diff options
author | emkael <emkael@tlen.pl> | 2017-01-18 20:07:16 +0100 |
---|---|---|
committer | emkael <emkael@tlen.pl> | 2017-01-18 20:07:16 +0100 |
commit | 9a9c04512e5dcb77c7fe5d850e3f2a0250cc160e (patch) | |
tree | fed46b5f4c2ed3a050bb1a7ad7c6d0a3ea844d55 /lib/querypath/src/QueryPath/CSS | |
parent | c5bcf8f74fb80b7e163663845b0d6e35cabface3 (diff) |
* Motor Sport Magazine feed provider
Diffstat (limited to 'lib/querypath/src/QueryPath/CSS')
-rw-r--r-- | lib/querypath/src/QueryPath/CSS/DOMTraverser.php | 775 | ||||
-rw-r--r-- | lib/querypath/src/QueryPath/CSS/DOMTraverser/PseudoClass.php | 421 | ||||
-rw-r--r-- | lib/querypath/src/QueryPath/CSS/DOMTraverser/Util.php | 139 | ||||
-rw-r--r-- | lib/querypath/src/QueryPath/CSS/EventHandler.php | 171 | ||||
-rw-r--r-- | lib/querypath/src/QueryPath/CSS/InputStream.php | 57 | ||||
-rw-r--r-- | lib/querypath/src/QueryPath/CSS/NotImplementedException.php | 15 | ||||
-rw-r--r-- | lib/querypath/src/QueryPath/CSS/ParseException.php | 15 | ||||
-rw-r--r-- | lib/querypath/src/QueryPath/CSS/Parser.php | 575 | ||||
-rw-r--r-- | lib/querypath/src/QueryPath/CSS/QueryPathEventHandler.php | 1424 | ||||
-rw-r--r-- | lib/querypath/src/QueryPath/CSS/Scanner.php | 306 | ||||
-rw-r--r-- | lib/querypath/src/QueryPath/CSS/Selector.php | 144 | ||||
-rw-r--r-- | lib/querypath/src/QueryPath/CSS/SimpleSelector.php | 138 | ||||
-rw-r--r-- | lib/querypath/src/QueryPath/CSS/Token.php | 60 | ||||
-rw-r--r-- | lib/querypath/src/QueryPath/CSS/Traverser.php | 38 |
14 files changed, 4278 insertions, 0 deletions
diff --git a/lib/querypath/src/QueryPath/CSS/DOMTraverser.php b/lib/querypath/src/QueryPath/CSS/DOMTraverser.php new file mode 100644 index 0000000..be8c2af --- /dev/null +++ b/lib/querypath/src/QueryPath/CSS/DOMTraverser.php @@ -0,0 +1,775 @@ +<?php +/** @file + * Traverse a DOM. + */ + +namespace QueryPath\CSS; + +use \QueryPath\CSS\DOMTraverser\Util; +use \QueryPath\CSS\DOMTraverser\PseudoClass; +use \QueryPath\CSS\DOMTraverser\PseudoElement; + +/** + * Traverse a DOM, finding matches to the selector. + * + * This traverses a DOMDocument and attempts to find + * matches to the provided selector. + * + * \b How this works + * + * This performs a bottom-up search. On the first pass, + * it attempts to find all of the matching elements for the + * last simple selector in a selector. + * + * Subsequent passes attempt to eliminate matches from the + * initial matching set. + * + * Example: + * + * Say we begin with the selector `foo.bar baz`. This is processed + * as follows: + * + * - First, find all baz elements. + * - Next, for any baz element that does not have foo as an ancestor, + * eliminate it from the matches. + * - Finally, for those that have foo as an ancestor, does that foo + * also have a class baz? If not, it is removed from the matches. + * + * \b Extrapolation + * + * Partial simple selectors are almost always expanded to include an + * element. + * + * Examples: + * + * - `:first` is expanded to `*:first` + * - `.bar` is expanded to `*.bar`. + * - `.outer .inner` is expanded to `*.outer *.inner` + * + * The exception is that IDs are sometimes not expanded, e.g.: + * + * - `#myElement` does not get expanded + * - `#myElement .class` \i may be expanded to `*#myElement *.class` + * (which will obviously not perform well). + */ +class DOMTraverser implements Traverser { + + protected $matches = array(); + protected $selector; + protected $dom; + protected $initialized = TRUE; + protected $psHandler; + protected $scopeNode; + + /** + * Build a new DOMTraverser. + * + * This requires a DOM-like object or collection of DOM nodes. + */ + public function __construct(\SPLObjectStorage $splos, $initialized = FALSE, $scopeNode = NULL) { + + $this->psHandler = new \QueryPath\CSS\DOMTraverser\PseudoClass(); + $this->initialized = $initialized; + + // Re-use the initial splos + $this->matches = $splos; + + if (count($splos) != 0) { + $splos->rewind(); + $first = $splos->current(); + if ($first instanceof \DOMDocument) { + $this->dom = $first;//->documentElement; + } + else { + $this->dom = $first->ownerDocument;//->documentElement; + } + if (empty($scopeNode)) { + $this->scopeNode = $this->dom->documentElement; + } + else { + $this->scopeNode = $scopeNode; + } + } + + // This assumes a DOM. Need to also accomodate the case + // where we get a set of elements. + /* + $this->dom = $dom; + $this->matches = new \SplObjectStorage(); + $this->matches->attach($this->dom); + */ + } + + public function debug($msg) { + fwrite(STDOUT, PHP_EOL . $msg); + } + + /** + * Given a selector, find the matches in the given DOM. + * + * This is the main function for querying the DOM using a CSS + * selector. + * + * @param string $selector + * The selector. + * @return \SPLObjectStorage + * An SPLObjectStorage containing a list of matched + * DOMNode objects. + */ + public function find($selector) { + // Setup + $handler = new Selector(); + $parser = new Parser($selector, $handler); + $parser->parse(); + $this->selector = $handler; + + //$selector = $handler->toArray(); + $found = $this->newMatches(); + foreach ($handler as $selectorGroup) { + // fprintf(STDOUT, "Selector group.\n"); + // Initialize matches if necessary. + if ($this->initialized) { + $candidates = $this->matches; + } + else { + //if (empty($selectorGroup)) { + // fprintf(STDOUT, "%s", print_r($handler->toArray(), TRUE)); + //} + $candidates = $this->initialMatch($selectorGroup[0], $this->matches); + //$this->initialized = TRUE; + } + + foreach ($candidates as $candidate) { + // fprintf(STDOUT, "Testing %s against %s.\n", $candidate->tagName, $selectorGroup[0]); + if ($this->matchesSelector($candidate, $selectorGroup)) { + // $this->debug('Attaching ' . $candidate->nodeName); + $found->attach($candidate); + } + } + } + $this->setMatches($found); + + + return $this; + } + + public function matches() { + return $this->matches; + } + + /** + * Check whether the given node matches the given selector. + * + * A selector is a group of one or more simple selectors combined + * by combinators. This determines if a given selector + * matches the given node. + * + * @attention + * Evaluation of selectors is done recursively. Thus the length + * of the selector is limited to the recursion depth allowed by + * the PHP configuration. This should only cause problems for + * absolutely huge selectors or for versions of PHP tuned to + * strictly limit recursion depth. + * + * @param object DOMNode + * The DOMNode to check. + * @param array Selector->toArray() + * The Selector to check. + * @return boolean + * A boolean TRUE if the node matches, false otherwise. + */ + public function matchesSelector($node, $selector) { + return $this->matchesSimpleSelector($node, $selector, 0); + } + + /** + * Performs a match check on a SimpleSelector. + * + * Where matchesSelector() does a check on an entire selector, + * this checks only a simple selector (plus an optional + * combinator). + * + * @param object DOMNode + * The DOMNode to check. + * @param object SimpleSelector + * The Selector to check. + * @return boolean + * A boolean TRUE if the node matches, false otherwise. + */ + public function matchesSimpleSelector($node, $selectors, $index) { + $selector = $selectors[$index]; + // Note that this will short circuit as soon as one of these + // returns FALSE. + $result = $this->matchElement($node, $selector->element, $selector->ns) + && $this->matchAttributes($node, $selector->attributes) + && $this->matchId($node, $selector->id) + && $this->matchClasses($node, $selector->classes) + && $this->matchPseudoClasses($node, $selector->pseudoClasses) + && $this->matchPseudoElements($node, $selector->pseudoElements); + + $isNextRule = isset($selectors[++$index]); + // If there is another selector, we process that if there a match + // hasn't been found. + /* + if ($isNextRule && $selectors[$index]->combinator == SimpleSelector::anotherSelector) { + // We may need to re-initialize the match set for the next selector. + if (!$this->initialized) { + $this->initialMatch($selectors[$index]); + } + if (!$result) fprintf(STDOUT, "Element: %s, Next selector: %s\n", $node->tagName, $selectors[$index]); + return $result || $this->matchesSimpleSelector($node, $selectors, $index); + } + // If we have a match and we have a combinator, we need to + // recurse up the tree. + else*/if ($isNextRule && $result) { + $result = $this->combine($node, $selectors, $index); + } + + return $result; + } + + /** + * Combine the next selector with the given match + * using the next combinator. + * + * If the next selector is combined with another + * selector, that will be evaluated too, and so on. + * So if this function returns TRUE, it means that all + * child selectors are also matches. + * + * @param DOMNode $node + * The DOMNode to test. + * @param array $selectors + * The array of simple selectors. + * @param int $index + * The index of the current selector. + * @return boolean + * TRUE if the next selector(s) match. + */ + public function combine($node, $selectors, $index) { + $selector = $selectors[$index]; + //$this->debug(implode(' ', $selectors)); + switch ($selector->combinator) { + case SimpleSelector::adjacent: + return $this->combineAdjacent($node, $selectors, $index); + case SimpleSelector::sibling: + return $this->combineSibling($node, $selectors, $index); + case SimpleSelector::directDescendant: + return $this->combineDirectDescendant($node, $selectors, $index); + case SimpleSelector::anyDescendant: + return $this->combineAnyDescendant($node, $selectors, $index); + case SimpleSelector::anotherSelector: + // fprintf(STDOUT, "Next selector: %s\n", $selectors[$index]); + return $this->matchesSimpleSelector($node, $selectors, $index); + ; + } + return FALSE; + } + + /** + * Process an Adjacent Sibling. + * + * The spec does not indicate whether Adjacent should ignore non-Element + * nodes, so we choose to ignore them. + * + * @param DOMNode $node + * A DOM Node. + * @param array $selectors + * The selectors array. + * @param int $index + * The current index to the operative simple selector in the selectors + * array. + * @return boolean + * TRUE if the combination matches, FALSE otherwise. + */ + public function combineAdjacent($node, $selectors, $index) { + while (!empty($node->previousSibling)) { + $node = $node->previousSibling; + if ($node->nodeType == XML_ELEMENT_NODE) { + //$this->debug(sprintf('Testing %s against "%s"', $node->tagName, $selectors[$index])); + return $this->matchesSimpleSelector($node, $selectors, $index); + } + } + return FALSE; + } + + /** + * Check all siblings. + * + * According to the spec, this only tests elements LEFT of the provided + * node. + * + * @param DOMNode $node + * A DOM Node. + * @param array $selectors + * The selectors array. + * @param int $index + * The current index to the operative simple selector in the selectors + * array. + * @return boolean + * TRUE if the combination matches, FALSE otherwise. + */ + public function combineSibling($node, $selectors, $index) { + while (!empty($node->previousSibling)) { + $node = $node->previousSibling; + if ($node->nodeType == XML_ELEMENT_NODE && $this->matchesSimpleSelector($node, $selectors, $index)) { + return TRUE; + } + } + return FALSE; + } + + /** + * Handle a Direct Descendant combination. + * + * Check whether the given node is a rightly-related descendant + * of its parent node. + * + * @param DOMNode $node + * A DOM Node. + * @param array $selectors + * The selectors array. + * @param int $index + * The current index to the operative simple selector in the selectors + * array. + * @return boolean + * TRUE if the combination matches, FALSE otherwise. + */ + public function combineDirectDescendant($node, $selectors, $index) { + $parent = $node->parentNode; + if (empty($parent)) { + return FALSE; + } + return $this->matchesSimpleSelector($parent, $selectors, $index); + } + + /** + * Handle Any Descendant combinations. + * + * This checks to see if there are any matching routes from the + * selector beginning at the present node. + * + * @param DOMNode $node + * A DOM Node. + * @param array $selectors + * The selectors array. + * @param int $index + * The current index to the operative simple selector in the selectors + * array. + * @return boolean + * TRUE if the combination matches, FALSE otherwise. + */ + public function combineAnyDescendant($node, $selectors, $index) { + while (!empty($node->parentNode)) { + $node = $node->parentNode; + + // Catch case where element is child of something + // else. This should really only happen with a + // document element. + if ($node->nodeType != XML_ELEMENT_NODE) { + continue; + } + + if ($this->matchesSimpleSelector($node, $selectors, $index)) { + return TRUE; + } + } + } + + /** + * Get the intial match set. + * + * This should only be executed when not working with + * an existing match set. + */ + protected function initialMatch($selector, $matches) { + $element = $selector->element; + + // If no element is specified, we have to start with the + // entire document. + if ($element == NULL) { + $element = '*'; + } + + // fprintf(STDOUT, "Initial match using %s.\n", $selector); + + // We try to do some optimization here to reduce the + // number of matches to the bare minimum. This will + // reduce the subsequent number of operations that + // must be performed in the query. + + // Experimental: ID queries use XPath to match, since + // this should give us only a single matched element + // to work with. + if (/*$element == '*' &&*/ !empty($selector->id)) { + // fprintf(STDOUT, "ID Fastrack on %s\n", $selector); + $initialMatches = $this->initialMatchOnID($selector, $matches); + } + // If a namespace is set, find the namespace matches. + elseif (!empty($selector->ns)) { + $initialMatches = $this->initialMatchOnElementNS($selector, $matches); + } + // If the element is a wildcard, using class can + // substantially reduce the number of elements that + // we start with. + elseif ($element == '*' && !empty($selector->classes)) { + // fprintf(STDOUT, "Class Fastrack on %s\n", $selector); + $initialMatches = $this->initialMatchOnClasses($selector, $matches); + } + else { + $initialMatches = $this->initialMatchOnElement($selector, $matches); + } + + //fprintf(STDOUT, "Found %d nodes.\n", count($this->matches)); + return $initialMatches; + } + + /** + * Shortcut for finding initial match by ID. + * + * If the element is set to '*' and an ID is + * set, then this should be used to find by ID, + * which will drastically reduce the amount of + * comparison operations done in PHP. + * + */ + protected function initialMatchOnID($selector, $matches) { + $id = $selector->id; + $found = $this->newMatches(); + + // Issue #145: DOMXPath will through an exception if the DOM is + // not set. + if (!($this->dom instanceof \DOMDocument)) { + return $found; + } + $baseQuery = ".//*[@id='{$id}']"; + $xpath = new \DOMXPath($this->dom); + + // Now we try to find any matching IDs. + foreach ($matches as $node) { + if ($node->getAttribute('id') == $id) { + $found->attach($node); + } + $nl = $this->initialXpathQuery($xpath, $node, $baseQuery); + $this->attachNodeList($nl, $found); + } + // Unset the ID selector. + $selector->id = NULL; + return $found; + } + + /** + * Shortcut for setting the intial match. + * + * This shortcut should only be used when the initial + * element is '*' and there are classes set. + * + * In any other case, the element finding algo is + * faster and should be used instead. + */ + protected function initialMatchOnClasses($selector, $matches) { + $found = $this->newMatches(); + + // Issue #145: DOMXPath will through an exception if the DOM is + // not set. + if (!($this->dom instanceof \DOMDocument)) { + return $found; + } + $baseQuery = ".//*[@class]"; + $xpath = new \DOMXPath($this->dom); + + // Now we try to find any matching IDs. + foreach ($matches as $node) { + // Refactor me! + if ($node->hasAttribute('class')) { + $intersect = array_intersect($selector->classes, explode(' ', $node->getAttribute('class'))); + if (count($intersect) == count($selector->classes)) { + $found->attach($node); + } + } + + $nl = $this->initialXpathQuery($xpath, $node, $baseQuery); + foreach ($nl as $node) { + $classes = $node->getAttribute('class'); + $classArray = explode(' ', $classes); + + $intersect = array_intersect($selector->classes, $classArray); + if (count($intersect) == count($selector->classes)) { + $found->attach($node); + } + } + } + + // Unset the classes selector. + $selector->classes = array(); + + return $found; + } + + /** + * Internal xpath query. + * + * This is optimized for very specific use, and is not a general + * purpose function. + */ + private function initialXpathQuery($xpath, $node, $query) { + // This works around a bug in which the document element + // does not correctly search with the $baseQuery. + if ($node->isSameNode($this->dom->documentElement)) { + $query = substr($query, 1); + } + + return $xpath->query($query, $node); + } + + /** + * Shortcut for setting the initial match. + */ + protected function initialMatchOnElement($selector, $matches) { + $element = $selector->element; + if (is_null($element)) { + $element = '*'; + } + $found = $this->newMatches(); + foreach ($matches as $node) { + // Capture the case where the initial element is the root element. + if ($node->tagName == $element + || $element == '*' && $node->parentNode instanceof \DOMDocument) { + $found->attach($node); + } + $nl = $node->getElementsByTagName($element); + $this->attachNodeList($nl, $found); + } + + $selector->element = NULL; + return $found; + } + + /** + * Get elements and filter by namespace. + */ + protected function initialMatchOnElementNS($selector, $matches) { + $ns = $selector->ns; + + $elements = $this->initialMatchOnElement($selector, $matches); + + // "any namespace" matches anything. + if ($ns == '*') { + return $elements; + } + + // Loop through and make a list of items that need to be filtered + // out, then filter them. This is required b/c ObjectStorage iterates + // wrongly when an item is detached in an access loop. + $detach = array(); + foreach ($elements as $node) { + // This lookup must be done PER NODE. + $nsuri = $node->lookupNamespaceURI($ns); + if (empty($nsuri) || $node->namespaceURI != $nsuri) { + $detach[] = $node; + } + } + foreach ($detach as $rem) { + $elements->detach($rem); + } + $selector->ns = NULL; + return $elements; + } + + /** + * Checks to see if the DOMNode matches the given element selector. + * + * This handles the following cases: + * + * - element (foo) + * - namespaced element (ns|foo) + * - namespaced wildcard (ns|*) + * - wildcard (* or *|*) + */ + protected function matchElement($node, $element, $ns = NULL) { + if (empty($element)) { + return TRUE; + } + + // Handle namespace. + if (!empty($ns) && $ns != '*') { + // Check whether we have a matching NS URI. + $nsuri = $node->lookupNamespaceURI($ns); + if(empty($nsuri) || $node->namespaceURI !== $nsuri) { + return FALSE; + } + } + + // Compare local name to given element name. + return $element == '*' || $node->localName == $element; + } + + /** + * Checks to see if the given DOMNode matches an "any element" (*). + * + * This does not handle namespaced whildcards. + */ + /* + protected function matchAnyElement($node) { + $ancestors = $this->ancestors($node); + + return count($ancestors) > 0; + } + */ + + /** + * Get a list of ancestors to the present node. + */ + protected function ancestors($node) { + $buffer = array(); + $parent = $node; + while (($parent = $parent->parentNode) !== NULL) { + $buffer[] = $parent; + } + return $buffer; + } + + /** + * Check to see if DOMNode has all of the given attributes. + * + * This can handle namespaced attributes, including namespace + * wildcards. + */ + protected function matchAttributes($node, $attributes) { + if (empty($attributes)) { + return TRUE; + } + + foreach($attributes as $attr) { + $val = isset($attr['value']) ? $attr['value'] : NULL; + + // Namespaced attributes. + if (isset($attr['ns']) && $attr['ns'] != '*') { + $nsuri = $node->lookupNamespaceURI($attr['ns']); + if (empty($nsuri) || !$node->hasAttributeNS($nsuri, $attr['name'])) { + return FALSE; + } + $matches = Util::matchesAttributeNS($node, $attr['name'], $nsuri, $val, $attr['op']); + } + elseif (isset($attr['ns']) && $attr['ns'] == '*' && $node->hasAttributes()) { + // Cycle through all of the attributes in the node. Note that + // these are DOMAttr objects. + $matches = FALSE; + $name = $attr['name']; + foreach ($node->attributes as $attrNode) { + if ($attrNode->localName == $name) { + $nsuri = $attrNode->namespaceURI; + $matches = Util::matchesAttributeNS($node, $name, $nsuri, $val, $attr['op']); + } + } + } + // No namespace. + else { + $matches = Util::matchesAttribute($node, $attr['name'], $val, $attr['op']); + } + + if (!$matches) { + return FALSE; + } + } + return TRUE; + } + /** + * Check that the given DOMNode has the given ID. + */ + protected function matchId($node, $id) { + if (empty($id)) { + return TRUE; + } + return $node->hasAttribute('id') && $node->getAttribute('id') == $id; + } + /** + * Check that the given DOMNode has all of the given classes. + */ + protected function matchClasses($node, $classes) { + if (empty($classes)) { + return TRUE; + } + + if (!$node->hasAttribute('class')) { + return FALSE; + } + + $eleClasses = preg_split('/\s+/', $node->getAttribute('class')); + if (empty($eleClasses)) { + return FALSE; + } + + // The intersection should match the given $classes. + $missing = array_diff($classes, array_intersect($classes, $eleClasses)); + + return count($missing) == 0; + } + protected function matchPseudoClasses($node, $pseudoClasses) { + $ret = TRUE; + foreach ($pseudoClasses as $pseudoClass) { + $name = $pseudoClass['name']; + // Avoid E_STRICT violation. + $value = isset($pseudoClass['value']) ? $pseudoClass['value'] : NULL; + $ret &= $this->psHandler->elementMatches($name, $node, $this->scopeNode, $value); + } + return $ret; + } + /** + * Test whether the given node matches the pseudoElements. + * + * If any pseudo-elements are passed, this will test to see + * <i>if conditions obtain that would allow the pseudo-element + * to be created</i>. This does not modify the match in any way. + */ + protected function matchPseudoElements($node, $pseudoElements) { + if (empty($pseudoElements)) { + return TRUE; + } + + foreach ($pseudoElements as $pse) { + switch ($pse) { + case 'first-line': + case 'first-letter': + case 'before': + case 'after': + return strlen($node->textContent) > 0; + case 'selection': + throw new \QueryPath\CSS\NotImplementedException("::$name is not implemented."); + } + } + } + + protected function newMatches() { + return new \SplObjectStorage(); + } + + /** + * Get the internal match set. + * Internal utility function. + */ + protected function getMatches() { + return $this->matches(); + } + + /** + * Set the internal match set. + * + * Internal utility function. + */ + protected function setMatches($matches) { + $this->matches = $matches; + } + + /** + * Attach all nodes in a node list to the given \SplObjectStorage. + */ + public function attachNodeList(\DOMNodeList $nodeList, \SplObjectStorage $splos) { + foreach ($nodeList as $item) $splos->attach($item); + } + + public function getDocument() { + return $this->dom; + } + +} diff --git a/lib/querypath/src/QueryPath/CSS/DOMTraverser/PseudoClass.php b/lib/querypath/src/QueryPath/CSS/DOMTraverser/PseudoClass.php new file mode 100644 index 0000000..0bcaf79 --- /dev/null +++ b/lib/querypath/src/QueryPath/CSS/DOMTraverser/PseudoClass.php @@ -0,0 +1,421 @@ +<?php +/** + * @file + * + * PseudoClass class. + * + * This is the first pass in an experiment to break PseudoClass handling + * out of the normal traversal. Eventually, this should become a + * top-level pluggable registry that will allow custom pseudoclasses. + * For now, though, we just handle the core pseudoclasses. + */ +namespace QueryPath\CSS\DOMTraverser; + +use \QueryPath\CSS\NotImplementedException; +use \QueryPath\CSS\EventHandler; +/** + * The PseudoClass handler. + * + */ +class PseudoClass { + + /** + * Tests whether the given element matches the given pseudoclass. + * + * @param string $pseudoclass + * The string name of the pseudoclass + * @param resource $node + * The DOMNode to be tested. + * @param resource $scope + * The DOMElement that is the active root for this node. + * @param mixed $value + * The optional value string provided with this class. This is + * used, for example, in an+b psuedoclasses. + * @retval boolean + * TRUE if the node matches, FALSE otherwise. + */ + public function elementMatches($pseudoclass, $node, $scope, $value = NULL) { + $name = strtolower($pseudoclass); + // Need to handle known pseudoclasses. + switch($name) { + case 'current': + case 'past': + case 'future': + case 'visited': + case 'hover': + case 'active': + case 'focus': + case 'animated': // Last 3 are from jQuery + case 'visible': + case 'hidden': + // These require a UA, which we don't have. + case 'valid': + case 'invalid': + case 'required': + case 'optional': + case 'read-only': + case 'read-write': + // Since we don't know how to validate elements, + // we can't supply these. + case 'dir': + // FIXME: I don't know how to get directionality info. + case 'nth-column': + case 'nth-last-column': + // We don't know what a column is in most documents. + // FIXME: Can we do this for HTML? + case 'target': + // This requires a location URL, which we don't have. + return FALSE; + case 'indeterminate': + // Because sometimes screwing with people is fun. + return (boolean) mt_rand(0, 1); + case 'lang': + // No value = exception. + if (!isset($value)) { + throw new NotImplementedException(":lang() requires a value."); + } + return $this->lang($node, $value); + case 'any-link': + return Util::matchesAttribute($node, 'href') + || Util::matchesAttribute($node, 'src') + || Util::matchesAttribute($node, 'link'); + case 'link': + return Util::matchesAttribute($node, 'href'); + case 'local-link': + return $this->isLocalLink($node); + case 'root': + return $node->isSameNode($node->ownerDocument->documentElement); + + // CSS 4 declares the :scope pseudo-class, which describes what was + // the :x-root QueryPath extension. + case 'x-root': + case 'x-reset': + case 'scope': + return $node->isSameNode($scope); + // NON-STANDARD extensions for simple support of even and odd. These + // are supported by jQuery, FF, and other user agents. + case 'even': + return $this->isNthChild($node, 'even'); + case 'odd': + return $this->isNthChild($node, 'odd'); + case 'nth-child': + return $this->isNthChild($node, $value); + case 'nth-last-child': + return $this->isNthChild($node, $value, TRUE); + case 'nth-of-type': + return $this->isNthChild($node, $value, FALSE, TRUE); + case 'nth-last-of-type': + return $this->isNthChild($node, $value, TRUE, TRUE); + case 'first-of-type': + return $this->isFirstOfType($node); + case 'last-of-type': + return $this->isLastOfType($node); + case 'only-of-type': + return $this->isFirstOfType($node) && $this->isLastOfType($node); + + // Additional pseudo-classes defined in jQuery: + case 'lt': + // I'm treating this as "less than or equal to". + $rule = sprintf('-n + %d', (int) $value); + // $rule = '-n+15'; + return $this->isNthChild($node, $rule); + case 'gt': + // I'm treating this as "greater than" + // return $this->nodePositionFromEnd($node) > (int) $value; + return $this->nodePositionFromStart($node) > (int) $value; + case 'nth': + case 'eq': + $rule = (int)$value; + return $this->isNthChild($node, $rule); + case 'first': + return $this->isNthChild($node, 1); + case 'first-child': + return $this->isFirst($node); + case 'last': + case 'last-child': + return $this->isLast($node); + case 'only-child': + return $this->isFirst($node) && $this->isLast($node); + case 'empty': + return $this->isEmpty($node); + case 'parent': + return !$this->isEmpty($node); + + case 'enabled': + case 'disabled': + case 'checked': + return Util::matchesAttribute($node, $name); + case 'text': + case 'radio': + case 'checkbox': + case 'file': + case 'password': + case 'submit': + case 'image': + case 'reset': + case 'button': + return Util::matchesAttribute($node, 'type', $name); + + case 'header': + return $this->header($node); + case 'has': + case 'matches': + return $this->has($node, $value); + break; + case 'not': + if (empty($value)) { + throw new ParseException(":not() requires a value."); + } + return $this->isNot($node, $value); + // Contains == text matches. + // In QP 2.1, this was changed. + case 'contains': + return $this->contains($node, $value); + // Since QP 2.1 + case 'contains-exactly': + return $this->containsExactly($node, $value); + default: + throw new \QueryPath\CSS\ParseException("Unknown Pseudo-Class: " . $name); + } + $this->findAnyElement = FALSE; + } + + /** + * Pseudo-class handler for :lang + * + * Note that this does not implement the spec in its entirety because we do + * not presume to "know the language" of the document. If anyone is interested + * in making this more intelligent, please do so. + */ + protected function lang($node, $value) { + // TODO: This checks for cases where an explicit language is + // set. The spec seems to indicate that an element should inherit + // language from the parent... but this is unclear. + $operator = (strpos($value, '-') !== FALSE) ? EventHandler::isExactly : EventHandler::containsWithHyphen; + + $match = TRUE; + foreach ($node->attributes as $attrNode) { + if ($attrNode->localName == 'lang') { + + if ($attrNode->nodeName == $attrNode->localName) { + // fprintf(STDOUT, "%s in NS %s\n", $attrNode->name, $attrNode->nodeName); + return Util::matchesAttribute($node, 'lang', $value, $operator); + } + else { + $nsuri = $attrNode->namespaceURI; + // fprintf(STDOUT, "%s in NS %s\n", $attrNode->name, $nsuri); + return Util::matchesAttributeNS($node, 'lang', $nsuri, $value, $operator); + } + + } + } + return FALSE; + } + + /** + * Provides jQuery pseudoclass ':header'. + */ + protected function header($node) { + return preg_match('/^h[1-9]$/i', $node->tagName) == 1; + } + + /** + * Provides pseudoclass :empty. + */ + protected function isEmpty($node) { + foreach ($node->childNodes as $kid) { + // We don't want to count PIs and comments. From the spec, it + // appears that CDATA is also not counted. + if ($kid->nodeType == XML_ELEMENT_NODE || $kid->nodeType == XML_TEXT_NODE) { + // As soon as we hit a FALSE, return. + return FALSE; + } + } + return TRUE; + } + + /** + * Provides jQuery pseudoclass :first. + * + * @todo + * This can be replaced by isNthChild(). + */ + protected function isFirst($node) { + while (isset($node->previousSibling)) { + $node = $node->previousSibling; + if ($node->nodeType == XML_ELEMENT_NODE) { + return FALSE; + } + } + return TRUE; + } + /** + * Fast version of first-of-type. + */ + protected function isFirstOfType($node) { + $type = $node->tagName; + while (isset($node->previousSibling)) { + $node = $node->previousSibling; + if ($node->nodeType == XML_ELEMENT_NODE && $node->tagName == $type) { + return FALSE; + } + } + return TRUE; + } + /** + * Fast version of jQuery :last. + */ + protected function isLast($node) { + while (isset($node->nextSibling)) { + $node = $node->nextSibling; + if ($node->nodeType == XML_ELEMENT_NODE) { + return FALSE; + } + } + return TRUE; + } + /** + * Provides last-of-type. + */ + protected function isLastOfType($node) { + $type = $node->tagName; + while (isset($node->nextSibling)) { + $node = $node->nextSibling; + if ($node->nodeType == XML_ELEMENT_NODE && $node->tagName == $type) { + return FALSE; + } + } + return TRUE; + } + /** + * Provides :contains() as the original spec called for. + * + * This is an INEXACT match. + */ + protected function contains($node, $value) { + $text = $node->textContent; + $value = Util::removeQuotes($value); + return isset($text) && (stripos($text, $value) !== FALSE); + } + /** + * Provides :contains-exactly QueryPath pseudoclass. + * + * This is an EXACT match. + */ + protected function containsExactly($node, $value) { + $text = $node->textContent; + $value = Util::removeQuotes($value); + return isset($text) && $text == $value; + } + + /** + * Provides :has pseudoclass. + */ + protected function has($node, $selector) { + $splos = new \SPLObjectStorage(); + $splos->attach($node); + $traverser = new \QueryPath\CSS\DOMTraverser($splos, TRUE); + $results = $traverser->find($selector)->matches(); + return count($results) > 0; + } + + /** + * Provides :not pseudoclass. + */ + protected function isNot($node, $selector) { + return !$this->has($node, $selector); + } + + /** + * Get the relative position of a node in its sibling set. + */ + protected function nodePositionFromStart($node, $byType = FALSE) { + $i = 1; + $tag = $node->tagName; + while (isset($node->previousSibling)) { + $node = $node->previousSibling; + if ($node->nodeType == XML_ELEMENT_NODE && (!$byType || $node->tagName == $tag)) { + ++$i; + } + } + return $i; + } + /** + * Get the relative position of a node in its sibling set. + */ + protected function nodePositionFromEnd($node, $byType = FALSE) { + $i = 1; + $tag = $node->tagName; + while (isset($node->nextSibling)) { + $node = $node->nextSibling; + if ($node->nodeType == XML_ELEMENT_NODE && (!$byType || $node->tagName == $tag)) { + ++$i; + } + } + return $i; + } + + /** + * Provides functionality for all "An+B" rules. + * Provides nth-child and also the functionality required for: + * + *- nth-last-child + *- even + *- odd + *- first + *- last + *- eq + *- nth + *- nth-of-type + *- first-of-type + *- last-of-type + *- nth-last-of-type + * + * See also QueryPath::CSS::DOMTraverser::Util::parseAnB(). + */ + protected function isNthChild($node, $value, $reverse = FALSE, $byType = FALSE) { + list($groupSize, $elementInGroup) = Util::parseAnB($value); + $parent = $node->parentNode; + if (empty($parent) + || ($groupSize == 0 && $elementInGroup == 0) + || ($groupSize > 0 && $elementInGroup > $groupSize) + ) { + return FALSE; + } + + // First we need to find the position of $node in other elements. + if ($reverse) { + $pos = $this->nodePositionFromEnd($node, $byType); + } + else { + $pos = $this->nodePositionFromStart($node, $byType); + } + + // If group size is 0, we just check to see if this + // is the nth element: + if ($groupSize == 0) { + return $pos == $elementInGroup; + } + + // Next, we normalize $elementInGroup + if ($elementInGroup < 0) { + $elementInGroup = $groupSize + $elementInGroup; + } + + + $prod = ($pos - $elementInGroup) / $groupSize; + // fprintf(STDOUT, "%d n + %d on %d is %3.5f\n", $groupSize, $elementInGroup, $pos, $prod); + + return is_int($prod) && $prod >= 0; + } + + protected function isLocalLink($node) { + if (!$node->hasAttribute('href')) { + return FALSE; + } + $url = $node->getAttribute('href'); + $scheme = parse_url($url, PHP_URL_SCHEME); + return empty($scheme) || $scheme == 'file'; + } + +} diff --git a/lib/querypath/src/QueryPath/CSS/DOMTraverser/Util.php b/lib/querypath/src/QueryPath/CSS/DOMTraverser/Util.php new file mode 100644 index 0000000..ec01d8f --- /dev/null +++ b/lib/querypath/src/QueryPath/CSS/DOMTraverser/Util.php @@ -0,0 +1,139 @@ +<?php +/** + * @file + * + * Utilities for DOM traversal. + */ +namespace QueryPath\CSS\DOMTraverser; + +use \QueryPath\CSS\EventHandler; + +/** + * Utilities for DOM Traversal. + */ +class Util { + + /** + * Check whether the given DOMElement has the given attribute. + */ + public static function matchesAttribute($node, $name, $value = NULL, $operation = EventHandler::isExactly) { + if (!$node->hasAttribute($name)) { + return FALSE; + } + + if (is_null($value)) { + return TRUE; + } + + return self::matchesAttributeValue($value, $node->getAttribute($name), $operation); + } + /** + * Check whether the given DOMElement has the given namespaced attribute. + */ + public static function matchesAttributeNS($node, $name, $nsuri, $value = NULL, $operation = EventHandler::isExactly) { + if (!$node->hasAttributeNS($nsuri, $name)) { + return FALSE; + } + + if (is_null($value)) { + return TRUE; + } + + return self::matchesAttributeValue($value, $node->getAttributeNS($nsuri, $name), $operation); + } + + /** + * Check for attr value matches based on an operation. + */ + public static function matchesAttributeValue($needle, $haystack, $operation) { + + if (strlen($haystack) < strlen($needle)) return FALSE; + + // According to the spec: + // "The case-sensitivity of attribute names in selectors depends on the document language." + // (6.3.2) + // To which I say, "huh?". We assume case sensitivity. + switch ($operation) { + case EventHandler::isExactly: + return $needle == $haystack; + case EventHandler::containsWithSpace: + // XXX: This needs testing! + return preg_match('/\b/', $haystack) == 1; + //return in_array($needle, explode(' ', $haystack)); + case EventHandler::containsWithHyphen: + return in_array($needle, explode('-', $haystack)); + case EventHandler::containsInString: + return strpos($haystack, $needle) !== FALSE; + case EventHandler::beginsWith: + return strpos($haystack, $needle) === 0; + case EventHandler::endsWith: + //return strrpos($haystack, $needle) === strlen($needle) - 1; + return preg_match('/' . $needle . '$/', $haystack) == 1; + } + return FALSE; // Shouldn't be able to get here. + } + + /** + * Remove leading and trailing quotes. + */ + public static function removeQuotes($str) { + $f = substr($str, 0, 1); + $l = substr($str, -1); + if ($f === $l && ($f == '"' || $f == "'")) { + $str = substr($str, 1, -1); + } + return $str; + } + + /** + * Parse an an+b rule for CSS pseudo-classes. + * + * Invalid rules return `array(0, 0)`. This is per the spec. + * + * @param $rule + * Some rule in the an+b format. + * @retval array + * `array($aVal, $bVal)` of the two values. + */ + public static function parseAnB($rule) { + if ($rule == 'even') { + return array(2, 0); + } + elseif ($rule == 'odd') { + return array(2, 1); + } + elseif ($rule == 'n') { + return array(1, 0); + } + elseif (is_numeric($rule)) { + return array(0, (int)$rule); + } + + $regex = '/^\s*([+\-]?[0-9]*)n\s*([+\-]?)\s*([0-9]*)\s*$/'; + $matches = array(); + $res = preg_match($regex, $rule, $matches); + + // If it doesn't parse, return 0, 0. + if (!$res) { + return array(0, 0); + } + + $aVal = isset($matches[1]) ? $matches[1] : 1; + if ($aVal == '-') { + $aVal = -1; + } + else { + $aVal = (int) $aVal; + } + + $bVal = 0; + if (isset($matches[3])) { + $bVal = (int) $matches[3]; + if (isset($matches[2]) && $matches[2] == '-') { + $bVal *= -1; + } + } + return array($aVal, $bVal); + } + +} diff --git a/lib/querypath/src/QueryPath/CSS/EventHandler.php b/lib/querypath/src/QueryPath/CSS/EventHandler.php new file mode 100644 index 0000000..a003a0a --- /dev/null +++ b/lib/querypath/src/QueryPath/CSS/EventHandler.php @@ -0,0 +1,171 @@ +<?php +/** @file + * CSS selector parsing classes. + * + * This file contains the tools necessary for parsing CSS 3 selectors. + * In the future it may be expanded to handle all of CSS 3. + * + * The parser contained herein is has an event-based API. Implementors should + * begin by implementing the {@link EventHandler} interface. For an example + * of how this is done, see {@link EventHandler.php}. + * + * @author M Butcher <matt@aleph-null.tv> + * @license MIT + */ +namespace QueryPath\CSS; + +/** @addtogroup querypath_css CSS Parsing + * QueryPath includes a CSS 3 Selector parser. + * + * + * Typically the parser is not accessed directly. Most developers will use it indirectly from + * qp(), htmlqp(), or one of the methods on a QueryPath object. + * + * This parser is modular and is not tied to QueryPath, so you can use it in your + * own (non-QueryPath) projects if you wish. To dive in, start with EventHandler, the + * event interface that works like a SAX API for CSS selectors. If you want to check out + * the details, check out the parser (QueryPath::CSS::Parser), scanner + * (QueryPath::CSS::Scanner), and token list (QueryPath::CSS::Token). + */ + +/** + * An event handler for handling CSS 3 Selector parsing. + * + * This provides a standard interface for CSS 3 Selector event handling. As the + * parser parses a selector, it will fire events. Implementations of EventHandler + * can then handle the events. + * + * This library is inspired by the SAX2 API for parsing XML. Each component of a + * selector fires an event, passing the necessary data on to the event handler. + * + * @ingroup querypath_css + */ +interface EventHandler { + /** The is-exactly (=) operator. */ + const isExactly = 0; // = + /** The contains-with-space operator (~=). */ + const containsWithSpace = 1; // ~= + /** The contains-with-hyphen operator (!=). */ + const containsWithHyphen = 2; // |= + /** The contains-in-string operator (*=). */ + const containsInString = 3; // *= + /** The begins-with operator (^=). */ + const beginsWith = 4; // ^= + /** The ends-with operator ($=). */ + const endsWith = 5; // $= + /** The any-element operator (*). */ + const anyElement = '*'; + + /** + * This event is fired when a CSS ID is encountered. + * An ID begins with an octothorp: #name. + * + * @param string $id + * The ID passed in. + */ + public function elementID($id); // #name + /** + * Handle an element name. + * Example: name + * @param string $name + * The name of the element. + */ + public function element($name); // name + /** + * Handle a namespaced element name. + * example: namespace|name + * @param string $name + * The tag name. + * @param string $namespace + * The namespace identifier (Not the URI) + */ + public function elementNS($name, $namespace = NULL); + /** + * Handle an any-element (*) operator. + * Example: * + */ + public function anyElement(); // * + /** + * Handle an any-element operator that is constrained to a namespace. + * Example: ns|* + * @param string $ns + * The namespace identifier (not the URI). + */ + public function anyElementInNS($ns); // ns|* + /** + * Handle a CSS class selector. + * Example: .name + * @param string $name + * The name of the class. + */ + public function elementClass($name); // .name + /** + * Handle an attribute selector. + * Example: [name=attr] + * Example: [name~=attr] + * @param string $name + * The attribute name. + * @param string $value + * The value of the attribute, if given. + * @param int $operation + * The operation to be used for matching. See {@link EventHandler} + * constants for a list of supported operations. + */ + public function attribute($name, $value = NULL, $operation = EventHandler::isExactly); // [name=attr] + /** + * Handle an attribute selector bound to a specific namespace. + * Example: [ns|name=attr] + * Example: [ns|name~=attr] + * @param string $name + * The attribute name. + * @param string $ns + * The namespace identifier (not the URI). + * @param string $value + * The value of the attribute, if given. + * @param int $operation + * The operation to be used for matching. See {@link EventHandler} + * constants for a list of supported operations. + */ + public function attributeNS($name, $ns, $value = NULL, $operation = EventHandler::isExactly); + /** + * Handle a pseudo-class. + * Example: :name(value) + * @param string $name + * The pseudo-class name. + * @param string $value + * The value, if one is found. + */ + public function pseudoClass($name, $value = NULL); //:name(value) + /** + * Handle a pseudo-element. + * Example: ::name + * @param string $name + * The pseudo-element name. + */ + public function pseudoElement($name); // ::name + /** + * Handle a direct descendant combinator. + * Example: > + */ + public function directDescendant(); // > + /** + * Handle a adjacent combinator. + * Example: + + */ + public function adjacent(); // + + /** + * Handle an another-selector combinator. + * Example: , + */ + public function anotherSelector(); // , + /** + * Handle a sibling combinator. + * Example: ~ + */ + public function sibling(); // ~ combinator + /** + * Handle an any-descendant combinator. + * Example: ' ' + */ + public function anyDescendant(); // ' ' (space) operator. +} diff --git a/lib/querypath/src/QueryPath/CSS/InputStream.php b/lib/querypath/src/QueryPath/CSS/InputStream.php new file mode 100644 index 0000000..2967845 --- /dev/null +++ b/lib/querypath/src/QueryPath/CSS/InputStream.php @@ -0,0 +1,57 @@ +<?php +/** + * @file + * + * The CSS Input Stream abstraction. + */ + +namespace QueryPath\CSS; + +/** + * Simple wrapper to turn a string into an input stream. + * This provides a standard interface on top of an array of + * characters. + */ +class InputStream { + protected $stream = NULL; + public $position = 0; + /** + * Build a new CSS input stream from a string. + * + * @param string + * String to turn into an input stream. + */ + function __construct($string) { + $this->stream = str_split($string); + } + /** + * Look ahead one character. + * + * @return char + * Returns the next character, but does not remove it from + * the stream. + */ + function peek() { + return $this->stream[0]; + } + /** + * Get the next unconsumed character in the stream. + * This will remove that character from the front of the + * stream and return it. + */ + function consume() { + $ret = array_shift($this->stream); + if (!empty($ret)) { + $this->position++; + } + return $ret; + } + /** + * Check if the stream is empty. + * @return boolean + * Returns TRUE when the stream is empty, FALSE otherwise. + */ + function isEmpty() { + return count($this->stream) == 0; + } +} diff --git a/lib/querypath/src/QueryPath/CSS/NotImplementedException.php b/lib/querypath/src/QueryPath/CSS/NotImplementedException.php new file mode 100644 index 0000000..6705f30 --- /dev/null +++ b/lib/querypath/src/QueryPath/CSS/NotImplementedException.php @@ -0,0 +1,15 @@ +<?php +/** + * @file + * An exception for CSS errors. + */ +namespace QueryPath\CSS; +/** + * Exception thrown for unimplemented CSS. + * + * This is thrown in cases where some feature is expected, but the current + * implementation does not support that feature. + * + * @ingroup querypath_css + */ +class NotImplementedException extends \Exception {} diff --git a/lib/querypath/src/QueryPath/CSS/ParseException.php b/lib/querypath/src/QueryPath/CSS/ParseException.php new file mode 100644 index 0000000..957857a --- /dev/null +++ b/lib/querypath/src/QueryPath/CSS/ParseException.php @@ -0,0 +1,15 @@ +<?php +/** + * @file + * + * The CSS parsing exception class. + */ + +namespace QueryPath\CSS; + +/** + * Exception indicating an error in CSS parsing. + * + * @ingroup querypath_css + */ +class ParseException extends \QueryPath\Exception {} diff --git a/lib/querypath/src/QueryPath/CSS/Parser.php b/lib/querypath/src/QueryPath/CSS/Parser.php new file mode 100644 index 0000000..f041612 --- /dev/null +++ b/lib/querypath/src/QueryPath/CSS/Parser.php @@ -0,0 +1,575 @@ +<?php +/** + * @file + * + * The CSS parser + */ + +namespace QueryPath\CSS; + +/** + * Parse a CSS selector. + * + * In CSS, a selector is used to identify which element or elements + * in a DOM are being selected for the application of a particular style. + * Effectively, selectors function as a query language for a structured + * document -- almost always HTML or XML. + * + * This class provides an event-based parser for CSS selectors. It can be + * used, for example, as a basis for writing a DOM query engine based on + * CSS. + * + * @ingroup querypath_css + */ +class Parser { + protected $scanner = NULL; + protected $buffer = ''; + protected $handler = NULL; + protected $strict = FALSE; + + protected $DEBUG = FALSE; + + /** + * Construct a new CSS parser object. This will attempt to + * parse the string as a CSS selector. As it parses, it will + * send events to the EventHandler implementation. + */ + public function __construct($string, EventHandler $handler) { + $this->originalString = $string; + $is = new InputStream($string); + $this->scanner = new Scanner($is); + $this->handler = $handler; + } + + /** + * Parse the selector. + * + * This begins an event-based parsing process that will + * fire events as the selector is handled. A EventHandler + * implementation will be responsible for handling the events. + * @throws ParseException + */ + public function parse() { + + $this->scanner->nextToken(); + while ($this->scanner->token !== FALSE) { + // Primitive recursion detection. + $position = $this->scanner->position(); + + if ($this->DEBUG) { + print "PARSE " . $this->scanner->token. "\n"; + } + $this->selector(); + + $finalPosition = $this->scanner->position(); + + if ($this->scanner->token !== FALSE && $finalPosition == $position) { + // If we get here, then the scanner did not pop a single character + // off of the input stream during a full run of the parser, which + // means that the current input does not match any recognizable + // pattern. + throw new ParseException('CSS selector is not well formed.'); + } + + } + + } + + /** + * A restricted parser that can only parse simple selectors. + * The pseudoClass handler for this parser will throw an + * exception if it encounters a pseudo-element or the + * negation pseudo-class. + * + * @deprecated This is not used anywhere in QueryPath and + * may be removed. + *//* + public function parseSimpleSelector() { + while ($this->scanner->token !== FALSE) { + if ($this->DEBUG) print "SIMPLE SELECTOR\n"; + $this->allElements(); + $this->elementName(); + $this->elementClass(); + $this->elementID(); + $this->pseudoClass(TRUE); // Operate in restricted mode. + $this->attribute(); + + // TODO: Need to add failure conditions here. + } + }*/ + + /** + * Handle an entire CSS selector. + */ + private function selector() { + if ($this->DEBUG) print "SELECTOR{$this->scanner->position()}\n"; + $this->consumeWhitespace(); // Remove leading whitespace + $this->simpleSelectors(); + $this->combinator(); + } + + /** + * Consume whitespace and return a count of the number of whitespace consumed. + */ + private function consumeWhitespace() { + if ($this->DEBUG) print "CONSUME WHITESPACE\n"; + $white = 0; + while ($this->scanner->token == Token::white) { + $this->scanner->nextToken(); + ++$white; + } + return $white; + } + + /** + * Handle one of the five combinators: '>', '+', ' ', '~', and ','. + * This will call the appropriate event handlers. + * @see EventHandler::directDescendant(), + * @see EventHandler::adjacent(), + * @see EventHandler::anyDescendant(), + * @see EventHandler::anotherSelector(). + */ + private function combinator() { + if ($this->DEBUG) print "COMBINATOR\n"; + /* + * Problem: ' ' and ' > ' are both valid combinators. + * So we have to track whitespace consumption to see + * if we are hitting the ' ' combinator or if the + * selector just has whitespace padding another combinator. + */ + + // Flag to indicate that post-checks need doing + $inCombinator = FALSE; + $white = $this->consumeWhitespace(); + $t = $this->scanner->token; + + if ($t == Token::rangle) { + $this->handler->directDescendant(); + $this->scanner->nextToken(); + $inCombinator = TRUE; + //$this->simpleSelectors(); + } + elseif ($t == Token::plus) { + $this->handler->adjacent(); + $this->scanner->nextToken(); + $inCombinator = TRUE; + //$this->simpleSelectors(); + } + elseif ($t == Token::comma) { + $this->handler->anotherSelector(); + $this->scanner->nextToken(); + $inCombinator = TRUE; + //$this->scanner->selectors(); + } + elseif ($t == Token::tilde) { + $this->handler->sibling(); + $this->scanner->nextToken(); + $inCombinator = TRUE; + } + + // Check that we don't get two combinators in a row. + if ($inCombinator) { + $white = 0; + if ($this->DEBUG) print "COMBINATOR: " . Token::name($t) . "\n"; + $this->consumeWhitespace(); + if ($this->isCombinator($this->scanner->token)) { + throw new ParseException("Illegal combinator: Cannot have two combinators in sequence."); + } + } + // Check to see if we have whitespace combinator: + elseif ($white > 0) { + if ($this->DEBUG) print "COMBINATOR: any descendant\n"; + $inCombinator = TRUE; + $this->handler->anyDescendant(); + } + else { + if ($this->DEBUG) print "COMBINATOR: no combinator found.\n"; + } + } + + /** + * Check if the token is a combinator. + */ + private function isCombinator($tok) { + $combinators = array(Token::plus, Token::rangle, Token::comma, Token::tilde); + return in_array($tok, $combinators); + } + + /** + * Handle a simple selector. + */ + private function simpleSelectors() { + if ($this->DEBUG) print "SIMPLE SELECTOR\n"; + $this->allElements(); + $this->elementName(); + $this->elementClass(); + $this->elementID(); + $this->pseudoClass(); + $this->attribute(); + } + + /** + * Handles CSS ID selectors. + * This will call EventHandler::elementID(). + */ + private function elementID() { + if ($this->DEBUG) print "ELEMENT ID\n"; + if ($this->scanner->token == Token::octo) { + $this->scanner->nextToken(); + if ($this->scanner->token !== Token::char) { + throw new ParseException("Expected string after #"); + } + $id = $this->scanner->getNameString(); + $this->handler->elementID($id); + } + } + + /** + * Handles CSS class selectors. + * This will call the EventHandler::elementClass() method. + */ + private function elementClass() { + if ($this->DEBUG) print "ELEMENT CLASS\n"; + if ($this->scanner->token == Token::dot) { + $this->scanner->nextToken(); + $this->consumeWhitespace(); // We're very fault tolerent. This should prob through error. + $cssClass = $this->scanner->getNameString(); + $this->handler->elementClass($cssClass); + } + } + + /** + * Handle a pseudo-class and pseudo-element. + * + * CSS 3 selectors support separate pseudo-elements, using :: instead + * of : for separator. This is now supported, and calls the pseudoElement + * handler, EventHandler::pseudoElement(). + * + * This will call EventHandler::pseudoClass() when a + * pseudo-class is parsed. + */ + private function pseudoClass($restricted = FALSE) { + if ($this->DEBUG) print "PSEUDO-CLASS\n"; + if ($this->scanner->token == Token::colon) { + + // Check for CSS 3 pseudo element: + $isPseudoElement = FALSE; + if ($this->scanner->nextToken() === Token::colon) { + $isPseudoElement = TRUE; + $this->scanner->nextToken(); + } + + $name = $this->scanner->getNameString(); + if ($restricted && $name == 'not') { + throw new ParseException("The 'not' pseudo-class is illegal in this context."); + } + + $value = NULL; + if ($this->scanner->token == Token::lparen) { + if ($isPseudoElement) { + throw new ParseException("Illegal left paren. Pseudo-Element cannot have arguments."); + } + $value = $this->pseudoClassValue(); + } + + // FIXME: This should throw errors when pseudo element has values. + if ($isPseudoElement) { + if ($restricted) { + throw new ParseException("Pseudo-Elements are illegal in this context."); + } + $this->handler->pseudoElement($name); + $this->consumeWhitespace(); + + // Per the spec, pseudo-elements must be the last items in a selector, so we + // check to make sure that we are either at the end of the stream or that a + // new selector is starting. Only one pseudo-element is allowed per selector. + if ($this->scanner->token !== FALSE && $this->scanner->token !== Token::comma) { + throw new ParseException("A Pseudo-Element must be the last item in a selector."); + } + } + else { + $this->handler->pseudoClass($name, $value); + } + } + } + + /** + * Get the value of a pseudo-classes. + * + * @return string + * Returns the value found from a pseudo-class. + * + * @todo Pseudoclasses can be passed pseudo-elements and + * other pseudo-classes as values, which means :pseudo(::pseudo) + * is legal. + */ + private function pseudoClassValue() { + if ($this->scanner->token == Token::lparen) { + $buf = ''; + + // For now, just leave pseudoClass value vague. + /* + // We have to peek to see if next char is a colon because + // pseudo-classes and pseudo-elements are legal strings here. + print $this->scanner->peek(); + if ($this->scanner->peek() == ':') { + print "Is pseudo\n"; + $this->scanner->nextToken(); + + // Pseudo class + if ($this->scanner->token == Token::colon) { + $buf .= ':'; + $this->scanner->nextToken(); + // Pseudo element + if ($this->scanner->token == Token::colon) { + $buf .= ':'; + $this->scanner->nextToken(); + } + // Ident + $buf .= $this->scanner->getNameString(); + } + } + else { + print "fetching string.\n"; + $buf .= $this->scanner->getQuotedString(); + if ($this->scanner->token != Token::rparen) { + $this->throwError(Token::rparen, $this->scanner->token); + } + $this->scanner->nextToken(); + } + return $buf; + */ + //$buf .= $this->scanner->getQuotedString(); + $buf .= $this->scanner->getPseudoClassString(); + return $buf; + } + } + + /** + * Handle element names. + * This will call the EventHandler::elementName(). + * + * This handles: + * <code> + * name (EventHandler::element()) + * |name (EventHandler::element()) + * ns|name (EventHandler::elementNS()) + * ns|* (EventHandler::elementNS()) + * </code> + */ + private function elementName() { + if ($this->DEBUG) print "ELEMENT NAME\n"; + if ($this->scanner->token === Token::pipe) { + // We have '|name', which is equiv to 'name' + $this->scanner->nextToken(); + $this->consumeWhitespace(); + $elementName = $this->scanner->getNameString(); + $this->handler->element($elementName); + } + elseif ($this->scanner->token === Token::char) { + $elementName = $this->scanner->getNameString(); + if ($this->scanner->token == Token::pipe) { + // Get ns|name + $elementNS = $elementName; + $this->scanner->nextToken(); + $this->consumeWhitespace(); + if ($this->scanner->token === Token::star) { + // We have ns|* + $this->handler->anyElementInNS($elementNS); + $this->scanner->nextToken(); + } + elseif ($this->scanner->token !== Token::char) { + $this->throwError(Token::char, $this->scanner->token); + } + else { + $elementName = $this->scanner->getNameString(); + // We have ns|name + $this->handler->elementNS($elementName, $elementNS); + } + + } + else { + $this->handler->element($elementName); + } + } + } + + /** + * Check for all elements designators. Due to the new CSS 3 namespace + * support, this is slightly more complicated, now, as it handles + * the *|name and *|* cases as well as *. + * + * Calls EventHandler::anyElement() or EventHandler::elementName(). + */ + private function allElements() { + if ($this->scanner->token === Token::star) { + $this->scanner->nextToken(); + if ($this->scanner->token === Token::pipe) { + $this->scanner->nextToken(); + if ($this->scanner->token === Token::star) { + // We got *|*. According to spec, this requires + // that the element has a namespace, so we pass it on + // to the handler: + $this->scanner->nextToken(); + $this->handler->anyElementInNS('*'); + } + else { + // We got *|name, which means the name MUST be in a namespce, + // so we pass this off to elementNameNS(). + $name = $this->scanner->getNameString(); + $this->handler->elementNS($name, '*'); + } + } + else { + $this->handler->anyElement(); + } + } + } + + /** + * Handler an attribute. + * An attribute can be in one of two forms: + * <code>[attrName]</code> + * or + * <code>[attrName="AttrValue"]</code> + * + * This may call the following event handlers: EventHandler::attribute(). + */ + private function attribute() { + if($this->scanner->token == Token::lsquare) { + $attrVal = $op = $ns = NULL; + + $this->scanner->nextToken(); + $this->consumeWhitespace(); + + if ($this->scanner->token === Token::at) { + if ($this->strict) { + throw new ParseException('The @ is illegal in attributes.'); + } + else { + $this->scanner->nextToken(); + $this->consumeWhitespace(); + } + } + + if ($this->scanner->token === Token::star) { + // Global namespace... requires that attr be prefixed, + // so we pass this on to a namespace handler. + $ns = '*'; + $this->scanner->nextToken(); + } + if ($this->scanner->token === Token::pipe) { + // Skip this. It's a global namespace. + $this->scanner->nextToken(); + $this->consumeWhitespace(); + } + + $attrName = $this->scanner->getNameString(); + $this->consumeWhitespace(); + + // Check for namespace attribute: ns|attr. We have to peek() to make + // sure that we haven't hit the |= operator, which looks the same. + if ($this->scanner->token === Token::pipe && $this->scanner->peek() !== '=') { + // We have a namespaced attribute. + $ns = $attrName; + $this->scanner->nextToken(); + $attrName = $this->scanner->getNameString(); + $this->consumeWhitespace(); + } + + // Note: We require that operators do not have spaces + // between characters, e.g. ~= , not ~ =. + + // Get the operator: + switch ($this->scanner->token) { + case Token::eq: + $this->consumeWhitespace(); + $op = EventHandler::isExactly; + break; + case Token::tilde: + if ($this->scanner->nextToken() !== Token::eq) { + $this->throwError(Token::eq, $this->scanner->token); + } + $op = EventHandler::containsWithSpace; + break; + case Token::pipe: + if ($this->scanner->nextToken() !== Token::eq) { + $this->throwError(Token::eq, $this->scanner->token); + } + $op = EventHandler::containsWithHyphen; + break; + case Token::star: + if ($this->scanner->nextToken() !== Token::eq) { + $this->throwError(Token::eq, $this->scanner->token); + } + $op = EventHandler::containsInString; + break; + case Token::dollar; + if ($this->scanner->nextToken() !== Token::eq) { + $this->throwError(Token::eq, $this->scanner->token); + } + $op = EventHandler::endsWith; + break; + case Token::carat: + if ($this->scanner->nextToken() !== Token::eq) { + $this->throwError(Token::eq, $this->scanner->token); + } + $op = EventHandler::beginsWith; + break; + } + + if (isset($op)) { + // Consume '=' and go on. + $this->scanner->nextToken(); + $this->consumeWhitespace(); + + // So... here we have a problem. The grammer suggests that the + // value here is String1 or String2, both of which are enclosed + // in quotes of some sort, and both of which allow lots of special + // characters. But the spec itself includes examples like this: + // [lang=fr] + // So some bareword support is assumed. To get around this, we assume + // that bare words follow the NAME rules, while quoted strings follow + // the String1/String2 rules. + + if ($this->scanner->token === Token::quote || $this->scanner->token === Token::squote) { + $attrVal = $this->scanner->getQuotedString(); + } + else { + $attrVal = $this->scanner->getNameString(); + } + + if ($this->DEBUG) { + print "ATTR: $attrVal AND OP: $op\n"; + } + } + + $this->consumeWhitespace(); + + if ($this->scanner->token != Token::rsquare) { + $this->throwError(Token::rsquare, $this->scanner->token); + } + + if (isset($ns)) { + $this->handler->attributeNS($attrName, $ns, $attrVal, $op); + } + elseif (isset($attrVal)) { + $this->handler->attribute($attrName, $attrVal, $op); + } + else { + $this->handler->attribute($attrName); + } + $this->scanner->nextToken(); + } + } + + /** + * Utility for throwing a consistantly-formatted parse error. + */ + private function throwError($expected, $got) { + $filter = sprintf('Expected %s, got %s', Token::name($expected), Token::name($got)); + throw new ParseException($filter); + } + +} + diff --git a/lib/querypath/src/QueryPath/CSS/QueryPathEventHandler.php b/lib/querypath/src/QueryPath/CSS/QueryPathEventHandler.php new file mode 100644 index 0000000..2dcfd57 --- /dev/null +++ b/lib/querypath/src/QueryPath/CSS/QueryPathEventHandler.php @@ -0,0 +1,1424 @@ +<?php +/** @file + * This file contains a full implementation of the EventHandler interface. + * + * The tools in this package initiate a CSS selector parsing routine and then + * handle all of the callbacks. + * + * The implementation provided herein adheres to the CSS 3 Selector specification + * with the following caveats: + * + * - The negation (:not()) and containment (:has()) pseudo-classes allow *full* + * selectors and not just simple selectors. + * - There are a variety of additional pseudo-classes supported by this + * implementation that are not part of the spec. Most of the jQuery + * pseudo-classes are supported. The :x-root pseudo-class is also supported. + * - Pseudo-classes that require a User Agent to function have been disabled. + * Thus there is no :hover pseudo-class. + * - All pseudo-elements require the double-colon (::) notation. This breaks + * backward compatibility with the 2.1 spec, but it makes visible the issue + * that pseudo-elements cannot be effectively used with most of the present + * library. They return <b>stdClass objects with a text property</b> (QP > 1.3) + * instead of elements. + * - The pseudo-classes first-of-type, nth-of-type and last-of-type may or may + * not conform to the specification. The spec is unclear. + * - pseudo-class filters of the form -an+b do not function as described in the + * specification. However, they do behave the same way here as they do in + * jQuery. + * - This library DOES provide XML namespace aware tools. Selectors can use + * namespaces to increase specificity. + * - This library does nothing with the CSS 3 Selector specificity rating. Of + * course specificity is preserved (to the best of our abilities), but there + * is no calculation done. + * + * For detailed examples of how the code works and what selectors are supported, + * see the CssEventTests file, which contains the unit tests used for + * testing this implementation. + * + * @author M Butcher <matt@aleph-null.tv> + * @license MIT + */ + +namespace QueryPath\CSS; + +/** + * Handler that tracks progress of a query through a DOM. + * + * The main idea is that we keep a copy of the tree, and then use an + * array to keep track of matches. To handle a list of selectors (using + * the comma separator), we have to track both the currently progressing + * match and the previously matched elements. + * + * To use this handler: + * @code + * $filter = '#id'; // Some CSS selector + * $handler = new QueryPathEventHandler(DOMNode $dom); + * $parser = new Parser(); + * $parser->parse($filter, $handler); + * $matches = $handler->getMatches(); + * @endcode + * + * $matches will be an array of zero or more DOMElement objects. + * + * @ingroup querypath_css + */ +class QueryPathEventHandler implements EventHandler, Traverser { + protected $dom = NULL; // Always points to the top level. + protected $matches = NULL; // The matches + protected $alreadyMatched = NULL; // Matches found before current selector. + protected $findAnyElement = TRUE; + + + /** + * Create a new event handler. + */ + public function __construct($dom) { + $this->alreadyMatched = new \SplObjectStorage(); + $matches = new \SplObjectStorage(); + + // Array of DOMElements + if (is_array($dom) || $dom instanceof \SplObjectStorage) { + //$matches = array(); + foreach($dom as $item) { + if ($item instanceof \DOMNode && $item->nodeType == XML_ELEMENT_NODE) { + //$matches[] = $item; + $matches->attach($item); + } + } + //$this->dom = count($matches) > 0 ? $matches[0] : NULL; + if ($matches->count() > 0) { + $matches->rewind(); + $this->dom = $matches->current(); + } + else { + //throw new Exception("Setting DOM to Null"); + $this->dom = NULL; + } + $this->matches = $matches; + } + // DOM Document -- we get the root element. + elseif ($dom instanceof \DOMDocument) { + $this->dom = $dom->documentElement; + $matches->attach($dom->documentElement); + } + // DOM Element -- we use this directly + elseif ($dom instanceof \DOMElement) { + $this->dom = $dom; + $matches->attach($dom); + } + // NodeList -- We turn this into an array + elseif ($dom instanceof \DOMNodeList) { + $a = array(); // Not sure why we are doing this.... + foreach ($dom as $item) { + if ($item->nodeType == XML_ELEMENT_NODE) { + $matches->attach($item); + $a[] = $item; + } + } + $this->dom = $a; + } + // FIXME: Handle SimpleXML! + // Uh-oh... we don't support anything else. + else { + throw new \QueryPath\Exception("Unhandled type: " . get_class($dom)); + } + $this->matches = $matches; + } + + /** + * Generic finding method. + * + * This is the primary searching method used throughout QueryPath. + * + * @param string $filter + * A valid CSS 3 filter. + * @return QueryPathEventHandler + * Returns itself. + */ + public function find($filter) { + $parser = new Parser($filter, $this); + $parser->parse(); + return $this; + } + + /** + * Get the elements that match the evaluated selector. + * + * This should be called after the filter has been parsed. + * + * @return array + * The matched items. This is almost always an array of + * {@link DOMElement} objects. It is always an instance of + * {@link DOMNode} objects. + */ + public function getMatches() { + //$result = array_merge($this->alreadyMatched, $this->matches); + $result = new \SplObjectStorage(); + foreach($this->alreadyMatched as $m) $result->attach($m); + foreach($this->matches as $m) $result->attach($m); + return $result; + } + + public function matches() { + return $this->getMatches(); + } + + /** + * Find any element with the ID that matches $id. + * + * If this finds an ID, it will immediately quit. Essentially, it doesn't + * enforce ID uniqueness, but it assumes it. + * + * @param $id + * String ID for an element. + */ + public function elementID($id) { + $found = new \SplObjectStorage(); + $matches = $this->candidateList(); + foreach ($matches as $item) { + // Check if any of the current items has the desired ID. + if ($item->hasAttribute('id') && $item->getAttribute('id') === $id) { + $found->attach($item); + break; + } + } + $this->matches = $found; + $this->findAnyElement = FALSE; + } + + // Inherited + public function element($name) { + $matches = $this->candidateList(); + $this->findAnyElement = FALSE; + $found = new \SplObjectStorage(); + foreach ($matches as $item) { + // Should the existing item be included? + // In some cases (e.g. element is root element) + // it definitely should. But what about other cases? + if ($item->tagName == $name) { + $found->attach($item); + } + // Search for matching kids. + //$nl = $item->getElementsByTagName($name); + //$found = array_merge($found, $this->nodeListToArray($nl)); + } + + $this->matches = $found; + } + + // Inherited + public function elementNS($lname, $namespace = NULL) { + $this->findAnyElement = FALSE; + $found = new \SplObjectStorage(); + $matches = $this->candidateList(); + foreach ($matches as $item) { + // Looking up NS URI only works if the XMLNS attributes are declared + // at a level equal to or above the searching doc. Normalizing a doc + // should fix this, but it doesn't. So we have to use a fallback + // detection scheme which basically searches by lname and then + // does a post hoc check on the tagname. + + //$nsuri = $item->lookupNamespaceURI($namespace); + $nsuri = $this->dom->lookupNamespaceURI($namespace); + + // XXX: Presumably the base item needs to be checked. Spec isn't + // too clear, but there are three possibilities: + // - base should always be checked (what we do here) + // - base should never be checked (only children) + // - base should only be checked if it is the root node + if ($item instanceof \DOMNode + && $item->namespaceURI == $nsuri + && $lname == $item->localName) { + $found->attach($item); + } + + if (!empty($nsuri)) { + $nl = $item->getElementsByTagNameNS($nsuri, $lname); + // If something is found, merge them: + //if (!empty($nl)) $found = array_merge($found, $this->nodeListToArray($nl)); + if (!empty($nl)) $this->attachNodeList($nl, $found); + } + else { + //$nl = $item->getElementsByTagName($namespace . ':' . $lname); + $nl = $item->getElementsByTagName($lname); + $tagname = $namespace . ':' . $lname; + $nsmatches = array(); + foreach ($nl as $node) { + if ($node->tagName == $tagname) { + //$nsmatches[] = $node; + $found->attach($node); + } + } + // If something is found, merge them: + //if (!empty($nsmatches)) $found = array_merge($found, $nsmatches); + } + } + $this->matches = $found; + } + + public function anyElement() { + $found = new \SplObjectStorage(); + //$this->findAnyElement = TRUE; + $matches = $this->candidateList(); + foreach ($matches as $item) { + $found->attach($item); // Add self + // See issue #20 or section 6.2 of this: + // http://www.w3.org/TR/2009/PR-css3-selectors-20091215/#universal-selector + //$nl = $item->getElementsByTagName('*'); + //$this->attachNodeList($nl, $found); + } + + $this->matches = $found; + $this->findAnyElement = FALSE; + } + public function anyElementInNS($ns) { + //$this->findAnyElement = TRUE; + $nsuri = $this->dom->lookupNamespaceURI($ns); + $found = new \SplObjectStorage(); + if (!empty($nsuri)) { + $matches = $this->candidateList(); + foreach ($matches as $item) { + if ($item instanceOf \DOMNode && $nsuri == $item->namespaceURI) { + $found->attach($item); + } + } + } + $this->matches = $found;//UniqueElementList::get($found); + $this->findAnyElement = FALSE; + } + public function elementClass($name) { + + $found = new \SplObjectStorage(); + $matches = $this->candidateList(); + foreach ($matches as $item) { + if ($item->hasAttribute('class')) { + $classes = explode(' ', $item->getAttribute('class')); + if (in_array($name, $classes)) $found->attach($item); + } + } + + $this->matches = $found;//UniqueElementList::get($found); + $this->findAnyElement = FALSE; + } + + public function attribute($name, $value = NULL, $operation = EventHandler::isExactly) { + $found = new \SplObjectStorage(); + $matches = $this->candidateList(); + foreach ($matches as $item) { + if ($item->hasAttribute($name)) { + if (isset($value)) { + // If a value exists, then we need a match. + if($this->attrValMatches($value, $item->getAttribute($name), $operation)) { + $found->attach($item); + } + } + else { + // If no value exists, then we consider it a match. + $found->attach($item); + } + } + } + $this->matches = $found; //UniqueElementList::get($found); + $this->findAnyElement = FALSE; + } + + /** + * Helper function to find all elements with exact matches. + * + * @deprecated All use cases seem to be covered by attribute(). + */ + protected function searchForAttr($name, $value = NULL) { + $found = new \SplObjectStorage(); + $matches = $this->candidateList(); + foreach ($matches as $candidate) { + if ($candidate->hasAttribute($name)) { + // If value is required, match that, too. + if (isset($value) && $value == $candidate->getAttribute($name)) { + $found->attach($candidate); + } + // Otherwise, it's a match on name alone. + else { + $found->attach($candidate); + } + } + } + + $this->matches = $found; + } + + public function attributeNS($lname, $ns, $value = NULL, $operation = EventHandler::isExactly) { + $matches = $this->candidateList(); + $found = new \SplObjectStorage(); + if (count($matches) == 0) { + $this->matches = $found; + return; + } + + // Get the namespace URI for the given label. + //$uri = $matches[0]->lookupNamespaceURI($ns); + $matches->rewind(); + $e = $matches->current(); + $uri = $e->lookupNamespaceURI($ns); + + foreach ($matches as $item) { + //foreach ($item->attributes as $attr) { + // print "$attr->prefix:$attr->localName ($attr->namespaceURI), Value: $attr->nodeValue\n"; + //} + if ($item->hasAttributeNS($uri, $lname)) { + if (isset($value)) { + if ($this->attrValMatches($value, $item->getAttributeNS($uri, $lname), $operation)) { + $found->attach($item); + } + } + else { + $found->attach($item); + } + } + } + $this->matches = $found; + $this->findAnyElement = FALSE; + } + + /** + * This also supports the following nonstandard pseudo classes: + * - :x-reset/:x-root (reset to the main item passed into the constructor. Less drastic than :root) + * - :odd/:even (shorthand for :nth-child(odd)/:nth-child(even)) + */ + public function pseudoClass($name, $value = NULL) { + $name = strtolower($name); + // Need to handle known pseudoclasses. + switch($name) { + case 'visited': + case 'hover': + case 'active': + case 'focus': + case 'animated': // Last 3 are from jQuery + case 'visible': + case 'hidden': + // These require a UA, which we don't have. + case 'target': + // This requires a location URL, which we don't have. + $this->matches = new \SplObjectStorage(); + break; + case 'indeterminate': + // The assumption is that there is a UA and the format is HTML. + // I don't know if this should is useful without a UA. + throw new NotImplementedException(":indeterminate is not implemented."); + break; + case 'lang': + // No value = exception. + if (!isset($value)) { + throw new NotImplementedException("No handler for lang pseudoclass without value."); + } + $this->lang($value); + break; + case 'link': + $this->searchForAttr('href'); + break; + case 'root': + $found = new \SplObjectStorage(); + if (empty($this->dom)) { + $this->matches = $found; + } + elseif (is_array($this->dom)) { + $found->attach($this->dom[0]->ownerDocument->documentElement); + $this->matches = $found; + } + elseif ($this->dom instanceof \DOMNode) { + $found->attach($this->dom->ownerDocument->documentElement); + $this->matches = $found; + } + elseif ($this->dom instanceof \DOMNodeList && $this->dom->length > 0) { + $found->attach($this->dom->item(0)->ownerDocument->documentElement); + $this->matches = $found; + } + else { + // Hopefully we never get here: + $found->attach($this->dom); + $this->matches = $found; + } + break; + + // NON-STANDARD extensions for reseting to the "top" items set in + // the constructor. + case 'x-root': + case 'x-reset': + $this->matches = new \SplObjectStorage(); + $this->matches->attach($this->dom); + break; + + // NON-STANDARD extensions for simple support of even and odd. These + // are supported by jQuery, FF, and other user agents. + case 'even': + $this->nthChild(2, 0); + break; + case 'odd': + $this->nthChild(2, 1); + break; + + // Standard child-checking items. + case 'nth-child': + list($aVal, $bVal) = $this->parseAnB($value); + $this->nthChild($aVal, $bVal); + break; + case 'nth-last-child': + list($aVal, $bVal) = $this->parseAnB($value); + $this->nthLastChild($aVal, $bVal); + break; + case 'nth-of-type': + list($aVal, $bVal) = $this->parseAnB($value); + $this->nthOfTypeChild($aVal, $bVal, FALSE); + break; + case 'nth-last-of-type': + list($aVal, $bVal) = $this->parseAnB($value); + $this->nthLastOfTypeChild($aVal, $bVal); + break; + case 'first-child': + $this->nthChild(0, 1); + break; + case 'last-child': + $this->nthLastChild(0, 1); + break; + case 'first-of-type': + $this->firstOfType(); + break; + case 'last-of-type': + $this->lastOfType(); + break; + case 'only-child': + $this->onlyChild(); + break; + case 'only-of-type': + $this->onlyOfType(); + break; + case 'empty': + $this->emptyElement(); + break; + case 'not': + if (empty($value)) { + throw new ParseException(":not() requires a value."); + } + $this->not($value); + break; + // Additional pseudo-classes defined in jQuery: + case 'lt': + case 'gt': + case 'nth': + case 'eq': + case 'first': + case 'last': + //case 'even': + //case 'odd': + $this->getByPosition($name, $value); + break; + case 'parent': + $matches = $this->candidateList(); + $found = new \SplObjectStorage(); + foreach ($matches as $match) { + if (!empty($match->firstChild)) { + $found->attach($match); + } + } + $this->matches = $found; + break; + + case 'enabled': + case 'disabled': + case 'checked': + $this->attribute($name); + break; + case 'text': + case 'radio': + case 'checkbox': + case 'file': + case 'password': + case 'submit': + case 'image': + case 'reset': + case 'button': + $this->attribute('type', $name); + break; + + case 'header': + $matches = $this->candidateList(); + $found = new \SplObjectStorage(); + foreach ($matches as $item) { + $tag = $item->tagName; + $f = strtolower(substr($tag, 0, 1)); + if ($f == 'h' && strlen($tag) == 2 && ctype_digit(substr($tag, 1, 1))) { + $found->attach($item); + } + } + $this->matches = $found; + break; + case 'has': + $this->has($value); + break; + // Contains == text matches. + // In QP 2.1, this was changed. + case 'contains': + $value = $this->removeQuotes($value); + + $matches = $this->candidateList(); + $found = new \SplObjectStorage(); + foreach ($matches as $item) { + if (strpos($item->textContent, $value) !== FALSE) { + $found->attach($item); + } + } + $this->matches = $found; + break; + + // Since QP 2.1 + case 'contains-exactly': + $value = $this->removeQuotes($value); + + $matches = $this->candidateList(); + $found = new \SplObjectStorage(); + foreach ($matches as $item) { + if ($item->textContent == $value) { + $found->attach($item); + } + } + $this->matches = $found; + break; + default: + throw new ParseException("Unknown Pseudo-Class: " . $name); + } + $this->findAnyElement = FALSE; + } + + /** + * Remove leading and trailing quotes. + */ + private function removeQuotes($str) { + $f = substr($str, 0, 1); + $l = substr($str, -1); + if ($f === $l && ($f == '"' || $f == "'")) { + $str = substr($str, 1, -1); + } + return $str; + } + + /** + * Pseudo-class handler for a variety of jQuery pseudo-classes. + * Handles lt, gt, eq, nth, first, last pseudo-classes. + */ + private function getByPosition($operator, $pos) { + $matches = $this->candidateList(); + $found = new \SplObjectStorage(); + if ($matches->count() == 0) { + return; + } + + switch ($operator) { + case 'nth': + case 'eq': + if ($matches->count() >= $pos) { + //$found[] = $matches[$pos -1]; + foreach ($matches as $match) { + // CSS is 1-based, so we pre-increment. + if ($matches->key() + 1 == $pos) { + $found->attach($match); + break; + } + } + } + break; + case 'first': + if ($matches->count() > 0) { + $matches->rewind(); // This is necessary to init. + $found->attach($matches->current()); + } + break; + case 'last': + if ($matches->count() > 0) { + + // Spin through iterator. + foreach ($matches as $item) {}; + + $found->attach($item); + } + break; + // case 'even': + // for ($i = 1; $i <= count($matches); ++$i) { + // if ($i % 2 == 0) { + // $found[] = $matches[$i]; + // } + // } + // break; + // case 'odd': + // for ($i = 1; $i <= count($matches); ++$i) { + // if ($i % 2 == 0) { + // $found[] = $matches[$i]; + // } + // } + // break; + case 'lt': + $i = 0; + foreach ($matches as $item) { + if (++$i < $pos) { + $found->attach($item); + } + } + break; + case 'gt': + $i = 0; + foreach ($matches as $item) { + if (++$i > $pos) { + $found->attach($item); + } + } + break; + } + + $this->matches = $found; + } + + /** + * Parse an an+b rule for CSS pseudo-classes. + * @param $rule + * Some rule in the an+b format. + * @return + * Array (list($aVal, $bVal)) of the two values. + * @throws ParseException + * If the rule does not follow conventions. + */ + protected function parseAnB($rule) { + if ($rule == 'even') { + return array(2, 0); + } + elseif ($rule == 'odd') { + return array(2, 1); + } + elseif ($rule == 'n') { + return array(1, 0); + } + elseif (is_numeric($rule)) { + return array(0, (int)$rule); + } + + $rule = explode('n', $rule); + if (count($rule) == 0) { + throw new ParseException("nth-child value is invalid."); + } + + // Each of these is legal: 1, -1, and -. '-' is shorthand for -1. + $aVal = trim($rule[0]); + $aVal = ($aVal == '-') ? -1 : (int)$aVal; + + $bVal = !empty($rule[1]) ? (int)trim($rule[1]) : 0; + return array($aVal, $bVal); + } + + /** + * Pseudo-class handler for nth-child and all related pseudo-classes. + * + * @param int $groupSize + * The size of the group (in an+b, this is a). + * @param int $elementInGroup + * The offset in a group. (in an+b this is b). + * @param boolean $lastChild + * Whether counting should begin with the last child. By default, this is false. + * Pseudo-classes that start with the last-child can set this to true. + */ + protected function nthChild($groupSize, $elementInGroup, $lastChild = FALSE) { + // EXPERIMENTAL: New in Quark. This should be substantially faster + // than the old (jQuery-ish) version. It still has E_STRICT violations + // though. + $parents = new \SplObjectStorage(); + $matches = new \SplObjectStorage(); + + $i = 0; + foreach ($this->matches as $item) { + $parent = $item->parentNode; + + // Build up an array of all of children of this parent, and store the + // index of each element for reference later. We only need to do this + // once per parent, though. + if (!$parents->contains($parent)) { + + $c = 0; + foreach ($parent->childNodes as $child) { + // We only want nodes, and if this call is preceded by an element + // selector, we only want to match elements with the same tag name. + // !!! This last part is a grey area in the CSS 3 Selector spec. It seems + // necessary to make the implementation match the examples in the spec. However, + // jQuery 1.2 does not do this. + if ($child->nodeType == XML_ELEMENT_NODE && ($this->findAnyElement || $child->tagName == $item->tagName)) { + // This may break E_STRICT. + $child->nodeIndex = ++$c; + } + } + // This may break E_STRICT. + $parent->numElements = $c; + $parents->attach($parent); + } + + // If we are looking for the last child, we count from the end of a list. + // Note that we add 1 because CSS indices begin at 1, not 0. + if ($lastChild) { + $indexToMatch = $item->parentNode->numElements - $item->nodeIndex + 1; + } + // Otherwise we count from the beginning of the list. + else { + $indexToMatch = $item->nodeIndex; + } + + // If group size is 0, then we return element at the right index. + if ($groupSize == 0) { + if ($indexToMatch == $elementInGroup) + $matches->attach($item); + } + // If group size != 0, then we grab nth element from group offset by + // element in group. + else { + if (($indexToMatch - $elementInGroup) % $groupSize == 0 + && ($indexToMatch - $elementInGroup) / $groupSize >= 0) { + $matches->attach($item); + } + } + + // Iterate. + ++$i; + } + $this->matches = $matches; + } + + /** + * Reverse a set of matches. + * + * This is now necessary because internal matches are no longer represented + * as arrays. + * @since QueryPath 2.0 + *//* + private function reverseMatches() { + // Reverse the candidate list. There must be a better way of doing + // this. + $arr = array(); + foreach ($this->matches as $m) array_unshift($arr, $m); + + $this->found = new \SplObjectStorage(); + foreach ($arr as $item) $this->found->attach($item); + }*/ + + /** + * Pseudo-class handler for :nth-last-child and related pseudo-classes. + */ + protected function nthLastChild($groupSize, $elementInGroup) { + // New in Quark. + $this->nthChild($groupSize, $elementInGroup, TRUE); + } + + /** + * Get a list of peer elements. + * If $requireSameTag is TRUE, then only peer elements with the same + * tagname as the given element will be returned. + * + * @param $element + * A DomElement. + * @param $requireSameTag + * Boolean flag indicating whether all matches should have the same + * element name (tagName) as $element. + * @return + * Array of peer elements. + *//* + protected function listPeerElements($element, $requireSameTag = FALSE) { + $peers = array(); + $parent = $element->parentNode; + foreach ($parent->childNodes as $node) { + if ($node->nodeType == XML_ELEMENT_NODE) { + if ($requireSameTag) { + // Need to make sure that the tag matches: + if ($element->tagName == $node->tagName) { + $peers[] = $node; + } + } + else { + $peers[] = $node; + } + } + } + return $peers; + } + */ + /** + * Get the nth child (by index) from matching candidates. + * + * This is used by pseudo-class handlers. + */ + /* + protected function childAtIndex($index, $tagName = NULL) { + $restrictToElement = !$this->findAnyElement; + $matches = $this->candidateList(); + $defaultTagName = $tagName; + + // XXX: Added in Quark: I believe this should return an empty + // match set if no child was found tat the index. + $this->matches = new \SplObjectStorage(); + + foreach ($matches as $item) { + $parent = $item->parentNode; + + // If a default tag name is supplied, we always use it. + if (!empty($defaultTagName)) { + $tagName = $defaultTagName; + } + // If we are inside of an element selector, we use the + // tag name of the given elements. + elseif ($restrictToElement) { + $tagName = $item->tagName; + } + // Otherwise, we skip the tag name match. + else { + $tagName = NULL; + } + + // Loop through all children looking for matches. + $i = 0; + foreach ($parent->childNodes as $child) { + if ($child->nodeType !== XML_ELEMENT_NODE) { + break; // Skip non-elements + } + + // If type is set, then we do type comparison + if (!empty($tagName)) { + // Check whether tag name matches the type. + if ($child->tagName == $tagName) { + // See if this is the index we are looking for. + if ($i == $index) { + //$this->matches = new \SplObjectStorage(); + $this->matches->attach($child); + return; + } + // If it's not the one we are looking for, increment. + ++$i; + } + } + // We don't care about type. Any tagName will match. + else { + if ($i == $index) { + $this->matches->attach($child); + return; + } + ++$i; + } + } // End foreach + } + + }*/ + + /** + * Pseudo-class handler for nth-of-type-child. + * Not implemented. + */ + protected function nthOfTypeChild($groupSize, $elementInGroup, $lastChild) { + // EXPERIMENTAL: New in Quark. This should be substantially faster + // than the old (jQuery-ish) version. It still has E_STRICT violations + // though. + $parents = new \SplObjectStorage(); + $matches = new \SplObjectStorage(); + + $i = 0; + foreach ($this->matches as $item) { + $parent = $item->parentNode; + + // Build up an array of all of children of this parent, and store the + // index of each element for reference later. We only need to do this + // once per parent, though. + if (!$parents->contains($parent)) { + + $c = 0; + foreach ($parent->childNodes as $child) { + // This doesn't totally make sense, since the CSS 3 spec does not require that + // this pseudo-class be adjoined to an element (e.g. ' :nth-of-type' is allowed). + if ($child->nodeType == XML_ELEMENT_NODE && $child->tagName == $item->tagName) { + // This may break E_STRICT. + $child->nodeIndex = ++$c; + } + } + // This may break E_STRICT. + $parent->numElements = $c; + $parents->attach($parent); + } + + // If we are looking for the last child, we count from the end of a list. + // Note that we add 1 because CSS indices begin at 1, not 0. + if ($lastChild) { + $indexToMatch = $item->parentNode->numElements - $item->nodeIndex + 1; + } + // Otherwise we count from the beginning of the list. + else { + $indexToMatch = $item->nodeIndex; + } + + // If group size is 0, then we return element at the right index. + if ($groupSize == 0) { + if ($indexToMatch == $elementInGroup) + $matches->attach($item); + } + // If group size != 0, then we grab nth element from group offset by + // element in group. + else { + if (($indexToMatch - $elementInGroup) % $groupSize == 0 + && ($indexToMatch - $elementInGroup) / $groupSize >= 0) { + $matches->attach($item); + } + } + + // Iterate. + ++$i; + } + $this->matches = $matches; + } + + /** + * Pseudo-class handler for nth-last-of-type-child. + * Not implemented. + */ + protected function nthLastOfTypeChild($groupSize, $elementInGroup) { + $this->nthOfTypeChild($groupSize, $elementInGroup, TRUE); + } + + /** + * Pseudo-class handler for :lang + */ + protected function lang($value) { + // TODO: This checks for cases where an explicit language is + // set. The spec seems to indicate that an element should inherit + // language from the parent... but this is unclear. + $operator = (strpos($value, '-') !== FALSE) ? self::isExactly : self::containsWithHyphen; + + $orig = $this->matches; + $origDepth = $this->findAnyElement; + + // Do first pass: attributes in default namespace + $this->attribute('lang', $value, $operator); + $lang = $this->matches; // Temp array for merging. + + // Reset + $this->matches = $orig; + $this->findAnyElement = $origDepth; + + // Do second pass: attributes in 'xml' namespace. + $this->attributeNS('lang', 'xml', $value, $operator); + + + // Merge results. + // FIXME: Note that we lose natural ordering in + // the document because we search for xml:lang separately + // from lang. + foreach ($this->matches as $added) $lang->attach($added); + $this->matches = $lang; + } + + /** + * Pseudo-class handler for :not(filter). + * + * This does not follow the specification in the following way: The CSS 3 + * selector spec says the value of not() must be a simple selector. This + * function allows complex selectors. + * + * @param string $filter + * A CSS selector. + */ + protected function not($filter) { + $matches = $this->candidateList(); + //$found = array(); + $found = new \SplObjectStorage(); + foreach ($matches as $item) { + $handler = new QueryPathEventHandler($item); + $not_these = $handler->find($filter)->getMatches(); + if ($not_these->count() == 0) { + $found->attach($item); + } + } + // No need to check for unique elements, since the list + // we began from already had no duplicates. + $this->matches = $found; + } + + /** + * Pseudo-class handler for :has(filter). + * This can also be used as a general filtering routine. + */ + public function has($filter) { + $matches = $this->candidateList(); + //$found = array(); + $found = new \SplObjectStorage(); + foreach ($matches as $item) { + $handler = new QueryPathEventHandler($item); + $these = $handler->find($filter)->getMatches(); + if (count($these) > 0) { + $found->attach($item); + } + } + $this->matches = $found; + return $this; + } + + /** + * Pseudo-class handler for :first-of-type. + */ + protected function firstOfType() { + $matches = $this->candidateList(); + $found = new \SplObjectStorage(); + foreach ($matches as $item) { + $type = $item->tagName; + $parent = $item->parentNode; + foreach ($parent->childNodes as $kid) { + if ($kid->nodeType == XML_ELEMENT_NODE && $kid->tagName == $type) { + if (!$found->contains($kid)) { + $found->attach($kid); + } + break; + } + } + } + $this->matches = $found; + } + + /** + * Pseudo-class handler for :last-of-type. + */ + protected function lastOfType() { + $matches = $this->candidateList(); + $found = new \SplObjectStorage(); + foreach ($matches as $item) { + $type = $item->tagName; + $parent = $item->parentNode; + for ($i = $parent->childNodes->length - 1; $i >= 0; --$i) { + $kid = $parent->childNodes->item($i); + if ($kid->nodeType == XML_ELEMENT_NODE && $kid->tagName == $type) { + if (!$found->contains($kid)) { + $found->attach($kid); + } + break; + } + } + } + $this->matches = $found; + } + + /** + * Pseudo-class handler for :only-child. + */ + protected function onlyChild() { + $matches = $this->candidateList(); + $found = new \SplObjectStorage(); + foreach($matches as $item) { + $parent = $item->parentNode; + $kids = array(); + foreach($parent->childNodes as $kid) { + if ($kid->nodeType == XML_ELEMENT_NODE) { + $kids[] = $kid; + } + } + // There should be only one child element, and + // it should be the one being tested. + if (count($kids) == 1 && $kids[0] === $item) { + $found->attach($kids[0]); + } + } + $this->matches = $found; + } + + /** + * Pseudo-class handler for :empty. + */ + protected function emptyElement() { + $found = new \SplObjectStorage(); + $matches = $this->candidateList(); + foreach ($matches as $item) { + $empty = TRUE; + foreach($item->childNodes as $kid) { + // From the spec: Elements and Text nodes are the only ones to + // affect emptiness. + if ($kid->nodeType == XML_ELEMENT_NODE || $kid->nodeType == XML_TEXT_NODE) { + $empty = FALSE; + break; + } + } + if ($empty) { + $found->attach($item); + } + } + $this->matches = $found; + } + + /** + * Pseudo-class handler for :only-of-type. + */ + protected function onlyOfType() { + $matches = $this->candidateList(); + $found = new \SplObjectStorage(); + foreach ($matches as $item) { + if (!$item->parentNode) { + $this->matches = new \SplObjectStorage(); + } + $parent = $item->parentNode; + $onlyOfType = TRUE; + + // See if any peers are of the same type + foreach($parent->childNodes as $kid) { + if ($kid->nodeType == XML_ELEMENT_NODE + && $kid->tagName == $item->tagName + && $kid !== $item) { + //$this->matches = new \SplObjectStorage(); + $onlyOfType = FALSE; + break; + } + } + + // If no others were found, attach this one. + if ($onlyOfType) $found->attach($item); + } + $this->matches = $found; + } + + /** + * Check for attr value matches based on an operation. + */ + protected function attrValMatches($needle, $haystack, $operation) { + + if (strlen($haystack) < strlen($needle)) return FALSE; + + // According to the spec: + // "The case-sensitivity of attribute names in selectors depends on the document language." + // (6.3.2) + // To which I say, "huh?". We assume case sensitivity. + switch ($operation) { + case EventHandler::isExactly: + return $needle == $haystack; + case EventHandler::containsWithSpace: + return in_array($needle, explode(' ', $haystack)); + case EventHandler::containsWithHyphen: + return in_array($needle, explode('-', $haystack)); + case EventHandler::containsInString: + return strpos($haystack, $needle) !== FALSE; + case EventHandler::beginsWith: + return strpos($haystack, $needle) === 0; + case EventHandler::endsWith: + //return strrpos($haystack, $needle) === strlen($needle) - 1; + return preg_match('/' . $needle . '$/', $haystack) == 1; + } + return FALSE; // Shouldn't be able to get here. + } + + /** + * As the spec mentions, these must be at the end of a selector or + * else they will cause errors. Most selectors return elements. Pseudo-elements + * do not. + */ + public function pseudoElement($name) { + // process the pseudoElement + switch ($name) { + // XXX: Should this return an array -- first line of + // each of the matched elements? + case 'first-line': + $matches = $this->candidateList(); + $found = new \SplObjectStorage(); + $o = new \stdClass(); + foreach ($matches as $item) { + $str = $item->textContent; + $lines = explode("\n", $str); + if (!empty($lines)) { + $line = trim($lines[0]); + if (!empty($line)) { + $o->textContent = $line; + $found->attach($o);//trim($lines[0]); + } + } + } + $this->matches = $found; + break; + // XXX: Should this return an array -- first letter of each + // of the matched elements? + case 'first-letter': + $matches = $this->candidateList(); + $found = new \SplObjectStorage(); + $o = new \stdClass(); + foreach ($matches as $item) { + $str = $item->textContent; + if (!empty($str)) { + $str = substr($str,0, 1); + $o->textContent = $str; + $found->attach($o); + } + } + $this->matches = $found; + break; + case 'before': + case 'after': + // There is nothing in a DOM to return for the before and after + // selectors. + case 'selection': + // With no user agent, we don't have a concept of user selection. + throw new NotImplementedException("The $name pseudo-element is not implemented."); + break; + } + $this->findAnyElement = FALSE; + } + public function directDescendant() { + $this->findAnyElement = FALSE; + + $kids = new \SplObjectStorage(); + foreach ($this->matches as $item) { + $kidsNL = $item->childNodes; + foreach ($kidsNL as $kidNode) { + if ($kidNode->nodeType == XML_ELEMENT_NODE) { + $kids->attach($kidNode); + } + } + } + $this->matches = $kids; + } + /** + * For an element to be adjacent to another, it must be THE NEXT NODE + * in the node list. So if an element is surrounded by pcdata, there are + * no adjacent nodes. E.g. in <a/>FOO<b/>, the a and b elements are not + * adjacent. + * + * In a strict DOM parser, line breaks and empty spaces are nodes. That means + * nodes like this will not be adjacent: <test/> <test/>. The space between + * them makes them non-adjacent. If this is not the desired behavior, pass + * in the appropriate flags to your parser. Example: + * <code> + * $doc = new DomDocument(); + * $doc->loadXML('<test/> <test/>', LIBXML_NOBLANKS); + * </code> + */ + public function adjacent() { + $this->findAnyElement = FALSE; + // List of nodes that are immediately adjacent to the current one. + //$found = array(); + $found = new \SplObjectStorage(); + foreach ($this->matches as $item) { + while (isset($item->nextSibling)) { + if (isset($item->nextSibling) && $item->nextSibling->nodeType === XML_ELEMENT_NODE) { + $found->attach($item->nextSibling); + break; + } + $item = $item->nextSibling; + } + } + $this->matches = $found; + } + + public function anotherSelector() { + $this->findAnyElement = FALSE; + // Copy old matches into buffer. + if ($this->matches->count() > 0) { + //$this->alreadyMatched = array_merge($this->alreadyMatched, $this->matches); + foreach ($this->matches as $item) $this->alreadyMatched->attach($item); + } + + // Start over at the top of the tree. + $this->findAnyElement = TRUE; // Reset depth flag. + $this->matches = new \SplObjectStorage(); + $this->matches->attach($this->dom); + } + + /** + * Get all nodes that are siblings to currently selected nodes. + * + * If two passed in items are siblings of each other, neither will + * be included in the list of siblings. Their status as being candidates + * excludes them from being considered siblings. + */ + public function sibling() { + $this->findAnyElement = FALSE; + // Get the nodes at the same level. + + if ($this->matches->count() > 0) { + $sibs = new \SplObjectStorage(); + foreach ($this->matches as $item) { + /*$candidates = $item->parentNode->childNodes; + foreach ($candidates as $candidate) { + if ($candidate->nodeType === XML_ELEMENT_NODE && $candidate !== $item) { + $sibs->attach($candidate); + } + } + */ + while ($item->nextSibling != NULL) { + $item = $item->nextSibling; + if ($item->nodeType === XML_ELEMENT_NODE) $sibs->attach($item); + } + } + $this->matches = $sibs; + } + } + + /** + * Get any descendant. + */ + public function anyDescendant() { + // Get children: + $found = new \SplObjectStorage(); + foreach ($this->matches as $item) { + $kids = $item->getElementsByTagName('*'); + //$found = array_merge($found, $this->nodeListToArray($kids)); + $this->attachNodeList($kids, $found); + } + $this->matches = $found; + + // Set depth flag: + $this->findAnyElement = TRUE; + } + + /** + * Determine what candidates are in the current scope. + * + * This is a utility method that gets the list of elements + * that should be evaluated in the context. If $this->findAnyElement + * is TRUE, this will return a list of every element that appears in + * the subtree of $this->matches. Otherwise, it will just return + * $this->matches. + */ + private function candidateList() { + if ($this->findAnyElement) { + return $this->getAllCandidates($this->matches); + } + return $this->matches; + } + + /** + * Get a list of all of the candidate elements. + * + * This is used when $this->findAnyElement is TRUE. + * @param $elements + * A list of current elements (usually $this->matches). + * + * @return + * A list of all candidate elements. + */ + private function getAllCandidates($elements) { + $found = new \SplObjectStorage(); + foreach ($elements as $item) { + $found->attach($item); // put self in + $nl = $item->getElementsByTagName('*'); + //foreach ($nl as $node) $found[] = $node; + $this->attachNodeList($nl, $found); + } + return $found; + } + /* + public function nodeListToArray($nodeList) { + $array = array(); + foreach ($nodeList as $node) { + if ($node->nodeType == XML_ELEMENT_NODE) { + $array[] = $node; + } + } + return $array; + } + */ + + /** + * Attach all nodes in a node list to the given \SplObjectStorage. + */ + public function attachNodeList(\DOMNodeList $nodeList, \SplObjectStorage $splos) { + foreach ($nodeList as $item) $splos->attach($item); + } + +} diff --git a/lib/querypath/src/QueryPath/CSS/Scanner.php b/lib/querypath/src/QueryPath/CSS/Scanner.php new file mode 100644 index 0000000..3513a0b --- /dev/null +++ b/lib/querypath/src/QueryPath/CSS/Scanner.php @@ -0,0 +1,306 @@ +<?php +/** @file + * The scanner. + */ +namespace QueryPath\CSS; +/** + * Scanner for CSS selector parsing. + * + * This provides a simple scanner for traversing an input stream. + * + * @ingroup querypath_css + */ +final class Scanner { + var $is = NULL; + public $value = NULL; + public $token = NULL; + + var $recurse = FALSE; + var $it = 0; + + /** + * Given a new input stream, tokenize the CSS selector string. + * @see InputStream + * @param InputStream $in + * An input stream to be scanned. + */ + public function __construct(InputStream $in) { + $this->is = $in; + } + + /** + * Return the position of the reader in the string. + */ + public function position() { + return $this->is->position; + } + + /** + * See the next char without removing it from the stack. + * + * @return char + * Returns the next character on the stack. + */ + public function peek() { + return $this->is->peek(); + } + + /** + * Get the next token in the input stream. + * + * This sets the current token to the value of the next token in + * the stream. + * + * @return int + * Returns an int value corresponding to one of the Token constants, + * or FALSE if the end of the string is reached. (Remember to use + * strong equality checking on FALSE, since 0 is a valid token id.) + */ + public function nextToken() { + $tok = -1; + ++$this->it; + if ($this->is->isEmpty()) { + if ($this->recurse) { + throw new \QueryPath\Exception("Recursion error detected at iteration " . $this->it . '.'); + exit(); + } + //print "{$this->it}: All done\n"; + $this->recurse = TRUE; + $this->token = FALSE; + return FALSE; + } + $ch = $this->is->consume(); + //print __FUNCTION__ . " Testing $ch.\n"; + if (ctype_space($ch)) { + $this->value = ' '; // Collapse all WS to a space. + $this->token = $tok = Token::white; + //$ch = $this->is->consume(); + return $tok; + } + + if (ctype_alnum($ch) || $ch == '-' || $ch == '_') { + // It's a character + $this->value = $ch; //strtolower($ch); + $this->token = $tok = Token::char; + return $tok; + } + + $this->value = $ch; + + switch($ch) { + case '*': + $tok = Token::star; + break; + case chr(ord('>')): + $tok = Token::rangle; + break; + case '.': + $tok = Token::dot; + break; + case '#': + $tok = Token::octo; + break; + case '[': + $tok = Token::lsquare; + break; + case ']': + $tok = Token::rsquare; + break; + case ':': + $tok = Token::colon; + break; + case '(': + $tok = Token::lparen; + break; + case ')': + $tok = Token::rparen; + break; + case '+': + $tok = Token::plus; + break; + case '~': + $tok = Token::tilde; + break; + case '=': + $tok = Token::eq; + break; + case '|': + $tok = Token::pipe; + break; + case ',': + $tok = Token::comma; + break; + case chr(34): + $tok = Token::quote; + break; + case "'": + $tok = Token::squote; + break; + case '\\': + $tok = Token::bslash; + break; + case '^': + $tok = Token::carat; + break; + case '$': + $tok = Token::dollar; + break; + case '@': + $tok = Token::at; + break; + } + + + // Catch all characters that are legal within strings. + if ($tok == -1) { + // TODO: This should be UTF-8 compatible, but PHP doesn't + // have a native UTF-8 string. Should we use external + // mbstring library? + + $ord = ord($ch); + // Characters in this pool are legal for use inside of + // certain strings. Extended ASCII is used here, though I + // Don't know if these are really legal. + if (($ord >= 32 && $ord <= 126) || ($ord >= 128 && $ord <= 255)) { + $tok = Token::stringLegal; + } + else { + throw new ParseException('Illegal character found in stream: ' . $ord); + } + } + + $this->token = $tok; + return $tok; + } + + /** + * Get a name string from the input stream. + * A name string must be composed of + * only characters defined in Token:char: -_a-zA-Z0-9 + */ + public function getNameString() { + $buf = ''; + while ($this->token === Token::char) { + $buf .= $this->value; + $this->nextToken(); + //print '_'; + } + return $buf; + } + + /** + * This gets a string with any legal 'string' characters. + * See CSS Selectors specification, section 11, for the + * definition of string. + * + * This will check for string1, string2, and the case where a + * string is unquoted (Oddly absent from the "official" grammar, + * though such strings are present as examples in the spec.) + * + * Note: + * Though the grammar supplied by CSS 3 Selectors section 11 does not + * address the contents of a pseudo-class value, the spec itself indicates + * that a pseudo-class value is a "value between parenthesis" [6.6]. The + * examples given use URLs among other things, making them closer to the + * definition of 'string' than to 'name'. So we handle them here as strings. + */ + public function getQuotedString() { + if ($this->token == Token::quote || $this->token == Token::squote || $this->token == Token::lparen) { + $end = ($this->token == Token::lparen) ? Token::rparen : $this->token; + $buf = ''; + $escape = FALSE; + + $this->nextToken(); // Skip the opening quote/paren + + // The second conjunct is probably not necessary. + while ($this->token !== FALSE && $this->token > -1) { + //print "Char: $this->value \n"; + if ($this->token == Token::bslash && !$escape) { + // XXX: The backslash (\) is removed here. + // Turn on escaping. + //$buf .= $this->value; + $escape = TRUE; + } + elseif ($escape) { + // Turn off escaping + $buf .= $this->value; + $escape = FALSE; + } + elseif ($this->token === $end) { + // At end of string; skip token and break. + $this->nextToken(); + break; + } + else { + // Append char. + $buf .= $this->value; + } + $this->nextToken(); + } + return $buf; + } + } + + // Get the contents inside of a pseudoClass(). + public function getPseudoClassString() { + if ($this->token == Token::quote || $this->token == Token::squote || $this->token == Token::lparen) { + $end = ($this->token == Token::lparen) ? Token::rparen : $this->token; + $buf = ''; + $escape = FALSE; + + $this->nextToken(); // Skip the opening quote/paren + + // The second conjunct is probably not necessary. + while ($this->token !== FALSE && $this->token > -1) { + //print "Char: $this->value \n"; + if ($this->token == Token::bslash && !$escape) { + // XXX: The backslash (\) is removed here. + // Turn on escaping. + //$buf .= $this->value; + $escape = TRUE; + } + elseif ($escape) { + // Turn off escaping + $buf .= $this->value; + $escape = FALSE; + } + // Allow nested pseudoclasses. + elseif ($this->token == Token::lparen) { + $buf .= "("; + $buf .= $this->getPseudoClassString(); + $buf .= ")"; + } + elseif ($this->token === $end) { + // At end of string; skip token and break. + $this->nextToken(); + break; + } + else { + // Append char. + $buf .= $this->value; + } + $this->nextToken(); + } + return $buf; + } + } + + /** + * Get a string from the input stream. + * This is a convenience function for getting a string of + * characters that are either alphanumber or whitespace. See + * the Token::white and Token::char definitions. + * + * @deprecated This is not used anywhere in QueryPath. + *//* + public function getStringPlusWhitespace() { + $buf = ''; + if($this->token === FALSE) {return '';} + while ($this->token === Token::char || $this->token == Token::white) { + $buf .= $this->value; + $this->nextToken(); + } + return $buf; + }*/ + +} diff --git a/lib/querypath/src/QueryPath/CSS/Selector.php b/lib/querypath/src/QueryPath/CSS/Selector.php new file mode 100644 index 0000000..4b538bd --- /dev/null +++ b/lib/querypath/src/QueryPath/CSS/Selector.php @@ -0,0 +1,144 @@ +<?php +/** @file + * A selector. + */ + +namespace QueryPath\CSS; + +/** + * A CSS Selector. + * + * A CSS selector is made up of one or more Simple Selectors + * (SimpleSelector). + * + * @attention + * The Selector data structure is a LIFO (Last in, First out). This is + * because CSS selectors are best processed "bottom up". Thus, when + * iterating over 'a>b>c', the iterator will produce: + * - c + * - b + * - a + * It is assumed, therefore, that any suitable querying engine will + * traverse from the bottom (`c`) back up. + * + * @b Usage + * + * This class is an event handler. It can be plugged into an Parser and + * receive the events the Parser generates. + * + * This class is also an iterator. Once the parser has completed, the + * captured selectors can be iterated over. + * + * @code + * <?php + * $selectorList = new \QueryPath\CSS\Selector(); + * $parser = new \QueryPath\CSS\Parser($selector, $selectorList); + * + * $parser->parse(); + * + * foreach ($selectorList as $simpleSelector) { + * // Do something with the SimpleSelector. + * print_r($simpleSelector); + * } + * ?> + * @endode + * + * + * @since QueryPath 3.0.0 + */ +class Selector implements EventHandler, \IteratorAggregate, \Countable { + protected $selectors = array(); + protected $currSelector; + protected $selectorGroups = array(); + protected $groupIndex = 0; + + public function __construct() { + $this->currSelector = new SimpleSelector(); + + $this->selectors[$this->groupIndex][] = $this->currSelector; + } + + public function getIterator() { + return new \ArrayIterator($this->selectors); + } + + /** + * Get the array of SimpleSelector objects. + * + * Normally, one iterates over a Selector. However, if it is + * necessary to get the selector array and manipulate it, this + * method can be used. + */ + public function toArray() { + return $this->selectors; + } + + public function count() { + return count($this->selectors); + } + + public function elementID($id) { + $this->currSelector->id = $id; + } + public function element($name) { + $this->currSelector->element = $name; + } + public function elementNS($name, $namespace = NULL) { + $this->currSelector->ns = $namespace; + $this->currSelector->element = $name; + } + public function anyElement() { + $this->currSelector->element = '*'; + } + public function anyElementInNS($ns) { + $this->currSelector->ns = $ns; + $this->currSelector->element = '*'; + } + public function elementClass($name) { + $this->currSelector->classes[] = $name; + } + public function attribute($name, $value = NULL, $operation = EventHandler::isExactly) { + $this->currSelector->attributes[] = array( + 'name' => $name, + 'value' => $value, + 'op' => $operation, + ); + } + public function attributeNS($name, $ns, $value = NULL, $operation = EventHandler::isExactly) { + $this->currSelector->attributes[] = array( + 'name' => $name, + 'value' => $value, + 'op' => $operation, + 'ns' => $ns, + ); + } + public function pseudoClass($name, $value = NULL) { + $this->currSelector->pseudoClasses[] = array('name' => $name, 'value' => $value); + } + public function pseudoElement($name) { + $this->currSelector->pseudoElements[] = $name; + } + public function combinator($combinatorName) { + $this->currSelector->combinator = $combinatorName; + $this->currSelector = new SimpleSelector(); + array_unshift($this->selectors[$this->groupIndex], $this->currSelector); + //$this->selectors[]= $this->currSelector; + } + public function directDescendant() { + $this->combinator(SimpleSelector::directDescendant); + } + public function adjacent() { + $this->combinator(SimpleSelector::adjacent); + } + public function anotherSelector() { + $this->groupIndex++; + $this->currSelector = new SimpleSelector(); + $this->selectors[$this->groupIndex] = array($this->currSelector); + } + public function sibling() { + $this->combinator(SimpleSelector::sibling); + } + public function anyDescendant() { + $this->combinator(SimpleSelector::anyDescendant); + } +} diff --git a/lib/querypath/src/QueryPath/CSS/SimpleSelector.php b/lib/querypath/src/QueryPath/CSS/SimpleSelector.php new file mode 100644 index 0000000..3fcc796 --- /dev/null +++ b/lib/querypath/src/QueryPath/CSS/SimpleSelector.php @@ -0,0 +1,138 @@ +<?php +/** @file + * + * A simple selector. + * + */ + +namespace QueryPath\CSS; + +/** + * Models a simple selector. + * + * CSS Selectors are composed of one or more simple selectors, where + * each simple selector may have any of the following components: + * + * - An element name (or wildcard *) + * - An ID (#foo) + * - One or more classes (.foo.bar) + * - One or more attribute matchers ([foo=bar]) + * - One or more pseudo-classes (:foo) + * - One or more pseudo-elements (::first) + * + * For performance reasons, this object has been kept as sparse as + * possible. + * + * @since QueryPath 3.x + * @author M Butcher + * + */ +class SimpleSelector { + + const adjacent = 1; + const directDescendant = 2; + const anotherSelector = 4; + const sibling = 8; + const anyDescendant = 16; + + public $element; + public $ns; + public $id; + public $classes = array(); + public $attributes = array(); + public $pseudoClasses = array(); + public $pseudoElements = array(); + public $combinator; + + public static function attributeOperator($code) { + switch($code) { + case EventHandler::containsWithSpace: + return '~='; + case EventHandler::containsWithHyphen: + return '|='; + case EventHandler::containsInString: + return '*='; + case EventHandler::beginsWith: + return '^='; + case EventHandler::endsWith: + return '$='; + default: + return '='; + } + } + + public static function combinatorOperator($code) { + switch ($code) { + case self::adjacent: + return '+'; + case self::directDescendant: + return '>'; + case self::sibling: + return '~'; + case self::anotherSelector: + return ', '; + case self::anyDescendant: + return ' '; + } + } + + public function __construct() { + } + + public function notEmpty() { + return !empty($element) + && !empty($id) + && !empty($classes) + && !empty($combinator) + && !empty($attributes) + && !empty($pseudoClasses) + && !empty($pseudoElements) + ; + } + + public function __tostring() { + $buffer = array(); + try { + + if (!empty($this->ns)) { + $buffer[] = $this->ns; $buffer[] = '|'; + } + if (!empty($this->element)) $buffer[] = $this->element; + if (!empty($this->id)) $buffer[] = '#' . $this->id; + if (!empty($this->attributes)) { + foreach ($this->attributes as $attr) { + $buffer[] = '['; + if(!empty($attr['ns'])) $buffer[] = $attr['ns'] . '|'; + $buffer[] = $attr['name']; + if (!empty($attr['value'])) { + $buffer[] = self::attributeOperator($attr['op']); + $buffer[] = $attr['value']; + } + $buffer[] = ']'; + } + } + if (!empty($this->pseudoClasses)) { + foreach ($this->pseudoClasses as $ps) { + $buffer[] = ':' . $ps['name']; + if (isset($ps['value'])) { + $buffer[] = '(' . $ps['value'] . ')'; + } + } + } + foreach ($this->pseudoElements as $pe) { + $buffer[] = '::' . $pe; + } + + if (!empty($this->combinator)) { + $buffer[] = self::combinatorOperator($this->combinator); + } + + } + catch (\Exception $e) { + return $e->getMessage(); + } + + return implode('', $buffer); + } + +} diff --git a/lib/querypath/src/QueryPath/CSS/Token.php b/lib/querypath/src/QueryPath/CSS/Token.php new file mode 100644 index 0000000..3c31ef4 --- /dev/null +++ b/lib/querypath/src/QueryPath/CSS/Token.php @@ -0,0 +1,60 @@ +<?php +/** @file + * Parser tokens. + */ +namespace QueryPath\CSS; +/** + * Tokens for CSS. + * This class defines the recognized tokens for the parser, and also + * provides utility functions for error reporting. + * + * @ingroup querypath_css + */ +final class Token { + const char = 0; + const star = 1; + const rangle = 2; + const dot = 3; + const octo = 4; + const rsquare = 5; + const lsquare = 6; + const colon = 7; + const rparen = 8; + const lparen = 9; + const plus = 10; + const tilde = 11; + const eq = 12; + const pipe = 13; + const comma = 14; + const white = 15; + const quote = 16; + const squote = 17; + const bslash = 18; + const carat = 19; + const dollar = 20; + const at = 21; // This is not in the spec. Apparently, old broken CSS uses it. + + // In legal range for string. + const stringLegal = 99; + + /** + * Get a name for a given constant. Used for error handling. + */ + static function name($const_int) { + $a = array('character', 'star', 'right angle bracket', + 'dot', 'octothorp', 'right square bracket', 'left square bracket', + 'colon', 'right parenthesis', 'left parenthesis', 'plus', 'tilde', + 'equals', 'vertical bar', 'comma', 'space', 'quote', 'single quote', + 'backslash', 'carat', 'dollar', 'at'); + if (isset($a[$const_int]) && is_numeric($const_int)) { + return $a[$const_int]; + } + elseif ($const_int == 99) { + return 'a legal non-alphanumeric character'; + } + elseif ($const_int == FALSE) { + return 'end of file'; + } + return sprintf('illegal character (%s)', $const_int); + } +} diff --git a/lib/querypath/src/QueryPath/CSS/Traverser.php b/lib/querypath/src/QueryPath/CSS/Traverser.php new file mode 100644 index 0000000..6d16d63 --- /dev/null +++ b/lib/querypath/src/QueryPath/CSS/Traverser.php @@ -0,0 +1,38 @@ +<?php +/** + * @file + * The main Traverser interface. + */ + +namespace QueryPath\CSS; + +/** + * An object capable of walking (and searching) a datastructure. + */ +interface Traverser { + /** + * Process a CSS selector and find matches. + * + * This specifies a query to be run by the Traverser. A given + * Traverser may, in practice, delay the finding until some later time + * but must return the found results when getMatches() is called. + * + * @param string $selector + * A selector. Typically this is a CSS 3 Selector. + * @return \Traverser + * The Traverser that can return matches. + */ + public function find($selector); + /** + * Get the results of a find() operation. + * + * Return an array of matching items. + * + * @return array + * An array of matched values. The specific data type in the matches + * will differ depending on the data type searched, but in the core + * QueryPath implementation, this will be an array of DOMNode + * objects. + */ + public function matches(); +} |