From 924663e865fcd8aee10b18a40df46473f3c5c4df Mon Sep 17 00:00:00 2001
From: pezcurrel <pezcurrel@tiscali.it>
Date: Wed, 21 Aug 2024 10:24:41 +0200
Subject: [PATCH] Added socks5 proxy support; minor changes

---
 web/clitools/getinstinfo.php | 126 +++++++++++++++++++----------------
 1 file changed, 70 insertions(+), 56 deletions(-)

diff --git a/web/clitools/getinstinfo.php b/web/clitools/getinstinfo.php
index 150ba52..b8076c2 100755
--- a/web/clitools/getinstinfo.php
+++ b/web/clitools/getinstinfo.php
@@ -57,13 +57,15 @@ $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,
 	'bothost'=>null,
-	'bottoken'=>null
+	'bottoken'=>null,
+	'botmaxchars'=>null,
+	'_sendtoot'=>false,
+	'proxy'=>null
 ];
 
 $msglevs=['Debug', 'Info', 'Warning', 'Error', 'None'];
@@ -106,10 +108,6 @@ OPTIONS
   DEFAULT: '.ght($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
@@ -131,13 +129,18 @@ TIME SPECIFICATION
 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.
+ It will try to do it if it finds a readable «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 (must be >= 10).
+
+PROXY SUPPORT
+
+ This script supports socks5 proxy to connect to an instance. It will try to
+ use a proxy if it finds a readable «getinstinfo.ini» file in the same
+ directory it lives in, with a «proxy» parameter with this syntax:
+ [user:pass@]<host>[:port].
 
 LICENSE
 
@@ -145,6 +148,34 @@ LICENSE
  This is free software, and you are welcome to redistribute it under certain
  conditions; see <http://www.gnu.org/licenses/> for details.'.N;
 
+foreach ($argv as $val) {
+	if ($val=='-h' || $val=='--help') {
+		echo($help);
+		exit(0);
+	}
+}
+
+$inifp=__DIR__.'/getinstinfo.ini';
+$iniarr=@parse_ini_file($inifp);
+if (is_array($iniarr)) {
+	if (isset($iniarr['bothost']) && !isempty($iniarr['bothost'])) $opts['bothost']=$iniarr['bothost'];
+	if (isset($iniarr['bottoken']) && !isempty($iniarr['bottoken'])) $opts['bottoken']=$iniarr['bottoken'];
+	if (isset($iniarr['botmaxchars'])) {
+		if (preg_match('/^[0-9]+$/',$iniarr['botmaxchars'],$matches)!=1 || $matches[0]+0<10)
+			mexit('config file «'.$inifp.'»: value for «botmaxchars» must be an integer >= 10.'.N,1);
+		$opts['botmaxchars']=$iniarr['botmaxchars']+0;
+	}
+	if (!is_null($opts['bothost']) && !is_null($opts['bottoken']) && !is_null($opts['botmaxchars']))
+		$opts['_sendtoot']=true;
+	if (isset($iniarr['proxy'])) $opts['proxy']=$iniarr['proxy'];
+} else {
+	eecho(1,"Could not open «{$inifp}» (it does not exist or is not readable).\n");
+}
+
+$inifp=__DIR__.'/../conf/mustard.ini';
+$iniarr=@parse_ini_file($inifp)
+	or mexit('could not open config file «'.$inifp.'».'.N,1);
+
 for ($i=1; $i<$argc; $i++) {
 	if ($argv[$i]=='-f' || $argv[$i]=='--fetchusers') {
 		$opts['fetchusers']=true;
@@ -175,16 +206,11 @@ for ($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);
 		$i++;
 		$opts['minmsgimplev']=array_search(ucfirst(strtolower($argv[$i])),$msglevs);
-	} elseif ($argv[$i]=='-h' || $argv[$i]=='--help') {
-		echo($help);
-		exit(0);
 	} elseif (is_null($opts['hostname']) && $argv[$i][0]!=='-') {
 		$opts['hostname']=$argv[$i];
 	} else {
@@ -194,10 +220,6 @@ for ($i=1; $i<$argc; $i++) {
 
 if (is_null($opts['hostname'])) mexit('you didn’t specify an hostname (you can read the help text using «-h» or «--help»).'.N,1);
 
-$inifp=__DIR__.'/../conf/mustard.ini';
-$iniarr=@parse_ini_file($inifp)
-	or mexit('could not open config file «'.$inifp.'»'.N,1);
-
 try { $link=@mysqli_connect($iniarr['db_host'],$iniarr['db_admin_name'],$iniarr['db_admin_password'],$iniarr['db_name'],$iniarr['db_port'],$iniarr['db_socket']); }
 catch (Exception $error) { mexit('could not connect to MySQL server: '.mysqli_connect_error().'.'.N,1,true); }
 // for php versions < 8
@@ -207,14 +229,6 @@ 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))
@@ -267,10 +281,10 @@ if ($count>1) {
 	$oidata=null;
 }
 eecho(0,'«'.$opts['hostname'].'»: trying to fetch nodeinfo specs on https...'.N);
-$buf=@gurl('https://'.$opts['hostname'].'/.well-known/nodeinfo',$opts['conntimeout'],$opts['functimeout'],['Accept: application/json']);
+$buf=@gurl('https://'.$opts['hostname'].'/.well-known/nodeinfo',$opts['conntimeout'],$opts['functimeout'],['Accept: application/json'],$opts['proxy']);
 if ($buf['cont']===false) {
 	eecho(0,'«'.$opts['hostname'].'»: trying to fetch nodeinfo specs on http...'.N);
-	$buf=@gurl('http://'.$opts['hostname'].'/.well-known/nodeinfo',$opts['conntimeout'],$opts['functimeout'],['Accept: application/json']);
+	$buf=@gurl('http://'.$opts['hostname'].'/.well-known/nodeinfo',$opts['conntimeout'],$opts['functimeout'],['Accept: application/json'],$opts['proxy']);
 }
 if ($buf['cont']!==false) {
 	$buf=@json_decode($buf['cont'],true);
@@ -291,7 +305,7 @@ if ($buf['cont']!==false) {
 				$niref=array_shift($nirefs);
 				eecho(1,'«'.$opts['hostname'].'»: got and successfully parsed nodeinfo specs :-)'.N);
 				eecho(0,'«'.$opts['hostname'].'»: trying to fetch nodeinfo data...'.N);
-				$buf=@gurl($niref,$opts['conntimeout'],$opts['functimeout'],['Accept: application/json']);
+				$buf=@gurl($niref,$opts['conntimeout'],$opts['functimeout'],['Accept: application/json'],$opts['proxy']);
 				if ($buf['cont']!==false) {
 					$buf=@json_decode($buf['cont'],true);
 					if (is_array($buf)) {
@@ -336,7 +350,7 @@ if ($buf['cont']!==false) {
 }
 if ($idata['IsMastodon'] && !is_null($idata['Version']) && $idata['Version']>='4.0.0') {
 	eecho(0,'«'.$opts['hostname'].'»: trying to fetch instance info from API v2...'.N);
-	$buf=@gurl('https://'.$opts['hostname'].'/api/v2/instance',$opts['conntimeout'],$opts['functimeout'],['Accept: application/json']);
+	$buf=@gurl('https://'.$opts['hostname'].'/api/v2/instance',$opts['conntimeout'],$opts['functimeout'],['Accept: application/json'],$opts['proxy']);
 	if ($buf['cont']!==false) {
 		ckratelimit($buf['headers'],"«{$opts['hostname']}»");
 		$buf=@json_decode($buf['cont'],true);
@@ -397,7 +411,7 @@ if ($idata['IsMastodon'] && !is_null($idata['Version']) && $idata['Version']>='4
 		eecho(2,'«'.$opts['hostname'].'»: could not fetch instance info from API v2: '.$buf['emsg'].'.'.N);
 	}
 	eecho(0,'«'.$opts['hostname'].'»: trying to fetch instance extended description from API v1...'.N);
-	$buf=@gurl('https://'.$opts['hostname'].'/api/v1/instance/extended_description',$opts['conntimeout'],$opts['functimeout'],['Accept: application/json']);
+	$buf=@gurl('https://'.$opts['hostname'].'/api/v1/instance/extended_description',$opts['conntimeout'],$opts['functimeout'],['Accept: application/json'],$opts['proxy']);
 	if ($buf['cont']!==false) {
 		ckratelimit($buf['headers'],"«{$opts['hostname']}»");
 		$buf=@json_decode($buf['cont'],true);
@@ -413,7 +427,7 @@ if ($idata['IsMastodon'] && !is_null($idata['Version']) && $idata['Version']>='4
 		eecho(2,'«'.$opts['hostname'].'»: could not fetch instance extended description from API v1: '.$buf['emsg'].'.'.N);
 	}
 	eecho(0,'«'.$opts['hostname'].'»: trying to fetch instance domain blocks from API v1...'.N);
-	$buf=@gurl('https://'.$opts['hostname'].'/api/v1/instance/domain_blocks',$opts['conntimeout'],$opts['functimeout'],['Accept: application/json']);
+	$buf=@gurl('https://'.$opts['hostname'].'/api/v1/instance/domain_blocks',$opts['conntimeout'],$opts['functimeout'],['Accept: application/json'],$opts['proxy']);
 	if ($buf['cont']!==false) {
 		ckratelimit($buf['headers'],"«{$opts['hostname']}»");
 		$buf=@json_decode($buf['cont'],true);
@@ -454,7 +468,7 @@ if ($idata['IsMastodon'] && !is_null($idata['Version']) && $idata['Version']>='4
 	}
 } else {// we still try to fetch instance info from api v1, if ver. < 4.0.0, since it could be a mastodon instance older than 2.1.2, when nodeinfo was introduced
 	eecho(0,'«'.$opts['hostname'].'»: trying to fetch instance info from API v1...'.N);
-	$buf=@gurl('https://'.$opts['hostname'].'/api/v1/instance',$opts['conntimeout'],$opts['functimeout'],['Accept: application/json']);
+	$buf=@gurl('https://'.$opts['hostname'].'/api/v1/instance',$opts['conntimeout'],$opts['functimeout'],['Accept: application/json'],$opts['proxy']);
 	if ($buf['cont']!==false) {
 		ckratelimit($buf['headers'],"«{$opts['hostname']}»");
 		$buf=@json_decode($buf['cont'],true);
@@ -532,7 +546,7 @@ if ($idata['IsMastodon'] && !is_null($idata['Version']) && $idata['Version']>='4
 
 if ($idata['IsMastodon'] && !is_null($idata['Version']) && $idata['Version']>='2.1.2') {
 	eecho(0,'«'.$opts['hostname'].'»: trying to fetch instance activity info from API v1...'.N);
-	$buf=@gurl('https://'.$opts['hostname'].'/api/v1/instance/activity',$opts['conntimeout'],$opts['functimeout'],['Accept: application/json']);
+	$buf=@gurl('https://'.$opts['hostname'].'/api/v1/instance/activity',$opts['conntimeout'],$opts['functimeout'],['Accept: application/json'],$opts['proxy']);
 	if ($buf['cont']!==false) {
 		ckratelimit($buf['headers'],"«{$opts['hostname']}»");
 		$buf=@json_decode($buf['cont'],true);
@@ -551,7 +565,7 @@ if ($idata['IsMastodon'] && !is_null($idata['Version']) && $idata['Version']>='3
 	eecho(0,'«'.$opts['hostname'].'»: trying to fetch instance tags trends info from API v1...'.N);
 	$url='https://'.$opts['hostname'].'/api/v1/trends';
 	if ($idata['Version']>='3.5.0') $url.='/tags';
-	$buf=@gurl($url,$opts['conntimeout'],$opts['functimeout'],['Accept: application/json']);
+	$buf=@gurl($url,$opts['conntimeout'],$opts['functimeout'],['Accept: application/json'],$opts['proxy']);
 	if ($buf['cont']!==false) {
 		ckratelimit($buf['headers'],"«{$opts['hostname']}»");
 		$buf=@json_decode($buf['cont'],true);
@@ -784,11 +798,11 @@ if ($instanswered && !$opts['dryrun'] && isset($idata['blocks'])) {
 	}
 }
 
-if ($opts['sendtoot'] && $instanswered && $idata['IsMastodon'] && $idata['FirstSeen']==$now) {
+if ($opts['_sendtoot'] && $instanswered && $idata['IsMastodon'] && $idata['FirstSeen']==$now) {
 	$lc=['decimal_point'=>'.', 'thousands_sep'=>','];
 	gettlds();
 	eecho(0,'«'.$opts['hostname'].'»: this instance is new, trying to send an announcement toot about it...'.N);
-	$endpoint='https://'.$iniarr['bothost'].'/api/v1/statuses';
+	$endpoint='https://'.$opts['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'];
 	$toot.=N.N.'Languages: ';
@@ -818,15 +832,15 @@ if ($opts['sendtoot'] && $instanswered && $idata['IsMastodon'] && $idata['FirstS
 	(is_null($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($toot)+13>$iniarr['botmaxchars'])
+	if (postlen($toot)>$opts['botmaxchars']) {
+		while (postlen($toot)+13>$opts['botmaxchars'])
 			$toot=preg_replace('#\s+(\S|\n)+$#u','',$toot);
 		$toot.=' [continues…]';
 	}
 	$context=[
 		'http'=>[
-			'header'=>'Authorization: Bearer '.$iniarr['bottoken'].RN.
-				'Idempotency-Key'.md5($iniarr['bothost'].$now.rand(1000,9999)).RN.
+			'header'=>'Authorization: Bearer '.$opts['bottoken'].RN.
+				'Idempotency-Key'.md5($opts['bothost'].$now.rand(1000,9999)).RN.
 				'Content-type: application/x-www-form-urlencoded'.RN,
 			'method'=>'POST',
 			'content'=>http_build_query([
@@ -840,16 +854,16 @@ if ($opts['sendtoot'] && $instanswered && $idata['IsMastodon'] && $idata['FirstS
 	$context=stream_context_create($context);
 	$res=@file_get_contents($endpoint,false,$context);
 	if ($res===false) {
-		eecho(2,'«'.$opts['hostname'].'»: trying to post a toot about this new instance, could not connect to «'.$iniarr['bothost'].'».'.N);
+		eecho(2,'«'.$opts['hostname'].'»: trying to post a toot about this new instance, could not connect to «'.$opts['bothost'].'».'.N);
 	} else {
 		$httprsc=gethttpcode($http_response_header);
 		$res=@json_decode($res,true);
 		if ($httprsc!=200)
-			eecho(2,'«'.$opts['hostname'].'»: trying to post a toot about this new instance, «'.$iniarr['bothost'].'» has returned HTTP code «'.$httprsc.'».'.N);
+			eecho(2,'«'.$opts['hostname'].'»: trying to post a toot about this new instance, «'.$opts['bothost'].'» has returned HTTP code «'.$httprsc.'».'.N);
 		elseif ($res===false)
-			eecho(2,'«'.$opts['hostname'].'»: trying to post a toot about this new instance, «'.$iniarr['bothost'].'» has not returned valid JSON data.'.N);
+			eecho(2,'«'.$opts['hostname'].'»: trying to post a toot about this new instance, «'.$opts['bothost'].'» has not returned valid JSON data.'.N);
 		else
-			eecho(1,'«'.$opts['hostname'].'»: correctly posted announcement about this new instance on «'.$iniarr['bothost'].'» :-)'.N);
+			eecho(1,'«'.$opts['hostname'].'»: correctly posted announcement about this new instance on «'.$opts['bothost'].'» :-)'.N);
 	}
 }
 
@@ -863,7 +877,7 @@ if ($instanswered && $opts['fetchusers'] && $idata['IsMastodon'] && !is_null($id
 		$offset=$chunk*$limit;
 		for ($att=0; $att<$opts['udiratts']; $att++) {
 			eecho(0,'«'.$opts['hostname'].'»: trying to fetch chunk '.($chunk+1).' of users info from directory API (attempt '.($att+1).'/'.$opts['udiratts'].')...'.N);
-			$buf=@gurl('https://'.$opts['hostname'].'/api/v1/directory?local=1&order=new&limit='.$limit.'&offset='.$offset,$opts['conntimeout'],$opts['functimeout'],['Accept: application/json']);
+			$buf=@gurl('https://'.$opts['hostname'].'/api/v1/directory?local=1&order=new&limit='.$limit.'&offset='.$offset,$opts['conntimeout'],$opts['functimeout'],['Accept: application/json'],$opts['proxy']);
 			if ($buf['cont']!==false) {
 				$xrlr=ckratelimit($buf['headers'],"«{$opts['hostname']}»");
 				eecho(1,'«'.$opts['hostname'].'»: got chunk '.($chunk+1).' of users info from directory API on attempt '.($att+1).'/'.$opts['udiratts'].' (xrlr: '.$xrlr.') :-)'.N);
@@ -885,7 +899,7 @@ if ($instanswered && $opts['fetchusers'] && $idata['IsMastodon'] && !is_null($id
 							/*if (!isset($user['noindex'])) {
 								$user['noindex']=true;
 								eecho(0,'«'.$opts['hostname'].'»: «'.$user['username'].'»: «noindex» is undefined, trying to define it by fetching user’s profile page...'.N);
-								$page=gurl($user['url'],$opts['conntimeout'],$opts['functimeout'],['Accept: application/json']);
+								$page=gurl($user['url'],$opts['conntimeout'],$opts['functimeout'],['Accept: application/json'],$opts['proxy']);
 								// here ckratelimit is not needed because it's a normal web page, not json from mastodon api
 								if ($page['cont']!==false) {
 									//<meta content='noindex, noarchive' name='robots'>
@@ -906,7 +920,7 @@ if ($instanswered && $opts['fetchusers'] && $idata['IsMastodon'] && !is_null($id
 							/*$user['tags']=[];
 							if (!$user['noindex'] && !is_null($idata['Version']) && $idata['Version']>='3.3.0') {
 								eecho(0,'«'.$opts['hostname'].'»: trying to fetch tags for user «'.$user['username'].'»...'.N);
-								$tags=@gurl('https://'.$opts['hostname'].'/api/v1/accounts/'.$user['id'].'/featured_tags',$opts['conntimeout'],$opts['functimeout'],['Accept: application/json']);
+								$tags=@gurl('https://'.$opts['hostname'].'/api/v1/accounts/'.$user['id'].'/featured_tags',$opts['conntimeout'],$opts['functimeout'],['Accept: application/json'],$opts['proxy']);
 								if ($tags['cont']!==false) {
 									ckratelimit($tags['headers'],"«{$opts['hostname']}»");
 									$tags=@json_decode($tags['cont'],true);
@@ -1093,7 +1107,7 @@ function truncn($num,$tab,$col,$ctx) {
 		notify($ctx.': function «truncn»: expecting a number, got something else; returning «0».',3);
 		$num=0;
 	}
-	return($num);
+	return $num;
 }
 
 function nocrnl($str) {
@@ -1102,12 +1116,12 @@ function nocrnl($str) {
 
 function b2i($bool) {
 	($bool)	? $r=1 : $r=0;
-	return($r);
+	return $r;
 }
 
 function isempty($str) {
 	(preg_match('/^\s*$/',$str)===1) ? $r=true : $r=false;
-	return($r);
+	return $r;
 }
 
 function notify($msg,$lev,$doecho=true) {
@@ -1196,7 +1210,7 @@ function ckratelimit($headers,$ctx) {
  */
 function get_api($host, $path) {
 	global $opts;
-	$buf = @gurl('https://'.$host.$path,$opts['conntimeout'],$opts['functimeout'],['Accept: application/json']);
+	$buf = @gurl('https://'.$host.$path,$opts['conntimeout'],$opts['functimeout'],['Accept: application/json'],$opts['proxy']);
 	if ($buf['cont']!==false) {
 		ckratelimit($buf['headers'],"«{$host}»");
 		$data = json_decode($buf['cont'], true);