Add top_pages plugin

conf.py

@@ -16,9 +16,9 @@ DB_ROOT = './output/'
 DISPLAY_ROOT = './output/'
 
 pre_analysis_hooks = ['page_to_hit', 'robots']
-post_analysis_hooks = ['referers']
+post_analysis_hooks = ['referers', 'top_pages']
 # post_analysis_hooks = ['top_visitors', 'reverse_dns']
-display_hooks = ['top_visitors', 'all_visits', 'referers']
+display_hooks = ['top_visitors', 'all_visits', 'referers', 'top_pages']
 
 reverse_dns_timeout = 0.2
 page_to_hit_conf = [r'^.+/logo/$']

iwla.py

@@ -140,9 +140,8 @@ class IWLA(object):
         request = hit['extract_request']
 
         if 'extract_uri' in request.keys():
+            uri = request['extract_uri'] = request['http_uri']
         uri = request['extract_uri']
-        else:
-            uri = request['http_uri']
 
         hit['is_page'] = self.isPage(uri)
 

plugins/display/top_pages.py

@@ -0,0 +1,42 @@
+import time
+
+from iwla import IWLA
+from iplugin import IPlugin
+from display import *
+
+class IWLADisplayTopPages(IPlugin):
+    def __init__(self, iwla):
+        super(IWLADisplayTopPages, self).__init__(iwla)
+        self.API_VERSION = 1
+        self.requires = ['IWLAPostAnalysisTopPages']
+
+    def hook(self):
+        top_pages = self.iwla.getMonthStats()['top_pages']
+
+        top_pages = sorted(top_pages.items(), key=lambda t: t[1], reverse=True)
+
+        index = self.iwla.getDisplayIndex()
+        
+        table = DisplayHTMLBlockTable('Top Pages', ['URI', 'Entrance'])        
+        for (uri, entrance) in top_pages[:10]:
+            table.appendRow([uri, entrance])
+        index.appendBlock(table)
+
+        cur_time = self.iwla.getCurTime()
+        title = time.strftime('Top Pages - %B %Y', cur_time)
+
+        filename = 'top_pages_%d.html' % (cur_time.tm_mon)
+        path = '%d/%s' % (cur_time.tm_year, filename)
+
+        page = DisplayHTMLPage(title, path)
+        table = DisplayHTMLBlockTable('Top Pages', ['URI', 'Entrance'])        
+        for (uri, entrance) in top_pages:
+            table.appendRow([uri, entrance])
+        page.appendBlock(table)
+                        
+        display = self.iwla.getDisplay()
+        display.addPage(page)
+
+        block = DisplayHTMLRawBlock()
+        block.setRawHTML('<a href=\'%s\'>All pages</a>' % (filename))
+        index.appendBlock(block)

plugins/post_analysis/referers.py

@@ -70,6 +70,7 @@ class IWLAPostAnalysisReferers(IPlugin):
         start_time = time.mktime(start_time)
         stats = self.iwla.getCurrentVisists()
         month_stats = self.iwla.getMonthStats()
+        
         referers = month_stats.get('referers', {})
         robots_referers = month_stats.get('robots_referers', {})
         search_engine_referers = month_stats.get('search_engine_referers', {})

plugins/post_analysis/top_pages.py

@@ -0,0 +1,43 @@
+import time
+import re
+
+from iwla import IWLA
+from iplugin import IPlugin
+
+class IWLAPostAnalysisTopPages(IPlugin):
+    def __init__(self, iwla):
+        super(IWLAPostAnalysisTopPages, self).__init__(iwla)
+        self.API_VERSION = 1
+
+    def load(self):
+        self.index_re = re.compile(r'/index.*')
+        return True
+        
+    def hook(self):
+        start_time = self.iwla.getStartAnalysisTime()
+        start_time = time.mktime(start_time)
+
+        stats = self.iwla.getCurrentVisists()
+        month_stats = self.iwla.getMonthStats()
+        
+        top_pages = month_stats.get('top_pages', {})
+
+        for (k, super_hit) in stats.items():
+            if super_hit['robot']: continue
+            for r in super_hit['requests']:
+                if not r['is_page']: continue
+
+                if time.mktime(r['time_decoded']) < start_time: continue
+
+                uri = r['extract_request']['extract_uri']
+                if self.index_re.match(uri):
+                    uri = '/'
+                
+                uri = "%s%s" % (r.get('server_name', ''), uri)
+
+                if not uri in top_pages.keys():
+                    top_pages[uri] = 1
+                else:
+                    top_pages[uri] += 1
+
+        month_stats['top_pages'] = top_pages