Improve UTF8 detection within the regexp, update tests for non-ascii char, fix #55

This commit is contained in:
Eduardo Gulias Davis
2015-04-26 17:06:10 +02:00
parent aabed98198
commit c383717824
3 changed files with 15 additions and 31 deletions
+2 -21
View File
@@ -75,8 +75,6 @@ class EmailLexer extends AbstractLexer
'\0' => self::C_NUL,
);
protected $invalidASCII = array(226 => 1,);
protected $hasInvalidTokens = false;
protected $previous;
@@ -138,7 +136,8 @@ class EmailLexer extends AbstractLexer
protected function getCatchablePatterns()
{
return array(
'[a-zA-Z_]+[46]?',
'[a-zA-Z_]+[46]?', //ASCII and domain literal
'[^\x00-\x7F]', //UTF-8
'[0-9]+',
'\r\n',
'::',
@@ -179,11 +178,6 @@ class EmailLexer extends AbstractLexer
return self::INVALID;
}
if ($this->isASCIIInvalid($value)) {
$this->hasInvalidTokens = true;
return self::INVALID;
}
return self::GENERIC;
}
@@ -209,19 +203,6 @@ class EmailLexer extends AbstractLexer
return false;
}
/**
* @param $value
* @return bool
*/
protected function isASCIIInvalid($value)
{
if (isset($this->invalidASCII[ord($value)])) {
return true;
}
return false;
}
/**
* @param $value
* @return bool
@@ -103,6 +103,17 @@ class EmailLexerTests extends \PHPUnit_Framework_TestCase
$this->assertEquals(EmailLexer::S_HTAB, $lexer->token['type']);
}
public function testLexerForUTF8()
{
$lexer = new EmailLexer();
$lexer->setInput("áÇ@bar.com");
$lexer->moveNext();
$lexer->moveNext();
$this->assertEquals(EmailLexer::GENERIC, $lexer->token['type']);
$lexer->moveNext();
$this->assertEquals(EmailLexer::GENERIC, $lexer->token['type']);
}
public function testLexerSearchToken()
{
$lexer = new EmailLexer();
@@ -111,15 +122,6 @@ class EmailLexerTests extends \PHPUnit_Framework_TestCase
$this->assertTrue($lexer->find(EmailLexer::S_HTAB));
}
public function testLexerHasInvalidTokens()
{
$lexer = new EmailLexer();
$lexer->setInput(chr(226));
$lexer->moveNext();
$lexer->moveNext();
$this->assertTrue($lexer->hasInvalidTokens());
}
public function getTokens()
{
return array(
@@ -152,7 +154,7 @@ class EmailLexerTests extends \PHPUnit_Framework_TestCase
array('}', EmailLexer::S_CLOSEQBRACKET),
array('', EmailLexer::S_EMPTY),
array(chr(31), EmailLexer::INVALID),
array(chr(226), EmailLexer::INVALID),
array(chr(226), EmailLexer::GENERIC),
array(chr(0), EmailLexer::C_NUL)
);
}
@@ -37,6 +37,7 @@ class EmailValidatorTest extends \PHPUnit_Framework_TestCase
public function getValidEmails()
{
return array(
array('â@iana.org'),
array('fabien@symfony.com'),
array('example@example.co.uk'),
array('fabien_potencier@example.fr'),