@@ -1,49 +1,107 @@
<?php
+
/*
- Thanks to Jonty (http://i.am.jonty.co.uk/) for this script.
+ Thanks to Jonty (http://jonty.co.uk/) for this script.
We use it to see a feed of twitter search results for a given term in IRC.
Needs php5.
- Run like: php twitter-feeder.php | nc -q0 localhost 12345
+ Run like: php twitter-feed.php #twitters 'monkey' | nc -q0 localhost 12345
+ (Where #twitters is the channel to pipe to, and monkey is the search term)
+
+ If you want to filter things from the feed you can optionally add regexes
+ to the end of the params.
*/
- $originalUrl = "?since_id=0&q=last.fm"; // <-- term in there.
- $mainUrl = "http://search.twitter.com/search";
+ if (count($argv) < 3) {
+ print "Usage: twitter-feed.php '#channel' 'searchterm' [url filter regexp, [url-filter regexp, ...]]\n";
+ print "(Quotes are important, remember # is a comment in bash!)\n";
+ exit;
+ }
+
+ array_shift($argv);
+ $channel = array_shift($argv);
+ $search = array_shift($argv);
+ $filters = $argv ? $argv : array();
+
+ print "{$channel} Starting twitter.com search feed for {$search}...\n";
+
+ $originalUrl = "?since_id=4028479400&q=".urlencode($search);
+ $baseUrl = "http://search.twitter.com/search";
+
+ $firstRun = true;
+ $refreshUrl = $originalUrl;
- print "#playdar Starting search.twitter.com feed...\n";
+ while (1) {
+ $source = file("{$baseUrl}{$refreshUrl}&refresh=true");
+ if ($source) {
+ $json = json_decode(implode('', $source));
- $firstRun = 1;
- $nextUrl = $originalUrl;
- while(1){
- $pageContent = file($mainUrl.$nextUrl.'&refresh=true');
- if ($pageContent) {
- $source = implode("", $pageContent);
- if (preg_match("/\"refresh_url\":\"(.*?)\"/", $source, $urlMatches)) {
- $nextUrl = $urlMatches[1];
+ if ($json && isset($json->refresh_url)) {
+ $refreshUrl = $json->refresh_url;
- preg_match_all("/{\"text\":\"(.*?)\",\"to_user_id\":.*?,\"from_user\":\"(.*?)\".*?}/sm", $source, $postMatches, PREG_SET_ORDER);
- foreach($postMatches as $match){
- $message = cleanString($match[1]);
- if($firstRun == 0) print "{$match[2]}: ".$message."\n";
+ if ($firstRun) {
+ $firstRun = false;
+ continue;
}
- if($firstRun == 1) $firstRun = 0;
+
+ foreach ($json->results as $result) {
+ if ($message = processMessage($result->text, $filters)) {
+ print "{$result->from_user}: {$message}\n";
+ }
+ }
+
} else {
// On error, reset to front page and retry
- $nextUrl = $originalUrl;
+ $refreshUrl = $originalUrl;
}
}
sleep(60);
}
- function cleanString($string) {
+ function processMessage ($string, $filters) {
+ $string = utf8_decode($string);
+
+ // Resolve shortened URL's to the full thing for filtering
+ preg_match_all('/http:\/\/[^ $)]+/i', $string, $urls);
+ if ($urls) {
+ foreach ($urls[0] as $url) {
+ if ($redirect = getRedirect($url)) {
+ $string = str_replace($url, $redirect, $string);
+ }
+ }
+ }
+
+ $string = cleanString($string);
+ if ($filters) {
+ foreach ($filters as $filter) {
+ if (preg_match("/{$filter}/i", $string)) {
+ $string = '';
+ break;
+ }
+ }
+ }
+
+ return $string;
+ }
+
+ function cleanString ($string) {
$string = preg_replace('/\\\u([[:alnum:]]{4})/u', '?', $string);
$string = str_replace('\/','/', urldecode($string));
$string = html_entity_decode($string, null, 'UTF-8');
- $string = htmlspecialchars_decode($string);
- $string = utf8_decode($string);
- return $string;
+ return htmlspecialchars_decode($string);
}
-?>
+ function getRedirect ($url) {
+ $curl = curl_init();
+ curl_setopt($curl, CURLOPT_RETURNTRANSFER, true);
+ curl_setopt($curl, CURLOPT_URL, $url);
+ curl_setopt($curl, CURLOPT_HEADER, true);
+ curl_setopt($curl, CURLOPT_NOBODY, true);
+ $headers = curl_exec($curl);
+ curl_close($curl);
+
+ preg_match('/Location:(.*?)[\r\n]/i', $headers, $redirect);
+ return $redirect ? urlencode(trim($redirect[1])) : '';
+ }
examples/twitter-feed.php
ce2e69c1469ec499a1b98ede6a472949ed2c9b24
be5e6854b92fb4a37d87df1b151a8cb298c8e503
RJ
rj@IS-2229.fordirectadmin.com.fordirectadmin.com
http://github.com/RJ/irccat/commit/fbb4df3a17e55d37ffc53081936ea68b70c85a68
fbb4df3a17e55d37ffc53081936ea68b70c85a68
2009-10-09T07:55:54-07:00
2009-10-09T07:55:54-07:00
Merge branch 'master' of git@github.com:RJ/irccat
7ed4e1b6577daec30ccf6cf38afe4392a3354c09
RJ
rj@IS-2229.fordirectadmin.com.fordirectadmin.com