/**
* Create a namespace optimizer for the provided document. The provided
* document will be copied.
*
* The second argument allows to provide namespaces and prefixes. The
* keys of the array are the namespace uri, the values are the prefixes.
*
* If a namespace is not provided it is read from the source node.
*
* You can use the same prefix for multiple namespace uris. Empty prefixes
* are possible (default namespace for an element).
*
* It is highly recommend that you always use a non-empty prefix if the
* here are attributes in that namespace. Attributes always need a prefix
* to make use of the namespace.
*
* @param \DOMDocument $document
* @param array $namespaces
*/
public function __construct(\DOMDocument $document, array $namespaces = [])
{
$this->_document = new Document();
foreach ($document->childNodes as $node) {
$this->_document->appendChild($this->_document->importNode($node, TRUE));
}
$this->_namespaceUris = $namespaces;
}
/**
* @see Loadable::load
* @param \SimpleXMLElement $source
* @param string $contentType
* @param array|\Traversable|Options $options
* @return Document|Result|NULL
*/
public function load($source, $contentType, $options = [])
{
if ($source instanceof \SimpleXMLElement) {
$document = new Document();
$document->appendChild($document->importNode(dom_import_simplexml($source), TRUE));
return new Result($document, 'text/xml');
}
return NULL;
}
/**
* Load the json string into an DOMDocument
*
* @param mixed $source
* @param string $contentType
* @param array|\Traversable|Options $options
* @return Document|Result|NULL
*/
public function load($source, $contentType, $options = [])
{
if (FALSE !== ($json = $this->getJson($source, $contentType))) {
$document = new Document('1.0', 'UTF-8');
$document->appendChild($root = $document->createElementNS(self::XMLNS, 'json:json'));
$this->transferTo($root, $json);
return $document;
}
return NULL;
}
/**
* @see Loadable::loadFragment
*
* @param string $source
* @param string $contentType
* @param array|\Traversable|Options $options
* @return DocumentFragment|NULL
*/
public function loadFragment($source, $contentType, $options = [])
{
if (FALSE !== ($json = $this->getJson($source, $contentType))) {
$document = new Document('1.0', 'UTF-8');
$fragment = $document->createDocumentFragment();
$this->transferTo($fragment, $json);
return $fragment;
}
return NULL;
}
/**
* @see LoadableFragment::loadFragment
* @param string $source
* @param string $contentType
* @param array|\Traversable|Options $options
* @return DocumentFragment|NULL
*/
public function loadFragment($source, $contentType, $options = [])
{
if ($this->supports($contentType)) {
$dom = new Document();
$fragment = $dom->createDocumentFragment();
$fragment->appendXml($source);
return $fragment;
}
return NULL;
}
/**
* @see Loadable::load
* @param mixed $source
* @param string $contentType
* @param array $options
* @return Document|NULL
*/
public function load($source, $contentType, array $options = [])
{
if ($this->supports($contentType) && ($this->_lines = $this->getLines($source))) {
$dom = new Document('1.0', 'UTF-8');
$dom->registerNamespace('', $this->_namespace);
$dom->appendElement($this->_nodeNames['root'])->append($this);
return $dom;
}
return NULL;
}
/**
* @see Loadable::loadFragment
*
* @param string $source
* @param string $contentType
* @param array|\Traversable|Options $options
* @return DocumentFragment|NULL
*/
public function loadFragment($source, $contentType, $options = [])
{
$hasHeaderLine = isset($options['FIELDS']) ? FALSE : isset($options['HEADER']) && $options['HEADER'];
$this->configure($options);
if ($this->supports($contentType) && ($lines = $this->getLines($source))) {
$document = new Document('1.0', 'UTF-8');
$fragment = $document->createDocumentFragment();
$this->appendLines($fragment, $lines, $hasHeaderLine, isset($options['FIELDS']) ? $options['FIELDS'] : NULL);
return $fragment;
}
return NULL;
}
/**
* @see Loadable::load
* @param \PDOStatement $source
* @param string $contentType
* @param array|\Traversable|Options $options
* @return Document|Result|NULL
*/
public function load($source, $contentType, $options = [])
{
if ($source instanceof \PDOStatement) {
$document = new Document('1.0', 'UTF-8');
$document->registerNamespace('json', self::XMLNS);
$root = $document->appendElement('json:json');
$source->setFetchMode(\PDO::FETCH_OBJ);
foreach ($source as $row) {
$child = $root->appendElement('_');
$this->transferTo($child, $row, 1);
}
return new Result($document, 'text/xml');
}
return NULL;
}
/**
* Load the YAML string into an DOMDocument
*
* @param mixed $source
* @param string $contentType
* @param array $options
* @return Document|NULL
*/
public function load($source, $contentType, array $options = [])
{
if ($this->supports($contentType)) {
if (FALSE === strpos($source, "\n")) {
$source = file_get_contents($source);
}
$yaml = Dipper::parse($source);
if (!empty($yaml) || is_array($yaml)) {
$dom = new Document('1.0', 'UTF-8');
$dom->appendChild($root = $dom->createElementNS(self::XMLNS, 'json:json'));
$this->transferTo($root, $yaml);
return $dom;
}
}
return NULL;
}
/**
* @param Document|Element $parent
* @param \DOMElement $node
* @param bool $addNameAttribute
*/
public function addNode($parent, \DOMElement $node, $addNameAttribute = FALSE)
{
switch ($this->getType($node)) {
case 'object':
$result = $parent->appendElement('json:object');
$this->appendChildNodes($result, $node, TRUE);
break;
case 'array':
$result = $parent->appendElement('json:array');
$this->appendChildNodes($result, $node, FALSE);
break;
case 'number':
$result = $parent->appendElement('json:number', $node->nodeValue);
break;
case 'boolean':
$result = $parent->appendElement('json:boolean', $node->nodeValue);
break;
case 'null':
$result = $parent->appendElement('json:null');
break;
default:
$result = $parent->appendElement('json:string', $node->nodeValue);
break;
}
if ($addNameAttribute) {
$name = $node->localName;
if ($node->hasAttributeNS(self::XMLNS_JSONDOM, 'name')) {
$name = $node->getAttributeNS(self::XMLNS_JSONDOM, 'name');
}
$result['name'] = $name;
}
}
/**
* @see Loadable::loadFragment
*
* @param string $source
* @param string $contentType
* @param array|\Traversable|Options $options
* @return DocumentFragment|NULL
*/
public function loadFragment($source, $contentType, $options = [])
{
if ($this->supports($contentType) && !empty($source)) {
$document = new Document();
$document->preserveWhiteSpace = FALSE;
$document->registerNamespace('jx', self::XMLNS_JSONX);
$sourceFragment = $document->createDocumentFragment();
$sourceFragment->appendXml($source);
$target = new Document();
$target->registerNamespace('json', self::XMLNS_JSONDOM);
$targetFragment = $target->createDocumentFragment();
foreach ($sourceFragment->childNodes as $node) {
$this->transferNode($node, $targetFragment);
}
return $targetFragment;
}
return NULL;
}
/**
* Load a HTML5 file and copy it into a FluentDOM\Document
*
* @codeCoverageIgnore
*
* @see Loadable::load
* @param string $source
* @param string $contentType
* @param array $options
* @return Document|NULL
*/
public function load($source, $contentType, array $options = [])
{
if ($this->supports($contentType)) {
$html5 = new HTML5Support();
if ($this->startsWith($source, '<')) {
$dom = $html5->loadHTML($source);
} else {
$dom = $html5->loadHTMLFile($source);
}
if (!$dom instanceof Document) {
$fd = new Document();
if ($dom->documentElement instanceof \DOMElement) {
$fd->appendChild($fd->importNode($dom->documentElement, TRUE));
}
$dom = $fd;
}
$dom->registerNamespace('html', 'http://www.w3.org/1999/xhtml');
return $dom;
}
return NULL;
}
/**
* Given a source (URL), retrieve raw data
* and turn it into an array of arrays of product data.
*
* Each sub array should contain the following keys:
* - title (string) - the title of the product
* - size (int) - bytes of raw data
* - unit_price (float) - the price of the product
* - description - description of the product
*
* @param string $source Information about where to load data from.
*
* @throws InvalidDataSourceException If the source is invalid.
* @throws MalformedDataException If the data is malformed.
*
* @return array
*/
public function getProductData($source)
{
// We will save errors to an array.
libxml_use_internal_errors(true);
if (false === $this->looksLikeUrl($source)) {
throw new InvalidDataSourceException("This isn't a URL.");
}
// Create a DOM parser & load the HTML.
$dom = new DOMParser();
$listHtml = $this->scraper->getHtml($source);
$return = array();
$dom->loadHTML($listHtml);
foreach ($dom->querySelectorAll("div.product") as $product) {
// Get the title element & price element
$titleElement = $product->querySelector("div.productInfo")->querySelector("a");
$priceElement = $product->querySelector("p.pricePerUnit");
// Load the product-specific page.
$productUrl = $titleElement->getAttributeNode("href")->textContent;
$productHtml = $this->scraper->getHtml($productUrl);
$productDom = new DOMParser();
$productDom->loadHTML($productHtml);
// Get the product description.
// Note: this does not have a unique ID, this is very volatile.
$descriptionElement = $productDom->querySelector("div#information")->querySelector("div.productText");
// Set up the product data.
$productData = array();
$productData['title'] = trim($titleElement->textContent);
$productData['description'] = trim($descriptionElement->textContent);
$productData['size'] = strlen($productHtml);
$productData['unit_price'] = (double) preg_replace('/[^0-9.]/', '', $priceElement->textContent);
$return[] = $productData;
}
// Put errors in $this->errors.
$this->errors = libxml_get_errors();
libxml_clear_errors();
return $return;
}
private function loadFragmentIntoDom(\DOMDocument $document, $source, $loadOptions)
{
$htmlDom = new Document();
$htmlDom->loadHTML('<html-fragment>' . $source . '</html-fragment>', $loadOptions);
$nodes = $htmlDom->evaluate('//html-fragment[1]/node()');
foreach ($nodes as $node) {
if ($importedNode = $document->importNode($node, TRUE)) {
$document->appendChild($importedNode);
}
}
}
/**
* @param string $target
* @param string $content
* @return \DOMProcessingInstruction
*/
public function pi($target, $content)
{
return $this->_document->createProcessingInstruction($target, $content);
}
/**
* @param string $source
* @param string $contentType
* @param array|\Traversable|Options $options
* @return \FluentDOM\DocumentFragment|null
*/
public function loadFragment($source, $contentType, $options = [])
{
if ($this->supports($contentType)) {
$dom = new Document('1.0', 'UTF-8');
$fragment = $dom->createDocumentFragment();
$onMapKey = $this->_onMapKey;
if (isset($options[self::ON_MAP_KEY]) && is_callable($options[self::ON_MAP_KEY])) {
$this->onMapKey($options[self::ON_MAP_KEY]);
}
$this->transferTo($fragment, json_decode($source), $this->_recursions);
$this->_onMapKey = $onMapKey;
return $fragment;
}
return NULL;
}
