Added “new instance announcement toot” function

web/clitools/getinstinfo.php

@@ -17,6 +17,7 @@
 */
 
 const N="\n";
+const RN="\r\n";
 define('SNAME',basename(__FILE__));
 const LIBDP='/../lib';
 
@@ -28,6 +29,8 @@ require __DIR__.LIBDP.'/mb_lcfirst.php';
 require __DIR__.LIBDP.'/ghs.php';
 require __DIR__.LIBDP.'/ght.php';
 require __DIR__.LIBDP.'/supplangs.php';
+require __DIR__.LIBDP.'/gettlds.php';
+require __DIR__.LIBDP.'/mastodon_postlen.php';
 require __DIR__.LIBDP.'/vendor/autoload.php';
 use LanguageDetection\Language;
 
@@ -52,10 +55,13 @@ $opts=[
 	'functimeout'=>20,
 	'ldtoots'=>40,// number of toots to check with the automatic language detection function
 	'dryrun'=>false,
+	'sendtoot'=>true,
 	'fetchusers'=>false,
 	'udiratts'=>5,
 	'udirfailst'=>90,
-	'minmsgimplev'=>1
+	'minmsgimplev'=>1,
+	'bothost'=>null,
+	'bottoken'=>null
 ];
 
 $msglevs=['Debug', 'Info', 'Warning', 'Error', 'None'];
@@ -93,11 +99,15 @@ OPTIONS
   SPECIFICATION» below to see how to specify time.
   DEFAULT: '.ght($opts['conntimeout'],null,0).'
  -T, --functimeout <time>
-  Sets the timeout for every downloa. See section «TIME SPECIFICATION» below
+  Sets the timeout for every download. See section «TIME SPECIFICATION» below
   to see how to specify time.
   DEFAULT: '.ght($opts['functimeout'],null,0).'
  -d, --dryrun
   If this option is set, the script won’t write anything in the database.
+ -S, --dontoot
+  If this option is set, the script won’t try to read «getinstinfo.ini» file
+  and won’t post an announcement toot about a new instance.
+  See NEW INSTANCE ANNOUNCEMENT TOOT below for more info.
  -m, --minmsgimplev <«debug»|«info»|«warning»|«error»|«none»>
   Defines the minimum “importance level” of messages to be written to the
   text user interface. There are 4 “importance levels”, in  this order of
@@ -116,6 +126,17 @@ TIME SPECIFICATION
  7 minutes and 12 seconds you can use «1y,6M,2w,3d,5h,7m,12s»; but you can
  also use «12s,7m,5h,3d,2w,6M,1y», or even «18M,1w,1w,2d,1d,3h,2h,7m,12s».
 
+NEW INSTANCE ANNOUNCEMENT TOOT
+
+ This script can send an announcement toot when the instance it checks is new.
+ You can disable this function with the «-S» or «--dontoot» options (see
+ above). If new instance announcement toot is enabled, this script expects
+ to find a «getinstinfo.ini» file in the same directory it lives in, with
+ a «bothost» parameter defining the instance to use to send the toot,
+ a «bottoken» parameter defining the token to be used to post, and
+ a «botmaxchars» parameter defining the maximum number of characters allowed
+ for toots on the defined instance.
+
 LICENSE
 
  This program comes with ABSOLUTELY NO WARRANTY; for details see the source.
@@ -152,6 +173,8 @@ for ($i=1; $i<$argc; $i++) {
 		$opts['ldtoots']=$argv[$i]+0;
 	} elseif ($argv[$i]=='-d' || $argv[$i]=='--dryrun') {
 		$opts['dryrun']=true;
+	} elseif ($argv[$i]=='-S' || $argv[$i]=='--dontoot') {
+		$opts['sendtoot']=false;
 	} elseif ($argv[$i]=='-m' || $argv[$i]=='--minmsgimplev') {
 		if ($i+1>=$argc || !in_array(ucfirst(strtolower($argv[$i+1])),$msglevs))
 			mexit('option «'.$argv[$i].'» requires a “message importance level” value as an argument (use «-h» to read help).'.N,1);
@@ -182,6 +205,14 @@ catch (Exception $error) { mexit('could not set «utf8mb4» charset for MySQL: '
 // for php versions < 8
 if ($res===false) mexit('could not set MySQL charset: '.mysqli_error($link).' ['.mysqli_errno($link).'].'.N,1,true);
 
+if ($opts['sendtoot']) {
+	$inifp=__DIR__.'/getinstinfo.ini';
+	$iniarr=@parse_ini_file($inifp)
+		or mexit('could not open config file «'.$inifp.'»'.N,1);
+	if (!isset($iniarr['bothost']) || !isset($iniarr['bottoken']) || !isset($iniarr['botmaxchars']) || preg_match('/^[0-9]+$/',$iniarr['botmaxchars'])!=1)
+		mexit('config file «'.$inifp.'» is malformed, please check its format in the help text (you can read it using the «-h» or «--help» options).'.N,1);
+}
+
 $mastodons=[];
 $res=myq($link,'SELECT Name FROM Platforms WHERE Consider=1',__LINE__);
 while ($row=mysqli_fetch_assoc($res))
@@ -756,6 +787,66 @@ if (isset($idata['blocks']) && is_array($idata['blocks'])) {
 	}
 }
 
+if ($opts['sendtoot'] && $instanswered && $idata['IsMastodon'] && $idata['FirstSeen']==$now) {
+	gettlds();
+	eecho(0,'«'.$opts['hostname'].'»: this instance is new, trying to send a toot about it...'.N);
+	$endpoint='https://'.$iniarr['bothost'].'/api/v1/statuses';
+	$toot='A new Mastodon instance, https://'.$opts['hostname'].', has been found by the crawler at https://mastodon.help'.N.N.'Name: ';
+	(is_null($idata['Title'])) ? $toot.='unspecified' : $toot.=$idata['Title'];
+	$toot.=N.N.'Languages: ';
+	(array_key_exists('languages',$idata) && is_array($idata['languages']) && count($idata['languages'])>0) ? $toot.=implode($idata['languages']) : $toot.='unspecified/undetectable';
+	$toot.=N.N.'Users: ';
+	(is_null($idata['UserCount'])) ? $toot.='unspecified' : $toot.=$idata['UserCount'];
+	$toot.=N.N.'Max. characters per post: ';
+	(is_null($idata['MaxTootChars']) || $idata['MaxTootChars']==500) ? $toot.='500' : $toot.=$idata['MaxTootChars'];
+	$toot.=N.N.'Registration: ';
+	if (is_null($idata['RegOpen'])) {
+		$toot.='unspecified';
+	} elseif ($idata['RegOpen']==1) {
+		$toot.='open';
+		if ($idata['RegReqApproval']==1)
+			$toot.=' (requires admins approval)';
+		elseif ($idata['RegReqApproval']==0)
+			$toot.=' (does not require admins approval)';
+	} elseif ($idata['RegOpen']==0) {
+		$toot.='closed';
+	}
+	$toot.=N.N.'Short description: ';
+	(is_null($idata['ShortDesc'])) ? $toot.='unspecified' : $toot.=$idata['ShortDesc'];
+	$toot.=N.N.'Long description: ';
+	(is_null($idata['LongDesc'])) ? $toot.='unspecified' : $toot.=html2text($idata['LongDesc']);
+	if (postlen($toot)>$iniarr['botmaxchars']) {
+		while (postlen($toot)+2>$iniarr['botmaxchars'])
+			$toot=preg_replace('#\s+(\S|\n)+$#u','',$toot);
+		$toot.=' …';
+	}
+	$context=[
+		'http'=>[
+			'header'=>'Authorization: Bearer '.$iniarr['bottoken'].RN.
+				'Idempotency-Key'.md5($iniarr['bothost'].$now.rand(1000,9999)).RN.
+				'Content-type: application/x-www-form-urlencoded'.RN,
+			'method'=>'POST',
+			'content'=>http_build_query([
+				'status'=>$toot,
+				'visibility'=>'direct',
+				'language'=>'en'
+			]),
+			'timeout'=>$opts['conntimeout']
+		]
+	];
+	$context=stream_context_create($context);
+	$res=@file_get_contents($endpoint,false,$context);
+	if ($res===false) {
+		eecho(2,'«'.$opts['hostname'].'»: could not connect to «'.$iniarr['bothost'].'».'.N);
+	} else {
+		$res=@json_decode($res,true);
+		if ($res===false)
+			eecho(2,'«'.$opts['hostname'].'»: trying to post a toot about this new instance, «'.$iniarr['bothost'].'» has not returned valid JSON data.'.N);
+		else
+			eecho(1,'«'.$opts['hostname'].'»: correctly posted announcement about this new instance on «'.$iniarr['bothost'].'» :-)'.N);
+	}
+}
+
 if ($instanswered && $opts['fetchusers'] && $idata['IsMastodon'] && !is_null($idata['Version']) && $idata['Version']>='4.0.0') {
 	eecho(0,'«'.$opts['hostname'].'»: trying to fetch users info from directory API...'.N);
 	$users=[];// array of users in this instance's directory
@@ -1294,4 +1385,31 @@ function waituntilonline() {
 	if ($gotoff) eecho(1,'it seems we are back online! :-)'.N);
 }
 
+function headcb($arr) {
+	$lpad='';
+	$rpad='';
+	for ($i=0; $i<7-$arr[2]; $i++) {
+		$lpad.='[';
+		$rpad.=']';
+	}
+	return "{$lpad} ".mb_strtoupper($arr[4])." {$rpad}\n\n";
+}
+
+function html2text($html,$keeplinks=false) {
+	$html=preg_replace('#\r#','',$html);
+	//$html=preg_replace('#\n|\r#','',$html);
+	if ($keeplinks) $html=preg_replace('#<a\s.*href="(.*)".*>#iU',' [link: $1] ',$html);
+	$html=preg_replace('#<li>#i',' * ',$html);
+	$html=preg_replace_callback('#(<h)(\d)(>)(.*)(</h\d>)#iU','headcb',$html);
+	$html=preg_replace('#<p>|<div>|<ul>|<ol>#i',"\n\n",$html);
+	$html=preg_replace('#</p>|</div>|</ul>|</ol>|</li>#i',"\n\n",$html);
+	$html=preg_replace('#<br>|<br />#i',"\n",$html);
+	$html=strip_tags($html);
+	$html=preg_replace('#\s\n#',"\n",$html);
+	$html=preg_replace('# +#',' ',$html);
+	$html=preg_replace('#\n+#',"\n\n",$html);
+	$html=trim($html)."\n";
+	return $html;
+}
+
 ?>