From 9571bf09b650f7a7b12ecf731a9d418e77992ea7 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Gr=C3=A9gory=20Soutad=C3=A9?= <soutade@gmail.com>
Date: Wed, 26 Nov 2014 19:53:00 +0100
Subject: [PATCH] Work with time

---
 iwla.py                             | 18 +++++++++++++-----
 plugins/pre_analysis/page_to_hit.py | 10 +++++++---
 2 files changed, 20 insertions(+), 8 deletions(-)

diff --git a/iwla.py b/iwla.py
index 1a3cf0f..9201baa 100755
--- a/iwla.py
+++ b/iwla.py
@@ -72,6 +72,9 @@ class IWLA(object):
     def getCurTime(self):
         return self.meta_infos['last_time']
 
+    def getStartAnalysisTime(self):
+        return self.meta_infos['start_analysis_time']
+
     def _clearMeta(self):
         self.meta_infos = {
             'last_time' : None
@@ -200,6 +203,7 @@ class IWLA(object):
 
     def _decodeTime(self, hit):
         hit['time_decoded'] = time.strptime(hit['time_local'], conf.time_format)
+        return hit['time_decoded']
 
     def getDisplayIndex(self):
         cur_time = self.meta_infos['last_time']
@@ -337,9 +341,7 @@ class IWLA(object):
         self.current_analysis['days_stats'][cur_time.tm_mday] = stats
 
     def _newHit(self, hit):
-        self._decodeTime(hit)
-
-        t = hit['time_decoded']
+        t = self._decodeTime(hit)
 
         cur_time = self.meta_infos['last_time']
 
@@ -360,6 +362,9 @@ class IWLA(object):
 
         self.meta_infos['last_time'] = t
 
+        if not self.meta_infos['start_analysis_time']:
+            self.meta_infos['start_analysis_time'] = t
+
         if not self._decodeHTTPRequest(hit): return False
 
         for k in hit.keys():
@@ -370,8 +375,6 @@ class IWLA(object):
         return True
 
     def start(self):
-        self.cache_plugins = preloadPlugins(self.plugins, self)
-
         print '==> Analyse previous database'
 
         self.meta_infos = self._deserialize(conf.META_PATH) or self._clearMeta()
@@ -380,6 +383,10 @@ class IWLA(object):
         else:
             self._clearVisits()
 
+        self.meta_infos['start_analysis_time'] = None
+
+        self.cache_plugins = preloadPlugins(self.plugins, self)
+
         print '==> Analysing log'
 
         with open(conf.analyzed_filename) as f:
@@ -398,6 +405,7 @@ class IWLA(object):
         if self.analyse_started:
             self._generateDayStats()
             self._generateMonthStats()
+            del self.meta_infos['start_analysis_time']
             self._serialize(self.meta_infos, conf.META_PATH)
         else:
             print '==> Analyse not started : nothing to do'
diff --git a/plugins/pre_analysis/page_to_hit.py b/plugins/pre_analysis/page_to_hit.py
index 8c046b6..d704b36 100644
--- a/plugins/pre_analysis/page_to_hit.py
+++ b/plugins/pre_analysis/page_to_hit.py
@@ -1,4 +1,5 @@
 import re
+import time
 
 from iwla import IWLA
 from iplugin import IPlugin
@@ -20,15 +21,18 @@ class IWLAPreAnalysisPageToHit(IPlugin):
         return True
 
     def hook(self, iwla):
-        hits = iwla.getCurrentVisists()
+        start_time = self.iwla.getStartAnalysisTime()
+        start_time = time.mktime(start_time)
 
+        hits = iwla.getCurrentVisists()
+        viewed_http_codes = self.iwla.getConfValue('viewed_http_codes', [200, 304])
         for (k, super_hit) in hits.items():
             if super_hit['robot']: continue
 
             for p in super_hit['requests']:
                 if not p['is_page']: continue
-                if int(p['status']) != 200: continue
-                if p['time_decoded'].tm_mday != super_hit['last_access'].tm_mday: continue
+                if int(p['status']) not in viewed_http_codes: continue
+                if time.mktime(p['time_decoded']) < start_time: continue
                 uri = p['extract_request']['extract_uri']
                 for r in self.regexps:
                     if r.match(uri):