public function testSetSchemeOnProtocolRelativeUrl()
{
$url = new \webignition\Url\Url('//example.com');
$url->setScheme('http');
$this->assertEquals('http', $url->getScheme());
$this->assertEquals('http://example.com', (string) $url);
}
public function testRemove()
{
$url = new \webignition\Url\Url('http://example.com/?key1=value&key2=value2&key3=value3');
$url->getQuery()->remove('key3');
$this->assertFalse($url->getQuery()->contains('key3'));
$this->assertEquals('key1=value&key2=value2', (string) $url->getQuery());
}
public function testWithoutFilename()
{
$url = new \webignition\Url\Url('http://www.example.com/path/is/here');
$this->assertFalse($url->getPath()->hasFilename());
$this->assertEquals('/path/is/here', $url->getPath()->getDirectory());
$this->assertEquals('', $url->getPath()->getFilename());
}
public function testMinimalEncodingWithEncodableKeysContainingVerySpecialCharacters()
{
$query = 'a%23a=1&b%26b=2&c!c=3';
$url = new \webignition\Url\Url('http://example.com/?' . $query);
$url->getConfiguration()->disableFullyEncodeQueryStringKeys();
$this->assertEquals($query, (string) $url->getQuery());
}
public function testProtocolRelativeHasUsernameNoPassword()
{
$url = new \webignition\Url\Url('//username@example.com');
$this->assertTrue($url->hasUser());
$this->assertEquals('username', $url->getUser());
$this->assertFalse($url->hasPass());
}
public function testSetPathWhenExistingPathIsUrlEncoded()
{
$url = new \webignition\Url\Url('js/scriptaculous.js?load=effects,builder');
$this->assertEquals('js/scriptaculous.js?load=effects%2Cbuilder', (string) $url);
$url->setPath('/js/scriptaculous.js');
$this->assertEquals('/js/scriptaculous.js?load=effects%2Cbuilder', (string) $url);
}
public function testRootUrlIsConstantWithRespectToPath()
{
$url = new \webignition\Url\Url('http://example.com/');
$this->assertEquals('http://example.com', $url->getRoot());
$url = new \webignition\Url\Url('http://example.com/index.html');
$this->assertEquals('http://example.com', $url->getRoot());
$url = new \webignition\Url\Url('http://example.com/path/to/application.php');
$this->assertEquals('http://example.com', $url->getRoot());
}
public function testUnset()
{
$url = new \webignition\Url\Url('http://example.com:8080/');
$this->assertTrue($url->hasPort());
$this->assertEquals(8080, $url->getPort());
$url->setPort(null);
$this->assertFalse($url->hasPort());
$this->assertEquals('http://example.com/', (string) $url);
}
public function testAddPathToRootUrl()
{
$url = new \webignition\Url\Url('http://example.com');
$url->setPath('/index.html');
$this->assertEquals('http://example.com/index.html', (string) $url);
$url = new \webignition\Url\Url('http://example.com/');
$url->setPath('/index.html');
$this->assertEquals('http://example.com/index.html', (string) $url);
}
public function testIdnEquivalence()
{
$idnUrl1 = new \webignition\Url\Url('http://econom.ía.com');
$idnUrl2 = new \webignition\Url\Url('http://econom.xn--a-iga.com');
$idnUrl3 = new \webignition\Url\Url('http://ヒキワリ.ナットウ.ニホン');
$idnUrl4 = new \webignition\Url\Url('http://xn--nckwd5cta.xn--gckxcpg.xn--idk6a7d');
$idnUrl5 = new \webignition\Url\Url('http://транспорт.com');
$idnUrl6 = new \webignition\Url\Url('http://xn--80a0addceeeh.com');
$this->assertTrue($idnUrl1->getHost()->isEquivalentTo($idnUrl2->getHost()));
$this->assertTrue($idnUrl3->getHost()->isEquivalentTo($idnUrl4->getHost()));
$this->assertTrue($idnUrl5->getHost()->isEquivalentTo($idnUrl6->getHost()));
}
/**
* Checks if the given href is not part
* of the domain we are crawling
*
* @return boolean
*/
public function isLeavingOriginDomain()
{
if ($this->origin_domain === null) {
throw new \Exception('origin is null');
}
$domain = $this->origin_domain->getParser();
$link = new \webignition\Url\Url($this->link_href);
$domain_host = $domain->getHost();
$link_host = $link->getHost();
if ($link_host === null) {
return false;
}
return !$domain_host->equals($link_host);
}
public function testSet()
{
$url = new \webignition\Url\Url('http://example.com/?key1=value&key2=value2');
$url->getQuery()->set('key3', 'value3');
$this->assertTrue($url->getQuery()->contains('key3'));
$this->assertEquals('key1=value&key2=value2&key3=value3', (string) $url->getQuery());
$url->getQuery()->set('key4', 'value4');
$this->assertEquals('key1=value&key2=value2&key3=value3&key4=value4', (string) $url->getQuery());
}
public function testAbsoluteUrlIsRecognisedAsAbsolute()
{
$url = new \webignition\Url\Url('http://example.com/absolute/url');
$this->assertTrue($url->isAbsolute());
}
public function testAbsoluteUrlIsRecognisedAsNonRelative()
{
$url = new \webignition\Url\Url('http://example.com/absolute/url');
$this->assertFalse($url->isProtocolRelative());
}
public function testQueryContains()
{
$url = new \webignition\Url\Url('http://example.com/?key1=value&key2=value2');
$this->assertTrue($url->getQuery()->contains('key1'));
$this->assertFalse($url->getQuery()->contains('key3'));
}
public function testNullifySchemeMakesUrlProtocolRelative()
{
$url = new \webignition\Url\Url('http://example.com/foo/bar/index.html');
$url->setScheme(null);
$this->assertEquals('//example.com/foo/bar/index.html', (string) $url);
}
/**
*
* @return string
*/
private function findSitemapUrlsFromRobotsTxt()
{
$robotsTxtParser = new \webignition\RobotsTxt\File\Parser();
$robotsTxtParser->setSource($this->getRobotsTxtContent());
$robotsTxtFile = $robotsTxtParser->getFile();
$urls = array();
if ($robotsTxtFile->directiveList()->containsField('sitemap')) {
$sitemapDirectives = $robotsTxtFile->directiveList()->filter(array('field' => 'sitemap'));
foreach ($sitemapDirectives->get() as $sitemapDirective) {
/* @var $sitemapDirective \webignition\RobotsTxt\Directive\Directive */
$sitemapUrl = new \webignition\Url\Url((string) $sitemapDirective->getValue());
if ($sitemapUrl->isRelative()) {
$absoluteUrlDeriver = new \webignition\AbsoluteUrlDeriver\AbsoluteUrlDeriver((string) $sitemapUrl, $this->getConfiguration()->getRootUrl());
$urls[] = (string) $absoluteUrlDeriver->getAbsoluteUrl();
} else {
$urls[] = (string) $sitemapUrl;
}
}
}
return $urls;
}
public function testChangeFragmentWithHash()
{
$url = new \webignition\Url\Url('http://example.com/#fragment');
$url->setFragment('#new-fragment');
$this->assertEquals('new-fragment', $url->getFragment());
$this->assertEquals('http://example.com/#new-fragment', (string) $url);
}
public function testRelativeUrl()
{
$url = new \webignition\Url\Url($this->urls['relative']);
$this->assertFalse($url->hasScheme());
$this->assertNull($url->getScheme());
$this->assertFalse($url->hasHost());
$this->assertNull($url->getHost());
$this->assertFalse($url->hasPort());
$this->assertNull($url->getPort());
$this->assertFalse($url->hasUser());
$this->assertNull($url->getUser());
$this->assertFalse($url->hasPass());
$this->assertNull($url->getPass());
$this->assertTrue($url->hasPath());
$this->assertEquals($this->completePath(), $url->getPath());
$this->assertTrue($url->hasQuery());
$this->assertEquals($this->completeUrlQueryString(), $url->getQuery());
$this->assertTrue($url->hasFragment());
$this->assertEquals(self::FRAGMENT, $url->getFragment());
$this->assertEquals($this->urls['relative'], (string) $url);
}
public function testSetPathThenHostThenSchemeThenUser()
{
$url = new \webignition\Url\Url('example.php');
$this->assertEquals('example.php', (string) $url);
$url->setPath('/pathOne/Path2/path2/example2.php');
$this->assertEquals('/pathOne/Path2/path2/example2.php', (string) $url);
$url->setHost('www.example.com');
$this->assertEquals('//www.example.com/pathOne/Path2/path2/example2.php', (string) $url);
$url->setScheme('http');
$this->assertEquals('http://www.example.com/pathOne/Path2/path2/example2.php', (string) $url);
$url->setUser('user');
$this->assertEquals('http://user:@www.example.com/pathOne/Path2/path2/example2.php', (string) $url);
}
public function testParseHashAndIdentifierOnly()
{
$url = new \webignition\Url\Url('#foo');
$this->assertTrue($url->hasFragment());
}
public function testInit()
{
$url = new \webignition\Url\Url();
$url->init('http://example.com/foo/bar.html?foo=bar&foobar=boofar#identity');
$this->assertTrue($url->hasScheme());
$this->assertEquals('http', $url->getScheme());
$this->assertTrue($url->hasHost());
$this->assertEquals('example.com', $url->getHost());
$this->assertTrue($url->hasPath());
$this->assertEquals('/foo/bar.html', $url->getPath());
$this->assertTrue($url->hasQuery());
$this->assertEquals('foo=bar&foobar=boofar', (string) $url->getQuery());
$this->assertTrue($url->hasFragment());
$this->assertEquals('identity', $url->getFragment());
}
/**
* When setting a fragment to null in a url that had a fragment
* and then setting the query to null where there was no query
* was resulting in the fragment containing the string '?', this is incorrect
*/
public function testReplaceFragmentWithNullSetNullQuery()
{
$url = new \webignition\Url\Url('http://example.com/#fragment');
$url->setFragment(null);
$url->setQuery(null);
$this->assertNull($url->getFragment());
$this->assertNull($url->getQuery());
$this->assertEquals("http://example.com/", (string) $url);
}
public function testAddPathToHashAndIdentifier()
{
$url = new \webignition\Url\Url('#foo');
$url->setPath('/index.html');
$this->assertEquals('/index.html#foo', (string) $url);
}
public function isValid(Link $link)
{
$link = new \webignition\Url\Url($link->getLinkHref());
return $link->getScheme() === 'http' || $link->getScheme() === 'https' || $link->getHost() !== null;
}