Merge remote-tracking branch 'upstream/master'

[nominatim.git] / utils / cron_banip.py

diff --git a/utils/cron_banip.py b/utils/cron_banip.py
index 36d0be4ebc07130066a831e5d9b9bfec331c97b2..53f5e5f13ab64a1b5527466bc2d84de35b679e73 100755 (executable)
--- a/utils/cron_banip.py
+++ b/utils/cron_banip.py
@@ -52,6 +52,9 @@ LOGFILE= BASEDIR + '/log/restricted_ip.log'
 WHITELIST = ''
 # space-separated list of IPs manually blocked
 BLACKLIST = ''
+# user-agents that should be blocked from bulk mode
+# (matched with startswith)
+UA_BLOCKLIST = ()
 
 # time before a automatically blocked IP is allowed back
 BLOCKCOOLOFF_PERIOD='1 hour'
@@ -111,6 +114,11 @@ BLOCK_LIMIT = max(BLOCK_LOWER, BLOCK_UPPER - BLOCK_LOADFAC * (dbload - 75))
 BULKLONG_LIMIT = max(BULK_LOWER, BULKLONG_LIMIT - BULK_LOADFAC * (avgload - 14))
 if len(prevbulks) > MAX_BULK_IPS:
     BLOCK_LIMIT = max(3600, BLOCK_LOWER - (len(prevbulks) - MAX_BULK_IPS)*10)
+# if the bulk pool is still empty, clients will be faster, avoid having
+# them blocked in this case
+if len(prevbulks) < 10:
+    BLOCK_LIMIT = 2*BLOCK_UPPER
+
 
 # get the new block candidates
 cur.execute("""
@@ -129,13 +137,22 @@ cur.execute("""
 
 bulkips = {}
 emergencyblocks = []
+useragentblocks = []
 
 for c in cur:
     if c[0] not in WHITELIST and c[0] not in BLACKLIST:
-        missing_agent = not c[2] or c[2].startswith('Java/1.')
+        # check for user agents that receive an immediate block
+        missing_agent = not c[2]
+        if not missing_agent:
+            for ua in UA_BLOCKLIST:
+                if c[2].startswith(ua):
+                    missing_agent = True
+                    break
         if (missing_agent or c[1] > BLOCK_UPPER) and c[0] not in prevblocks:
             newblocks.add(c[0])
-            if c[0] not in prevbulks:
+            if missing_agent:
+                useragentblocks.append(c[0])
+            else:
                 emergencyblocks.append(c[0])
         else:
             bulkips[c[0]] = c[1]
@@ -156,15 +173,16 @@ for ip in prevblocks:
         deblockcandidates.add(ip)    
         
 for ip in prevbulks:
-    if ip in bulkips:
-        if bulkips[ip] > BLOCK_LIMIT:
-            newblocks.add(ip)
-            newlyblocked.append(ip)
+    if ip not in newblocks:
+        if ip in bulkips:
+            if bulkips[ip] > BLOCK_LIMIT:
+                newblocks.add(ip)
+                newlyblocked.append(ip)
+            else:
+                newbulks.add(ip)
+            del bulkips[ip]
         else:
-            newbulks.add(ip)
-        del bulkips[ip]
-    else:
-        debulkcandidates.add(ip)
+            debulkcandidates.add(ip)
 
 # cross-check deblock candidates
 if deblockcandidates:
@@ -218,6 +236,8 @@ if bulkips:
     fd.write(logstr % ('new bulks:', ', '.join(bulkips.keys())))
 if emergencyblocks:
     fd.write(logstr % ('dir.block:', ', '.join(emergencyblocks)))
+if useragentblocks:
+    fd.write(logstr % (' ua block:', ', '.join(useragentblocks)))
 if newlyblocked:
     fd.write(logstr % ('new block:', ', '.join(newlyblocked)))
 fd.close()