 2:1",
Chris@0:                 "Unexpected character \"\1\" (ASCII 1) from 2:2 to 2:2",
Chris@0:                 "Unterminated comment from 2:8 to 2:14"
Chris@0:             )),
Chris@0:         );
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * @dataProvider provideTestLex
Chris@0:      */
Chris@0:     public function testLex($code, $options, $tokens) {
Chris@0:         $lexer = $this->getLexer($options);
Chris@0:         $lexer->startLexing($code);
Chris@0:         while ($id = $lexer->getNextToken($value, $startAttributes, $endAttributes)) {
Chris@0:             $token = array_shift($tokens);
Chris@0: 
Chris@0:             $this->assertSame($token[0], $id);
Chris@0:             $this->assertSame($token[1], $value);
Chris@0:             $this->assertEquals($token[2], $startAttributes);
Chris@0:             $this->assertEquals($token[3], $endAttributes);
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     public function provideTestLex() {
Chris@0:         return array(
Chris@0:             // tests conversion of closing PHP tag and drop of whitespace and opening tags
Chris@0:             array(
Chris@0:                 '<?php tokens ?>plaintext',
Chris@0:                 array(),
Chris@0:                 array(
Chris@0:                     array(
Chris@0:                         Tokens::T_STRING, 'tokens',
Chris@0:                         array('startLine' => 1), array('endLine' => 1)
Chris@0:                     ),
Chris@0:                     array(
Chris@0:                         ord(';'), '?>',
Chris@0:                         array('startLine' => 1), array('endLine' => 1)
Chris@0:                     ),
Chris@0:                     array(
Chris@0:                         Tokens::T_INLINE_HTML, 'plaintext',
Chris@0:                         array('startLine' => 1, 'hasLeadingNewline' => false),
Chris@0:                         array('endLine' => 1)
Chris@0:                     ),
Chris@0:                 )
Chris@0:             ),
Chris@0:             // tests line numbers
Chris@0:             array(
Chris@0:                 '<?php' . "\n" . '$ token /** doc' . "\n" . 'comment */ $',
Chris@0:                 array(),
Chris@0:                 array(
Chris@0:                     array(
Chris@0:                         ord('$'), '$',
Chris@0:                         array('startLine' => 2), array('endLine' => 2)
Chris@0:                     ),
Chris@0:                     array(
Chris@0:                         Tokens::T_STRING, 'token',
Chris@0:                         array('startLine' => 2), array('endLine' => 2)
Chris@0:                     ),
Chris@0:                     array(
Chris@0:                         ord('$'), '$',
Chris@0:                         array(
Chris@0:                             'startLine' => 3,
Chris@0:                             'comments' => array(
Chris@0:                                 new Comment\Doc('/** doc' . "\n" . 'comment */', 2, 14),
Chris@0:                             )
Chris@0:                         ),
Chris@0:                         array('endLine' => 3)
Chris@0:                     ),
Chris@0:                 )
Chris@0:             ),
Chris@0:             // tests comment extraction
Chris@0:             array(
Chris@0:                 '<?php /* comment */ // comment' . "\n" . '/** docComment 1 *//** docComment 2 */ token',
Chris@0:                 array(),
Chris@0:                 array(
Chris@0:                     array(
Chris@0:                         Tokens::T_STRING, 'token',
Chris@0:                         array(
Chris@0:                             'startLine' => 2,
Chris@0:                             'comments' => array(
Chris@0:                                 new Comment('/* comment */', 1, 6),
Chris@0:                                 new Comment('// comment' . "\n", 1, 20),
Chris@0:                                 new Comment\Doc('/** docComment 1 */', 2, 31),
Chris@0:                                 new Comment\Doc('/** docComment 2 */', 2, 50),
Chris@0:                             ),
Chris@0:                         ),
Chris@0:                         array('endLine' => 2)
Chris@0:                     ),
Chris@0:                 )
Chris@0:             ),
Chris@0:             // tests differing start and end line
Chris@0:             array(
Chris@0:                 '<?php "foo' . "\n" . 'bar"',
Chris@0:                 array(),
Chris@0:                 array(
Chris@0:                     array(
Chris@0:                         Tokens::T_CONSTANT_ENCAPSED_STRING, '"foo' . "\n" . 'bar"',
Chris@0:                         array('startLine' => 1), array('endLine' => 2)
Chris@0:                     ),
Chris@0:                 )
Chris@0:             ),
Chris@0:             // tests exact file offsets
Chris@0:             array(
Chris@0:                 '<?php "a";' . "\n" . '// foo' . "\n" . '"b";',
Chris@0:                 array('usedAttributes' => array('startFilePos', 'endFilePos')),
Chris@0:                 array(
Chris@0:                     array(
Chris@0:                         Tokens::T_CONSTANT_ENCAPSED_STRING, '"a"',
Chris@0:                         array('startFilePos' => 6), array('endFilePos' => 8)
Chris@0:                     ),
Chris@0:                     array(
Chris@0:                         ord(';'), ';',
Chris@0:                         array('startFilePos' => 9), array('endFilePos' => 9)
Chris@0:                     ),
Chris@0:                     array(
Chris@0:                         Tokens::T_CONSTANT_ENCAPSED_STRING, '"b"',
Chris@0:                         array('startFilePos' => 18), array('endFilePos' => 20)
Chris@0:                     ),
Chris@0:                     array(
Chris@0:                         ord(';'), ';',
Chris@0:                         array('startFilePos' => 21), array('endFilePos' => 21)
Chris@0:                     ),
Chris@0:                 )
Chris@0:             ),
Chris@0:             // tests token offsets
Chris@0:             array(
Chris@0:                 '<?php "a";' . "\n" . '// foo' . "\n" . '"b";',
Chris@0:                 array('usedAttributes' => array('startTokenPos', 'endTokenPos')),
Chris@0:                 array(
Chris@0:                     array(
Chris@0:                         Tokens::T_CONSTANT_ENCAPSED_STRING, '"a"',
Chris@0:                         array('startTokenPos' => 1), array('endTokenPos' => 1)
Chris@0:                     ),
Chris@0:                     array(
Chris@0:                         ord(';'), ';',
Chris@0:                         array('startTokenPos' => 2), array('endTokenPos' => 2)
Chris@0:                     ),
Chris@0:                     array(
Chris@0:                         Tokens::T_CONSTANT_ENCAPSED_STRING, '"b"',
Chris@0:                         array('startTokenPos' => 5), array('endTokenPos' => 5)
Chris@0:                     ),
Chris@0:                     array(
Chris@0:                         ord(';'), ';',
Chris@0:                         array('startTokenPos' => 6), array('endTokenPos' => 6)
Chris@0:                     ),
Chris@0:                 )
Chris@0:             ),
Chris@0:             // tests all attributes being disabled
Chris@0:             array(
Chris@0:                 '<?php /* foo */ $bar;',
Chris@0:                 array('usedAttributes' => array()),
Chris@0:                 array(
Chris@0:                     array(
Chris@0:                         Tokens::T_VARIABLE, '$bar',
Chris@0:                         array(), array()
Chris@0:                     ),
Chris@0:                     array(
Chris@0:                         ord(';'), ';',
Chris@0:                         array(), array()
Chris@0:                     )
Chris@0:                 )
Chris@0:             ),
Chris@0:             // tests no tokens
Chris@0:             array(
Chris@0:                 '',
Chris@0:                 array(),
Chris@0:                 array()
Chris@0:             ),
Chris@0:         );
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * @dataProvider provideTestHaltCompiler
Chris@0:      */
Chris@0:     public function testHandleHaltCompiler($code, $remaining) {
Chris@0:         $lexer = $this->getLexer();
Chris@0:         $lexer->startLexing($code);
Chris@0: 
Chris@0:         while (Tokens::T_HALT_COMPILER !== $lexer->getNextToken());
Chris@0: 
Chris@0:         $this->assertSame($remaining, $lexer->handleHaltCompiler());
Chris@0:         $this->assertSame(0, $lexer->getNextToken());
Chris@0:     }
Chris@0: 
Chris@0:     public function provideTestHaltCompiler() {
Chris@0:         return array(
Chris@0:             array('<?php ... __halt_compiler();Remaining Text', 'Remaining Text'),
Chris@0:             array('<?php ... __halt_compiler ( ) ;Remaining Text', 'Remaining Text'),
Chris@0:             array('<?php ... __halt_compiler() ?>Remaining Text', 'Remaining Text'),
Chris@0:             //array('<?php ... __halt_compiler();' . "\0", "\0"),
Chris@0:             //array('<?php ... __halt_compiler /* */ ( ) ;Remaining Text', 'Remaining Text'),
Chris@0:         );
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * @expectedException \PhpParser\Error
Chris@0:      * @expectedExceptionMessage __HALT_COMPILER must be followed by "();"
Chris@0:      */
Chris@0:     public function testHandleHaltCompilerError() {
Chris@0:         $lexer = $this->getLexer();
Chris@0:         $lexer->startLexing('<?php ... __halt_compiler invalid ();');
Chris@0: 
Chris@0:         while (Tokens::T_HALT_COMPILER !== $lexer->getNextToken());
Chris@0:         $lexer->handleHaltCompiler();
Chris@0:     }
Chris@0: 
Chris@0:     public function testGetTokens() {
Chris@0:         $code = '<?php "a";' . "\n" . '// foo' . "\n" . '"b";';
Chris@0:         $expectedTokens = array(
Chris@0:             array(T_OPEN_TAG, '<?php ', 1),
Chris@0:             array(T_CONSTANT_ENCAPSED_STRING, '"a"', 1),
Chris@0:             ';',
Chris@0:             array(T_WHITESPACE, "\n", 1),
Chris@0:             array(T_COMMENT, '// foo' . "\n", 2),
Chris@0:             array(T_CONSTANT_ENCAPSED_STRING, '"b"', 3),
Chris@0:             ';',
Chris@0:         );
Chris@0: 
Chris@0:         $lexer = $this->getLexer();
Chris@0:         $lexer->startLexing($code);
Chris@0:         $this->assertSame($expectedTokens, $lexer->getTokens());
Chris@0:     }
Chris@0: }