[TwitterBridge] Fix the bridge using a brand new API

This commit is contained in:
Lyra 2020-06-05 10:17:53 +02:00
parent c4422bdbb5
commit 06891ae35f
3 changed files with 121 additions and 206 deletions

View file

@ -2,6 +2,7 @@
class TwitterBridge extends BridgeAbstract { class TwitterBridge extends BridgeAbstract {
const NAME = 'Twitter Bridge'; const NAME = 'Twitter Bridge';
const URI = 'https://twitter.com/'; const URI = 'https://twitter.com/';
const API_URI = 'https://api.twitter.com';
const CACHE_TIMEOUT = 300; // 5min const CACHE_TIMEOUT = 300; // 5min
const DESCRIPTION = 'returns tweets'; const DESCRIPTION = 'returns tweets';
const MAINTAINER = 'pmaziere'; const MAINTAINER = 'pmaziere';
@ -168,6 +169,27 @@ EOD
} }
} }
public function getApiURI() {
switch($this->queriedContext) {
case 'By keyword or hashtag':
return self::API_URI
. '/2/search/adaptive.json?q='
. urlencode($this->getInput('q'))
. '&tweet_mode=extended';
case 'By username':
return self::API_URI
. '/2/timeline/profile/'
. $this->getRestId($this->getInput('u'))
. '.json?tweet_mode=extended';
case 'By list':
return self::API_URI
. '/2/timeline/list.json?list_id='
. $this->getListId($this->getInput('user'), $this->getInput('list'))
. '&tweet_mode=extended';
default: returnServerError('Invalid query context !');
}
}
public function collectData(){ public function collectData(){
$html = ''; $html = '';
$page = $this->getURI(); $page = $this->getURI();
@ -176,14 +198,9 @@ EOD
'User-Agent: Mozilla/5.0 (Windows NT 9.0; WOW64; Trident/7.0; rv:11.0) like Gecko' 'User-Agent: Mozilla/5.0 (Windows NT 9.0; WOW64; Trident/7.0; rv:11.0) like Gecko'
); );
if(php_sapi_name() === 'cli' && empty(ini_get('curl.cainfo'))) { $data = json_decode($this->getApiContents($this->getApiURI()));
$cookies = $this->getCookies($page);
$html = getSimpleHTMLDOM($page, array_merge($header, array("Cookie: $cookies")));
} else {
$html = getSimpleHTMLDOM($page, $header, array(CURLOPT_COOKIEFILE => ''));
}
if(!$html) { if(!$data) {
switch($this->queriedContext) { switch($this->queriedContext) {
case 'By keyword or hashtag': case 'By keyword or hashtag':
returnServerError('No results for this query.'); returnServerError('No results for this query.');
@ -196,75 +213,33 @@ EOD
$hidePictures = $this->getInput('nopic'); $hidePictures = $this->getInput('nopic');
foreach($html->find('div.js-stream-tweet') as $tweet) { foreach($data->globalObjects->tweets as $tweet) {
// Skip retweets? // Skip retweets?
if($this->getInput('noretweet') if($this->getInput('noretweet')
&& $tweet->find('div.context span.js-retweet-text a', 0)) { && isset($tweet->retweeted_status_id_str)) {
continue;
}
// remove 'invisible' content
foreach($tweet->find('.invisible') as $invisible) {
$invisible->outertext = '';
}
// Skip protmoted tweets
$heading = $tweet->previousSibling();
if(!is_null($heading) &&
$heading->getAttribute('class') === 'promoted-tweet-heading'
) {
continue; continue;
} }
$item = array(); $item = array();
// extract username and sanitize // extract username and sanitize
$item['username'] = htmlspecialchars_decode($tweet->getAttribute('data-screen-name'), ENT_QUOTES); $user_info = $this->getUserInformation($tweet->user_id_str, $data->globalObjects);
// extract fullname (pseudonym)
$item['fullname'] = htmlspecialchars_decode($tweet->getAttribute('data-name'), ENT_QUOTES); $item['username'] = $user_info->name;
// get author $item['fullname'] = $user_info->screen_name;
$item['author'] = $item['fullname'] . ' (@' . $item['username'] . ')'; $item['author'] = $item['fullname'] . ' (@' . $item['username'] . ')';
if($rt = $tweet->find('div.context span.js-retweet-text a', 0)) { $item['avatar'] = $user_info->profile_image_url_https;
$item['author'] .= ' RT: @' . $rt->plaintext;
} $item['id'] = $tweet->id_str;
// get avatar link $item['uri'] = self::URI . $tweet->user_id_str . '/status/' . $item['id'];
$item['avatar'] = $tweet->find('img', 0)->src;
// get TweetID
$item['id'] = $tweet->getAttribute('data-tweet-id');
// get tweet link
$item['uri'] = self::URI . substr($tweet->find('a.js-permalink', 0)->getAttribute('href'), 1);
// extract tweet timestamp // extract tweet timestamp
$item['timestamp'] = $tweet->find('span.js-short-timestamp', 0)->getAttribute('data-time'); $item['timestamp'] = $tweet->created_at;
// generate the title // generate the title
$item['title'] = strip_tags($this->fixAnchorSpacing(htmlspecialchars_decode( $item['title'] = $tweet->full_text;
$tweet->find('p.js-tweet-text', 0), ENT_QUOTES), '<a>')); $cleanedTweet = $tweet->full_text;
switch($this->queriedContext) { // Add avatar
case 'By list':
// Check if filter applies to list (using raw content)
if($this->getInput('filter')) {
if(stripos($tweet->find('p.js-tweet-text', 0)->plaintext, $this->getInput('filter')) === false) {
continue 2; // switch + for-loop!
}
}
break;
default:
}
$this->processContentLinks($tweet);
$this->processEmojis($tweet);
// get tweet text
$cleanedTweet = str_replace(
'href="/',
'href="' . self::URI,
$tweet->find('p.js-tweet-text', 0)->innertext
);
// fix anchors missing spaces in-between
$cleanedTweet = $this->fixAnchorSpacing($cleanedTweet);
// Add picture to content
$picture_html = ''; $picture_html = '';
if(!$hidePictures) { if(!$hidePictures) {
$picture_html = <<<EOD $picture_html = <<<EOD
@ -278,31 +253,37 @@ EOD
EOD; EOD;
} }
// Add embeded image to content // Get images
$image_html = ''; $image_html = '';
$images = $this->getImageURI($tweet); if(isset($tweet->extended_entities->media) && !$this->getInput('noimg')) {
if(!$this->getInput('noimg') && !is_null($images)) { foreach($tweet->extended_entities->media as $media) {
$image = $media->media_url_https;
foreach ($images as $image) { $display_image = $media->display_url;
// Set image scaling
$image_orig = $this->getInput('noimgscaling') ? $image : $image . ':orig';
$image_thumb = $this->getInput('noimgscaling') ? $image : $image . ':thumb';
// add enclosures // add enclosures
$item['enclosures'][] = $image_orig; $item['enclosures'][] = $image;
$image_html .= <<<EOD $image_html .= <<<EOD
<a href="{$image_orig}"> <a href="{$image}">
<img <img
style="align:top; max-width:558px; border:1px solid black;" style="align:top; max-width:558px; border:1px solid black;"
src="{$image_thumb}" /> src="{$display_image}" />
</a> </a>
EOD; EOD;
} }
} }
// add content switch($this->queriedContext) {
case 'By list':
// Check if filter applies to list (using raw content)
if($this->getInput('filter')) {
if(stripos($cleanedTweet, $this->getInput('filter')) === false) {
continue 2; // switch + for-loop!
}
}
break;
default:
}
$item['content'] = <<<EOD $item['content'] = <<<EOD
<div style="display: inline-block; vertical-align: top;"> <div style="display: inline-block; vertical-align: top;">
{$picture_html} {$picture_html}
@ -315,151 +296,85 @@ EOD;
</div> </div>
EOD; EOD;
// add quoted tweet
$quotedTweet = $tweet->find('div.QuoteTweet', 0);
if($quotedTweet) {
// get tweet text
$cleanedQuotedTweet = str_replace(
'href="/',
'href="' . self::URI,
$quotedTweet->find('div.tweet-text', 0)->innertext
);
$this->processContentLinks($quotedTweet);
$this->processEmojis($quotedTweet);
// Add embeded image to content
$quotedImage_html = '';
$quotedImages = $this->getQuotedImageURI($tweet);
if(!$this->getInput('noimg') && !is_null($quotedImages)) {
foreach ($quotedImages as $image) {
// Set image scaling
$image_orig = $this->getInput('noimgscaling') ? $image : $image . ':orig';
$image_thumb = $this->getInput('noimgscaling') ? $image : $image . ':thumb';
// add enclosures
$item['enclosures'][] = $image_orig;
$quotedImage_html .= <<<EOD
<a href="{$image_orig}">
<img
style="align:top; max-width:558px; border:1px solid black;"
src="{$image_thumb}" />
</a>
EOD;
}
}
$item['content'] = <<<EOD
{$item['content']}
<hr>
<div style="display: inline-block; vertical-align: top;">
<blockquote>{$cleanedQuotedTweet}</blockquote>
</div>
<div style="display: block; vertical-align: top;">
<blockquote>{$quotedImage_html}</blockquote>
</div>
EOD;
}
$item['content'] = htmlspecialchars_decode($item['content'], ENT_QUOTES); $item['content'] = htmlspecialchars_decode($item['content'], ENT_QUOTES);
// put out // put out
$this->items[] = $item; $this->items[] = $item;
} }
usort($this->items, array('TwitterBridge', 'compareTweetId'));
} }
private function processEmojis($tweet){ private static function compareTweetId($tweet1, $tweet2) {
// process emojis (reduce size) return (intval($tweet1['id']) < intval($tweet2['id']) ? 1 : -1);
foreach($tweet->find('img.Emoji') as $img) { }
$img->style .= ' height: 1em;';
//The aim of this function is to get an API key and a guest token
//This function takes 2 requests, and therefore is cached
private function getApiKey() {
$cacheFac = new CacheFactory();
$cacheFac->setWorkingDir(PATH_LIB_CACHES);
$cache = $cacheFac->create(Configuration::getConfig('cache', 'type'));
$cache->setScope(get_called_class());
$cache->setKey(array('api_key'));
$data = $cache->loadData();
if($data === null) {
$twitterPage = getContents('https://twitter.com');
$jsMainRegex = '/(https:\/\/abs\.twimg\.com\/responsive-web\/web\/main\.[^\.]+\.js)/m';
preg_match_all($jsMainRegex, $twitterPage, $jsMainMatches, PREG_SET_ORDER, 0);
$jsLink = $jsMainMatches[0][0];
$guestTokenRegex = '/gt=([0-9]*)/m';
preg_match_all($guestTokenRegex, $twitterPage, $guestTokenMatches, PREG_SET_ORDER, 0);
$guestToken = $guestTokenMatches[0][1];
$jsContent = getContents($jsLink);
$apiKeyRegex = '/([a-zA-Z0-9]{59}%[a-zA-Z0-9]{44})/m';
preg_match_all($apiKeyRegex, $jsContent, $apiKeyMatches, PREG_SET_ORDER, 0);
$apiKey = $apiKeyMatches[0][0];
$cache->saveData(array($apiKey, $guestToken));
return array($apiKey, $guestToken);
} }
return $data;
} }
private function processContentLinks($tweet){ private function getApiContents($uri) {
// processing content links $apiKeys = $this->getApiKey();
foreach($tweet->find('a') as $link) { $headers = array('authorization: Bearer ' . $apiKeys[0],
if($link->hasAttribute('data-expanded-url')) { 'x-guest-token: ' . $apiKeys[1],
$link->href = $link->getAttribute('data-expanded-url'); );
} return getContents($uri, $headers);
$link->removeAttribute('data-expanded-url');
$link->removeAttribute('data-query-source');
$link->removeAttribute('rel');
$link->removeAttribute('class');
$link->removeAttribute('target');
$link->removeAttribute('title');
}
} }
private function fixAnchorSpacing($content){ private function getRestId($username) {
// fix anchors missing spaces in-between $searchparams = urlencode('{"screen_name":"' . strtolower($username) . '", "withHighlightedLabel":true}');
return str_replace( $searchURL = self::API_URI . '/graphql/-xfUfZsnR_zqjFd-IfrN5A/UserByScreenName?variables=' . $searchparams;
'<a', $searchResult = $this->getApiContents($searchURL);
' <a', $searchResult = json_decode($searchResult);
$content return $searchResult->data->user->rest_id;
);
} }
private function getImageURI($tweet){ private function getListId($username, $listName) {
// Find media in tweet $searchparams = urlencode('{"screenName":"'
$images = array(); . strtolower($username)
. '", "listSlug": "'
. $listName
. '", "withHighlightedLabel":false}');
$searchURL = self::API_URI . '/graphql/ErWsz9cObLel1BF-HjuBlA/ListBySlug?variables=' . $searchparams;
$searchResult = $this->getApiContents($searchURL);
$searchResult = json_decode($searchResult);
return $searchResult->data->user_by_screen_name->list->id_str;
}
$container = $tweet->find('div.AdaptiveMedia-container', 0); private function getUserInformation($userId, $apiData) {
foreach($apiData->users as $user) {
if($container && $container->find('img', 0)) { if($user->id_str == $userId) {
foreach ($container->find('img') as $img) { return $user;
$images[] = $img->src;
} }
} }
if (!empty($images)) {
return $images;
}
return null;
} }
private function getQuotedImageURI($tweet){
// Find media in tweet
$images = array();
$container = $tweet->find('div.QuoteMedia-container', 0);
if($container && $container->find('img', 0)) {
foreach ($container->find('img') as $img) {
$images[] = $img->src;
}
}
if (!empty($images)) {
return $images;
}
return null;
}
private function getCookies($pageURL){
$ctx = stream_context_create(array(
'http' => array(
'follow_location' => false
)
)
);
$a = file_get_contents($pageURL, 0, $ctx);
//First request to get the cookie
$cookies = '';
foreach($http_response_header as $hdr) {
if(stripos($hdr, 'Set-Cookie') !== false) {
$cLine = explode(':', $hdr)[1];
$cLine = explode(';', $cLine)[0];
$cookies .= ';' . $cLine;
}
}
return substr($cookies, 2);
}
} }

View file

View file