only block bulks with ignored redirect

[nominatim.git] / utils / cron_ipanalyse.py

diff --git a/utils/cron_ipanalyse.py b/utils/cron_ipanalyse.py
index 1ca2267fa4144081020db88545538fb01e76559c..35db77595ddcff0788fdfec89867499fdd28d635 100755 (executable)
--- a/utils/cron_ipanalyse.py
+++ b/utils/cron_ipanalyse.py
@@ -223,6 +223,7 @@ class BlockList:
 class IPstats:
 
     def __init__(self):
+        self.redirected = 0
         self.short_total = 0
         self.short_api = 0
         self.long_total = 0
@@ -232,6 +233,8 @@ class IPstats:
 
     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:
@@ -240,6 +243,9 @@ 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)
@@ -255,7 +261,9 @@ class IPstats:
             # 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 < 20 else 'block'
-        if self.long_api > BLOCK_UPPER or self.short_api > BLOCK_UPPER / 3:
+        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: