apply a regex solution instead of tokenizer

main
Rob Bast 9 years ago
parent e4118385a0
commit b5d286e27b

@ -192,115 +192,69 @@ class SpdxLicense
* @param string $license * @param string $license
* *
* @return bool * @return bool
*
* @throws \RuntimeException * @throws \RuntimeException
*/ */
private function isValidLicenseString($license) private function isValidLicenseString($license)
{ {
$tokens = array( $licenses = array_map('preg_quote', array_keys($this->licenses));
'po' => '\(', sort($licenses);
'pc' => '\)', $licenses = array_reverse($licenses);
'op' => '(?:or|OR|and|AND)', $licenses = implode('|', $licenses);
'wi' => '(?:with|WITH)',
'lix' => '(?:NONE|NOASSERTION)', $exceptions = array_map('preg_quote', array_keys($this->exceptions));
'lir' => 'LicenseRef-\d+', sort($exceptions);
'lic' => '[-_.a-zA-Z0-9]{3,}\+?', $exceptions = array_reverse($exceptions);
'ws' => '\s+', $exceptions = implode('|', $exceptions);
'_' => '.',
); $regex = "{
(?(DEFINE)
$next = function () use ($license, $tokens) { # idstring: 1*( ALPHA / DIGIT / - / . )
static $offset = 0; (?<idstring>[\pL\pN\-\.]{1,})
if ($offset >= strlen($license)) { # license-id: taken from list
return null; (?<licenseid>${licenses})
}
# license-exception-id: taken from list
foreach ($tokens as $name => $token) { (?<licenseexceptionid>${exceptions})
if (false === $r = preg_match('{' . $token . '}', $license, $matches, PREG_OFFSET_CAPTURE, $offset)) {
throw new \RuntimeException('Pattern for token %s failed (regex error).', $name); # license-ref: [DocumentRef-1*(idstring):]LicenseRef-1*(idstring)
} (?<licenseref>(?:DocumentRef-(?&idstring):)?LicenseRef-(?&idstring))
if ($r === 0) {
continue; # simple-expresssion: license-id / license-id+ / license-ref
} (?<simple_expression>(?&licenseid)\+? | (?&licenseid) | (?&licenseref))
if ($matches[0][1] !== $offset) {
continue; # compound expression: 1*(
} # simple-expression /
$offset += strlen($matches[0][0]); # simple-expression WITH license-exception-id /
# compound-expression AND compound-expression /
return array($name, $matches[0][0]); # compound-expression OR compound-expression
} # ) / ( compound-expression ) )
(?<compound_head>
throw new \RuntimeException( (?&simple_expression) ( \s+ (?:with|WITH) \s+ (?&licenseexceptionid))?
'At least the last pattern needs to match, but it did not (dot-match-all is missing?).' | \( \s* (?&compound_expression) \s*\)
); )
}; (?<compound_expression>
(?&compound_head) (?: \s+ (?:and|AND|or|OR) \s+ (?&compound_expression))?
$open = 0; )
$with = false;
$require = true; # license-expression: 1*1(simple-expression / compound-expression)
$lastop = null; (?<license_expression>NONE | NOASSERTION | (?&compound_expression) | (?&simple_expression))
) # end of define
while (list($token, $string) = $next()) {
switch ($token) { ^(?&license_expression)$
case 'po': }x";
if ($open || !$require || $with) {
return false; $match = preg_match($regex, $license);
}
$open = 1; if (0 === $match) {
break;
case 'pc':
if ($open !== 1 || $require || !$lastop || $with) {
return false;
}
$open = 2;
break;
case 'op':
if ($require || !$open || $with) {
return false; return false;
} }
$lastop || $lastop = $string;
if ($lastop !== $string) { if (false === $match) {
return false; throw new \RuntimeException('Regex failed to compile/run.');
}
$require = true;
break;
case 'wi':
$with = true;
break;
case 'lix':
if ($open || $with) {
return false;
}
goto lir;
case 'lic':
if ($with && $this->isValidExceptionIdentifier($string)) {
$require = true;
$with = false;
goto lir;
}
if ($with) {
return false;
}
if (!$this->isValidLicenseIdentifier(rtrim($string, '+'))) {
return false;
}
// Fall-through intended
case 'lir':
lir:
if (!$require) {
return false;
}
$require = false;
break;
case 'ws':
break;
case '_':
return false;
default:
throw new \RuntimeException(sprintf('Unparsed token: %s.', print_r($token, true)));
}
} }
return !($open % 2 || $require || $with); return true;
} }
} }

@ -39,6 +39,7 @@ class SpdxLicenseTest extends TestCase
"GPL-2.0 with Autoconf-exception-2.0", "GPL-2.0 with Autoconf-exception-2.0",
"GPL-2.0 WITH Autoconf-exception-2.0", "GPL-2.0 WITH Autoconf-exception-2.0",
"GPL-2.0+ WITH Autoconf-exception-2.0", "GPL-2.0+ WITH Autoconf-exception-2.0",
"(GPL-3.0 and GPL-2.0 or GPL-3.0+)",
), ),
$identifiers $identifiers
); );
@ -57,7 +58,6 @@ class SpdxLicenseTest extends TestCase
array(array()), array(array()),
array("The system pwns you"), array("The system pwns you"),
array("()"), array("()"),
array("(MIT)"),
array("(MIT"), array("(MIT"),
array("MIT)"), array("MIT)"),
array("MIT NONE"), array("MIT NONE"),
@ -66,8 +66,6 @@ class SpdxLicenseTest extends TestCase
array("(MIT and MIT) MIT"), array("(MIT and MIT) MIT"),
array(array("LGPL-2.0", "The system pwns you")), array(array("LGPL-2.0", "The system pwns you")),
array("and GPL-3.0+"), array("and GPL-3.0+"),
array("EUDatagrid and GPL-3.0+"),
array("(GPL-3.0 and GPL-2.0 or GPL-3.0+)"),
array("(EUDatagrid and GPL-3.0+ and )"), array("(EUDatagrid and GPL-3.0+ and )"),
array("(EUDatagrid xor GPL-3.0+)"), array("(EUDatagrid xor GPL-3.0+)"),
array("(MIT Or MIT)"), array("(MIT Or MIT)"),

Loading…
Cancel
Save