Add new-lines to better format the whole thing

* Added user-agent and more headers
* Design has changed from: 1) search acct on instance-to-check 2) check API lookup of instance.

mastodon/check-instance-ban.py

@@ -14,35 +14,41 @@ def main():
 	if len(sys.argv) > 1:
 		DOMAIN = sys.argv[1]
 
-	res = http_get('https://api.joinmastodon.org/servers?language=en&category=&region=&ownership=&registrations=')
+	res = http_get('https://api.joinmastodon.org/servers?language=&category=&region=&ownership=&registrations=')
 	res = json.loads(res.text)
 
+	domains = [it['domain'] for it in res]
+	domains = sorted(set(domains))
+	acct = get_admin_account(DOMAIN)
+	if acct == False:
+		return 1
+
 	failed_servers = 0
 	banned_by = []
-	for it in res:
+	for domain in domains:
 		print('checking "{}" has banned "{}"... '\
-			.format(it['domain'], DOMAIN))
-		acct = get_admin_account(it['domain'])
-		if acct == False:
-			failed_servers += 1
-			continue
-
-		r = http_get('https://'+DOMAIN+'/'+acct, True)
+			.format(domain, DOMAIN))
+		r = http_get('https://{}/api/v1/accounts/lookup?acct={}'\
+			.format(domain, acct), True)
 		if r == False:
 			failed_servers += 1
 			continue
 		if r.status_code == 404:
-			banned_by.append(it['domain'])
+			banned_by.append(domain)
 			print('YES')
 		else:
 			print('NO')
 	print()
-	print('////////')
+	print('-'*30)
 	print('Instance: '+DOMAIN)
+	print()
 	print('Banned by: '+json.dumps(banned_by))
+	print()
 	print('Servers count: total={} | failed={} | used_in_percentage={}'.format(len(res), failed_servers, len(res)-failed_servers))
-	print('Percentage of bans: '+str( (len(banned_by) * 100) / (len(res)-failed_servers) )+'% has banned the instance')
 	print()
+	print('Percentage of bans: '+str(round( (len(banned_by) * 100) / (len(res)-failed_servers), 2))+'% has banned the instance')
+	print('-'*30)
+	return 0
 
 def http_get(url, show=False):
 	try:
@@ -56,7 +62,24 @@ def http_get(url, show=False):
 				return cache_read(fl)
 		if show:
 			print('  HTTP: '+url)
-		r = requests.get(url, timeout=5)
+		_headers = """
+User-Agent: Mozilla/5.0 (Windows NT 10.0; rv:102.0) Gecko/20100101 Firefox/102.0
+Accept-Language: en-US,en;q=0.5
+Connection: keep-alive
+Upgrade-Insecure-Requests: 1
+Pragma: no-cache
+Cache-Control: no-cache
+"""
+		headers = {}
+		for l in _headers.strip().splitlines():
+			l = l.strip()
+			if l == '':
+				continue
+			k = l[:l.index(':')]
+			v = l[l.index(':')+1:].strip()
+			headers[k] = v
+
+		r = requests.get(url, headers=headers, timeout=5)
 		cache_write(fl, r)
 		return r
 	except requests.exceptions.ConnectionError:
@@ -80,7 +103,7 @@ def get_admin_account(domain):
 	if r == False:
 		return False
 	js = json.loads(r.text)
-	return '@'+js['contact_account']['username']+'@'+domain
+	return js['contact_account']['username']+'@'+domain
 
 def md5(s):
 	return hashlib.md5(s.encode('utf8')).hexdigest()
@@ -100,4 +123,4 @@ def writef(f,c,t='w'):
 	w.close()
 
 if __name__ == '__main__':
-	main()
+	sys.exit(main())