parameters['xpath_expressions'] == '') { preg_match("/\(.*)\<\/body\>/isU", $html, $matches); $html = preg_replace('#(.*?)#is', '', $matches[1]); } else { $html = preg_replace('#(.*?)#is', '', $html); } switch ($this->parameters['mode_creation_items']) { case 'all_change': case 'by_change': $html = strip_tags($html); break; case 'all_links': preg_match_all("/\(.*)\<\/a\>/isU", strip_tags($html, ''), $link_matches); $links = array(); foreach ($link_matches[0] as $link_match) { $links[] = $link_match; } $html = implode(PHP_EOL, $links); break; } return $html; } protected function save_content($first=true) { if($first) { $query = "insert into docwatch_datasource_monitoring_website set datasource_monitoring_website_num_datasource = '".$this->id."', "; } else { $query = "update docwatch_datasource_monitoring_website set "; $where = "where datasource_monitoring_website_num_datasource = '".$this->id."' "; } $query .= " datasource_monitoring_website_upload_date = NOW(), datasource_monitoring_website_content = '".addslashes(serialize($this->content))."', datasource_monitoring_website_content_hash = '".addslashes($this->content_hash)."'"; $query .= $where; pmb_mysql_query($query); } protected function get_xdiff_change($xdiff_string) { $xdiff_change = array(); preg_match_all("/@@(.*)@@/isU", $xdiff_string, $matches); for($i=0; $i

"; $has_change = true; } } elseif(substr($line, 0, 1) == '+') { if(strip_tags(trim(substr($line, 1))) != '') { // $change[] = "

+ ".trim(substr($line, 1))."

"; $change[] = "

".trim(substr($line, 1))."

"; $has_change = true; } } else { if(strip_tags($line) != '') { $change[] = "

".$line."

"; } } } } if($has_change) { $xdiff_change[] = $change; } } return $xdiff_change; } protected function get_content_from_link($link) { global $charset; $content_from_link = array(); if($link){ $datas = array(); @ini_set("zend.ze1_compatibility_mode", "0"); $informations = array(); $loaded=false; $aCurl = new Curl(); $aCurl->timeout=2; $content = $aCurl->get($link); $html=$content->body; if($html && $content->headers['Status-Code'] == 200){ $this->content_headers = $content->headers; if(is_array($this->parameters['xpath_expressions']) && count($this->parameters['xpath_expressions'])) { $dom = new DOMDocument(); $old_errors_value = false; if(libxml_use_internal_errors(true)){ $old_errors_value = true; } $loaded = $dom->loadHTML($html); if($loaded) { $xpath = new DOMXPath($dom); foreach ($this->parameters['xpath_expressions'] as $i=>$xpath_expression) { $entries = $xpath->query($xpath_expression); $html_content = $this->clean_html($dom->saveHTML($entries->item(0))); if($this->parameters['xpath_expressions_for_title'][$i]) { $entries = $xpath->query($this->parameters['xpath_expressions_for_title'][$i]); $html_title = $this->clean_html($dom->saveHTML($entries->item(0))); } else { $html_title = ''; } $html_link = ''; if($this->parameters['xpath_expressions_for_link'][$i]) { $entries = $xpath->query($this->parameters['xpath_expressions_for_link'][$i]); preg_match("/\(.*)\<\/a\>/isU", strip_tags($dom->saveHTML($entries->item(0)), ''), $link_matches); if($link_matches[0]) { $html_link = $link_matches[0]; } } $content_from_link[] = array( 'content' => $html_content, 'title' => $html_title, 'link' => $html_link ); } } libxml_use_internal_errors($old_errors_value); } else { $content_from_link[] = array( 'content' => $this->clean_html($html), 'title' => '', 'link' => '' ); } } } //Dom renvoie de l'utf-8. Mais certains caractères windows peuvent être présents dans les pages... if($charset != "utf-8"){ foreach ($content_from_link as $key=>$content) { foreach ($content as $key_content=>$value_content) { $content_from_link[$key][$key_content] = utf8_decode(encoding_normalize::clean_cp1252($value_content,'utf-8')); } } } return $content_from_link; } protected function find_root_url($url) { $tmp = parse_url($url); $url = ($tmp["scheme"]?$tmp["scheme"]."://":"").$tmp["host"]."/"; return $url; } protected function get_constructed_remote_link($link, $link_for_construct = '') { if($link_for_construct) { if(substr($link_for_construct, 0, 7) == 'http://' || substr($link_for_construct, 0, 8) == 'https://') { $link = $link_for_construct; } else { preg_match("/\s+(?:[^\"'>]+|\"[^\"]*\"|'[^']*')*href=(\"[^\"]+\"|'[^']+'|[^<>\s]+)/i", ' '.$link_for_construct.' ', $matches); if($matches[1]) { $match_link = str_replace('"', '', $matches[1]); if(substr($match_link, 0, 7) == 'http://' || substr($match_link, 0, 8) == 'https://') { $link = $match_link; } else { if(strpos($match_link, '/') !== false) { $match_link = substr($match_link, strpos($match_link, '/')+1); } if($this->parameters['use_root_url']) { $link = $this->find_root_url($link).$match_link; } else { $link .= $match_link; } } } else { $link .= $link_for_construct; } } } else { $link .= '#'.strtotime($this->content_headers['Date']).rand(0,1000); } return $link; } protected function get_items_datas($link){ $items = array(); $content_from_link = $this->get_content_from_link($link); $content_hash_from_link = md5(serialize($content_from_link)); if(is_array($content_from_link) && count($content_from_link)) { $query = "select datasource_monitoring_website_upload_date, datasource_monitoring_website_content, datasource_monitoring_website_content_hash from docwatch_datasource_monitoring_website where datasource_monitoring_website_num_datasource = ".$this->id; $result = pmb_mysql_query($query); if($result && pmb_mysql_num_rows($result)) { $row = pmb_mysql_fetch_object($result); if($content_hash_from_link != $row->datasource_monitoring_website_content_hash) { $content_from_base = unserialize($row->datasource_monitoring_website_content); if(is_array($this->parameters['xpath_expressions']) && count($this->parameters['xpath_expressions'])) { $hash_from_base = array(); if(is_array($content_from_base)) { foreach ($content_from_base as $content) { $hash_from_base[] = md5($content['content']); } } foreach ($content_from_link as $i=>$content) { if(!in_array(md5($content['content']), $hash_from_base)) { $items[] = array( 'content' => $content['content'], 'title' => ($content['title'] ? $content['title'] : $this->get_title()), 'link' => $this->get_constructed_remote_link($link, $content['link']) ); } } } else { if(!extension_loaded('xdiff')) { return false; } if(md5($content_from_base[0]['content']) != md5($content_from_link[0]['content'])) { $xdiff_string = xdiff_string_diff($content_from_base[0]['content'] , $content_from_link[0]['content']); $xdiff_change = $this->get_xdiff_change($xdiff_string); if($this->parameters['mode_creation_items'] == 'by_change') { foreach ($xdiff_change as $i=>$change) { $items[] = array( 'content' => implode('', $change), 'title' => $this->get_title(), 'link' => $this->get_constructed_remote_link($link) ); } } else { $item_content = ''; foreach ($xdiff_change as $change) { $item_content .= implode('', $change); } $items[] = array( 'content' => $item_content, 'title' => $this->get_title(), 'link' => $this->get_constructed_remote_link($link) ); } } } foreach ($items as $item) { $data = array(); $data["type"] = "monitoring_website"; $data["title"] = $item['title']; $data["summary"] = $item['content']; $data["content"] = ''; $data["url"] = $item['link']; $data["publication_date"] = date( 'Y-m-d H:i:s', strtotime($this->content_headers['Date'])); $data["logo_url"] = ''; $data["descriptors"] = ""; $data["tags"] = ''; $datas[] = $data; } $this->content = $content_from_link; $this->content_hash = $content_hash_from_link; $this->save_content(false); } } else { $this->content = $content_from_link; $this->content_hash = $content_hash_from_link; $this->save_content(true); } return $datas; }else{ return false; } } public function get_available_selectors(){ global $msg; return array( "docwatch_selector_monitoring_website" => $msg['dsi_docwatch_selector_monitoring_website'] ); } protected function get_expression_xpath_content($parameter_name, $i, $expression) { global $msg; $form = " "; if($i == 0) { $form .= ""; } return $form; } protected function get_xpath_expressions_form($parameter_name) { global $msg,$charset; $form = "
"; if(is_array($this->parameters[$parameter_name]) && count($this->parameters[$parameter_name])) { foreach ($this->parameters[$parameter_name] as $i=>$expression) { $form .= "
".$this->get_expression_xpath_content($parameter_name, $i, $expression)."
"; } $form .= ""; } else { $form .= "
".$this->get_expression_xpath_content($parameter_name, 0, '')."
"; } $form .= "
"; return $form; } public function get_form_content(){ global $msg,$charset; $form = parent::get_form_content(); $form .= "
 
 
".$this->get_xpath_expressions_form('xpath_expressions')."
 
".$this->get_xpath_expressions_form('xpath_expressions_for_title')."
 
".$this->get_xpath_expressions_form('xpath_expressions_for_link')."
 
".$msg['39']."parameters['use_root_url'] ? "":"checked='checked'")." type='radio' data-dojo-type='dijit/form/RadioButton' name='docwatch_datasource_monitoring_website_use_root_url' value='0' />  ".$msg['40']."parameters['use_root_url'] ? "checked='checked'":"")." type='radio' data-dojo-type='dijit/form/RadioButton' name='docwatch_datasource_monitoring_website_use_root_url' value='1' />
"; return $form; } public function set_from_form() { global $docwatch_datasource_monitoring_website_mode_creation_items; global $docwatch_datasource_monitoring_website_xpath_expressions; global $docwatch_datasource_monitoring_website_xpath_expressions_for_title; global $docwatch_datasource_monitoring_website_xpath_expressions_for_link; global $docwatch_datasource_monitoring_website_use_root_url; $this->parameters['mode_creation_items'] = stripslashes($docwatch_datasource_monitoring_website_mode_creation_items); $this->parameters['xpath_expressions'] = array(); foreach ($docwatch_datasource_monitoring_website_xpath_expressions as $xpath_expression) { if($xpath_expression) { $this->parameters['xpath_expressions'][] = stripslashes($xpath_expression); } } $this->parameters['xpath_expressions_for_title'] = array(); foreach ($docwatch_datasource_monitoring_website_xpath_expressions_for_title as $xpath_expression_for_title) { if($xpath_expression_for_title) { $this->parameters['xpath_expressions_for_title'][] = stripslashes($xpath_expression_for_title); } } $this->parameters['xpath_expressions_for_link'] = array(); foreach ($docwatch_datasource_monitoring_website_xpath_expressions_for_link as $xpath_expression_for_link) { if($xpath_expression_for_link) { $this->parameters['xpath_expressions_for_link'][] = stripslashes($xpath_expression_for_link); } } $this->parameters['use_root_url'] = $docwatch_datasource_monitoring_website_use_root_url; parent::set_from_form(); } } // end of docwatch_datasource_monitoring_website