From 850bc96a7822955047794accc08a7ba6cfb897e1 Mon Sep 17 00:00:00 2001
From: =?utf8?q?Stefan=20K=C3=B6gl?= <stefan@skoegl.net>
Date: Sat, 13 Apr 2013 14:53:54 +0200
Subject: [PATCH] provide file type statistics for admins

---
 .../_design/episode_stats/views/filetypes/map.js   |  2 +-
 mygpo/admin/templates/admin/filetypes.html         | 36 ++++++++++++++++++++++
 mygpo/admin/urls.py                                |  7 ++++-
 mygpo/admin/views.py                               | 20 +++++++++++-
 mygpo/db/couchdb/episode.py                        | 14 +++++++++
 5 files changed, 76 insertions(+), 3 deletions(-)
 create mode 100644 mygpo/admin/templates/admin/filetypes.html

diff --git a/couchdb/_design/episode_stats/views/filetypes/map.js b/couchdb/_design/episode_stats/views/filetypes/map.js
index fa9a6d51..eadb4a93 100644
--- a/couchdb/_design/episode_stats/views/filetypes/map.js
+++ b/couchdb/_design/episode_stats/views/filetypes/map.js
@@ -15,7 +15,7 @@ function(doc)
                 /* make sure we exclude obvious non-extensions */
                 if (ext.length < 10)
                 {
-                    emit(ext, url);
+                    emit(ext.toLowerCase(), url);
                 }
             }
         }
diff --git a/mygpo/admin/templates/admin/filetypes.html b/mygpo/admin/templates/admin/filetypes.html
new file mode 100644
index 00000000..b4c06b56
--- /dev/null
+++ b/mygpo/admin/templates/admin/filetypes.html
@@ -0,0 +1,36 @@
+{% extends "base.html" %}
+{% load i18n %}
+{% load charts %}
+{% load utils %}
+
+{% load menu %}
+{% block mainmenu %}{{ "/admin/"|main_menu }}{% endblock %}
+{% block sectionmenu %}{{ "/admin/"|section_menu:"Admin" }}{% endblock %}
+
+{% block title %}{% trans "File Type Stats" %}{% endblock %}
+
+{% block header %}
+ <h1>{% trans "File Type Stats" %}</h1>
+{% endblock %}
+
+{% block content %}
+  <table>
+   <tr>
+    <th>{% trans "#" %}</th>
+    <th>{% trans "File Type" %}</th>
+    <th>{% trans "Episode URLs" %}</th>
+   </tr>
+
+   {% for filetype, num in stats %}
+    <tr>
+     <td class="numeric">{{ forloop.counter }}</td>
+     <td>{{ filetype }}</td>
+     <td>{% vertical_bar num max_num %}</td>
+    </tr>
+   {% endfor %}
+
+  </table>
+
+
+{% endblock %}
+
diff --git a/mygpo/admin/urls.py b/mygpo/admin/urls.py
index b6f631d9..ff475cb5 100644
--- a/mygpo/admin/urls.py
+++ b/mygpo/admin/urls.py
@@ -2,7 +2,8 @@ from django.conf.urls import *
 
 from mygpo.admin.views import Overview, MergeSelect, MergeVerify, \
          MergeProcess, MergeStatus, ClientStatsView, ClientStatsJsonView, \
-         UserAgentStatsView, StatsView, StatsJsonView, HostInfo
+         UserAgentStatsView, StatsView, StatsJsonView, HostInfo, \
+         FiletypeStatsView
 
 urlpatterns = patterns('mygpo.admin.views',
  url(r'^$',              Overview.as_view(),     name='admin-overview'),
@@ -21,4 +22,8 @@ urlpatterns = patterns('mygpo.admin.views',
 
  url(r'^stats$',        StatsView.as_view(), name='stats'),
  url(r'^stats\.json$',  StatsJsonView.as_view(), name='stats-json'),
+
+ url(r'^filetypes/$',
+     FiletypeStatsView.as_view(),
+     name='admin-filetypes'),
 )
diff --git a/mygpo/admin/views.py b/mygpo/admin/views.py
index 0ff7a3fc..a586c734 100644
--- a/mygpo/admin/views.py
+++ b/mygpo/admin/views.py
@@ -23,7 +23,7 @@ from mygpo.utils import get_git_head
 from mygpo.api.httpresponse import JsonResponse
 from mygpo.cel import celery
 from mygpo.db.couchdb import get_main_database
-from mygpo.db.couchdb.episode import episode_count
+from mygpo.db.couchdb.episode import episode_count, filetype_stats
 from mygpo.db.couchdb.podcast import podcast_count, podcast_for_url
 
 
@@ -276,3 +276,21 @@ class StatsJsonView(StatsView):
     def get(self, request):
         stats = self._get_stats()
         return JsonResponse(stats)
+
+
+class FiletypeStatsView(AdminView):
+
+    template_name = 'admin/filetypes.html'
+
+    def get(self, request):
+        stats = filetype_stats()
+
+        if len(stats):
+            max_num = stats.most_common(1)[0][1]
+        else:
+            max_num = 0
+
+        return self.render_to_response({
+            'max_num': max_num,
+            'stats': stats.most_common(),
+        })
diff --git a/mygpo/db/couchdb/episode.py b/mygpo/db/couchdb/episode.py
index e7b92ffd..f03d0a1f 100644
--- a/mygpo/db/couchdb/episode.py
+++ b/mygpo/db/couchdb/episode.py
@@ -1,5 +1,6 @@
 from hashlib import sha1
 from datetime import datetime
+from collections import Counter
 
 from django.core.cache import cache
 
@@ -384,6 +385,19 @@ def chapters_for_episode(episode_id):
     return map(_wrap_chapter, r)
 
 
+def filetype_stats():
+    """ Returns a filetype counter over all episodes """
+
+    db = get_main_database()
+    r = db.view('episode_stats/filetypes',
+        stale       = 'update_after',
+        reduce      = True,
+        group_level = 1,
+    )
+
+    return Counter({x['key']: x['value'] for x in r})
+
+
 def _wrap_chapter(res):
     from mygpo.users.models import Chapter
     user = res['key'][1]
-- 
2.11.4.GIT