* * For the full copyright and license information, please view the LICENSE * file that was distributed with this source code. */ namespace Symfony\Component\DomCrawler\Tests; use Symfony\Component\DomCrawler\Crawler; class CrawlerTest extends \PHPUnit_Framework_TestCase { public function testConstructor() { $crawler = new Crawler(); $this->assertCount(0, $crawler, '__construct() returns an empty crawler'); $crawler = new Crawler(new \DOMNode()); $this->assertCount(1, $crawler, '__construct() takes a node as a first argument'); } /** * @covers Symfony\Component\DomCrawler\Crawler::add */ public function testAdd() { $crawler = new Crawler(); $crawler->add($this->createDomDocument()); $this->assertEquals('foo', $crawler->filterXPath('//div')->attr('class'), '->add() adds nodes from a \DOMDocument'); $crawler = new Crawler(); $crawler->add($this->createNodeList()); $this->assertEquals('foo', $crawler->filterXPath('//div')->attr('class'), '->add() adds nodes from a \DOMNodeList'); foreach ($this->createNodeList() as $node) { $list[] = $node; } $crawler = new Crawler(); $crawler->add($list); $this->assertEquals('foo', $crawler->filterXPath('//div')->attr('class'), '->add() adds nodes from an array of nodes'); $crawler = new Crawler(); $crawler->add($this->createNodeList()->item(0)); $this->assertEquals('foo', $crawler->filterXPath('//div')->attr('class'), '->add() adds nodes from an \DOMNode'); $crawler = new Crawler(); $crawler->add('Foo'); $this->assertEquals('Foo', $crawler->filterXPath('//body')->text(), '->add() adds nodes from a string'); } /** * @expectedException \InvalidArgumentException */ public function testAddInvalidNode() { $crawler = new Crawler(); $crawler->add(1); } /** * @covers Symfony\Component\DomCrawler\Crawler::addHtmlContent */ public function testAddHtmlContent() { $crawler = new Crawler(); $crawler->addHtmlContent('
', 'UTF-8'); $this->assertEquals('foo', $crawler->filterXPath('//div')->attr('class'), '->addHtmlContent() adds nodes from an HTML string'); $crawler->addHtmlContent('', 'UTF-8'); $this->assertEquals('http://symfony.com', $crawler->filterXPath('//base')->attr('href'), '->addHtmlContent() adds nodes from an HTML string'); $this->assertEquals('http://symfony.com/contact', $crawler->filterXPath('//a')->link()->getUri(), '->addHtmlContent() adds nodes from an HTML string'); } /** * @covers Symfony\Component\DomCrawler\Crawler::addHtmlContent */ public function testAddHtmlContentCharset() { $crawler = new Crawler(); $crawler->addHtmlContent('
Tiếng Việt', 'UTF-8'); $this->assertEquals('Tiếng Việt', $crawler->filterXPath('//div')->text()); } /** * @covers Symfony\Component\DomCrawler\Crawler::addHtmlContent */ public function testAddHtmlContentInvalidBaseTag() { $crawler = new Crawler(null, 'http://symfony.com'); $crawler->addHtmlContent('', 'UTF-8'); $this->assertEquals('http://symfony.com/contact', current($crawler->filterXPath('//a')->links())->getUri(), '->addHtmlContent() correctly handles a non-existent base tag href attribute'); } /** * @covers Symfony\Component\DomCrawler\Crawler::addHtmlContent */ public function testAddHtmlContentUnsupportedCharset() { $crawler = new Crawler(); $crawler->addHtmlContent(file_get_contents(__DIR__.'/Fixtures/windows-1250.html'), 'Windows-1250'); $this->assertEquals('Žťčýů', $crawler->filterXPath('//p')->text()); } /** * @covers Symfony\Component\DomCrawler\Crawler::addHtmlContent */ public function testAddHtmlContentWithErrors() { libxml_use_internal_errors(true); $crawler = new Crawler(); $crawler->addHtmlContent(<< EOF , 'UTF-8'); $errors = libxml_get_errors(); $this->assertCount(1, $errors); $this->assertEquals("Tag nav invalid\n", $errors[0]->message); libxml_clear_errors(); libxml_use_internal_errors(false); } /** * @covers Symfony\Component\DomCrawler\Crawler::addXmlContent */ public function testAddXmlContent() { $crawler = new Crawler(); $crawler->addXmlContent('
', 'UTF-8'); $this->assertEquals('foo', $crawler->filterXPath('//div')->attr('class'), '->addXmlContent() adds nodes from an XML string'); } /** * @covers Symfony\Component\DomCrawler\Crawler::addXmlContent */ public function testAddXmlContentCharset() { $crawler = new Crawler(); $crawler->addXmlContent('
Tiếng Việt
', 'UTF-8'); $this->assertEquals('Tiếng Việt', $crawler->filterXPath('//div')->text()); } /** * @covers Symfony\Component\DomCrawler\Crawler::addXmlContent */ public function testAddXmlContentWithErrors() { libxml_use_internal_errors(true); $crawler = new Crawler(); $crawler->addXmlContent(<<
EOF , 'UTF-8'); $this->assertTrue(count(libxml_get_errors()) > 1); libxml_clear_errors(); libxml_use_internal_errors(false); } /** * @covers Symfony\Component\DomCrawler\Crawler::addContent */ public function testAddContent() { $crawler = new Crawler(); $crawler->addContent('
', 'text/html; charset=UTF-8'); $this->assertEquals('foo', $crawler->filterXPath('//div')->attr('class'), '->addContent() adds nodes from an HTML string'); $crawler = new Crawler(); $crawler->addContent('
', 'text/html; charset=UTF-8; dir=RTL'); $this->assertEquals('foo', $crawler->filterXPath('//div')->attr('class'), '->addContent() adds nodes from an HTML string with extended content type'); $crawler = new Crawler(); $crawler->addContent('
'); $this->assertEquals('foo', $crawler->filterXPath('//div')->attr('class'), '->addContent() uses text/html as the default type'); $crawler = new Crawler(); $crawler->addContent('
', 'text/xml; charset=UTF-8'); $this->assertEquals('foo', $crawler->filterXPath('//div')->attr('class'), '->addContent() adds nodes from an XML string'); $crawler = new Crawler(); $crawler->addContent('
', 'text/xml'); $this->assertEquals('foo', $crawler->filterXPath('//div')->attr('class'), '->addContent() adds nodes from an XML string'); $crawler = new Crawler(); $crawler->addContent('foo bar', 'text/plain'); $this->assertCount(0, $crawler, '->addContent() does nothing if the type is not (x|ht)ml'); } /** * @covers Symfony\Component\DomCrawler\Crawler::addDocument */ public function testAddDocument() { $crawler = new Crawler(); $crawler->addDocument($this->createDomDocument()); $this->assertEquals('foo', $crawler->filterXPath('//div')->attr('class'), '->addDocument() adds nodes from a \DOMDocument'); } /** * @covers Symfony\Component\DomCrawler\Crawler::addNodeList */ public function testAddNodeList() { $crawler = new Crawler(); $crawler->addNodeList($this->createNodeList()); $this->assertEquals('foo', $crawler->filterXPath('//div')->attr('class'), '->addNodeList() adds nodes from a \DOMNodeList'); } /** * @covers Symfony\Component\DomCrawler\Crawler::addNodes */ public function testAddNodes() { foreach ($this->createNodeList() as $node) { $list[] = $node; } $crawler = new Crawler(); $crawler->addNodes($list); $this->assertEquals('foo', $crawler->filterXPath('//div')->attr('class'), '->addNodes() adds nodes from an array of nodes'); } /** * @covers Symfony\Component\DomCrawler\Crawler::addNode */ public function testAddNode() { $crawler = new Crawler(); $crawler->addNode($this->createNodeList()->item(0)); $this->assertEquals('foo', $crawler->filterXPath('//div')->attr('class'), '->addNode() adds nodes from an \DOMNode'); } public function testClear() { $crawler = new Crawler(new \DOMNode()); $crawler->clear(); $this->assertCount(0, $crawler, '->clear() removes all the nodes from the crawler'); } public function testEq() { $crawler = $this->createTestCrawler()->filterXPath('//li'); $this->assertNotSame($crawler, $crawler->eq(0), '->eq() returns a new instance of a crawler'); $this->assertInstanceOf('Symfony\\Component\\DomCrawler\\Crawler', $crawler, '->eq() returns a new instance of a crawler'); $this->assertEquals('Two', $crawler->eq(1)->text(), '->eq() returns the nth node of the list'); $this->assertCount(0, $crawler->eq(100), '->eq() returns an empty crawler if the nth node does not exist'); } public function testEach() { $data = $this->createTestCrawler()->filterXPath('//ul[1]/li')->each(function ($node, $i) { return $i.'-'.$node->text(); }); $this->assertEquals(array('0-One', '1-Two', '2-Three'), $data, '->each() executes an anonymous function on each node of the list'); } public function testReduce() { $crawler = $this->createTestCrawler()->filterXPath('//ul[1]/li'); $nodes = $crawler->reduce(function ($node, $i) { return $i == 1 ? false : true; }); $this->assertNotSame($nodes, $crawler, '->reduce() returns a new instance of a crawler'); $this->assertInstanceOf('Symfony\\Component\\DomCrawler\\Crawler', $nodes, '->reduce() returns a new instance of a crawler'); $this->assertCount(2, $nodes, '->reduce() filters the nodes in the list'); } public function testAttr() { $this->assertEquals('first', $this->createTestCrawler()->filterXPath('//li')->attr('class'), '->attr() returns the attribute of the first element of the node list'); try { $this->createTestCrawler()->filterXPath('//ol')->attr('class'); $this->fail('->attr() throws an \InvalidArgumentException if the node list is empty'); } catch (\InvalidArgumentException $e) { $this->assertTrue(true, '->attr() throws an \InvalidArgumentException if the node list is empty'); } } public function testText() { $this->assertEquals('One', $this->createTestCrawler()->filterXPath('//li')->text(), '->text() returns the node value of the first element of the node list'); try { $this->createTestCrawler()->filterXPath('//ol')->text(); $this->fail('->text() throws an \InvalidArgumentException if the node list is empty'); } catch (\InvalidArgumentException $e) { $this->assertTrue(true, '->text() throws an \InvalidArgumentException if the node list is empty'); } } public function testHtml() { $this->assertEquals('Bar', $this->createTestCrawler()->filterXPath('//a[5]')->html()); $this->assertEquals('' , trim($this->createTestCrawler()->filterXPath('//form[@id="FooFormId"]')->html())); try { $this->createTestCrawler()->filterXPath('//ol')->html(); $this->fail('->html() throws an \InvalidArgumentException if the node list is empty'); } catch (\InvalidArgumentException $e) { $this->assertTrue(true, '->html() throws an \InvalidArgumentException if the node list is empty'); } } public function testExtract() { $crawler = $this->createTestCrawler()->filterXPath('//ul[1]/li'); $this->assertEquals(array('One', 'Two', 'Three'), $crawler->extract('_text'), '->extract() returns an array of extracted data from the node list'); $this->assertEquals(array(array('One', 'first'), array('Two', ''), array('Three', '')), $crawler->extract(array('_text', 'class')), '->extract() returns an array of extracted data from the node list'); $this->assertEquals(array(), $this->createTestCrawler()->filterXPath('//ol')->extract('_text'), '->extract() returns an empty array if the node list is empty'); } /** * @covers Symfony\Component\DomCrawler\Crawler::filterXPath */ public function testFilterXPath() { $crawler = $this->createTestCrawler(); $this->assertNotSame($crawler, $crawler->filterXPath('//li'), '->filterXPath() returns a new instance of a crawler'); $this->assertInstanceOf('Symfony\\Component\\DomCrawler\\Crawler', $crawler, '->filterXPath() returns a new instance of a crawler'); $crawler = $this->createTestCrawler()->filterXPath('//ul'); $this->assertCount(6, $crawler->filterXPath('//li'), '->filterXPath() filters the node list with the XPath expression'); } /** * @covers Symfony\Component\DomCrawler\Crawler::filter */ public function testFilter() { if (!class_exists('Symfony\Component\CssSelector\CssSelector')) { $this->markTestSkipped('The "CssSelector" component is not available'); } $crawler = $this->createTestCrawler(); $this->assertNotSame($crawler, $crawler->filter('li'), '->filter() returns a new instance of a crawler'); $this->assertInstanceOf('Symfony\\Component\\DomCrawler\\Crawler', $crawler, '->filter() returns a new instance of a crawler'); $crawler = $this->createTestCrawler()->filter('ul'); $this->assertCount(6, $crawler->filter('li'), '->filter() filters the node list with the CSS selector'); } public function testSelectLink() { $crawler = $this->createTestCrawler(); $this->assertNotSame($crawler, $crawler->selectLink('Foo'), '->selectLink() returns a new instance of a crawler'); $this->assertInstanceOf('Symfony\\Component\\DomCrawler\\Crawler', $crawler, '->selectLink() returns a new instance of a crawler'); $this->assertCount(1, $crawler->selectLink('Fabien\'s Foo'), '->selectLink() selects links by the node values'); $this->assertCount(1, $crawler->selectLink('Fabien\'s Bar'), '->selectLink() selects links by the alt attribute of a clickable image'); $this->assertCount(2, $crawler->selectLink('Fabien"s Foo'), '->selectLink() selects links by the node values'); $this->assertCount(2, $crawler->selectLink('Fabien"s Bar'), '->selectLink() selects links by the alt attribute of a clickable image'); $this->assertCount(1, $crawler->selectLink('\' Fabien"s Foo'), '->selectLink() selects links by the node values'); $this->assertCount(1, $crawler->selectLink('\' Fabien"s Bar'), '->selectLink() selects links by the alt attribute of a clickable image'); $this->assertCount(4, $crawler->selectLink('Foo'), '->selectLink() selects links by the node values'); $this->assertCount(4, $crawler->selectLink('Bar'), '->selectLink() selects links by the node values'); } public function testSelectButton() { $crawler = $this->createTestCrawler(); $this->assertNotSame($crawler, $crawler->selectButton('FooValue'), '->selectButton() returns a new instance of a crawler'); $this->assertInstanceOf('Symfony\\Component\\DomCrawler\\Crawler', $crawler, '->selectButton() returns a new instance of a crawler'); $this->assertEquals(1, $crawler->selectButton('FooValue')->count(), '->selectButton() selects buttons'); $this->assertEquals(1, $crawler->selectButton('FooName')->count(), '->selectButton() selects buttons'); $this->assertEquals(1, $crawler->selectButton('FooId')->count(), '->selectButton() selects buttons'); $this->assertEquals(1, $crawler->selectButton('BarValue')->count(), '->selectButton() selects buttons'); $this->assertEquals(1, $crawler->selectButton('BarName')->count(), '->selectButton() selects buttons'); $this->assertEquals(1, $crawler->selectButton('BarId')->count(), '->selectButton() selects buttons'); $this->assertEquals(1, $crawler->selectButton('FooBarValue')->count(), '->selectButton() selects buttons with form attribute too'); $this->assertEquals(1, $crawler->selectButton('FooBarName')->count(), '->selectButton() selects buttons with form attribute too'); } public function testLink() { $crawler = $this->createTestCrawler('http://example.com/bar/')->selectLink('Foo'); $this->assertInstanceOf('Symfony\\Component\\DomCrawler\\Link', $crawler->link(), '->link() returns a Link instance'); $this->assertEquals('POST', $crawler->link('post')->getMethod(), '->link() takes a method as its argument'); $crawler = $this->createTestCrawler('http://example.com/bar')->selectLink('GetLink'); $this->assertEquals('http://example.com/bar?get=param', $crawler->link()->getUri(), '->link() returns a Link instance'); try { $this->createTestCrawler()->filterXPath('//ol')->link(); $this->fail('->link() throws an \InvalidArgumentException if the node list is empty'); } catch (\InvalidArgumentException $e) { $this->assertTrue(true, '->link() throws an \InvalidArgumentException if the node list is empty'); } } public function testLinks() { $crawler = $this->createTestCrawler('http://example.com/bar/')->selectLink('Foo'); $this->assertInternalType('array', $crawler->links(), '->links() returns an array'); $this->assertCount(4, $crawler->links(), '->links() returns an array'); $links = $crawler->links(); $this->assertInstanceOf('Symfony\\Component\\DomCrawler\\Link', $links[0], '->links() returns an array of Link instances'); $this->assertEquals(array(), $this->createTestCrawler()->filterXPath('//ol')->links(), '->links() returns an empty array if the node selection is empty'); } public function testForm() { $testCrawler = $this->createTestCrawler('http://example.com/bar/'); $crawler = $testCrawler->selectButton('FooValue'); $crawler2 = $testCrawler->selectButton('FooBarValue'); $this->assertInstanceOf('Symfony\\Component\\DomCrawler\\Form', $crawler->form(), '->form() returns a Form instance'); $this->assertInstanceOf('Symfony\\Component\\DomCrawler\\Form', $crawler2->form(), '->form() returns a Form instance'); $this->assertEquals($crawler->form()->getFormNode()->getAttribute('id'), $crawler2->form()->getFormNode()->getAttribute('id'), '->form() works on elements with form attribute'); $this->assertEquals(array('FooName' => 'FooBar', 'TextName' => 'TextValue', 'FooTextName' => 'FooTextValue'), $crawler->form(array('FooName' => 'FooBar'))->getValues(), '->form() takes an array of values to submit as its first argument'); $this->assertEquals(array('FooName' => 'FooValue', 'TextName' => 'TextValue', 'FooTextName' => 'FooTextValue'), $crawler->form()->getValues(), '->form() takes an array of values to submit as its first argument'); $this->assertEquals(array('FooBarName' => 'FooBarValue', 'TextName' => 'TextValue', 'FooTextName' => 'FooTextValue'), $crawler2->form()->getValues(), '->form() takes an array of values to submit as its first argument'); try { $this->createTestCrawler()->filterXPath('//ol')->form(); $this->fail('->form() throws an \InvalidArgumentException if the node list is empty'); } catch (\InvalidArgumentException $e) { $this->assertTrue(true, '->form() throws an \InvalidArgumentException if the node list is empty'); } } public function testLast() { $crawler = $this->createTestCrawler()->filterXPath('//ul[1]/li'); $this->assertNotSame($crawler, $crawler->last(), '->last() returns a new instance of a crawler'); $this->assertInstanceOf('Symfony\\Component\\DomCrawler\\Crawler', $crawler, '->last() returns a new instance of a crawler'); $this->assertEquals('Three', $crawler->last()->text()); } public function testFirst() { $crawler = $this->createTestCrawler()->filterXPath('//li'); $this->assertNotSame($crawler, $crawler->first(), '->first() returns a new instance of a crawler'); $this->assertInstanceOf('Symfony\\Component\\DomCrawler\\Crawler', $crawler, '->first() returns a new instance of a crawler'); $this->assertEquals('One', $crawler->first()->text()); } public function testSiblings() { $crawler = $this->createTestCrawler()->filterXPath('//li')->eq(1); $this->assertNotSame($crawler, $crawler->siblings(), '->siblings() returns a new instance of a crawler'); $this->assertInstanceOf('Symfony\\Component\\DomCrawler\\Crawler', $crawler, '->siblings() returns a new instance of a crawler'); $nodes = $crawler->siblings(); $this->assertEquals(2, $nodes->count()); $this->assertEquals('One', $nodes->eq(0)->text()); $this->assertEquals('Three', $nodes->eq(1)->text()); $nodes = $this->createTestCrawler()->filterXPath('//li')->eq(0)->siblings(); $this->assertEquals(2, $nodes->count()); $this->assertEquals('Two', $nodes->eq(0)->text()); $this->assertEquals('Three', $nodes->eq(1)->text()); try { $this->createTestCrawler()->filterXPath('//ol')->siblings(); $this->fail('->siblings() throws an \InvalidArgumentException if the node list is empty'); } catch (\InvalidArgumentException $e) { $this->assertTrue(true, '->siblings() throws an \InvalidArgumentException if the node list is empty'); } } public function testNextAll() { $crawler = $this->createTestCrawler()->filterXPath('//li')->eq(1); $this->assertNotSame($crawler, $crawler->nextAll(), '->nextAll() returns a new instance of a crawler'); $this->assertInstanceOf('Symfony\\Component\\DomCrawler\\Crawler', $crawler, '->nextAll() returns a new instance of a crawler'); $nodes = $crawler->nextAll(); $this->assertEquals(1, $nodes->count()); $this->assertEquals('Three', $nodes->eq(0)->text()); try { $this->createTestCrawler()->filterXPath('//ol')->nextAll(); $this->fail('->nextAll() throws an \InvalidArgumentException if the node list is empty'); } catch (\InvalidArgumentException $e) { $this->assertTrue(true, '->nextAll() throws an \InvalidArgumentException if the node list is empty'); } } public function testPreviousAll() { $crawler = $this->createTestCrawler()->filterXPath('//li')->eq(2); $this->assertNotSame($crawler, $crawler->previousAll(), '->previousAll() returns a new instance of a crawler'); $this->assertInstanceOf('Symfony\\Component\\DomCrawler\\Crawler', $crawler, '->previousAll() returns a new instance of a crawler'); $nodes = $crawler->previousAll(); $this->assertEquals(2, $nodes->count()); $this->assertEquals('Two', $nodes->eq(0)->text()); try { $this->createTestCrawler()->filterXPath('//ol')->previousAll(); $this->fail('->previousAll() throws an \InvalidArgumentException if the node list is empty'); } catch (\InvalidArgumentException $e) { $this->assertTrue(true, '->previousAll() throws an \InvalidArgumentException if the node list is empty'); } } public function testChildren() { $crawler = $this->createTestCrawler()->filterXPath('//ul'); $this->assertNotSame($crawler, $crawler->children(), '->children() returns a new instance of a crawler'); $this->assertInstanceOf('Symfony\\Component\\DomCrawler\\Crawler', $crawler, '->children() returns a new instance of a crawler'); $nodes = $crawler->children(); $this->assertEquals(3, $nodes->count()); $this->assertEquals('One', $nodes->eq(0)->text()); $this->assertEquals('Two', $nodes->eq(1)->text()); $this->assertEquals('Three', $nodes->eq(2)->text()); try { $this->createTestCrawler()->filterXPath('//ol')->children(); $this->fail('->children() throws an \InvalidArgumentException if the node list is empty'); } catch (\InvalidArgumentException $e) { $this->assertTrue(true, '->children() throws an \InvalidArgumentException if the node list is empty'); } try { $crawler = new Crawler('

'); $crawler->filter('p')->children(); $this->assertTrue(true, '->children() does not trigger a notice if the node has no children'); } catch (\PHPUnit_Framework_Error_Notice $e) { $this->fail('->children() does not trigger a notice if the node has no children'); } } public function testParents() { $crawler = $this->createTestCrawler()->filterXPath('//li[1]'); $this->assertNotSame($crawler, $crawler->parents(), '->parents() returns a new instance of a crawler'); $this->assertInstanceOf('Symfony\\Component\\DomCrawler\\Crawler', $crawler, '->parents() returns a new instance of a crawler'); $nodes = $crawler->parents(); $this->assertEquals(3, $nodes->count()); $nodes = $this->createTestCrawler()->filterXPath('//html')->parents(); $this->assertEquals(0, $nodes->count()); try { $this->createTestCrawler()->filterXPath('//ol')->parents(); $this->fail('->parents() throws an \InvalidArgumentException if the node list is empty'); } catch (\InvalidArgumentException $e) { $this->assertTrue(true, '->parents() throws an \InvalidArgumentException if the node list is empty'); } } public function testBaseTag() { $crawler = new Crawler('
'); $this->assertEquals('http://base.com/link', $crawler->filterXPath('//a')->link()->getUri()); $crawler = new Crawler('', 'https://domain.com'); $this->assertEquals('https://base.com/link', $crawler->filterXPath('//a')->link()->getUri(), ' tag can use a schema-less url'); $crawler = new Crawler('', 'https://domain.com'); $this->assertEquals('https://domain.com/path/link', $crawler->filterXPath('//a')->link()->getUri(), ' tag can set a path'); } public function createTestCrawler($uri = null) { $dom = new \DOMDocument(); $dom->loadHTML(' Foo Fabien\'s Foo Fabien"s Foo \' Fabien"s Foo Bar    Fabien\'s Bar   Fabien"s Bar \' Fabien"s Bar GetLink