Merge remote-tracking branch 'upstream/master'

[nominatim.git] / utils / cron_ipanalyse.py

diff --git a/utils/cron_ipanalyse.py b/utils/cron_ipanalyse.py
index b344a849d08093030f074804c497bada55223aa4..66d122cd0c110432bb7820600821c6205868dd4c 100755 (executable)
--- a/utils/cron_ipanalyse.py
+++ b/utils/cron_ipanalyse.py
@@ -9,6 +9,7 @@
 import re
 import os
 import sys
+import subprocess
 from datetime import datetime, timedelta
 from collections import defaultdict
 
@@ -33,6 +34,8 @@ UA_BLOCKLIST = ()
 BLOCKCOOLOFF_DELTA=timedelta(hours=1)
 # quiet time before an IP is released from the bulk pool
 BULKCOOLOFF_DELTA=timedelta(minutes=15)
+# time to check if new accesses appear despite being blocked
+BLOCKCHECK_DELTA=timedelta(minutes=1)
 
 BULKLONG_LIMIT=8000
 BULKSHORT_LIMIT=2000
@@ -84,12 +87,16 @@ class LogEntry:
             if qp[0] == 'OPTIONS':
                 self.request = None
             else:
-                if '/search' in qp[1]:
+                if '/?' in qp[1]:
+                    self.request = 'S'
+                elif '/search' in qp[1]:
                     self.request = 'S'
                 elif '/reverse' in qp[1]:
                     self.request = 'R'
                 elif '/details' in qp[1]:
                     self.request = 'D'
+                elif '/lookup' in qp[1]:
+                    self.request = 'L'
                 else:
                     self.request = None
         self.query = e['query']
@@ -216,14 +223,18 @@ class BlockList:
 class IPstats:
 
     def __init__(self):
+        self.redirected = 0
         self.short_total = 0
         self.short_api = 0
         self.long_total = 0
         self.long_api = 0
+        self.block_total = 0
         self.bad_ua = False
 
     def add_long(self, logentry):
         self.long_total += 1
+        if logentry.retcode == 301:
+            return
         if logentry.request is not None:
             self.long_api += 1
         if not self.bad_ua:
@@ -232,20 +243,31 @@ class IPstats:
 
     def add_short(self, logentry):
         self.short_total += 1
+        if logentry.retcode == 301:
+            self.redirected += 1
+            return
         if logentry.request is not None:
             self.short_api += 1
         self.add_long(logentry)
 
+    def add_block(self, logentry):
+        self.block_total += 1
+
+    def ignores_warnings(self, wasblocked):
+        return self.block_total > 5 or (wasblocked and self.redirected > 5)
+
     def new_state(self, was_blocked, was_bulked):
         if was_blocked:
             # deblock only if the IP has been really quiet
             # (properly catches the ones that simply ignore the HTTP error)
-            return None if self.long_total < 5 else 'block'
-        if self.long_api > BLOCK_UPPER or self.short_api > BLOCK_UPPER / 3:
+            return None if self.long_total < 20 else 'block'
+        if self.long_api > BLOCK_UPPER \
+            or self.short_api > BLOCK_UPPER / 3 \
+            or (self.redirected > 100 and self.short_total == self.redirected):
                 # client totally overdoing it
                 return 'block'
         if was_bulked:
-            if self.short_total < 5:
+            if self.short_total < 20:
                 # client has stopped, debulk
                 return None
             if self.long_api > BLOCK_LIMIT or self.short_api > BLOCK_LIMIT / 3:
@@ -254,8 +276,8 @@ class IPstats:
             return 'bulk'
 
         if self.long_api > BULKLONG_LIMIT or self.short_api > BULKSHORT_LIMIT:
-            if self.bad_ua:
-                return 'uablock' # bad useragent
+            #if self.bad_ua:
+            #    return 'uablock' # bad useragent
             return 'bulk'
 
         return None
@@ -282,6 +304,7 @@ if __name__ == '__main__':
     bl = BlockList()
 
     shortstart = dt + BLOCKCOOLOFF_DELTA - BULKCOOLOFF_DELTA
+    blockstart = dt + BLOCKCOOLOFF_DELTA - BLOCKCHECK_DELTA
     notlogged = bl.whitelist | bl.blacklist
 
     stats = defaultdict(IPstats)
@@ -298,6 +321,8 @@ if __name__ == '__main__':
             stats[l.ip].add_short(l)
         if l.request is not None and l.retcode == 200:
             total200 += 1
+        if l.date > blockstart and l.retcode in (403, 429):
+            stats[l.ip].add_block(l)
 
     # adapt limits according to CPU and DB load
     fd = open("/proc/loadavg")
@@ -316,7 +341,8 @@ if __name__ == '__main__':
     # if the bulk pool is still empty, clients will be faster, avoid having
     # them blocked in this case
     if numbulks < 10:
-        BLOCK_LIMIT = 2*BLOCK_UPPER
+        BLOCK_UPPER *= 2
+        BLOCK_LIMIT = BLOCK_UPPER
 
 
     # collecting statistics
@@ -356,7 +382,7 @@ if __name__ == '__main__':
     fd.close()
 
     # TODO write logs (need to collect some statistics)
-    logstr = datetime.now().strftime('%Y-%m-%d %H:%M') + ' %s %s\n'
+    logstr = datetime.now().strftime('%d/%b/%Y:%H:%M:%S') + ' %s %s\n'
     fd = open(LOGFILE, 'a')
     if unblocked:
         fd.write(logstr % ('unblocked:', ', '.join(unblocked)))
@@ -370,4 +396,7 @@ if __name__ == '__main__':
         fd.write(logstr % (' ua block:', ', '.join(uablocked)))
     if blocked:
         fd.write(logstr % ('new block:', ', '.join(blocked)))
+    for k,v in stats.items():
+        if v.ignores_warnings(k in bl.prevblocks) and k not in notlogged and ':' not in k:
+            fd.write(logstr % ('Warning ignored:', k))
     fd.close()