From e0d6b5dbbcfff919cf0c8dfcd71d7f4e51dc729b Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Gr=C3=A9gory=20Soutad=C3=A9?= <gsoutade@neotion.com>
Date: Thu, 27 Nov 2014 12:35:41 +0100
Subject: [PATCH] Add top_downloads plugin

---
 plugins/display/top_downloads.py       | 42 ++++++++++++++++++++++
 plugins/post_analysis/top_downloads.py | 48 ++++++++++++++++++++++++++
 2 files changed, 90 insertions(+)
 create mode 100644 plugins/display/top_downloads.py
 create mode 100644 plugins/post_analysis/top_downloads.py

diff --git a/plugins/display/top_downloads.py b/plugins/display/top_downloads.py
new file mode 100644
index 0000000..47fedaf
--- /dev/null
+++ b/plugins/display/top_downloads.py
@@ -0,0 +1,42 @@
+import time
+
+from iwla import IWLA
+from iplugin import IPlugin
+from display import *
+
+class IWLADisplayTopDownloads(IPlugin):
+    def __init__(self, iwla):
+        super(IWLADisplayTopDownloads, self).__init__(iwla)
+        self.API_VERSION = 1
+        self.requires = ['IWLAPostAnalysisTopDownloads']
+
+    def hook(self):
+        top_downloads = self.iwla.getMonthStats()['top_downloads']
+
+        top_downloads = sorted(top_downloads.items(), key=lambda t: t[1], reverse=True)
+
+        index = self.iwla.getDisplayIndex()
+        
+        table = DisplayHTMLBlockTable('Top Downloads', ['URI', 'Hits'])        
+        for (uri, entrance) in top_downloads[:10]:
+            table.appendRow([uri, entrance])
+        index.appendBlock(table)
+
+        cur_time = self.iwla.getCurTime()
+        title = time.strftime('Top Downloads - %B %Y', cur_time)
+
+        filename = 'top_downloads_%d.html' % (cur_time.tm_mon)
+        path = '%d/%s' % (cur_time.tm_year, filename)
+
+        page = DisplayHTMLPage(title, path)
+        table = DisplayHTMLBlockTable('Top Downloads', ['URI', 'Hit'])        
+        for (uri, entrance) in top_downloads:
+            table.appendRow([uri, entrance])
+        page.appendBlock(table)
+                        
+        display = self.iwla.getDisplay()
+        display.addPage(page)
+
+        block = DisplayHTMLRawBlock()
+        block.setRawHTML('<a href=\'%s\'>All Downloads</a>' % (filename))
+        index.appendBlock(block)
diff --git a/plugins/post_analysis/top_downloads.py b/plugins/post_analysis/top_downloads.py
new file mode 100644
index 0000000..3ea7a38
--- /dev/null
+++ b/plugins/post_analysis/top_downloads.py
@@ -0,0 +1,48 @@
+import re
+
+from iwla import IWLA
+from iplugin import IPlugin
+
+class IWLAPostAnalysisTopDownloads(IPlugin):
+    def __init__(self, iwla):
+        super(IWLAPostAnalysisTopDownloads, self).__init__(iwla)
+        self.API_VERSION = 1
+        self.conf_requires = ['multimedia_files', 'viewed_http_codes']
+
+    def hook(self):
+        stats = self.iwla.getCurrentVisists()
+        month_stats = self.iwla.getMonthStats()
+
+        multimedia_files = self.iwla.getConfValue('multimedia_files')
+        viewed_http_codes = self.iwla.getConfValue('viewed_http_codes')
+
+        top_downloads = month_stats.get('top_downloads', {})
+
+        for (k, super_hit) in stats.items():
+            if super_hit['robot']: continue
+            for r in super_hit['requests']:
+                if r['is_page']: continue
+
+                if not self.iwla.isValidForCurrentAnalysis(r): continue
+
+                if not int(r['status']) in viewed_http_codes: continue
+
+                uri = r['extract_request']['extract_uri'].lower()
+
+                isMultimedia = False
+                for ext in multimedia_files:
+                    if uri.endswith(ext):
+                        isMultimedia = True
+                        break
+
+                if isMultimedia: continue
+
+                uri = "%s%s" % (r.get('server_name', ''),
+                                r['extract_request']['extract_uri'])
+
+                if not uri in top_downloads.keys():
+                    top_downloads[uri] = 1
+                else:
+                    top_downloads[uri] += 1
+
+        month_stats['top_downloads'] = top_downloads