2017-02-08 12:21:59 +01:00
|
|
|
<?php
|
2017-02-11 16:16:56 +01:00
|
|
|
class FB2Bridge extends BridgeAbstract {
|
2017-02-08 12:21:59 +01:00
|
|
|
|
2017-02-11 16:16:56 +01:00
|
|
|
const MAINTAINER = 'teromene';
|
|
|
|
const NAME = 'Facebook Alternate';
|
|
|
|
const URI = 'https://www.facebook.com/';
|
2017-02-08 12:21:59 +01:00
|
|
|
const CACHE_TIMEOUT = 1000;
|
2017-02-11 16:16:56 +01:00
|
|
|
const DESCRIPTION = 'Input a page title or a profile log. For a profile log,
|
|
|
|
please insert the parameter as follow : myExamplePage/132621766841117';
|
2017-02-08 12:21:59 +01:00
|
|
|
|
2017-02-11 16:16:56 +01:00
|
|
|
const PARAMETERS = array( array(
|
|
|
|
'u' => array(
|
|
|
|
'name' => 'Username',
|
|
|
|
'required' => true
|
|
|
|
)
|
|
|
|
));
|
2017-02-08 12:21:59 +01:00
|
|
|
|
2017-02-11 16:16:56 +01:00
|
|
|
public function collectData(){
|
2017-02-08 12:21:59 +01:00
|
|
|
|
2017-02-11 16:16:56 +01:00
|
|
|
function extractFromDelimiters($string, $start, $end){
|
2017-07-29 19:28:00 +02:00
|
|
|
if(strpos($string, $start) !== false) {
|
2017-02-08 12:21:59 +01:00
|
|
|
$section_retrieved = substr($string, strpos($string, $start) + strlen($start));
|
|
|
|
$section_retrieved = substr($section_retrieved, 0, strpos($section_retrieved, $end));
|
|
|
|
return $section_retrieved;
|
2017-02-11 16:16:56 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
2017-02-08 12:21:59 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
//Utility function for cleaning a Facebook link
|
2017-02-11 16:16:56 +01:00
|
|
|
$unescape_fb_link = function($matches){
|
2017-07-29 19:28:00 +02:00
|
|
|
if(is_array($matches) && count($matches) > 1) {
|
2017-02-08 12:21:59 +01:00
|
|
|
$link = $matches[1];
|
2017-02-11 16:16:56 +01:00
|
|
|
if(strpos($link, '/') === 0)
|
|
|
|
$link = self::URI . $link . '"';
|
|
|
|
if(strpos($link, 'facebook.com/l.php?u=') !== false)
|
|
|
|
$link = urldecode(extractFromDelimiters($link, 'facebook.com/l.php?u=', '&'));
|
|
|
|
return ' href="' . $link . '"';
|
2017-02-08 12:21:59 +01:00
|
|
|
}
|
|
|
|
};
|
|
|
|
|
|
|
|
//Utility function for converting facebook emoticons
|
2017-02-11 16:16:56 +01:00
|
|
|
$unescape_fb_emote = function($matches){
|
2017-02-08 12:21:59 +01:00
|
|
|
static $facebook_emoticons = array(
|
|
|
|
'smile' => ':)',
|
|
|
|
'frown' => ':(',
|
|
|
|
'tongue' => ':P',
|
|
|
|
'grin' => ':D',
|
|
|
|
'gasp' => ':O',
|
|
|
|
'wink' => ';)',
|
|
|
|
'pacman' => ':<',
|
|
|
|
'grumpy' => '>_<',
|
|
|
|
'unsure' => ':/',
|
|
|
|
'cry' => ':\'(',
|
|
|
|
'kiki' => '^_^',
|
|
|
|
'glasses' => '8-)',
|
|
|
|
'sunglasses' => 'B-)',
|
|
|
|
'heart' => '<3',
|
|
|
|
'devil' => ']:D',
|
|
|
|
'angel' => '0:)',
|
|
|
|
'squint' => '-_-',
|
|
|
|
'confused' => 'o_O',
|
|
|
|
'upset' => 'xD',
|
|
|
|
'colonthree' => ':3',
|
|
|
|
'like' => '👍');
|
|
|
|
$len = count($matches);
|
|
|
|
if ($len > 1)
|
|
|
|
for ($i = 1; $i < $len; $i++)
|
|
|
|
foreach ($facebook_emoticons as $name => $emote)
|
|
|
|
if ($matches[$i] === $name)
|
|
|
|
return $emote;
|
|
|
|
return $matches[0];
|
|
|
|
};
|
|
|
|
|
2017-07-29 19:28:00 +02:00
|
|
|
if($this->getInput('u') !== null) {
|
2017-02-11 16:16:56 +01:00
|
|
|
$page = 'https://touch.facebook.com/' . $this->getInput('u');
|
2017-02-08 12:21:59 +01:00
|
|
|
$cookies = $this->getCookies($page);
|
|
|
|
$pageID = $this->getPageID($page, $cookies);
|
|
|
|
|
2017-07-29 19:28:00 +02:00
|
|
|
if($pageID === null) {
|
2017-02-11 16:16:56 +01:00
|
|
|
echo <<<EOD
|
|
|
|
Unable to get the page id. You should consider getting the ID by hand, then importing it into FB2Bridge
|
2017-06-15 12:51:11 +02:00
|
|
|
EOD;
|
|
|
|
die();
|
|
|
|
} elseif($pageID == -1) {
|
|
|
|
echo <<<EOD
|
|
|
|
This page is not accessible without being logged in.
|
2017-02-11 16:16:56 +01:00
|
|
|
EOD;
|
2017-02-08 12:21:59 +01:00
|
|
|
die();
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
//Build the string for the first request
|
2017-02-11 16:16:56 +01:00
|
|
|
$requestString = 'https://touch.facebook.com/pages_reaction_units/more/?page_id='
|
|
|
|
. $pageID
|
|
|
|
. '&cursor={"card_id"%3A"videos"%2C"has_next_page"%3Atrue}&surface=mobile_page_home&unit_count=8';
|
2017-02-08 12:21:59 +01:00
|
|
|
|
|
|
|
$fileContent = file_get_contents($requestString);
|
|
|
|
|
|
|
|
$articleIndex = 0;
|
|
|
|
$maxArticle = 3;
|
|
|
|
|
|
|
|
$html = $this->buildContent($fileContent);
|
|
|
|
$author = $this->getInput('u');
|
2017-06-15 12:51:11 +02:00
|
|
|
|
2017-07-29 19:28:00 +02:00
|
|
|
foreach($html->find("article") as $content) {
|
2017-02-08 12:21:59 +01:00
|
|
|
|
|
|
|
$item = array();
|
2017-06-15 12:51:11 +02:00
|
|
|
|
2017-02-11 16:16:56 +01:00
|
|
|
$item['uri'] = "http://touch.facebook.com"
|
2017-06-15 12:42:59 +02:00
|
|
|
. $content->find("div[class='_52jc _5qc4 _24u0 _36xo']", 0)->find("a", 0)->getAttribute("href");
|
2017-02-08 12:21:59 +01:00
|
|
|
|
2017-06-15 12:42:59 +02:00
|
|
|
if($content->find("header", 0) !== null) {
|
|
|
|
$content->find("header", 0)->innertext = "";
|
|
|
|
}
|
2017-06-15 12:51:11 +02:00
|
|
|
|
2017-06-15 12:42:59 +02:00
|
|
|
if($content->find("footer", 0) !== null) {
|
|
|
|
$content->find("footer", 0)->innertext = "";
|
|
|
|
}
|
2017-02-08 12:21:59 +01:00
|
|
|
|
|
|
|
//Remove html nodes, keep only img, links, basic formatting
|
2017-02-11 16:16:56 +01:00
|
|
|
$content = strip_tags($content, '<a><img><i><u><br><p>');
|
2017-02-08 12:21:59 +01:00
|
|
|
|
|
|
|
//Adapt link hrefs: convert relative links into absolute links and bypass external link redirection
|
|
|
|
$content = preg_replace_callback('/ href=\"([^"]+)\"/i', $unescape_fb_link, $content);
|
|
|
|
|
|
|
|
//Clean useless html tag properties and fix link closing tags
|
2017-02-11 16:16:56 +01:00
|
|
|
foreach (array(
|
|
|
|
'onmouseover',
|
|
|
|
'onclick',
|
|
|
|
'target',
|
|
|
|
'ajaxify',
|
|
|
|
'tabindex',
|
|
|
|
'class',
|
|
|
|
'style',
|
|
|
|
'data-[^=]*',
|
|
|
|
'aria-[^=]*',
|
|
|
|
'role',
|
|
|
|
'rel',
|
|
|
|
'id') as $property_name)
|
|
|
|
$content = preg_replace('/ ' . $property_name . '=\"[^"]*\"/i', '', $content);
|
2017-02-08 12:21:59 +01:00
|
|
|
$content = preg_replace('/<\/a [^>]+>/i', '</a>', $content);
|
|
|
|
|
2017-02-11 16:16:56 +01:00
|
|
|
//Convert textual representation of emoticons eg
|
|
|
|
// "<i><u>smile emoticon</u></i>" back to ASCII emoticons eg ":)"
|
2017-02-08 12:21:59 +01:00
|
|
|
$content = preg_replace_callback('/<i><u>([^ <>]+) ([^<>]+)<\/u><\/i>/i', $unescape_fb_emote, $content);
|
|
|
|
|
|
|
|
$item['content'] = $content;
|
|
|
|
|
|
|
|
$title = $author;
|
|
|
|
if (strlen($title) > 24)
|
2017-02-11 16:16:56 +01:00
|
|
|
$title = substr($title, 0, strpos(wordwrap($title, 24), "\n")) . '...';
|
|
|
|
$title = $title . ' | ' . strip_tags($content);
|
2017-02-08 12:21:59 +01:00
|
|
|
if (strlen($title) > 64)
|
2017-02-11 16:16:56 +01:00
|
|
|
$title = substr($title, 0, strpos(wordwrap($title, 64), "\n")) . '...';
|
2017-02-08 12:21:59 +01:00
|
|
|
|
|
|
|
$item['title'] = $title;
|
|
|
|
$item['author'] = $author;
|
|
|
|
|
|
|
|
array_push($this->items, $item);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
// Currently not used. Is used to get more than only 3 elements, as they appear on another page.
|
2017-02-11 16:16:56 +01:00
|
|
|
private function computeNextLink($string, $pageID){
|
2017-02-08 12:21:59 +01:00
|
|
|
|
2017-02-12 13:58:42 +01:00
|
|
|
$regex = implode(
|
|
|
|
'',
|
|
|
|
array(
|
|
|
|
"/timeline_unit",
|
|
|
|
"\\\\\\\\u00253A1",
|
|
|
|
"\\\\\\\\u00253A([0-9]*)",
|
|
|
|
"\\\\\\\\u00253A([0-9]*)",
|
|
|
|
"\\\\\\\\u00253A([0-9]*)",
|
|
|
|
"\\\\\\\\u00253A([0-9]*)/"
|
|
|
|
)
|
|
|
|
);
|
|
|
|
|
2017-02-08 12:21:59 +01:00
|
|
|
preg_match($regex, $string, $result);
|
|
|
|
|
2017-02-12 13:58:42 +01:00
|
|
|
return implode(
|
|
|
|
'',
|
|
|
|
array(
|
|
|
|
"https://touch.facebook.com/pages_reaction_units/more/?page_id=",
|
|
|
|
$pageID,
|
|
|
|
"&cursor=%7B%22timeline_cursor%22%3A%22timeline_unit%3A1%3A",
|
|
|
|
$result[1],
|
|
|
|
"%3A",
|
|
|
|
$result[2],
|
|
|
|
"%3A",
|
|
|
|
$result[3],
|
|
|
|
"%3A",
|
|
|
|
$result[4],
|
|
|
|
"%22%2C%22timeline_section_cursor%22%3A%7B%7D%2C%22",
|
|
|
|
"has_next_page%22%3Atrue%7D&surface=mobile_page_home&unit_count=3"
|
|
|
|
)
|
|
|
|
);
|
2017-02-08 12:21:59 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
//Builds the HTML from the encoded JS that Facebook provides.
|
2017-02-11 16:16:56 +01:00
|
|
|
private function buildContent($pageContent){
|
2017-02-08 12:21:59 +01:00
|
|
|
|
|
|
|
$regex = "/\\\"html\\\":\\\"(.*?)\\\",\\\"replace/";
|
|
|
|
preg_match($regex, $pageContent, $result);
|
|
|
|
|
|
|
|
return str_get_html(html_entity_decode(json_decode('"' . $result[1] . '"')));
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2017-02-11 16:16:56 +01:00
|
|
|
//Builds the cookie from the page, as Facebook sometimes refuses to give
|
|
|
|
//the page if no cookie is provided.
|
|
|
|
private function getCookies($pageURL){
|
2017-02-08 12:21:59 +01:00
|
|
|
|
|
|
|
$ctx = stream_context_create(array(
|
|
|
|
'http' => array(
|
|
|
|
'user_agent' => "Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:46.0) Gecko/20100101 Firefox/46.0",
|
|
|
|
'Accept' => 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8'
|
|
|
|
)
|
|
|
|
)
|
2017-02-11 16:16:56 +01:00
|
|
|
);
|
2017-02-08 12:21:59 +01:00
|
|
|
$a = file_get_contents($pageURL, 0, $ctx);
|
|
|
|
|
|
|
|
//First request to get the cookie
|
|
|
|
$cookies = "";
|
2017-07-29 19:28:00 +02:00
|
|
|
foreach($http_response_header as $hdr) {
|
|
|
|
if(strpos($hdr, "Set-Cookie") !== false) {
|
2017-02-08 12:21:59 +01:00
|
|
|
$cLine = explode(":", $hdr)[1];
|
|
|
|
$cLine = explode(";", $cLine)[0];
|
2017-02-11 16:16:56 +01:00
|
|
|
$cookies .= ";" . $cLine;
|
2017-02-08 12:21:59 +01:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return substr($cookies, 1);
|
|
|
|
}
|
|
|
|
|
|
|
|
//Get the page ID from the Facebook page.
|
2017-02-11 16:16:56 +01:00
|
|
|
private function getPageID($page, $cookies){
|
2017-02-08 12:21:59 +01:00
|
|
|
|
|
|
|
$context = stream_context_create(array(
|
|
|
|
'http' => array(
|
|
|
|
'user_agent' => "Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:46.0) Gecko/20100101 Firefox/46.0",
|
2017-02-11 16:16:56 +01:00
|
|
|
'header' => 'Cookie: ' . $cookies
|
2017-02-08 12:21:59 +01:00
|
|
|
)
|
|
|
|
)
|
2017-02-11 16:16:56 +01:00
|
|
|
);
|
2017-02-08 12:21:59 +01:00
|
|
|
|
|
|
|
$pageContent = file_get_contents($page, 0, $context);
|
|
|
|
|
2017-06-15 12:51:11 +02:00
|
|
|
if(strpos($pageContent, "signup-button") != false) {
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
2017-02-08 12:21:59 +01:00
|
|
|
//Get the page ID if we don't have a captcha
|
2017-02-11 16:16:56 +01:00
|
|
|
$regex = "/page_id=([0-9]*)&/";
|
2017-02-08 12:21:59 +01:00
|
|
|
preg_match($regex, $pageContent, $matches);
|
|
|
|
|
2017-07-29 19:28:00 +02:00
|
|
|
if(count($matches) > 0) {
|
2017-02-08 12:21:59 +01:00
|
|
|
return $matches[1];
|
|
|
|
}
|
|
|
|
|
|
|
|
//Get the page ID if we do have a captcha
|
2017-02-11 16:16:56 +01:00
|
|
|
$regex = "/\"pageID\":\"([0-9]*)\"/";
|
2017-02-08 12:21:59 +01:00
|
|
|
preg_match($regex, $pageContent, $matches);
|
|
|
|
|
|
|
|
return $matches[1];
|
|
|
|
|
|
|
|
}
|
|
|
|
|
2017-02-11 16:16:56 +01:00
|
|
|
public function getName(){
|
|
|
|
return (isset($this->name) ? $this->name . ' - ' : '') . 'Facebook Bridge';
|
2017-02-08 12:21:59 +01:00
|
|
|
}
|
|
|
|
|
2017-02-11 16:16:56 +01:00
|
|
|
public function getURI(){
|
2017-02-08 12:21:59 +01:00
|
|
|
return 'http://facebook.com';
|
|
|
|
}
|
|
|
|
|
2017-02-11 16:16:56 +01:00
|
|
|
public function getCacheDuration(){
|
2017-02-08 12:21:59 +01:00
|
|
|
return 60 * 60 * 3; // 5 minutes
|
|
|
|
}
|
|
|
|
}
|