From 35b3576ef8022d1cbacdb8db847583ca284c18bd Mon Sep 17 00:00:00 2001
From: =?utf8?q?Stefan=20K=C3=B6gl?= <stefan@skoegl.net>
Date: Sat, 11 Oct 2014 11:19:38 +0200
Subject: [PATCH] [Migration] remove / rewrite get_num_listened_episodes

---
 mygpo/db/couchdb/user.py                           | 24 ----------------------
 .../migrations/0006_episodehistory_index.py        | 18 ++++++++++++++++
 mygpo/history/models.py                            |  4 ++++
 mygpo/history/stats.py                             | 17 +++++++++++++++
 mygpo/users/subscriptions.py                       |  6 +++---
 5 files changed, 42 insertions(+), 27 deletions(-)
 create mode 100644 mygpo/history/migrations/0006_episodehistory_index.py
 create mode 100644 mygpo/history/stats.py

diff --git a/mygpo/db/couchdb/user.py b/mygpo/db/couchdb/user.py
index 1802adff..93f7c9ed 100644
--- a/mygpo/db/couchdb/user.py
+++ b/mygpo/db/couchdb/user.py
@@ -4,30 +4,6 @@ from mygpo.db import QueryParameterMissing
 
 
 @cache_result(timeout=60)
-def get_num_listened_episodes(user):
-
-    if not user:
-        raise QueryParameterMissing('user')
-
-    udb = get_userdata_database()
-    r = udb.view('listeners/by_user_podcast',
-            startkey    = [user.profile.uuid.hex, None],
-            endkey      = [user.profile.uuid.hex, {}],
-            reduce      = True,
-            group_level = 2,
-            stale       = 'update_after',
-        )
-
-    return map(_wrap_num_listened, r)
-
-
-def _wrap_num_listened(obj):
-    count = obj['value']
-    podcast = obj['key'][1]
-    return (podcast, count)
-
-
-@cache_result(timeout=60)
 def get_num_played_episodes(user, since=None, until={}):
     """ Number of played episodes in interval """
 
diff --git a/mygpo/history/migrations/0006_episodehistory_index.py b/mygpo/history/migrations/0006_episodehistory_index.py
new file mode 100644
index 00000000..57d8e3c5
--- /dev/null
+++ b/mygpo/history/migrations/0006_episodehistory_index.py
@@ -0,0 +1,18 @@
+# -*- coding: utf-8 -*-
+from __future__ import unicode_literals
+
+from django.db import models, migrations
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('history', '0005_episodehistoryentry'),
+    ]
+
+    operations = [
+        migrations.AlterIndexTogether(
+            name='episodehistoryentry',
+            index_together=set([('user', 'action', 'episode'), ('user', 'client', 'episode', 'action', 'timestamp')]),
+        ),
+    ]
diff --git a/mygpo/history/models.py b/mygpo/history/models.py
index e0e28316..1f201501 100644
--- a/mygpo/history/models.py
+++ b/mygpo/history/models.py
@@ -1,4 +1,5 @@
 from __future__ import unicode_literals
+from collections import Counter
 
 from django.db import models
 from django.conf import settings
@@ -105,6 +106,9 @@ class EpisodeHistoryEntry(models.Model):
     class Meta:
         index_together = [
             ['user', 'client', 'episode', 'action', 'timestamp'],
+
+            # see query in played_episode_counts()
+            ['user', 'action', 'episode'],
         ]
 
         ordering = ['-timestamp']
diff --git a/mygpo/history/stats.py b/mygpo/history/stats.py
new file mode 100644
index 00000000..5752d3cc
--- /dev/null
+++ b/mygpo/history/stats.py
@@ -0,0 +1,17 @@
+from collections import Counter
+
+from mygpo.history.models import EpisodeHistoryEntry
+
+
+def played_episode_counts(user):
+    """ number of played episodes per podcast for the given user """
+    # retrieve list of unique episodes that the user has played.
+    # for each episode only it's podcast is returned, because we don't care
+    # about which episodes exactly have been played, only the number
+    podcasts = EpisodeHistoryEntry.objects\
+                                  .filter(user=user,
+                                          action=EpisodeHistoryEntry.PLAY)\
+                                  .order_by('episode__id')\
+                                  .distinct('episode__id')\
+                                  .values_list('episode__podcast', flat=True)
+    return Counter(podcasts)
diff --git a/mygpo/users/subscriptions.py b/mygpo/users/subscriptions.py
index 780453b6..e8460a50 100644
--- a/mygpo/users/subscriptions.py
+++ b/mygpo/users/subscriptions.py
@@ -3,7 +3,7 @@ from operator import itemgetter
 
 from mygpo.utils import linearize
 from mygpo.podcasts.models import Podcast
-from mygpo.db.couchdb.user import get_num_listened_episodes
+from mygpo.history.stats import played_episode_counts
 
 
 class PodcastSorter(object):
@@ -52,9 +52,9 @@ class PodcastPercentageListenedSorter(PodcastSorter):
 
         SORT_KEY = lambda podcast: podcast.percent_listened
 
-        counts = dict(get_num_listened_episodes(self.user))
+        counts = played_episode_counts(self.user)
         for podcast in self.podcasts:
-            c = counts.get(podcast.get_id(), 0)
+            c = counts.get(podcast.id, 0)
             if podcast.episode_count:
                 podcast.percent_listened = c / float(podcast.episode_count)
                 podcast.episodes_listened = c
-- 
2.11.4.GIT