From f593cc78d93d02e080dcd3f71b2993eb4f8dc96b Mon Sep 17 00:00:00 2001
From: Gregory Soutade <gregory@soutade.fr>
Date: Thu, 20 Nov 2014 14:09:01 +0100
Subject: [PATCH] Basically seems to work

---
 hooks/pre_analysis/H001_soutade.py | 2 ++
 hooks/pre_analysis/H002_robot.py   | 5 ++++-
 iwla.py                            | 4 +++-
 3 files changed, 9 insertions(+), 2 deletions(-)

diff --git a/hooks/pre_analysis/H001_soutade.py b/hooks/pre_analysis/H001_soutade.py
index 50a7932..6d683be 100644
--- a/hooks/pre_analysis/H001_soutade.py
+++ b/hooks/pre_analysis/H001_soutade.py
@@ -13,6 +13,8 @@ def hook(hits):
 
         for p in super_hit['pages']:
             if not p['is_page']: continue
+            if int(p['status']) != 200: continue
+
             if logo_re.match(p['extract_request']['extract_uri']):
                 p['is_page'] = False
                 if super_hit['viewed_pages']:
diff --git a/hooks/pre_analysis/H002_robot.py b/hooks/pre_analysis/H002_robot.py
index 2e59ad5..8a6e721 100644
--- a/hooks/pre_analysis/H002_robot.py
+++ b/hooks/pre_analysis/H002_robot.py
@@ -19,7 +19,10 @@ def hook(hits):
         if not super_hit['viewed_hits']:
             super_hit['robot'] = 1
             continue
-
+        elif not super_hit['viewed_pages']:
+# Hit only
+            super_hit['hit_only'] = 1
+            
         for hit in super_hit['pages']:
 # 3) /robots.txt read
             if hit['extract_request']['http_uri'] == '/robots.txt':
diff --git a/iwla.py b/iwla.py
index 5dbb728..25478d9 100755
--- a/iwla.py
+++ b/iwla.py
@@ -161,6 +161,7 @@ def createUser(hit):
     super_hit['last_access'] = meta_visit['last_time']
     super_hit['pages'] = [];
     super_hit['robot'] = isRobot(hit);
+    super_hit['hit_only'] = 0;
     appendHit(hit)
 
 def isRobot(hit):
@@ -276,7 +277,8 @@ def generateStats(visits):
 
         print "[%s] =>\t%d/%d" % (k, super_hit['viewed_pages'], super_hit['viewed_hits'])
         
-        stats['nb_visitors'] += 1
+        if not super_hit['hit_only']:
+            stats['nb_visitors'] += 1
         stats['viewed_bandwidth'] += super_hit['bandwith']
         stats['viewed_pages'] += super_hit['viewed_pages']
         stats['viewed_hits'] += super_hit['viewed_hits']