From 639e51a0b4203a033a5202765a649a95abf80677 Mon Sep 17 00:00:00 2001
From: sneak <sneak@sneak.berlin>
Date: Mon, 20 Apr 2020 08:31:05 -0700
Subject: [PATCH] add AP instance finder

---
 findmastos/findmastos.py | 131 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 131 insertions(+)
 create mode 100755 findmastos/findmastos.py

diff --git a/findmastos/findmastos.py b/findmastos/findmastos.py
new file mode 100755
index 0000000..54c1d4f
--- /dev/null
+++ b/findmastos/findmastos.py
@@ -0,0 +1,131 @@
+#!/usr/bin/env python3
+
+import requests
+import time
+import json
+import pprint
+import socket
+import sys
+import re
+import random
+import os
+
+# shouts to @mewmew@blob.cat:
+# https://blob.cat/notice/9u8rxBOMWtDFFQG81Y
+SEEDS = [
+    "blob.cat",
+    "s.sneak.berlin"
+]
+
+TIMEOUT_SECS = 5
+
+fakehostre = re.compile('gab\.best$')
+
+simplehostre = re.compile('^[A-Za-z0-9\-\.]+$')
+
+def main():
+    os.environ['PYTHONUNBUFFERED'] = '1'
+    knownhosts = {}
+    goodhosts = {}
+    badhosts = {}
+    fetchqueue = {}
+
+    for seed in SEEDS:
+        fetchqueue[seed] = True
+
+    if os.path.exists('fetchqueue.json'):
+        with open('fetchqueue.json') as f:
+            t = json.load(f)
+            for i in t:
+                fetchqueue[i] = True
+            print(f"loaded {len(t)} fetchqueue items")
+
+    if os.path.exists('goodhosts.json'):
+        with open('goodhosts.json') as f:
+            t = json.load(f)
+            for i in t:
+                goodhosts[i] = True
+        print(f"loaded {len(t)} goodhosts")
+
+    if os.path.exists('badhosts.json'):
+        with open('badhosts.json') as f:
+            t = json.load(f)
+            for i in t:
+                badhosts[i] = True
+        print(f"loaded {len(t)} badhosts")
+
+    def fetchnext(hostname):
+        hostname = str(hostname)
+        del fetchqueue[hostname]
+
+        try:
+            addr = socket.gethostbyname(hostname)
+        except:
+            print(f"skipping fetch from {hostname}: dns lookup failed")
+            return
+
+        fail = False
+        try:
+            url = "https://" + hostname + "/api/v1/instance/peers"
+            print(f"GET {url}")
+            r = requests.get(url, timeout=TIMEOUT_SECS)
+            count = len(r.json())
+        except:
+            fail = True
+
+        if fail:
+            print(f"{hostname}: api fetch failed")
+            badhosts[hostname] = True
+        else:
+            print(f"{hostname}: api fetch got {count}")
+            goodhosts[hostname] = True
+            for newhn in r.json():
+                if newhn is None:
+                    continue
+                newhn = str(newhn)
+                if newhn in fetchqueue:
+                    continue
+                if newhn in goodhosts:
+                    continue
+                if newhn in badhosts:
+                    continue
+
+                if fakehostre.search(newhn):
+                    print(f"skipping add of {newhn} due to re spam match")
+                    continue
+
+                if simplehostre.search(newhn):
+                    print(f"got new hostname '{newhn}' from api on {hostname}")
+                    fetchqueue[newhn] = True
+                else:
+                    print(f"got weird hostname '{newhn}' from api on {hostname}, skipping")
+
+        writefiles()
+
+    def writefiles():
+        print(f"writing files. fetchqueue={len(fetchqueue.keys())} goodhosts={len(goodhosts.keys())} badhosts={len(badhosts.keys())}")
+        gh = list(goodhosts.keys())
+        gh.sort()
+        bh = list(badhosts.keys())
+        bh.sort()
+        fq = list(fetchqueue.keys())
+        fq.sort()
+        with open('.goodhosts.json.tmp', 'w') as outfile:
+            json.dump(gh, outfile)
+        os.rename('.goodhosts.json.tmp', 'goodhosts.json')
+        with open('.badhosts.json.tmp', 'w') as outfile:
+            json.dump(bh, outfile)
+        os.rename('.badhosts.json.tmp', 'badhosts.json')
+        with open('.fetchqueue.json.tmp', 'w') as outfile:
+            json.dump(fq, outfile)
+        os.rename('.fetchqueue.json.tmp', 'fetchqueue.json')
+
+    while len(fetchqueue.keys()):
+        tl = list(fetchqueue.keys())
+        random.shuffle(tl)
+        for h in tl:
+            print(f"{len(fetchqueue.keys())} hosts in fetchqueue")
+            fetchnext(h)
+
+if __name__ == "__main__":
+    main()