-<?php\r
-/**\r
- * Content Extractor\r
- * \r
- * Uses patterns specified in site config files and auto detection (hNews/PHP Readability) \r
- * to extract content from HTML files.\r
- * \r
- * @version 1.0\r
- * @date 2013-02-05\r
- * @author Keyvan Minoukadeh\r
- * @copyright 2013 Keyvan Minoukadeh\r
- * @license http://www.gnu.org/licenses/agpl-3.0.html AGPL v3\r
- */\r
-\r
-class ContentExtractor\r
-{\r
- protected static $tidy_config = array(\r
- 'clean' => true,\r
- 'output-xhtml' => true,\r
- 'logical-emphasis' => true,\r
- 'show-body-only' => false,\r
- 'new-blocklevel-tags' => 'article, aside, footer, header, hgroup, menu, nav, section, details, datagrid',\r
- 'new-inline-tags' => 'mark, time, meter, progress, data',\r
- 'wrap' => 0,\r
- 'drop-empty-paras' => true,\r
- 'drop-proprietary-attributes' => false,\r
- 'enclose-text' => true,\r
- 'enclose-block-text' => true,\r
- 'merge-divs' => true,\r
- 'merge-spans' => true,\r
- 'char-encoding' => 'utf8',\r
- 'hide-comments' => true\r
- );\r
- protected $html;\r
- protected $config;\r
- protected $title;\r
- protected $author = array();\r
- protected $language;\r
- protected $date;\r
- protected $body;\r
- protected $success = false;\r
- protected $nextPageUrl;\r
- public $allowedParsers = array('libxml', 'html5lib');\r
- public $fingerprints = array();\r
- public $readability;\r
- public $debug = false;\r
- public $debugVerbose = false;\r
-\r
- function __construct($path, $fallback=null) {\r
- SiteConfig::set_config_path($path, $fallback); \r
- }\r
- \r
- protected function debug($msg) {\r
- if ($this->debug) {\r
- $mem = round(memory_get_usage()/1024, 2);\r
- $memPeak = round(memory_get_peak_usage()/1024, 2);\r
- echo '* ',$msg;\r
- if ($this->debugVerbose) echo ' - mem used: ',$mem," (peak: $memPeak)";\r
- echo "\n";\r
- ob_flush();\r
- flush();\r
- }\r
- }\r
- \r
- public function reset() {\r
- $this->html = null;\r
- $this->readability = null;\r
- $this->config = null;\r
- $this->title = null;\r
- $this->body = null;\r
- $this->author = array();\r
- $this->language = null;\r
- $this->date = null;\r
- $this->nextPageUrl = null;\r
- $this->success = false;\r
- }\r
-\r
- public function findHostUsingFingerprints($html) {\r
- $this->debug('Checking fingerprints...');\r
- $head = substr($html, 0, 8000);\r
- foreach ($this->fingerprints as $_fp => $_fphost) {\r
- $lookin = 'html';\r
- if (is_array($_fphost)) {\r
- if (isset($_fphost['head']) && $_fphost['head']) {\r
- $lookin = 'head';\r
- }\r
- $_fphost = $_fphost['hostname'];\r
- }\r
- if (strpos($$lookin, $_fp) !== false) {\r
- $this->debug("Found match: $_fphost");\r
- return $_fphost;\r
- }\r
- }\r
- $this->debug('No fingerprint matches');\r
- return false;\r
- }\r
- \r
- // returns SiteConfig instance (joined in order: exact match, wildcard, fingerprint, global, default)\r
- public function buildSiteConfig($url, $html='', $add_to_cache=true) {\r
- // extract host name\r
- $host = @parse_url($url, PHP_URL_HOST);\r
- $host = strtolower($host);\r
- if (substr($host, 0, 4) == 'www.') $host = substr($host, 4);\r
- // is merged version already cached?\r
- if (SiteConfig::is_cached("$host.merged")) {\r
- $this->debug("Returning cached and merged site config for $host");\r
- return SiteConfig::build("$host.merged");\r
- }\r
- // let's build from site_config/custom/ and standard/\r
- $config = SiteConfig::build($host);\r
- if ($add_to_cache && $config && !SiteConfig::is_cached("$host")) {\r
- SiteConfig::add_to_cache($host, $config);\r
- }\r
- // if no match, use defaults\r
- if (!$config) $config = new SiteConfig();\r
- // load fingerprint config?\r
- if ($config->autodetect_on_failure()) {\r
- // check HTML for fingerprints\r
- if (!empty($this->fingerprints) && ($_fphost = $this->findHostUsingFingerprints($html))) {\r
- if ($config_fingerprint = SiteConfig::build($_fphost)) {\r
- $this->debug("Appending site config settings from $_fphost (fingerprint match)");\r
- $config->append($config_fingerprint);\r
- if ($add_to_cache && !SiteConfig::is_cached($_fphost)) {\r
- //$config_fingerprint->cache_in_apc = true;\r
- SiteConfig::add_to_cache($_fphost, $config_fingerprint);\r
- }\r
- }\r
- }\r
- }\r
- // load global config?\r
- if ($config->autodetect_on_failure()) {\r
- if ($config_global = SiteConfig::build('global', true)) {\r
- $this->debug('Appending site config settings from global.txt');\r
- $config->append($config_global);\r
- if ($add_to_cache && !SiteConfig::is_cached('global')) {\r
- //$config_global->cache_in_apc = true;\r
- SiteConfig::add_to_cache('global', $config_global);\r
- }\r
- }\r
- }\r
- // store copy of merged config\r
- if ($add_to_cache) {\r
- // do not store in APC if wildcard match\r
- $use_apc = ($host == $config->cache_key);\r
- $config->cache_key = null;\r
- SiteConfig::add_to_cache("$host.merged", $config, $use_apc);\r
- }\r
- return $config;\r
- }\r
- \r
- // returns true on success, false on failure\r
- // $smart_tidy indicates that if tidy is used and no results are produced, we will\r
- // try again without it. Tidy helps us deal with PHP's patchy HTML parsing most of the time\r
- // but it has problems of its own which we try to avoid with this option.\r
- public function process($html, $url, $smart_tidy=true) {\r
- $this->reset();\r
- $this->config = $this->buildSiteConfig($url, $html);\r
- \r
- // do string replacements\r
- if (!empty($this->config->find_string)) {\r
- if (count($this->config->find_string) == count($this->config->replace_string)) {\r
- $html = str_replace($this->config->find_string, $this->config->replace_string, $html, $_count);\r
- $this->debug("Strings replaced: $_count (find_string and/or replace_string)");\r
- } else {\r
- $this->debug('Skipped string replacement - incorrect number of find-replace strings in site config');\r
- }\r
- unset($_count);\r
- }\r
- \r
- // use tidy (if it exists)?\r
- // This fixes problems with some sites which would otherwise\r
- // trouble DOMDocument's HTML parsing. (Although sometimes it\r
- // makes matters worse, which is why you can override it in site config files.)\r
- $tidied = false;\r
- if ($this->config->tidy() && function_exists('tidy_parse_string') && $smart_tidy) {\r
- $this->debug('Using Tidy');\r
- $tidy = tidy_parse_string($html, self::$tidy_config, 'UTF8');\r
- if (tidy_clean_repair($tidy)) {\r
- $original_html = $html;\r
- $tidied = true;\r
- $html = $tidy->value;\r
- }\r
- unset($tidy);\r
- }\r
- \r
- // load and parse html\r
- $_parser = $this->config->parser();\r
- if (!in_array($_parser, $this->allowedParsers)) {\r
- $this->debug("HTML parser $_parser not listed, using libxml instead");\r
- $_parser = 'libxml';\r
- }\r
- $this->debug("Attempting to parse HTML with $_parser");\r
- $this->readability = new Readability($html, $url, $_parser);\r
- \r
- // we use xpath to find elements in the given HTML document\r
- // see http://en.wikipedia.org/wiki/XPath_1.0\r
- $xpath = new DOMXPath($this->readability->dom);\r
-\r
- // try to get next page link\r
- foreach ($this->config->next_page_link as $pattern) {\r
- $elems = @$xpath->evaluate($pattern, $this->readability->dom);\r
- if (is_string($elems)) {\r
- $this->nextPageUrl = trim($elems);\r
- break;\r
- } elseif ($elems instanceof DOMNodeList && $elems->length > 0) {\r
- foreach ($elems as $item) {\r
- if ($item instanceof DOMElement && $item->hasAttribute('href')) {\r
- $this->nextPageUrl = $item->getAttribute('href');\r
- break 2;\r
- } elseif ($item instanceof DOMAttr && $item->value) {\r
- $this->nextPageUrl = $item->value;\r
- break 2;\r
- }\r
- }\r
- }\r
- }\r
- \r
- // try to get title\r
- foreach ($this->config->title as $pattern) {\r
- // $this->debug("Trying $pattern");\r
- $elems = @$xpath->evaluate($pattern, $this->readability->dom);\r
- if (is_string($elems)) {\r
- $this->title = trim($elems);\r
- $this->debug('Title expression evaluated as string: '.$this->title);\r
- $this->debug("...XPath match: $pattern");\r
- break;\r
- } elseif ($elems instanceof DOMNodeList && $elems->length > 0) {\r
- $this->title = $elems->item(0)->textContent;\r
- $this->debug('Title matched: '.$this->title);\r
- $this->debug("...XPath match: $pattern");\r
- // remove title from document\r
- try {\r
- $elems->item(0)->parentNode->removeChild($elems->item(0));\r
- } catch (DOMException $e) {\r
- // do nothing\r
- }\r
- break;\r
- }\r
- }\r
- \r
- // try to get author (if it hasn't already been set)\r
- if (empty($this->author)) {\r
- foreach ($this->config->author as $pattern) {\r
- $elems = @$xpath->evaluate($pattern, $this->readability->dom);\r
- if (is_string($elems)) {\r
- if (trim($elems) != '') {\r
- $this->author[] = trim($elems);\r
- $this->debug('Author expression evaluated as string: '.trim($elems));\r
- $this->debug("...XPath match: $pattern");\r
- break;\r
- }\r
- } elseif ($elems instanceof DOMNodeList && $elems->length > 0) {\r
- foreach ($elems as $elem) {\r
- if (!isset($elem->parentNode)) continue;\r
- $this->author[] = trim($elem->textContent);\r
- $this->debug('Author matched: '.trim($elem->textContent));\r
- }\r
- if (!empty($this->author)) {\r
- $this->debug("...XPath match: $pattern");\r
- break;\r
- }\r
- }\r
- }\r
- }\r
- \r
- // try to get language\r
- $_lang_xpath = array('//html[@lang]/@lang', '//meta[@name="DC.language"]/@content');\r
- foreach ($_lang_xpath as $pattern) {\r
- $elems = @$xpath->evaluate($pattern, $this->readability->dom);\r
- if (is_string($elems)) {\r
- if (trim($elems) != '') {\r
- $this->language = trim($elems);\r
- $this->debug('Language matched: '.$this->language);\r
- break;\r
- }\r
- } elseif ($elems instanceof DOMNodeList && $elems->length > 0) {\r
- foreach ($elems as $elem) {\r
- if (!isset($elem->parentNode)) continue;\r
- $this->language = trim($elem->textContent);\r
- $this->debug('Language matched: '.$this->language); \r
- }\r
- if ($this->language) break;\r
- }\r
- }\r
- \r
- // try to get date\r
- foreach ($this->config->date as $pattern) {\r
- $elems = @$xpath->evaluate($pattern, $this->readability->dom);\r
- if (is_string($elems)) {\r
- $this->date = strtotime(trim($elems, "; \t\n\r\0\x0B")); \r
- } elseif ($elems instanceof DOMNodeList && $elems->length > 0) {\r
- $this->date = $elems->item(0)->textContent;\r
- $this->date = strtotime(trim($this->date, "; \t\n\r\0\x0B"));\r
- // remove date from document\r
- // $elems->item(0)->parentNode->removeChild($elems->item(0));\r
- }\r
- if (!$this->date) {\r
- $this->date = null;\r
- } else {\r
- $this->debug('Date matched: '.date('Y-m-d H:i:s', $this->date));\r
- $this->debug("...XPath match: $pattern");\r
- break;\r
- }\r
- }\r
-\r
- // strip elements (using xpath expressions)\r
- foreach ($this->config->strip as $pattern) {\r
- $elems = @$xpath->query($pattern, $this->readability->dom);\r
- // check for matches\r
- if ($elems && $elems->length > 0) {\r
- $this->debug('Stripping '.$elems->length.' elements (strip)');\r
- for ($i=$elems->length-1; $i >= 0; $i--) {\r
- $elems->item($i)->parentNode->removeChild($elems->item($i));\r
- }\r
- }\r
- }\r
- \r
- // strip elements (using id and class attribute values)\r
- foreach ($this->config->strip_id_or_class as $string) {\r
- $string = strtr($string, array("'"=>'', '"'=>''));\r
- $elems = @$xpath->query("//*[contains(@class, '$string') or contains(@id, '$string')]", $this->readability->dom);\r
- // check for matches\r
- if ($elems && $elems->length > 0) {\r
- $this->debug('Stripping '.$elems->length.' elements (strip_id_or_class)');\r
- for ($i=$elems->length-1; $i >= 0; $i--) {\r
- $elems->item($i)->parentNode->removeChild($elems->item($i));\r
- }\r
- }\r
- }\r
- \r
- // strip images (using src attribute values)\r
- foreach ($this->config->strip_image_src as $string) {\r
- $string = strtr($string, array("'"=>'', '"'=>''));\r
- $elems = @$xpath->query("//img[contains(@src, '$string')]", $this->readability->dom);\r
- // check for matches\r
- if ($elems && $elems->length > 0) {\r
- $this->debug('Stripping '.$elems->length.' image elements');\r
- for ($i=$elems->length-1; $i >= 0; $i--) {\r
- $elems->item($i)->parentNode->removeChild($elems->item($i));\r
- }\r
- }\r
- }\r
- // strip elements using Readability.com and Instapaper.com ignore class names\r
- // .entry-unrelated and .instapaper_ignore\r
- // See https://www.readability.com/publishers/guidelines/#view-plainGuidelines\r
- // and http://blog.instapaper.com/post/730281947\r
- $elems = @$xpath->query("//*[contains(concat(' ',normalize-space(@class),' '),' entry-unrelated ') or contains(concat(' ',normalize-space(@class),' '),' instapaper_ignore ')]", $this->readability->dom);\r
- // check for matches\r
- if ($elems && $elems->length > 0) {\r
- $this->debug('Stripping '.$elems->length.' .entry-unrelated,.instapaper_ignore elements');\r
- for ($i=$elems->length-1; $i >= 0; $i--) {\r
- $elems->item($i)->parentNode->removeChild($elems->item($i));\r
- }\r
- }\r
- \r
- // strip elements that contain style="display: none;"\r
- $elems = @$xpath->query("//*[contains(@style,'display:none')]", $this->readability->dom);\r
- // check for matches\r
- if ($elems && $elems->length > 0) {\r
- $this->debug('Stripping '.$elems->length.' elements with inline display:none style');\r
- for ($i=$elems->length-1; $i >= 0; $i--) {\r
- $elems->item($i)->parentNode->removeChild($elems->item($i));\r
- }\r
- }\r
- \r
- // try to get body\r
- foreach ($this->config->body as $pattern) {\r
- $elems = @$xpath->query($pattern, $this->readability->dom);\r
- // check for matches\r
- if ($elems && $elems->length > 0) {\r
- $this->debug('Body matched');\r
- $this->debug("...XPath match: $pattern");\r
- if ($elems->length == 1) { \r
- $this->body = $elems->item(0);\r
- // prune (clean up elements that may not be content)\r
- if ($this->config->prune()) {\r
- $this->debug('...pruning content');\r
- $this->readability->prepArticle($this->body);\r
- }\r
- break;\r
- } else {\r
- $this->body = $this->readability->dom->createElement('div');\r
- $this->debug($elems->length.' body elems found');\r
- foreach ($elems as $elem) {\r
- if (!isset($elem->parentNode)) continue;\r
- $isDescendant = false;\r
- foreach ($this->body->childNodes as $parent) {\r
- if ($this->isDescendant($parent, $elem)) {\r
- $isDescendant = true;\r
- break;\r
- }\r
- }\r
- if ($isDescendant) {\r
- $this->debug('...element is child of another body element, skipping.');\r
- } else {\r
- // prune (clean up elements that may not be content)\r
- if ($this->config->prune()) {\r
- $this->debug('Pruning content');\r
- $this->readability->prepArticle($elem);\r
- }\r
- $this->debug('...element added to body');\r
- $this->body->appendChild($elem);\r
- }\r
- }\r
- if ($this->body->hasChildNodes()) break;\r
- }\r
- }\r
- } \r
- \r
- // auto detect?\r
- $detect_title = $detect_body = $detect_author = $detect_date = false;\r
- // detect title?\r
- if (!isset($this->title)) {\r
- if (empty($this->config->title) || $this->config->autodetect_on_failure()) {\r
- $detect_title = true;\r
- }\r
- }\r
- // detect body?\r
- if (!isset($this->body)) {\r
- if (empty($this->config->body) || $this->config->autodetect_on_failure()) {\r
- $detect_body = true;\r
- }\r
- }\r
- // detect author?\r
- if (empty($this->author)) {\r
- if (empty($this->config->author) || $this->config->autodetect_on_failure()) {\r
- $detect_author = true;\r
- }\r
- }\r
- // detect date?\r
- if (!isset($this->date)) {\r
- if (empty($this->config->date) || $this->config->autodetect_on_failure()) {\r
- $detect_date = true;\r
- }\r
- }\r
-\r
- // check for hNews\r
- if ($detect_title || $detect_body) {\r
- // check for hentry\r
- $elems = @$xpath->query("//*[contains(concat(' ',normalize-space(@class),' '),' hentry ')]", $this->readability->dom);\r
- if ($elems && $elems->length > 0) {\r
- $this->debug('hNews: found hentry');\r
- $hentry = $elems->item(0);\r
- \r
- if ($detect_title) {\r
- // check for entry-title\r
- $elems = @$xpath->query(".//*[contains(concat(' ',normalize-space(@class),' '),' entry-title ')]", $hentry);\r
- if ($elems && $elems->length > 0) {\r
- $this->title = $elems->item(0)->textContent;\r
- $this->debug('hNews: found entry-title: '.$this->title);\r
- // remove title from document\r
- $elems->item(0)->parentNode->removeChild($elems->item(0));\r
- $detect_title = false;\r
- }\r
- }\r
- \r
- if ($detect_date) {\r
- // check for time element with pubdate attribute\r
- $elems = @$xpath->query(".//time[@pubdate] | .//abbr[contains(concat(' ',normalize-space(@class),' '),' published ')]", $hentry);\r
- if ($elems && $elems->length > 0) {\r
- $this->date = strtotime(trim($elems->item(0)->textContent));\r
- // remove date from document\r
- //$elems->item(0)->parentNode->removeChild($elems->item(0));\r
- if ($this->date) {\r
- $this->debug('hNews: found publication date: '.date('Y-m-d H:i:s', $this->date));\r
- $detect_date = false;\r
- } else {\r
- $this->date = null;\r
- }\r
- }\r
- }\r
-\r
- if ($detect_author) {\r
- // check for time element with pubdate attribute\r
- $elems = @$xpath->query(".//*[contains(concat(' ',normalize-space(@class),' '),' vcard ') and (contains(concat(' ',normalize-space(@class),' '),' author ') or contains(concat(' ',normalize-space(@class),' '),' byline '))]", $hentry);\r
- if ($elems && $elems->length > 0) {\r
- $author = $elems->item(0);\r
- $fn = @$xpath->query(".//*[contains(concat(' ',normalize-space(@class),' '),' fn ')]", $author);\r
- if ($fn && $fn->length > 0) {\r
- foreach ($fn as $_fn) {\r
- if (trim($_fn->textContent) != '') {\r
- $this->author[] = trim($_fn->textContent);\r
- $this->debug('hNews: found author: '.trim($_fn->textContent));\r
- }\r
- }\r
- } else {\r
- if (trim($author->textContent) != '') {\r
- $this->author[] = trim($author->textContent);\r
- $this->debug('hNews: found author: '.trim($author->textContent));\r
- }\r
- }\r
- $detect_author = empty($this->author);\r
- }\r
- }\r
- \r
- // check for entry-content.\r
- // according to hAtom spec, if there are multiple elements marked entry-content,\r
- // we include all of these in the order they appear - see http://microformats.org/wiki/hatom#Entry_Content\r
- if ($detect_body) {\r
- $elems = @$xpath->query(".//*[contains(concat(' ',normalize-space(@class),' '),' entry-content ')]", $hentry);\r
- if ($elems && $elems->length > 0) {\r
- $this->debug('hNews: found entry-content');\r
- if ($elems->length == 1) {\r
- // what if it's empty? (some sites misuse hNews - place their content outside an empty entry-content element)\r
- $e = $elems->item(0);\r
- if (($e->tagName == 'img') || (trim($e->textContent) != '')) {\r
- $this->body = $elems->item(0);\r
- // prune (clean up elements that may not be content)\r
- if ($this->config->prune()) {\r
- $this->debug('Pruning content');\r
- $this->readability->prepArticle($this->body);\r
- }\r
- $detect_body = false;\r
- } else {\r
- $this->debug('hNews: skipping entry-content - appears not to contain content');\r
- }\r
- unset($e);\r
- } else {\r
- $this->body = $this->readability->dom->createElement('div');\r
- $this->debug($elems->length.' entry-content elems found');\r
- foreach ($elems as $elem) {\r
- if (!isset($elem->parentNode)) continue;\r
- $isDescendant = false;\r
- foreach ($this->body->childNodes as $parent) {\r
- if ($this->isDescendant($parent, $elem)) {\r
- $isDescendant = true;\r
- break;\r
- }\r
- }\r
- if ($isDescendant) {\r
- $this->debug('Element is child of another body element, skipping.');\r
- } else {\r
- // prune (clean up elements that may not be content)\r
- if ($this->config->prune()) {\r
- $this->debug('Pruning content');\r
- $this->readability->prepArticle($elem);\r
- } \r
- $this->debug('Element added to body'); \r
- $this->body->appendChild($elem);\r
- }\r
- }\r
- $detect_body = false;\r
- }\r
- }\r
- }\r
- }\r
- }\r
-\r
- // check for elements marked with instapaper_title\r
- if ($detect_title) {\r
- // check for instapaper_title\r
- $elems = @$xpath->query("//*[contains(concat(' ',normalize-space(@class),' '),' instapaper_title ')]", $this->readability->dom);\r
- if ($elems && $elems->length > 0) {\r
- $this->title = $elems->item(0)->textContent;\r
- $this->debug('Title found (.instapaper_title): '.$this->title);\r
- // remove title from document\r
- $elems->item(0)->parentNode->removeChild($elems->item(0));\r
- $detect_title = false;\r
- }\r
- }\r
- // check for elements marked with instapaper_body\r
- if ($detect_body) {\r
- $elems = @$xpath->query("//*[contains(concat(' ',normalize-space(@class),' '),' instapaper_body ')]", $this->readability->dom);\r
- if ($elems && $elems->length > 0) {\r
- $this->debug('body found (.instapaper_body)');\r
- $this->body = $elems->item(0);\r
- // prune (clean up elements that may not be content)\r
- if ($this->config->prune()) {\r
- $this->debug('Pruning content');\r
- $this->readability->prepArticle($this->body);\r
- }\r
- $detect_body = false;\r
- }\r
- }\r
- \r
- // Find author in rel="author" marked element\r
- // We only use this if there's exactly one.\r
- // If there's more than one, it could indicate more than\r
- // one author, but it could also indicate that we're processing\r
- // a page listing different articles with different authors.\r
- if ($detect_author) {\r
- $elems = @$xpath->query("//a[contains(concat(' ',normalize-space(@rel),' '),' author ')]", $this->readability->dom);\r
- if ($elems && $elems->length == 1) {\r
- $author = trim($elems->item(0)->textContent);\r
- if ($author != '') {\r
- $this->debug("Author found (rel=\"author\"): $author");\r
- $this->author[] = $author;\r
- $detect_author = false;\r
- }\r
- }\r
- }\r
-\r
- // Find date in pubdate marked time element\r
- // For the same reason given above, we only use this\r
- // if there's exactly one element.\r
- if ($detect_date) {\r
- $elems = @$xpath->query("//time[@pubdate]", $this->readability->dom);\r
- if ($elems && $elems->length == 1) {\r
- $this->date = strtotime(trim($elems->item(0)->textContent));\r
- // remove date from document\r
- //$elems->item(0)->parentNode->removeChild($elems->item(0));\r
- if ($this->date) {\r
- $this->debug('Date found (pubdate marked time element): '.date('Y-m-d H:i:s', $this->date));\r
- $detect_date = false;\r
- } else {\r
- $this->date = null;\r
- }\r
- }\r
- }\r
-\r
- // still missing title or body, so we detect using Readability\r
- if ($detect_title || $detect_body) {\r
- $this->debug('Using Readability');\r
- // clone body if we're only using Readability for title (otherwise it may interfere with body element)\r
- if (isset($this->body)) $this->body = $this->body->cloneNode(true);\r
- $success = $this->readability->init();\r
- }\r
- if ($detect_title) {\r
- $this->debug('Detecting title');\r
- $this->title = $this->readability->getTitle()->textContent;\r
- }\r
- if ($detect_body && $success) {\r
- $this->debug('Detecting body');\r
- $this->body = $this->readability->getContent();\r
- if ($this->body->childNodes->length == 1 && $this->body->firstChild->nodeType === XML_ELEMENT_NODE) {\r
- $this->body = $this->body->firstChild;\r
- }\r
- // prune (clean up elements that may not be content)\r
- if ($this->config->prune()) {\r
- $this->debug('Pruning content');\r
- $this->readability->prepArticle($this->body);\r
- }\r
- }\r
- if (isset($this->body)) {\r
- // remove scripts\r
- $this->readability->removeScripts($this->body);\r
- // remove any h1-h6 elements that appear as first thing in the body\r
- // and which match our title\r
- if (isset($this->title) && ($this->title != '')) {\r
- $firstChild = $this->body->firstChild;\r
- while ($firstChild->nodeType && ($firstChild->nodeType !== XML_ELEMENT_NODE)) {\r
- $firstChild = $firstChild->nextSibling;\r
- }\r
- if (($firstChild->nodeType === XML_ELEMENT_NODE)\r
- && in_array(strtolower($firstChild->tagName), array('h1', 'h2', 'h3', 'h4', 'h5', 'h6'))\r
- && (strtolower(trim($firstChild->textContent)) == strtolower(trim($this->title)))) {\r
- $this->body->removeChild($firstChild);\r
- }\r
- }\r
- // prevent self-closing iframes\r
- $elems = $this->body->getElementsByTagName('iframe');\r
- for ($i = $elems->length-1; $i >= 0; $i--) {\r
- $e = $elems->item($i);\r
- if (!$e->hasChildNodes()) {\r
- $e->appendChild($this->body->ownerDocument->createTextNode('[embedded content]'));\r
- }\r
- }\r
- // remove image lazy loading - WordPress plugin http://wordpress.org/extend/plugins/lazy-load/\r
- // the plugin replaces the src attribute to point to a 1x1 gif and puts the original src\r
- // inside the data-lazy-src attribute. It also places the original image inside a noscript element \r
- // next to the amended one.\r
- $elems = @$xpath->query("//img[@data-lazy-src]", $this->body);\r
- for ($i = $elems->length-1; $i >= 0; $i--) {\r
- $e = $elems->item($i);\r
- // let's see if we can grab image from noscript\r
- if ($e->nextSibling !== null && $e->nextSibling->nodeName === 'noscript') {\r
- $_new_elem = $e->ownerDocument->createDocumentFragment();\r
- @$_new_elem->appendXML($e->nextSibling->innerHTML);\r
- $e->nextSibling->parentNode->replaceChild($_new_elem, $e->nextSibling);\r
- $e->parentNode->removeChild($e);\r
- } else {\r
- // Use data-lazy-src as src value\r
- $e->setAttribute('src', $e->getAttribute('data-lazy-src'));\r
- $e->removeAttribute('data-lazy-src');\r
- }\r
- }\r
- \r
- $this->success = true;\r
- }\r
- \r
- // if we've had no success and we've used tidy, there's a chance\r
- // that tidy has messed up. So let's try again without tidy...\r
- if (!$this->success && $tidied && $smart_tidy) {\r
- $this->debug('Trying again without tidy');\r
- $this->process($original_html, $url, false);\r
- }\r
-\r
- return $this->success;\r
- }\r
- \r
- private function isDescendant(DOMElement $parent, DOMElement $child) {\r
- $node = $child->parentNode;\r
- while ($node != null) {\r
- if ($node->isSameNode($parent)) return true;\r
- $node = $node->parentNode;\r
- }\r
- return false;\r
- }\r
-\r
- public function getContent() {\r
- return $this->body;\r
- }\r
- \r
- public function getTitle() {\r
- return $this->title;\r
- }\r
- \r
- public function getAuthors() {\r
- return $this->author;\r
- }\r
- \r
- public function getLanguage() {\r
- return $this->language;\r
- }\r
- \r
- public function getDate() {\r
- return $this->date;\r
- }\r
- \r
- public function getSiteConfig() {\r
- return $this->config;\r
- }\r
- \r
- public function getNextPageUrl() {\r
- return $this->nextPageUrl;\r
- }\r
-}\r
-?>
\ No newline at end of file
+<?php
+/**
+ * Content Extractor
+ *
+ * Uses patterns specified in site config files and auto detection (hNews/PHP Readability)
+ * to extract content from HTML files.
+ *
+ * @version 1.0
+ * @date 2013-02-05
+ * @author Keyvan Minoukadeh
+ * @copyright 2013 Keyvan Minoukadeh
+ * @license http://www.gnu.org/licenses/agpl-3.0.html AGPL v3
+ */
+
+class ContentExtractor
+{
+ protected static $tidy_config = array(
+ 'clean' => true,
+ 'output-xhtml' => true,
+ 'logical-emphasis' => true,
+ 'show-body-only' => false,
+ 'new-blocklevel-tags' => 'article, aside, footer, header, hgroup, menu, nav, section, details, datagrid',
+ 'new-inline-tags' => 'mark, time, meter, progress, data',
+ 'wrap' => 0,
+ 'drop-empty-paras' => true,
+ 'drop-proprietary-attributes' => false,
+ 'enclose-text' => true,
+ 'enclose-block-text' => true,
+ 'merge-divs' => true,
+ 'merge-spans' => true,
+ 'char-encoding' => 'utf8',
+ 'hide-comments' => true
+ );
+ protected $html;
+ protected $config;
+ protected $title;
+ protected $author = array();
+ protected $language;
+ protected $date;
+ protected $body;
+ protected $success = false;
+ protected $nextPageUrl;
+ public $allowedParsers = array('libxml', 'html5lib');
+ public $fingerprints = array();
+ public $readability;
+ public $debug = false;
+ public $debugVerbose = false;
+
+ function __construct($path, $fallback=null) {
+ SiteConfig::set_config_path($path, $fallback);
+ }
+
+ protected function debug($msg) {
+ if ($this->debug) {
+ $mem = round(memory_get_usage()/1024, 2);
+ $memPeak = round(memory_get_peak_usage()/1024, 2);
+ echo '* ',$msg;
+ if ($this->debugVerbose) echo ' - mem used: ',$mem," (peak: $memPeak)";
+ echo "\n";
+ ob_flush();
+ flush();
+ }
+ }
+
+ public function reset() {
+ $this->html = null;
+ $this->readability = null;
+ $this->config = null;
+ $this->title = null;
+ $this->body = null;
+ $this->author = array();
+ $this->language = null;
+ $this->date = null;
+ $this->nextPageUrl = null;
+ $this->success = false;
+ }
+
+ public function findHostUsingFingerprints($html) {
+ $this->debug('Checking fingerprints...');
+ $head = substr($html, 0, 8000);
+ foreach ($this->fingerprints as $_fp => $_fphost) {
+ $lookin = 'html';
+ if (is_array($_fphost)) {
+ if (isset($_fphost['head']) && $_fphost['head']) {
+ $lookin = 'head';
+ }
+ $_fphost = $_fphost['hostname'];
+ }
+ if (strpos($$lookin, $_fp) !== false) {
+ $this->debug("Found match: $_fphost");
+ return $_fphost;
+ }
+ }
+ $this->debug('No fingerprint matches');
+ return false;
+ }
+
+ // returns SiteConfig instance (joined in order: exact match, wildcard, fingerprint, global, default)
+ public function buildSiteConfig($url, $html='', $add_to_cache=true) {
+ // extract host name
+ $host = @parse_url($url, PHP_URL_HOST);
+ $host = strtolower($host);
+ if (substr($host, 0, 4) == 'www.') $host = substr($host, 4);
+ // is merged version already cached?
+ if (SiteConfig::is_cached("$host.merged")) {
+ $this->debug("Returning cached and merged site config for $host");
+ return SiteConfig::build("$host.merged");
+ }
+ // let's build from site_config/custom/ and standard/
+ $config = SiteConfig::build($host);
+ if ($add_to_cache && $config && !SiteConfig::is_cached("$host")) {
+ SiteConfig::add_to_cache($host, $config);
+ }
+ // if no match, use defaults
+ if (!$config) $config = new SiteConfig();
+ // load fingerprint config?
+ if ($config->autodetect_on_failure()) {
+ // check HTML for fingerprints
+ if (!empty($this->fingerprints) && ($_fphost = $this->findHostUsingFingerprints($html))) {
+ if ($config_fingerprint = SiteConfig::build($_fphost)) {
+ $this->debug("Appending site config settings from $_fphost (fingerprint match)");
+ $config->append($config_fingerprint);
+ if ($add_to_cache && !SiteConfig::is_cached($_fphost)) {
+ //$config_fingerprint->cache_in_apc = true;
+ SiteConfig::add_to_cache($_fphost, $config_fingerprint);
+ }
+ }
+ }
+ }
+ // load global config?
+ if ($config->autodetect_on_failure()) {
+ if ($config_global = SiteConfig::build('global', true)) {
+ $this->debug('Appending site config settings from global.txt');
+ $config->append($config_global);
+ if ($add_to_cache && !SiteConfig::is_cached('global')) {
+ //$config_global->cache_in_apc = true;
+ SiteConfig::add_to_cache('global', $config_global);
+ }
+ }
+ }
+ // store copy of merged config
+ if ($add_to_cache) {
+ // do not store in APC if wildcard match
+ $use_apc = ($host == $config->cache_key);
+ $config->cache_key = null;
+ SiteConfig::add_to_cache("$host.merged", $config, $use_apc);
+ }
+ return $config;
+ }
+
+ // returns true on success, false on failure
+ // $smart_tidy indicates that if tidy is used and no results are produced, we will
+ // try again without it. Tidy helps us deal with PHP's patchy HTML parsing most of the time
+ // but it has problems of its own which we try to avoid with this option.
+ public function process($html, $url, $smart_tidy=true) {
+ $this->reset();
+ $this->config = $this->buildSiteConfig($url, $html);
+
+ // do string replacements
+ if (!empty($this->config->find_string)) {
+ if (count($this->config->find_string) == count($this->config->replace_string)) {
+ $html = str_replace($this->config->find_string, $this->config->replace_string, $html, $_count);
+ $this->debug("Strings replaced: $_count (find_string and/or replace_string)");
+ } else {
+ $this->debug('Skipped string replacement - incorrect number of find-replace strings in site config');
+ }
+ unset($_count);
+ }
+
+ // use tidy (if it exists)?
+ // This fixes problems with some sites which would otherwise
+ // trouble DOMDocument's HTML parsing. (Although sometimes it
+ // makes matters worse, which is why you can override it in site config files.)
+ $tidied = false;
+ if ($this->config->tidy() && function_exists('tidy_parse_string') && $smart_tidy) {
+ $this->debug('Using Tidy');
+ $tidy = tidy_parse_string($html, self::$tidy_config, 'UTF8');
+ if (tidy_clean_repair($tidy)) {
+ $original_html = $html;
+ $tidied = true;
+ $html = $tidy->value;
+ }
+ unset($tidy);
+ }
+
+ // load and parse html
+ $_parser = $this->config->parser();
+ if (!in_array($_parser, $this->allowedParsers)) {
+ $this->debug("HTML parser $_parser not listed, using libxml instead");
+ $_parser = 'libxml';
+ }
+ $this->debug("Attempting to parse HTML with $_parser");
+ $this->readability = new Readability($html, $url, $_parser);
+
+ // we use xpath to find elements in the given HTML document
+ // see http://en.wikipedia.org/wiki/XPath_1.0
+ $xpath = new DOMXPath($this->readability->dom);
+
+ // try to get next page link
+ foreach ($this->config->next_page_link as $pattern) {
+ $elems = @$xpath->evaluate($pattern, $this->readability->dom);
+ if (is_string($elems)) {
+ $this->nextPageUrl = trim($elems);
+ break;
+ } elseif ($elems instanceof DOMNodeList && $elems->length > 0) {
+ foreach ($elems as $item) {
+ if ($item instanceof DOMElement && $item->hasAttribute('href')) {
+ $this->nextPageUrl = $item->getAttribute('href');
+ break 2;
+ } elseif ($item instanceof DOMAttr && $item->value) {
+ $this->nextPageUrl = $item->value;
+ break 2;
+ }
+ }
+ }
+ }
+
+ // try to get title
+ foreach ($this->config->title as $pattern) {
+ // $this->debug("Trying $pattern");
+ $elems = @$xpath->evaluate($pattern, $this->readability->dom);
+ if (is_string($elems)) {
+ $this->title = trim($elems);
+ $this->debug('Title expression evaluated as string: '.$this->title);
+ $this->debug("...XPath match: $pattern");
+ break;
+ } elseif ($elems instanceof DOMNodeList && $elems->length > 0) {
+ $this->title = $elems->item(0)->textContent;
+ $this->debug('Title matched: '.$this->title);
+ $this->debug("...XPath match: $pattern");
+ // remove title from document
+ try {
+ @$elems->item(0)->parentNode->removeChild($elems->item(0));
+ } catch (DOMException $e) {
+ // do nothing
+ }
+ break;
+ }
+ }
+
+ // try to get author (if it hasn't already been set)
+ if (empty($this->author)) {
+ foreach ($this->config->author as $pattern) {
+ $elems = @$xpath->evaluate($pattern, $this->readability->dom);
+ if (is_string($elems)) {
+ if (trim($elems) != '') {
+ $this->author[] = trim($elems);
+ $this->debug('Author expression evaluated as string: '.trim($elems));
+ $this->debug("...XPath match: $pattern");
+ break;
+ }
+ } elseif ($elems instanceof DOMNodeList && $elems->length > 0) {
+ foreach ($elems as $elem) {
+ if (!isset($elem->parentNode)) continue;
+ $this->author[] = trim($elem->textContent);
+ $this->debug('Author matched: '.trim($elem->textContent));
+ }
+ if (!empty($this->author)) {
+ $this->debug("...XPath match: $pattern");
+ break;
+ }
+ }
+ }
+ }
+
+ // try to get language
+ $_lang_xpath = array('//html[@lang]/@lang', '//meta[@name="DC.language"]/@content');
+ foreach ($_lang_xpath as $pattern) {
+ $elems = @$xpath->evaluate($pattern, $this->readability->dom);
+ if (is_string($elems)) {
+ if (trim($elems) != '') {
+ $this->language = trim($elems);
+ $this->debug('Language matched: '.$this->language);
+ break;
+ }
+ } elseif ($elems instanceof DOMNodeList && $elems->length > 0) {
+ foreach ($elems as $elem) {
+ if (!isset($elem->parentNode)) continue;
+ $this->language = trim($elem->textContent);
+ $this->debug('Language matched: '.$this->language);
+ }
+ if ($this->language) break;
+ }
+ }
+
+ // try to get date
+ foreach ($this->config->date as $pattern) {
+ $elems = @$xpath->evaluate($pattern, $this->readability->dom);
+ if (is_string($elems)) {
+ $this->date = strtotime(trim($elems, "; \t\n\r\0\x0B"));
+ } elseif ($elems instanceof DOMNodeList && $elems->length > 0) {
+ $this->date = $elems->item(0)->textContent;
+ $this->date = strtotime(trim($this->date, "; \t\n\r\0\x0B"));
+ // remove date from document
+ // $elems->item(0)->parentNode->removeChild($elems->item(0));
+ }
+ if (!$this->date) {
+ $this->date = null;
+ } else {
+ $this->debug('Date matched: '.date('Y-m-d H:i:s', $this->date));
+ $this->debug("...XPath match: $pattern");
+ break;
+ }
+ }
+
+ // strip elements (using xpath expressions)
+ foreach ($this->config->strip as $pattern) {
+ $elems = @$xpath->query($pattern, $this->readability->dom);
+ // check for matches
+ if ($elems && $elems->length > 0) {
+ $this->debug('Stripping '.$elems->length.' elements (strip)');
+ for ($i=$elems->length-1; $i >= 0; $i--) {
+ $elems->item($i)->parentNode->removeChild($elems->item($i));
+ }
+ }
+ }
+
+ // strip elements (using id and class attribute values)
+ foreach ($this->config->strip_id_or_class as $string) {
+ $string = strtr($string, array("'"=>'', '"'=>''));
+ $elems = @$xpath->query("//*[contains(@class, '$string') or contains(@id, '$string')]", $this->readability->dom);
+ // check for matches
+ if ($elems && $elems->length > 0) {
+ $this->debug('Stripping '.$elems->length.' elements (strip_id_or_class)');
+ for ($i=$elems->length-1; $i >= 0; $i--) {
+ $elems->item($i)->parentNode->removeChild($elems->item($i));
+ }
+ }
+ }
+
+ // strip images (using src attribute values)
+ foreach ($this->config->strip_image_src as $string) {
+ $string = strtr($string, array("'"=>'', '"'=>''));
+ $elems = @$xpath->query("//img[contains(@src, '$string')]", $this->readability->dom);
+ // check for matches
+ if ($elems && $elems->length > 0) {
+ $this->debug('Stripping '.$elems->length.' image elements');
+ for ($i=$elems->length-1; $i >= 0; $i--) {
+ $elems->item($i)->parentNode->removeChild($elems->item($i));
+ }
+ }
+ }
+ // strip elements using Readability.com and Instapaper.com ignore class names
+ // .entry-unrelated and .instapaper_ignore
+ // See https://www.readability.com/publishers/guidelines/#view-plainGuidelines
+ // and http://blog.instapaper.com/post/730281947
+ $elems = @$xpath->query("//*[contains(concat(' ',normalize-space(@class),' '),' entry-unrelated ') or contains(concat(' ',normalize-space(@class),' '),' instapaper_ignore ')]", $this->readability->dom);
+ // check for matches
+ if ($elems && $elems->length > 0) {
+ $this->debug('Stripping '.$elems->length.' .entry-unrelated,.instapaper_ignore elements');
+ for ($i=$elems->length-1; $i >= 0; $i--) {
+ $elems->item($i)->parentNode->removeChild($elems->item($i));
+ }
+ }
+
+ // strip elements that contain style="display: none;"
+ $elems = @$xpath->query("//*[contains(@style,'display:none')]", $this->readability->dom);
+ // check for matches
+ if ($elems && $elems->length > 0) {
+ $this->debug('Stripping '.$elems->length.' elements with inline display:none style');
+ for ($i=$elems->length-1; $i >= 0; $i--) {
+ $elems->item($i)->parentNode->removeChild($elems->item($i));
+ }
+ }
+
+ // try to get body
+ foreach ($this->config->body as $pattern) {
+ $elems = @$xpath->query($pattern, $this->readability->dom);
+ // check for matches
+ if ($elems && $elems->length > 0) {
+ $this->debug('Body matched');
+ $this->debug("...XPath match: $pattern");
+ if ($elems->length == 1) {
+ $this->body = $elems->item(0);
+ // prune (clean up elements that may not be content)
+ if ($this->config->prune()) {
+ $this->debug('...pruning content');
+ $this->readability->prepArticle($this->body);
+ }
+ break;
+ } else {
+ $this->body = $this->readability->dom->createElement('div');
+ $this->debug($elems->length.' body elems found');
+ foreach ($elems as $elem) {
+ if (!isset($elem->parentNode)) continue;
+ $isDescendant = false;
+ foreach ($this->body->childNodes as $parent) {
+ if ($this->isDescendant($parent, $elem)) {
+ $isDescendant = true;
+ break;
+ }
+ }
+ if ($isDescendant) {
+ $this->debug('...element is child of another body element, skipping.');
+ } else {
+ // prune (clean up elements that may not be content)
+ if ($this->config->prune()) {
+ $this->debug('Pruning content');
+ $this->readability->prepArticle($elem);
+ }
+ $this->debug('...element added to body');
+ $this->body->appendChild($elem);
+ }
+ }
+ if ($this->body->hasChildNodes()) break;
+ }
+ }
+ }
+
+ // auto detect?
+ $detect_title = $detect_body = $detect_author = $detect_date = false;
+ // detect title?
+ if (!isset($this->title)) {
+ if (empty($this->config->title) || $this->config->autodetect_on_failure()) {
+ $detect_title = true;
+ }
+ }
+ // detect body?
+ if (!isset($this->body)) {
+ if (empty($this->config->body) || $this->config->autodetect_on_failure()) {
+ $detect_body = true;
+ }
+ }
+ // detect author?
+ if (empty($this->author)) {
+ if (empty($this->config->author) || $this->config->autodetect_on_failure()) {
+ $detect_author = true;
+ }
+ }
+ // detect date?
+ if (!isset($this->date)) {
+ if (empty($this->config->date) || $this->config->autodetect_on_failure()) {
+ $detect_date = true;
+ }
+ }
+
+ // check for hNews
+ if ($detect_title || $detect_body) {
+ // check for hentry
+ $elems = @$xpath->query("//*[contains(concat(' ',normalize-space(@class),' '),' hentry ')]", $this->readability->dom);
+ if ($elems && $elems->length > 0) {
+ $this->debug('hNews: found hentry');
+ $hentry = $elems->item(0);
+
+ if ($detect_title) {
+ // check for entry-title
+ $elems = @$xpath->query(".//*[contains(concat(' ',normalize-space(@class),' '),' entry-title ')]", $hentry);
+ if ($elems && $elems->length > 0) {
+ $this->title = $elems->item(0)->textContent;
+ $this->debug('hNews: found entry-title: '.$this->title);
+ // remove title from document
+ $elems->item(0)->parentNode->removeChild($elems->item(0));
+ $detect_title = false;
+ }
+ }
+
+ if ($detect_date) {
+ // check for time element with pubdate attribute
+ $elems = @$xpath->query(".//time[@pubdate] | .//abbr[contains(concat(' ',normalize-space(@class),' '),' published ')]", $hentry);
+ if ($elems && $elems->length > 0) {
+ $this->date = strtotime(trim($elems->item(0)->textContent));
+ // remove date from document
+ //$elems->item(0)->parentNode->removeChild($elems->item(0));
+ if ($this->date) {
+ $this->debug('hNews: found publication date: '.date('Y-m-d H:i:s', $this->date));
+ $detect_date = false;
+ } else {
+ $this->date = null;
+ }
+ }
+ }
+
+ if ($detect_author) {
+ // check for time element with pubdate attribute
+ $elems = @$xpath->query(".//*[contains(concat(' ',normalize-space(@class),' '),' vcard ') and (contains(concat(' ',normalize-space(@class),' '),' author ') or contains(concat(' ',normalize-space(@class),' '),' byline '))]", $hentry);
+ if ($elems && $elems->length > 0) {
+ $author = $elems->item(0);
+ $fn = @$xpath->query(".//*[contains(concat(' ',normalize-space(@class),' '),' fn ')]", $author);
+ if ($fn && $fn->length > 0) {
+ foreach ($fn as $_fn) {
+ if (trim($_fn->textContent) != '') {
+ $this->author[] = trim($_fn->textContent);
+ $this->debug('hNews: found author: '.trim($_fn->textContent));
+ }
+ }
+ } else {
+ if (trim($author->textContent) != '') {
+ $this->author[] = trim($author->textContent);
+ $this->debug('hNews: found author: '.trim($author->textContent));
+ }
+ }
+ $detect_author = empty($this->author);
+ }
+ }
+
+ // check for entry-content.
+ // according to hAtom spec, if there are multiple elements marked entry-content,
+ // we include all of these in the order they appear - see http://microformats.org/wiki/hatom#Entry_Content
+ if ($detect_body) {
+ $elems = @$xpath->query(".//*[contains(concat(' ',normalize-space(@class),' '),' entry-content ')]", $hentry);
+ if ($elems && $elems->length > 0) {
+ $this->debug('hNews: found entry-content');
+ if ($elems->length == 1) {
+ // what if it's empty? (some sites misuse hNews - place their content outside an empty entry-content element)
+ $e = $elems->item(0);
+ if (($e->tagName == 'img') || (trim($e->textContent) != '')) {
+ $this->body = $elems->item(0);
+ // prune (clean up elements that may not be content)
+ if ($this->config->prune()) {
+ $this->debug('Pruning content');
+ $this->readability->prepArticle($this->body);
+ }
+ $detect_body = false;
+ } else {
+ $this->debug('hNews: skipping entry-content - appears not to contain content');
+ }
+ unset($e);
+ } else {
+ $this->body = $this->readability->dom->createElement('div');
+ $this->debug($elems->length.' entry-content elems found');
+ foreach ($elems as $elem) {
+ if (!isset($elem->parentNode)) continue;
+ $isDescendant = false;
+ foreach ($this->body->childNodes as $parent) {
+ if ($this->isDescendant($parent, $elem)) {
+ $isDescendant = true;
+ break;
+ }
+ }
+ if ($isDescendant) {
+ $this->debug('Element is child of another body element, skipping.');
+ } else {
+ // prune (clean up elements that may not be content)
+ if ($this->config->prune()) {
+ $this->debug('Pruning content');
+ $this->readability->prepArticle($elem);
+ }
+ $this->debug('Element added to body');
+ $this->body->appendChild($elem);
+ }
+ }
+ $detect_body = false;
+ }
+ }
+ }
+ }
+ }
+
+ // check for elements marked with instapaper_title
+ if ($detect_title) {
+ // check for instapaper_title
+ $elems = @$xpath->query("//*[contains(concat(' ',normalize-space(@class),' '),' instapaper_title ')]", $this->readability->dom);
+ if ($elems && $elems->length > 0) {
+ $this->title = $elems->item(0)->textContent;
+ $this->debug('Title found (.instapaper_title): '.$this->title);
+ // remove title from document
+ $elems->item(0)->parentNode->removeChild($elems->item(0));
+ $detect_title = false;
+ }
+ }
+ // check for elements marked with instapaper_body
+ if ($detect_body) {
+ $elems = @$xpath->query("//*[contains(concat(' ',normalize-space(@class),' '),' instapaper_body ')]", $this->readability->dom);
+ if ($elems && $elems->length > 0) {
+ $this->debug('body found (.instapaper_body)');
+ $this->body = $elems->item(0);
+ // prune (clean up elements that may not be content)
+ if ($this->config->prune()) {
+ $this->debug('Pruning content');
+ $this->readability->prepArticle($this->body);
+ }
+ $detect_body = false;
+ }
+ }
+
+ // Find author in rel="author" marked element
+ // We only use this if there's exactly one.
+ // If there's more than one, it could indicate more than
+ // one author, but it could also indicate that we're processing
+ // a page listing different articles with different authors.
+ if ($detect_author) {
+ $elems = @$xpath->query("//a[contains(concat(' ',normalize-space(@rel),' '),' author ')]", $this->readability->dom);
+ if ($elems && $elems->length == 1) {
+ $author = trim($elems->item(0)->textContent);
+ if ($author != '') {
+ $this->debug("Author found (rel=\"author\"): $author");
+ $this->author[] = $author;
+ $detect_author = false;
+ }
+ }
+ }
+
+ // Find date in pubdate marked time element
+ // For the same reason given above, we only use this
+ // if there's exactly one element.
+ if ($detect_date) {
+ $elems = @$xpath->query("//time[@pubdate]", $this->readability->dom);
+ if ($elems && $elems->length == 1) {
+ $this->date = strtotime(trim($elems->item(0)->textContent));
+ // remove date from document
+ //$elems->item(0)->parentNode->removeChild($elems->item(0));
+ if ($this->date) {
+ $this->debug('Date found (pubdate marked time element): '.date('Y-m-d H:i:s', $this->date));
+ $detect_date = false;
+ } else {
+ $this->date = null;
+ }
+ }
+ }
+
+ // still missing title or body, so we detect using Readability
+ if ($detect_title || $detect_body) {
+ $this->debug('Using Readability');
+ // clone body if we're only using Readability for title (otherwise it may interfere with body element)
+ if (isset($this->body)) $this->body = $this->body->cloneNode(true);
+ $success = $this->readability->init();
+ }
+ if ($detect_title) {
+ $this->debug('Detecting title');
+ $this->title = $this->readability->getTitle()->textContent;
+ }
+ if ($detect_body && $success) {
+ $this->debug('Detecting body');
+ $this->body = $this->readability->getContent();
+ if ($this->body->childNodes->length == 1 && $this->body->firstChild->nodeType === XML_ELEMENT_NODE) {
+ $this->body = $this->body->firstChild;
+ }
+ // prune (clean up elements that may not be content)
+ if ($this->config->prune()) {
+ $this->debug('Pruning content');
+ $this->readability->prepArticle($this->body);
+ }
+ }
+ if (isset($this->body)) {
+ // remove scripts
+ $this->readability->removeScripts($this->body);
+ // remove any h1-h6 elements that appear as first thing in the body
+ // and which match our title
+ if (isset($this->title) && ($this->title != '')) {
+ $firstChild = $this->body->firstChild;
+ while ($firstChild->nodeType && ($firstChild->nodeType !== XML_ELEMENT_NODE)) {
+ $firstChild = $firstChild->nextSibling;
+ }
+ if (($firstChild->nodeType === XML_ELEMENT_NODE)
+ && in_array(strtolower($firstChild->tagName), array('h1', 'h2', 'h3', 'h4', 'h5', 'h6'))
+ && (strtolower(trim($firstChild->textContent)) == strtolower(trim($this->title)))) {
+ $this->body->removeChild($firstChild);
+ }
+ }
+ // prevent self-closing iframes
+ $elems = $this->body->getElementsByTagName('iframe');
+ for ($i = $elems->length-1; $i >= 0; $i--) {
+ $e = $elems->item($i);
+ if (!$e->hasChildNodes()) {
+ $e->appendChild($this->body->ownerDocument->createTextNode('[embedded content]'));
+ }
+ }
+ // remove image lazy loading - WordPress plugin http://wordpress.org/extend/plugins/lazy-load/
+ // the plugin replaces the src attribute to point to a 1x1 gif and puts the original src
+ // inside the data-lazy-src attribute. It also places the original image inside a noscript element
+ // next to the amended one.
+ $elems = @$xpath->query("//img[@data-lazy-src]", $this->body);
+ for ($i = $elems->length-1; $i >= 0; $i--) {
+ $e = $elems->item($i);
+ // let's see if we can grab image from noscript
+ if ($e->nextSibling !== null && $e->nextSibling->nodeName === 'noscript') {
+ $_new_elem = $e->ownerDocument->createDocumentFragment();
+ @$_new_elem->appendXML($e->nextSibling->innerHTML);
+ $e->nextSibling->parentNode->replaceChild($_new_elem, $e->nextSibling);
+ $e->parentNode->removeChild($e);
+ } else {
+ // Use data-lazy-src as src value
+ $e->setAttribute('src', $e->getAttribute('data-lazy-src'));
+ $e->removeAttribute('data-lazy-src');
+ }
+ }
+
+ $this->success = true;
+ }
+
+ // if we've had no success and we've used tidy, there's a chance
+ // that tidy has messed up. So let's try again without tidy...
+ if (!$this->success && $tidied && $smart_tidy) {
+ $this->debug('Trying again without tidy');
+ $this->process($original_html, $url, false);
+ }
+
+ return $this->success;
+ }
+
+ private function isDescendant(DOMElement $parent, DOMElement $child) {
+ $node = $child->parentNode;
+ while ($node != null) {
+ if ($node->isSameNode($parent)) return true;
+ $node = $node->parentNode;
+ }
+ return false;
+ }
+
+ public function getContent() {
+ return $this->body;
+ }
+
+ public function getTitle() {
+ return $this->title;
+ }
+
+ public function getAuthors() {
+ return $this->author;
+ }
+
+ public function getLanguage() {
+ return $this->language;
+ }
+
+ public function getDate() {
+ return $this->date;
+ }
+
+ public function getSiteConfig() {
+ return $this->config;
+ }
+
+ public function getNextPageUrl() {
+ return $this->nextPageUrl;
+ }
+}
\ No newline at end of file