public function get($node)
{
Type::enforce($node, 'DOMNode');
$elements = self::findAll($node, $this->selector);
if (!empty($elements) && $elements->item(0)) {
$element = $elements->item(0);
do {
$element = $element->nextSibling;
} while ($element !== null && !Type::is($element, 'DOMElement'));
if ($element && Type::is($element, 'DOMElement')) {
if ($this->siblingSelector) {
$siblings = self::findAll($element, $this->siblingSelector);
if (!empty($siblings) && $siblings->item(0)) {
$siblingElement = $siblings->item(0);
} else {
// Returns null because sibling content doesn't match
return null;
}
} else {
$siblingElement = $element;
}
Transformer::markAsProcessed($siblingElement);
return Transformer::cloneNode($siblingElement);
}
}
return null;
}
public function apply($transformer, $context_element, $node)
{
$h2 = H2::create();
if (Type::is($context_element, array(Header::getClassName(), Caption::getClassName()))) {
$context_element->withSubTitle($h2);
} elseif (Type::is($context_element, InstantArticle::getClassName())) {
$context_element->addChild($h2);
}
if ($this->getProperty(Caption::POSITION_BELOW, $node)) {
$h2->withPosition(Caption::POSITION_BELOW);
}
if ($this->getProperty(Caption::POSITION_CENTER, $node)) {
$h2->withPosition(Caption::POSITION_CENTER);
}
if ($this->getProperty(Caption::POSITION_ABOVE, $node)) {
$h2->withPosition(Caption::POSITION_ABOVE);
}
if ($this->getProperty(Caption::ALIGN_LEFT, $node)) {
$h2->withTextAlignment(Caption::ALIGN_LEFT);
}
if ($this->getProperty(Caption::ALIGN_CENTER, $node)) {
$h2->withTextAlignment(Caption::ALIGN_CENTER);
}
if ($this->getProperty(Caption::ALIGN_RIGHT, $node)) {
$h2->withTextAlignment(Caption::ALIGN_RIGHT);
}
$transformer->transform($h2, $node);
return $context_element;
}
/**
* Adds a new item to the List
*
* @param string|ListItem $new_item The new item that will be pushed to the end of the list
*
* @return $this
*/
public function addItem($new_item)
{
Type::enforce($new_item, [ListItem::getClassName(), Type::STRING]);
if (Type::is($new_item, Type::STRING)) {
$new_item = ListItem::create()->appendText($new_item);
}
$this->items[] = $new_item;
return $this;
}
/**
* Sets the attribution string
*
* @param string|Cite $attribution The attribution text
*
* @return $this
*/
public function withAttribution($attribution)
{
Type::enforce($attribution, [Type::STRING, Cite::getClassName()]);
if (Type::is($attribution, Type::STRING)) {
$this->attribution = Cite::create()->appendText($attribution);
} else {
$this->attribution = $attribution;
}
return $this;
}
/**
* Sets the unescaped HTML.
*
* @param \DOMNode|string $html The unescaped HTML.
*
* @return $this
*/
public function withHTML($html)
{
Type::enforce($html, ['DOMNode', Type::STRING]);
// If this is raw HTML source, wrap in a CDATA section as it could contain JS etc. with characters (such as &) that are not allowed in unescaped form
if (Type::is($html, Type::STRING)) {
$html = new \DOMCdataSection($html);
}
$this->html = $html;
return $this;
}
public function matchesNode($node)
{
// Only matches DOMElements (ignore text and comments)
if (!Type::is($node, 'DOMElement')) {
return false;
}
// Handles selector = tag
if ($node->nodeName === $this->selector) {
return true;
}
// Handles selector = .class
if (preg_match('/^\\.[a-zA-Z][a-zA-Z0-9-]*$/', $this->selector) === 1) {
// Tries every class
$classNames = explode(' ', $node->getAttribute('class'));
foreach ($classNames as $className) {
if ('.' . $className === $this->selector) {
return true;
}
}
// No match!
return false;
}
// Handles selector = tag.class
if (preg_match('/^[a-zA-Z][a-zA-Z0-9-]*(\\.[a-zA-Z][a-zA-Z0-9-]*)?$/', $this->selector) === 1) {
// Tries every class
$classNames = explode(' ', $node->getAttribute('class'));
foreach ($classNames as $className) {
if ($node->nodeName . '.' . $className === $this->selector) {
return true;
}
}
// No match!
return false;
}
// Proceed with the more expensive XPath query
$document = $node->ownerDocument;
$domXPath = new \DOMXPath($document);
if (substr($this->selector, 0, 1) === '/') {
$xpath = $this->selector;
} else {
$converter = new CssSelectorConverter();
$xpath = $converter->toXPath($this->selector);
}
$results = $domXPath->query($xpath);
if (false !== $results) {
foreach ($results as $result) {
if ($result === $node) {
return true;
}
}
}
return false;
}
public function get($node)
{
$fragment = $node->ownerDocument->createDocumentFragment();
foreach ($this->children as $child) {
$cloned_node = $child->get($node);
if (Type::is($cloned_node, 'DOMNode')) {
$fragment->appendChild($cloned_node);
}
}
if ($fragment->hasChildNodes()) {
return $fragment;
}
return null;
}
public function apply($transformer, $context, $node)
{
$interactive = Interactive::create();
if (Type::is($context, InstantArticle::getClassName())) {
$instant_article = $context;
} elseif ($transformer->getInstantArticle()) {
$instant_article = $transformer->getInstantArticle();
$context->disableEmptyValidation();
$context = Paragraph::create();
$context->disableEmptyValidation();
} else {
$transformer->addWarning(new NoRootInstantArticleFoundWarning(null, $node));
return $context;
}
// Builds the interactive
$iframe = $this->getProperty(self::PROPERTY_IFRAME, $node);
$url = $this->getProperty(self::PROPERTY_URL, $node);
if ($iframe) {
$interactive->withHTML($iframe);
}
if ($url) {
$interactive->withSource($url);
}
if ($iframe || $url) {
$instant_article->addChild($interactive);
if ($instant_article !== $context) {
$instant_article->addChild($context);
}
} else {
$transformer->addWarning(new InvalidSelector(self::PROPERTY_IFRAME, $instant_article, $node, $this));
}
if ($this->getProperty(self::PROPERTY_WIDTH_COLUMN_WIDTH, $node)) {
$interactive->withMargin(Interactive::COLUMN_WIDTH);
} else {
$interactive->withMargin(Interactive::NO_MARGIN);
}
$width = $this->getProperty(self::PROPERTY_WIDTH, $node);
if ($width) {
$interactive->withWidth($width);
}
$height = $this->getProperty(self::PROPERTY_HEIGHT, $node);
if ($height) {
$interactive->withHeight($height);
}
$suppress_warnings = $transformer->suppress_warnings;
$transformer->suppress_warnings = true;
$transformer->transform($interactive, $node);
$transformer->suppress_warnings = $suppress_warnings;
return $context;
}
/**
* @param string|DOMDocument $document The document html of an Instant Article
*
* @return InstantArticle filled element that was parsed from the DOMDocument parameter
*/
public function parse($content)
{
if (Type::is($content, Type::STRING)) {
libxml_use_internal_errors(true);
$document = new \DOMDocument();
$document->loadHTML($content);
libxml_use_internal_errors(false);
} else {
$document = $content;
}
$json_file = file_get_contents(__DIR__ . '/instant-articles-rules.json');
$instant_article = InstantArticle::create();
$transformer = new Transformer();
$transformer->loadRules($json_file);
$transformer->transform($instant_article, $document);
return $instant_article;
}
public function apply($transformer, $context, $node)
{
$image = Image::create();
if (Type::is($context, InstantArticle::getClassName())) {
$instant_article = $context;
} elseif ($transformer->getInstantArticle()) {
$instant_article = $transformer->getInstantArticle();
$context->disableEmptyValidation();
$context = Paragraph::create();
$context->disableEmptyValidation();
} else {
$transformer->addWarning(new NoRootInstantArticleFoundWarning(null, $node));
return $context;
}
// Builds the image
$url = $this->getProperty(self::PROPERTY_IMAGE_URL, $node);
if ($url) {
$image->withURL($url);
$instant_article->addChild($image);
if ($instant_article !== $context) {
$instant_article->addChild($context);
}
} else {
$transformer->addWarning(new InvalidSelector(self::PROPERTY_IMAGE_URL, $instant_article, $node, $this));
}
if ($this->getProperty(Image::ASPECT_FIT, $node)) {
$image->withPresentation(Image::ASPECT_FIT);
} elseif ($this->getProperty(Image::ASPECT_FIT_ONLY, $node)) {
$image->withPresentation(Image::ASPECT_FIT_ONLY);
} elseif ($this->getProperty(Image::FULLSCREEN, $node)) {
$image->withPresentation(Image::FULLSCREEN);
} elseif ($this->getProperty(Image::NON_INTERACTIVE, $node)) {
$image->withPresentation(Image::NON_INTERACTIVE);
}
if ($this->getProperty(self::PROPERTY_LIKE, $node)) {
$image->enableLike();
}
if ($this->getProperty(self::PROPERTY_COMMENTS, $node)) {
$image->enableComments();
}
$suppress_warnings = $transformer->suppress_warnings;
$transformer->suppress_warnings = true;
$transformer->transform($image, $node);
$transformer->suppress_warnings = $suppress_warnings;
return $context;
}
/**
* @return string
*/
public function __toString()
{
$reflection = new \ReflectionClass(get_class($this->context));
$className = $reflection->getShortName();
$nodeName = $this->node->nodeName;
if (substr($nodeName, 0, 1) === '#') {
$nodeDescription = '"' . mb_strimwidth($this->node->textContent, 0, 30, '...') . '"';
} else {
$nodeDescription = '<';
$nodeDescription .= $nodeName;
if (Type::is($this->node, 'DOMElement')) {
$class = $this->node->getAttribute('class');
if ($class) {
$nodeDescription .= ' class="' . $class . '"';
}
}
$nodeDescription .= '>';
}
return "No rules defined for {$nodeDescription} in the context of {$className}";
}
/**
* Implements the Container::getContainerChildren().
*
* @see Container::getContainerChildren().
* @return array of TextContainer
*/
public function getContainerChildren()
{
$children = array();
foreach ($this->textChildren as $content) {
// Recursive check on TextContainer, if something inside is valid, this is valid.
if (Type::is($content, TextContainer::getClassName())) {
$children[] = $content;
}
}
return $children;
}
/**
* The caption credit. optional.
*
* @param string $credit the caption credit text that will be shown
*
* @return $this
*/
public function withCredit($credit)
{
Type::enforce($credit, [Type::STRING, Cite::getClassName()]);
if (Type::is($credit, Type::STRING)) {
$credit = Cite::create()->appendText($credit);
}
$this->credit = $credit;
return $this;
}
public function testIsNotString()
{
$result = Type::is(1, Type::STRING);
$this->assertFalse($result);
}
/**
* Kicker text for the article header.
*
* @param H3|string The kicker text to be set
*
* @return $this
*/
public function withKicker($kicker)
{
Type::enforce($kicker, array(Type::STRING, H3::getClassName()));
if (Type::is($kicker, Type::STRING)) {
$this->kicker = H3::create()->appendText($kicker);
} else {
$this->kicker = $kicker;
}
return $this;
}
/**
* Sets the geotag on the video.
*
* @see {link:http://geojson.org/}
*
* @param string $geoTag
*
* @return $this
*/
public function withGeoTag($geoTag)
{
Type::enforce($geoTag, [Type::STRING, GeoTag::getClassName()]);
if (Type::is($geoTag, Type::STRING)) {
$this->geoTag = GeoTag::create()->withScript($geoTag);
} elseif (Type::is($geoTag, GeoTag::getClassName())) {
$this->geoTag = $geoTag;
}
return $this;
}
/**
* @param InstantArticle $context
* @param \DOMNode $node
*
* @return mixed
*/
public function transform($context, $node)
{
if (Type::is($context, InstantArticle::getClassName())) {
$context->addMetaProperty('op:generator:transformer', 'facebook-instant-articles-sdk-php');
$context->addMetaProperty('op:generator:transformer:version', InstantArticle::CURRENT_VERSION);
$this->instantArticle = $context;
}
$log = \Logger::getLogger('facebook-instantarticles-transformer');
if (!$node) {
$e = new \Exception();
$log->error('Transformer::transform($context, $node) requires $node' . ' to be a valid one. Check on the stacktrace if this is ' . 'some nested transform operation and fix the selector.', $e->getTraceAsString());
}
$current_context = $context;
if ($node->hasChildNodes()) {
foreach ($node->childNodes as $child) {
if (self::isProcessed($child)) {
continue;
}
$matched = false;
$log->debug("===========================");
$log->debug($child->ownerDocument->saveHtml($child));
// Get all classes and interfaces this context extends/implements
$contextClassNames = self::getAllClassTypes($context->getClassName());
// Look for rules applying to any of them as context
$matchingContextRules = [];
foreach ($contextClassNames as $contextClassName) {
if (isset($this->rules[$contextClassName])) {
// Use array union (+) instead of merge to preserve
// indexes (as they represent the order of insertion)
$matchingContextRules = $matchingContextRules + $this->rules[$contextClassName];
}
}
// Sort by insertion order
ksort($matchingContextRules);
// Process in reverse order
$matchingContextRules = array_reverse($matchingContextRules);
foreach ($matchingContextRules as $rule) {
// We know context was matched, now check if it matches the node
if ($rule->matchesNode($child)) {
$current_context = $rule->apply($this, $current_context, $child);
$matched = true;
// Just a single rule for each node, so move on
break;
}
}
if (!$matched && !($child->nodeName === '#text' && trim($child->textContent) === '') && !($child->nodeName === '#comment') && !($child->nodeName === 'html' && Type::is($child, 'DOMDocumentType')) && !($child->nodeName === 'xml' && Type::is($child, 'DOMProcessingInstruction')) && !$this->suppress_warnings) {
$tag_content = $child->ownerDocument->saveXML($child);
$tag_trimmed = trim($tag_content);
if (!empty($tag_trimmed)) {
$log->debug('context class: ' . get_class($context));
$log->debug('node name: ' . $child->nodeName);
$log->debug("CONTENT NOT MATCHED: \n" . $tag_content);
} else {
$log->debug('empty content ignored');
}
$this->addWarning(new UnrecognizedElement($current_context, $child));
}
}
}
return $context;
}
public function toDOMElement($document = null)
{
if (!$document) {
$document = new \DOMDocument();
}
// Builds and appends head to the HTML document
$html = $document->createElement('html');
if ($this->isRTLEnabled) {
$html->setAttribute('dir', 'rtl');
}
$head = $document->createElement('head');
$html->appendChild($head);
$link = $document->createElement('link');
$link->setAttribute('rel', 'canonical');
$link->setAttribute('href', $this->canonicalURL);
$head->appendChild($link);
$charset = $document->createElement('meta');
$charset->setAttribute('charset', $this->charset);
$head->appendChild($charset);
$this->addMetaProperty('op:markup_version', $this->markupVersion);
if ($this->header && count($this->header->getAds()) > 0) {
$this->addMetaProperty('fb:use_automatic_ad_placement', $this->isAutomaticAdPlaced ? 'true' : 'false');
}
if ($this->style) {
$this->addMetaProperty('fb:article_style', $this->style);
}
// Adds all meta properties
foreach ($this->metaProperties as $property_name => $property_content) {
$head->appendChild($this->createMetaElement($document, $property_name, $property_content));
}
// Build and append body and article tags to the HTML document
$body = $document->createElement('body');
$article = $document->createElement('article');
$body->appendChild($article);
$html->appendChild($body);
if ($this->header && $this->header->isValid()) {
$article->appendChild($this->header->toDOMElement($document));
}
if ($this->children) {
foreach ($this->children as $child) {
if (Type::is($child, TextContainer::getClassName())) {
if (count($child->getTextChildren()) === 0) {
continue;
} elseif (count($child->getTextChildren()) === 1) {
if (Type::is($child->getTextChildren()[0], Type::STRING) && trim($child->getTextChildren()[0]) === '') {
continue;
}
}
}
$article->appendChild($child->toDOMElement($document));
}
if ($this->footer && $this->footer->isValid()) {
$article->appendChild($this->footer->toDOMElement($document));
}
} else {
$article->appendChild($document->createTextNode(''));
}
return $html;
}
/**
* Implements the Container::getContainerChildren().
*
* @see Container::getContainerChildren()
* @return array of Paragraph|RelatedArticles
*/
public function getContainerChildren()
{
$children = array();
if ($this->credits) {
if (is_array($this->credits)) {
foreach ($this->credits as $paragraph) {
if (Type::is($paragraph, Element::getClassName())) {
$children[] = $paragraph;
}
}
} else {
if (Type::is($this->credits, Element::getClassName())) {
$children[] = $this->credits;
}
}
}
if ($this->relatedArticles) {
$children[] = $this->relatedArticles;
}
return $children;
}
