Merge branch 'master' of https://github.com/Charcoal-SE/SmokeDetector

continuousintegration.py

@@ -1,7 +1,9 @@
 import socket
 import requests
-import sys
+import re
 from globalvars import GlobalVars
+import datetime
+import json
 
 def watchCi():
     HOST = ''
@@ -21,16 +23,23 @@ def watchCi():
 
     while 1:
         conn, addr = s.accept()
-        print 'Received request from ' + addr[0]
-        r=requests.get('https://api.github.com/repos/Charcoal-SE/SmokeDetector/git/refs/heads/master')
+        addr_host = socket.gethostbyaddr(addr[0])[0]
+        is_circleci = True if re.compile(r"ec2-\d{1,3}-\d{1,3}-\d{1,3}-\d{1,3}.compute-1.amazonaws.com").search(addr_host) else False
+        print 'Received request from ' + addr[0] + " ; " + "verified as CircleCI" if is_circleci else "NOT verified as CircleCI!"
+        if not is_circleci:
+            continue
+        r = requests.get('https://api.github.com/repos/Charcoal-SE/SmokeDetector/git/refs/heads/master')
         latest_sha = r.json()["object"]["sha"]
         r = requests.get('https://api.github.com/repos/Charcoal-SE/SmokeDetector/commits/' + latest_sha + '/statuses')
-        states = []
         for status in r.json():
             state = status["state"]
-            states.append(state)
-        if "success" in states:
-            GlobalVars.charcoal_hq.send_message("CI build passed. Ready to pull!")
-        elif "error" in states or "failure" in states:
-            GlobarVars.charcoal_hq.send_message("CI build failed, *someone* (prolly Undo) borked something!")
+            target_url = status["target_url"]
+            if state == "success":
+                if datetime.datetime.strptime(status["updated_at"], '%Y-%m-%dT%H:%M:%SZ') > datetime.datetime.now()-datetime.timedelta(seconds=10):
+                    GlobalVars.charcoal_hq.send_message("[CI build passed](%s). Ready to pull!" % target_url)
+                    continue
+            elif state == "error" or state == "failure":
+                if datetime.datetime.strptime(status["updated_at"], '%Y-%m-%dT%H:%M:%SZ') > datetime.datetime.now()-datetime.timedelta(seconds=10):
+                    GlobalVars.charcoal_hq.send_message("[CI build failed](%s), *someone* (prolly Undo) borked something!" % target_url)
+                    continue
     s.close()

findspam.py

@@ -4,10 +4,10 @@
 
 class FindSpam:
     rules = [
-     {'regex': u"(?i)\\b(baba(ji)?|fifa.*coin|nike|bajotz|vashi?k[ae]r[ae]n|sumer|kolcak|porn|molvi|judi bola|ituBola.com|lost lover|11s|acai|skin care|LifeForce|swtor2credits|me2.do|black magic|bam2u|Neuro(3X|flexyn)|Nutra|TesteroneXL|Bowtrol|Slim ?Genix|Cleanse EFX|Babyliss ?Pro|Forskolin|Blackline Elite|TestCore Pro|Xtreme Antler|Maxx Test 3000|Cheap Wigs?|(Improve )?Brain Power|aging skin|acne( prone)? skin|(skin )?eye serum|skin (serum|eye)|fake (passports?|driver'?s? licen[cs]e|ID cards?)|bagprada)\\b|ಌ|(support|service|helpline)( phone)? number|1[ -]?866[ -]?978[ -]?6819", 'all': True,
+     {'regex': u"(?i)\\b(baba(ji)?|fifa.*coin|nike|bajotz|vashi?k[ae]r[ae]n|sumer|kolcak|porn|molvi|judi bola|ituBola.com|lost lover|11s|acai|skin care|rejuvenated skin|LifeForce|swtor2credits|me2.do|black magic|bam2u|Neuro(3X|flexyn)|Nutra|TesteroneXL|Bowtrol|Slim ?Genix|Cleanse EFX|Babyliss ?Pro|Forskolin|Blackline Elite|TestCore Pro|Xtreme Antler|Maxx Test 3000|Cheap Wigs?|jivam|(Improve )?Brain Power|aging skin|acne( prone)? skin|(skin )?eye serum|skin (serum|eye)|fake (passports?|driver'?s? licen[cs]e|ID cards?)|bagprada)\\b|ಌ|(support|service|helpline)( phone)? number|1[ -]?866[ -]?978[ -]?6819", 'all': True,
         'sites': [], 'reason': "Bad keyword in {}", 'title': True, 'body': True, 'username': True},
      {'regex': u"(?i)\\b(fifabay)\\b", 'all': True, 'reason': "Bad keyword in {}", 'sites': [], 'title': True, 'body': True, 'username': True},
-     {'regex': u"(?i)\\b(weight (loo?s[es]|reduction)|muscles? build(ing)?|muscles?( (grow(th)?|diets?))?|anti aging|SkinCentric|loo?s[es] weight|wrinkles?)\\b", 'all': True,
+     {'regex': u"(?i)\\b(weight (body ?builder|loo?s[es]|reduction)|muscles? build(ing)?|muscles?( (grow(th)?|diets?))?|anti aging|SkinCentric|loo?s[es] weight|wrinkles?)\\b", 'all': True,
         'sites': ["fitness.stackexchange.com"], 'reason': "Bad keyword in {}", 'title': True, 'body': False, 'username': True},
      {'regex': u"(?i)^(?:(?=.*?\\b(?:online|hd)\\b)(?=.*?(?:free|full|unlimited)).*?movies?\\b|(?=.*?\\b(?:acai|kisn)\\b)(?=.*?care).*products?\\b|(?=.*?packer).*mover)", 'all': True,
         'sites': [], 'reason': "Bad keywords in {}", 'title': True, 'body': False, 'username': True},
@@ -20,7 +20,8 @@ class FindSpam:
      {'regex': u"^(?=.*[0-9])[^a-zA-Z]*$", 'all': True, 'sites': [], 'reason': "Numbers-only title", 'title': True, 'body': False, 'username': False},
      {'regex': u"https?://[a-zA-Z0-9_.-]+\\.[a-zA-Z]{2,4}(/[a-zA-Z0-9_/?=.-])?", 'all': True,
         'sites': ["stackoverflow.com", "superuser.com", "askubuntu.com"], 'reason': "URL in title", 'title': True, 'body': False, 'username': False},
-     {'regex': u"(?i)\\b(cfpchampionship2015playofflive|optimalstackfacts\.org|maletestosteronebooster\.org|x4facts\.com|tripleeffectseyeserum\.org|healthcaresup\.com|garciniacambogiaprofacts\.com|filerepairforum\.com)\\b", 'sites': [], 'all': True, 'reason': "Blacklisted website", 'title': True, 'body': True, 'username': True}
+     {'regex': u"(?i)(orabank|cfpchampionship2015playofflive|optimalstackfacts\.org|maletestosteronebooster\.org|x4facts\.com|tripleeffectseyeserum\.org|healthcaresup\.com|garciniacambogiaprofacts\.com|filerepairforum\.com)", 'sites': [], 'all': True, 'reason': "Blacklisted website", 'title': True, 'body': True, 'username': True},
+     {'regex': u"([a-zA-Z])\\1{10,}", 'all': True, 'sites': [], 'reason': "Repeating characters in {}", 'title': True, 'body': True, 'username': False}
     ]
 
     @staticmethod

test/test_regexes.py

@@ -18,7 +18,8 @@
      ('How to avoid objects when traveling at greater than .75 light speed. or How Not to Go SPLAT?', '', 'bowlturner', '', False),
      ('Is it unfair to regrade prior work after detecting cheating?', '', 'Village', '', False),
      ('Inner workings of muscles', '', '', 'fitness.stackexchange.com', False),
-     ('Cannot access http://stackoverflow.com/ with proxy enabled', '', '', 'superuser.com', False)
+     ('Cannot access http://stackoverflow.com/ with proxy enabled', '', '', 'superuser.com', False),
+     ('kkkkkkkkkkkkkkkkkkkkkkkkkkkk', '<p>bbbbbbbbbbbbbbbbbbbbbb</p>', '', 'stackoverflow.com', True)
 ])
 
 def test_regexes(title, body, username, site, match):