public function testTitleTransformedWithBold()
{
$transformer = new Transformer();
$json_file = file_get_contents(__DIR__ . '/wp-rules.json');
$transformer->loadRules($json_file);
$title_html_string = '<?xml encoding="utf-8" ?><h1>Title <b>in bold</b></h1>';
libxml_use_internal_errors(true);
$document = new \DOMDocument();
$document->loadHtml($title_html_string);
libxml_use_internal_errors(false);
$header = Header::create();
$transformer->transform($header, $document);
$this->assertEquals('<h1>Title <b>in bold</b></h1>', $header->getTitle()->render());
}
public function get($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 testTransformPullquote()
{
$transformer_rules = <<<'JSON'
{
"rules" : [
{
"class": "TextNodeRule"
},
{
"class": "ItalicRule",
"selector": "em"
},
{
"class": "ParagraphRule",
"selector": "p"
},
{
"class": "PassThroughRule",
"selector": "div.field-quote > p"
},
{
"class": "PassThroughRule",
"selector" : "div.field-quote"
},
{
"class" : "PullquoteRule",
"selector" : "blockquote.pull-quote"
},
{
"class" : "PullquoteCiteRule",
"selector" : "div.field-quote-author"
}
]
}
JSON;
$html = '<blockquote class="pull-quote">' . '<div class="field-quote">' . '<p>Here is a fancy pull quote for the <em>world</em> to see it all.</p>' . '</div>' . '<div class="field-quote-author">Matthew Oliveira</div>' . '</blockquote>';
$expected = "<aside>Here is a fancy pull quote for the <i>world</i> to see it all." . "<cite>Matthew Oliveira</cite>" . "</aside>\n";
$instant_article = InstantArticle::create();
$transformer = new Transformer();
$transformer->loadRules($transformer_rules);
$document = new \DOMDocument();
$document->loadXML($html);
$transformer->transform($instant_article, $document);
$pullquote = $instant_article->getChildren()[0];
$result = $pullquote->render('', true) . "\n";
$this->assertEquals($expected, $result);
}
/**
* @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 $instant_article;
}
public function get($node)
{
$elements = self::findAll($node, $this->selector);
if (!empty($elements) && property_exists($elements, 'length') && $elements->length !== 0) {
Transformer::markAsProcessed($elements->item(0));
return Transformer::cloneNode($elements->item(0));
}
return null;
}
public function get($node)
{
$elements = self::findAll($node, $this->selector);
if (!empty($elements)) {
Transformer::markAsProcessed($elements->item(0));
return Transformer::cloneNode($elements->item(0));
}
return null;
}
public function testSelfTransformerContent()
{
$json_file = file_get_contents(__DIR__ . '/simple-rules.json');
$instant_article = InstantArticle::create();
$transformer = new Transformer();
$transformer->loadRules($json_file);
$html_file = file_get_contents(__DIR__ . '/simple.html');
libxml_use_internal_errors(true);
$document = new \DOMDocument();
$document->loadHTML($html_file);
libxml_use_internal_errors(false);
$transformer->transform($instant_article, $document);
$instant_article->addMetaProperty('op:generator:version', '1.0.0');
$instant_article->addMetaProperty('op:generator:transformer:version', '1.0.0');
$result = $instant_article->render('', true) . "\n";
$expected = file_get_contents(__DIR__ . '/simple-ia.html');
//var_dump($result);
// print_r($warnings);
$this->assertEquals($expected, $result);
}
public function testTransformerCustomHTML()
{
$json_file = file_get_contents(__DIR__ . '/custom-html-rules.json');
$instant_article = InstantArticle::create();
$transformer = new Transformer();
$transformer->loadRules($json_file);
$html_file = file_get_contents(__DIR__ . '/custom.html');
libxml_use_internal_errors(true);
$document = new \DOMDocument();
$document->loadHTML($html_file);
libxml_use_internal_errors(false);
$instant_article->withCanonicalURL('http://localhost/article')->withHeader(Header::create()->withTitle('Peace on <b>earth</b>')->addAuthor(Author::create()->withName('bill'))->withPublishTime(Time::create(Time::PUBLISHED)->withDatetime(\DateTime::createFromFormat('j-M-Y G:i:s', '12-Apr-2016 19:46:51'))));
$transformer->transform($instant_article, $document);
$instant_article->addMetaProperty('op:generator:version', '1.0.0');
$instant_article->addMetaProperty('op:generator:transformer:version', '1.0.0');
$result = $instant_article->render('', true) . "\n";
$expected = file_get_contents(__DIR__ . '/custom-html-ia.xml');
$this->assertEquals($expected, $result);
// there must be 3 warnings related to <img> inside <li> that is not supported by IA
$this->assertEquals(3, count($transformer->getWarnings()));
}
public function get($node)
{
$element = parent::get($node);
if ($element) {
$fragment = $element->ownerDocument->createDocumentFragment();
foreach ($element->childNodes as $child) {
Transformer::markAsProcessed($child);
$fragment->appendChild(Transformer::cloneNode($child));
}
if ($fragment->hasChildNodes()) {
return $fragment;
}
}
return null;
}
public function testTransformerResetRules()
{
$transformer = new Transformer();
$rule1 = new ParagraphRule();
$rule2 = new ItalicRule();
$transformer->addRule($rule1);
$transformer->addRule($rule2);
$transformer->resetRules();
$this->assertEquals([], $transformer->getRules());
}
/**
* Render post
*
* @since 0.1
* @return InstantArticle
*/
public function to_instant_article()
{
/**
* Fires before the instant article is rendered.
*
* @since 0.1
* @param Instant_Article_Post $instant_article_post The instant article post.
*/
do_action('instant_articles_before_transform_post', $this);
// Get time zone configured in WordPress. Default to UTC if no time zone configured.
$date_time_zone = get_option('timezone_string') ? new DateTimeZone(get_option('timezone_string')) : new DateTimeZone('UTC');
// Initialize transformer
$file_path = plugin_dir_path(__FILE__) . 'rules-configuration.json';
$configuration = file_get_contents($file_path);
$transformer = new Transformer();
$this->transformer = $transformer;
$transformer->loadRules($configuration);
$transformer = apply_filters('instant_articles_transformer_rules_loaded', $transformer);
$settings_publishing = Instant_Articles_Option_Publishing::get_option_decoded();
if (isset($settings_publishing['custom_rules_enabled']) && !empty($settings_publishing['custom_rules_enabled']) && isset($settings_publishing['custom_rules']) && !empty($settings_publishing['custom_rules'])) {
$transformer->loadRules($settings_publishing['custom_rules']);
}
$transformer = apply_filters('instant_articles_transformer_custom_rules_loaded', $transformer);
$blog_charset = get_option('blog_charset');
$header = Header::create()->withPublishTime(Time::create(Time::PUBLISHED)->withDatetime(new DateTime($this->_post->post_date, $date_time_zone)))->withModifyTime(Time::create(Time::MODIFIED)->withDatetime(new DateTime($this->_post->post_modified, $date_time_zone)));
$title = $this->get_the_title();
if ($title) {
$document = new DOMDocument();
libxml_use_internal_errors(true);
$document->loadHTML('<?xml encoding="' . $blog_charset . '" ?><h1>' . $title . '</h1>');
libxml_use_internal_errors(false);
$transformer->transform($header, $document);
}
if ($this->has_subtitle()) {
$header->withSubTitle($this->get_the_subtitle());
}
$authors = $this->get_the_authors();
foreach ($authors as $author) {
$author_obj = Author::create();
if ($author->display_name) {
$author_obj->withName($author->display_name);
}
if ($author->bio) {
$author_obj->withDescription($author->bio);
}
if ($author->user_url) {
$author_obj->withURL($author->user_url);
}
$header->addAuthor($author_obj);
}
$kicker = $this->get_the_kicker();
if ($kicker) {
$header->withKicker($kicker);
}
$cover = $this->get_cover_media();
if ($cover->getUrl()) {
$header->withCover($cover);
}
$this->instant_article = InstantArticle::create()->withCanonicalUrl($this->get_canonical_url())->withHeader($header)->addMetaProperty('op:generator:application', 'facebook-instant-articles-wp')->addMetaProperty('op:generator:application:version', IA_PLUGIN_VERSION);
$settings_style = Instant_Articles_Option_Styles::get_option_decoded();
if (isset($settings_style['article_style']) && !empty($settings_style['article_style'])) {
$this->instant_article->withStyle($settings_style['article_style']);
} else {
$this->instant_article->withStyle('default');
}
$libxml_previous_state = libxml_use_internal_errors(true);
$document = new DOMDocument('1.0', get_option('blog_charset'));
$content = $this->get_the_content();
// DOMDocument isn’t handling encodings too well, so let’s help it a little.
if (function_exists('mb_convert_encoding')) {
$content = mb_convert_encoding($content, 'HTML-ENTITIES', get_option('blog_charset'));
} else {
$content = htmlspecialchars_decode(utf8_decode(htmlentities($content, ENT_COMPAT, 'utf-8', false)));
}
$result = $document->loadHTML('<!doctype html><html><body>' . $content . '</body></html>');
// We need to make sure that scripts use absolute URLs and not relative URLs.
$scripts = $document->getElementsByTagName('script');
if (!empty($scripts)) {
foreach ($scripts as $script) {
$src = $script->getAttribute('src');
$explode_src = parse_url($src);
if (is_array($explode_src) && empty($explode_src['scheme']) && !empty($explode_src['host']) && !empty($explode_src['path'])) {
$src = 'https://' . $explode_src['host'] . $explode_src['path'];
}
$script->setAttribute('src', $src);
}
}
libxml_clear_errors();
libxml_use_internal_errors($libxml_previous_state);
$document = apply_filters('instant_articles_parsed_document', $document);
if ($result) {
$transformer->transform($this->instant_article, $document);
}
$this->add_ads_from_settings();
$this->add_analytics_from_settings();
$this->instant_article = apply_filters('instant_articles_transformed_element', $this->instant_article);
/**
* Fires after the instant article is rendered.
*
* @since 0.1
* @param Instant_Article_Post $instant_article_post The instant article post.
*/
do_action('instant_articles_after_transform_post', $this);
return $this->instant_article;
}
/**
* Render post
*
* @since 0.1
* @return InstantArticle
*/
public function to_instant_article()
{
/**
* Fires before the instant article is rendered.
*
* @since 0.1
* @param Instant_Article_Post $instant_article_post The instant article post.
*/
do_action('instant_articles_before_transform_post', $this);
// Get time zone configured in WordPress. Default to UTC if no time zone configured.
$date_time_zone = get_option('timezone_string') ? new DateTimeZone(get_option('timezone_string')) : new DateTimeZone('UTC');
// Initialize transformer
$file_path = plugin_dir_path(__FILE__) . 'rules-configuration.json';
$configuration = file_get_contents($file_path);
$transformer = new Transformer();
$this->transformer = $transformer;
$transformer->loadRules($configuration);
$transformer = apply_filters('instant_articles_transformer_rules_loaded', $transformer);
$settings_publishing = Instant_Articles_Option_Publishing::get_option_decoded();
if (isset($settings_publishing['custom_rules_enabled']) && !empty($settings_publishing['custom_rules_enabled']) && isset($settings_publishing['custom_rules']) && !empty($settings_publishing['custom_rules'])) {
$transformer->loadRules($settings_publishing['custom_rules']);
}
$transformer = apply_filters('instant_articles_transformer_custom_rules_loaded', $transformer);
$blog_charset = get_option('blog_charset');
$header = Header::create()->withPublishTime(Time::create(Time::PUBLISHED)->withDatetime(new DateTime($this->_post->post_date, $date_time_zone)))->withModifyTime(Time::create(Time::MODIFIED)->withDatetime(new DateTime($this->_post->post_modified, $date_time_zone)));
$title = $this->get_the_title();
if ($title) {
$document = new DOMDocument();
libxml_use_internal_errors(true);
$document->loadHTML('<?xml encoding="' . $blog_charset . '" ?><h1>' . $title . '</h1>');
libxml_use_internal_errors(false);
$transformer->transform($header, $document);
}
if ($this->has_subtitle()) {
$header->withSubTitle($this->get_the_subtitle());
}
$authors = $this->get_the_authors();
foreach ($authors as $author) {
$author_obj = Author::create();
if ($author->display_name) {
$author_obj->withName($author->display_name);
}
if ($author->bio) {
$author_obj->withDescription($author->bio);
}
if ($author->user_url) {
$author_obj->withURL($author->user_url);
}
$header->addAuthor($author_obj);
}
$kicker = $this->get_the_kicker();
if ($kicker) {
$header->withKicker($kicker);
}
$cover = $this->get_cover_media();
if ($cover->getUrl()) {
$header->withCover($cover);
}
$this->instant_article = InstantArticle::create()->withCanonicalUrl($this->get_canonical_url())->withHeader($header)->addMetaProperty('op:generator:application', 'facebook-instant-articles-wp')->addMetaProperty('op:generator:application:version', IA_PLUGIN_VERSION);
$settings_style = Instant_Articles_Option_Styles::get_option_decoded();
if (isset($settings_style['article_style']) && !empty($settings_style['article_style'])) {
$this->instant_article->withStyle($settings_style['article_style']);
} else {
$this->instant_article->withStyle('default');
}
$transformer->transformString($this->instant_article, $this->get_the_content(), get_option('blog_charset'));
$this->add_ads_from_settings();
$this->add_analytics_from_settings();
$this->instant_article = apply_filters('instant_articles_transformed_element', $this->instant_article);
/**
* Fires after the instant article is rendered.
*
* @since 0.1
* @param Instant_Article_Post $instant_article_post The instant article post.
*/
do_action('instant_articles_after_transform_post', $this);
return $this->instant_article;
}