From b31046238760a50bc96864706bd5b4f65f644998 Mon Sep 17 00:00:00 2001
From: =?utf8?q?Stefan=20K=C3=B6gl?= <stefan@skoegl.net>
Date: Thu, 26 Jun 2014 18:18:05 +0200
Subject: [PATCH] [Migration] remove unused methods for unifying slugs

Slugs now must be unique, so there is no point in unifying them anymore.
---
 mygpo/administration/tasks.py                      | 103 ++++++---------------
 mygpo/administration/templates/admin/overview.html |   1 -
 .../templates/admin/unify-slugs-select.html        |  28 ------
 mygpo/administration/urls.py                       |  17 +---
 mygpo/administration/views.py                      |  51 +---------
 mygpo/core/slugs.py                                |  38 --------
 mygpo/core/tests.py                                |  28 ------
 7 files changed, 31 insertions(+), 235 deletions(-)
 rewrite mygpo/administration/tasks.py (71%)
 delete mode 100644 mygpo/administration/templates/admin/unify-slugs-select.html

diff --git a/mygpo/administration/tasks.py b/mygpo/administration/tasks.py
dissimilarity index 71%
index f7487d66..708ba67b 100644
--- a/mygpo/administration/tasks.py
+++ b/mygpo/administration/tasks.py
@@ -1,75 +1,28 @@
-from collections import Counter
-
-from mygpo.podcasts.models import Podcast
-from mygpo.cel import celery
-from mygpo.core.slugs import get_duplicate_slugs, EpisodeSlug
-from mygpo.maintenance.merge import PodcastMerger
-
-from celery.utils.log import get_task_logger
-logger = get_task_logger(__name__)
-
-
-@celery.task
-def merge_podcasts(podcast_ids, num_groups):
-    """ Task to merge some podcasts"""
-
-    logger.info('merging podcast ids %s', podcast_ids)
-
-    podcasts = Podcast.objects.filter(id__in=podcast_ids)
-
-    logger.info('merging podcasts %s', podcasts)
-
-    actions = Counter()
-
-    pm = PodcastMerger(podcasts, actions, num_groups)
-    podcast = pm.merge()
-
-    logger.info('merging result: %s', actions)
-
-    return actions, podcast
-
-
-@celery.task
-def unify_slugs(podcast):
-    """ Removes duplicate slugs of a podcast's episodes """
-
-    logger.warn('unifying slugs for podcast %s', podcast)
-    episodes = podcast.episode_set.all()
-    logger.info('found %d episodes', len(episodes))
-
-    common_title = podcast.get_common_episode_title()
-    actions = Counter()
-
-    # get episodes with duplicate slugs
-    for slug, dups in get_duplicate_slugs(episodes):
-        actions['dup-slugs'] += 1
-        # and remove their slugs
-        logger.info('Found %d duplicates for slug %s', len(dups), slug)
-        for dup in dups:
-            actions['dup-episodes'] += 1
-
-            # check if we're removing the "main" slug
-            if dup.slug == slug:
-
-                # if possible, replace it with a "merged" slug
-                if dup.merged_slugs:
-                    dup.slug = dup.merged_slugs.pop()
-                    actions['replaced-with-merged'] += 1
-                    logger.info('Replacing slug with merged slug %s', dup.slug)
-
-                # try to find a new slug
-                else:
-                    dup.slug = EpisodeSlug(dup, common_title,
-                        override_existing=True).get_slug()
-                    actions['replaced-with-new'] += 1
-                    logger.info('Replacing slug with new slug %s', dup.slug)
-
-            # if the problematic slug is a merged one, remove it
-            if slug in dup.merged_slugs:
-                actions['removed-merged'] += 1
-                logger.info('Removing merged slug %s', slug)
-                dup.merged_slugs.remove(slug)
-
-            dup.save()
-
-    return actions, podcast
+from collections import Counter
+
+from mygpo.podcasts.models import Podcast
+from mygpo.cel import celery
+from mygpo.maintenance.merge import PodcastMerger
+
+from celery.utils.log import get_task_logger
+logger = get_task_logger(__name__)
+
+
+@celery.task
+def merge_podcasts(podcast_ids, num_groups):
+    """ Task to merge some podcasts"""
+
+    logger.info('merging podcast ids %s', podcast_ids)
+
+    podcasts = Podcast.objects.filter(id__in=podcast_ids)
+
+    logger.info('merging podcasts %s', podcasts)
+
+    actions = Counter()
+
+    pm = PodcastMerger(podcasts, actions, num_groups)
+    podcast = pm.merge()
+
+    logger.info('merging result: %s', actions)
+
+    return actions, podcast
diff --git a/mygpo/administration/templates/admin/overview.html b/mygpo/administration/templates/admin/overview.html
index a6272055..1c542533 100644
--- a/mygpo/administration/templates/admin/overview.html
+++ b/mygpo/administration/templates/admin/overview.html
@@ -16,7 +16,6 @@
  <ul class="icons">
   <li><i class="icon-stethoscope"></i> <a href="{% url "admin-hostinfo" %}">{% trans "Host Information" %}</a></li>
   <li><i class="icon-resize-small"></i> <a href="{% url "admin-merge" %}?podcasts=2">{% trans "Merge Podcasts and Episodes" %}</a></li>
-  <li><i class="icon-resize-small"></i> <a href="{% url "admin-unify-slugs-select" %}">{% trans "Unify Duplicate Slugs" %}</a></li>
   <li><i class="icon-bar-chart"></i> <a href="{% url "clients" %}">{% trans "Client Stats" %}</a> (<a href="{% url "clients-json" %}">{% trans "JSON" %}</a>)</li>
   <li><i class="icon-bar-chart"></i> <a href="{% url "useragents" %}">{% trans "User-Agent Stats" %}</a></li>
   <li><i class="icon-signal"></i> <a href="{% url "stats" %}">{% trans "General Stats" %}</a> (<a href="{% url "stats-json" %}">{% trans "JSON" %}</a>)</li>
diff --git a/mygpo/administration/templates/admin/unify-slugs-select.html b/mygpo/administration/templates/admin/unify-slugs-select.html
deleted file mode 100644
index 57d3d455..00000000
--- a/mygpo/administration/templates/admin/unify-slugs-select.html
+++ /dev/null
@@ -1,28 +0,0 @@
-{% extends "base.html" %}
-{% load i18n %}
-{% load podcasts %}
-
-{% load menu %}
-{% block mainmenu %}{{ "/admin/"|main_menu }}{% endblock %}
-{% block sectionmenu %}{{ "/admin/"|section_menu:"Admin" }}{% endblock %}
-
-{% block title %}{% trans "Admin Area" %}{% endblock %}
-
-{% block header %}
- <h1>{% trans "Merge Podcasts and Episodes" %}</h1>
-{% endblock %}
-
-{% block content %}
- <form method="post" class="form-inline" action="{% url "admin-unify-slugs" %}" role="form">
-  {% csrf_token %}
-  <div class="input-group">
-   <span class="input-group-addon"><i class="icon-link"></i></span>
-   <input type="text" class="form-control" name="feed" id="feed" value="" placeholder="Feed URL">
-   <span class="input-group-btn">
-    <button class="btn btn-primary" type="submit">{% trans "OK" %}</button>
-   </span>
-  </div>
- </form>
-
-{% endblock %}
-
diff --git a/mygpo/administration/urls.py b/mygpo/administration/urls.py
index 23787a3e..d1d5c359 100644
--- a/mygpo/administration/urls.py
+++ b/mygpo/administration/urls.py
@@ -3,9 +3,8 @@ from django.conf.urls import *
 from mygpo.administration.views import Overview, MergeSelect, MergeVerify, \
          MergeProcess, MergeStatus, ClientStatsView, ClientStatsJsonView, \
          UserAgentStatsView, StatsView, StatsJsonView, HostInfo, \
-         ActivateUserView, UnifyDuplicateSlugsSelect, \
-         UnifyDuplicateSlugs, UnifySlugsStatus, MakePublisherInput, \
-         MakePublisher, MakePublisherResult
+         ActivateUserView, MakePublisherInput, MakePublisher, \
+         MakePublisherResult
 
 urlpatterns = patterns('mygpo.admin.views',
  url(r'^$',              Overview.as_view(),     name='admin-overview'),
@@ -29,18 +28,6 @@ urlpatterns = patterns('mygpo.admin.views',
      ActivateUserView.as_view(),
      name='admin-activate-user'),
 
- url(r'^unify-slugs/select$',
-     UnifyDuplicateSlugsSelect.as_view(),
-     name='admin-unify-slugs-select'),
-
- url(r'^unify-slugs/$',
-     UnifyDuplicateSlugs.as_view(),
-     name='admin-unify-slugs'),
-
- url(r'^unify-slugs/status/(?P<task_id>[^/]+)$',
-     UnifySlugsStatus.as_view(),
-     name='admin-unify-slugs-status'),
-
  url(r'^make-publisher/input$',
      MakePublisherInput.as_view(),
      name='admin-make-publisher-input'),
diff --git a/mygpo/administration/views.py b/mygpo/administration/views.py
index 74f424e8..ac51aeaa 100644
--- a/mygpo/administration/views.py
+++ b/mygpo/administration/views.py
@@ -24,7 +24,7 @@ from mygpo.administration.group import PodcastGrouper
 from mygpo.maintenance.merge import PodcastMerger, IncorrectMergeException
 from mygpo.users.models import User
 from mygpo.administration.clients import UserAgentStats, ClientStats
-from mygpo.administration.tasks import merge_podcasts, unify_slugs
+from mygpo.administration.tasks import merge_podcasts
 from mygpo.utils import get_git_head
 from mygpo.api.httpresponse import JsonResponse
 from mygpo.cel import celery
@@ -338,55 +338,6 @@ class ActivateUserView(AdminView):
 
 
 
-class UnifyDuplicateSlugsSelect(AdminView):
-    """ select a podcast for which to unify slugs """
-    template_name = 'admin/unify-slugs-select.html'
-
-
-class UnifyDuplicateSlugs(AdminView):
-    """ start slug-unification task """
-
-    def post(self, request):
-        podcast_url = request.POST.get('feed')
-
-        try:
-            podcast = Podcast.objects.get(urls__url=podcast_url)
-        except Podcast.DoesNotExist:
-            messages.error(request, _('Podcast with URL "%s" does not exist' %
-                                      (podcast_url,)))
-            return HttpResponseRedirect(reverse('admin-unify-slugs-select'))
-
-        res = unify_slugs.delay(podcast)
-        return HttpResponseRedirect(reverse('admin-unify-slugs-status',
-                    args=[res.task_id]))
-
-
-class UnifySlugsStatus(AdminView):
-    """ Displays the status of the unify-slugs operation """
-
-    template_name = 'admin/task-status.html'
-
-    def get(self, request, task_id):
-        result = merge_podcasts.AsyncResult(task_id)
-
-        if not result.ready():
-            return self.render_to_response({
-                'ready': False,
-            })
-
-        # clear cache to make merge result visible
-        # TODO: what to do with multiple frontends?
-        cache.clear()
-
-        actions, podcast = result.get()
-
-        return self.render_to_response({
-            'ready': True,
-            'actions': actions.items(),
-            'podcast': podcast,
-        })
-
-
 class MakePublisherInput(AdminView):
     """ Get all information necessary for making someone publisher """
 
diff --git a/mygpo/core/slugs.py b/mygpo/core/slugs.py
index 3e1378d4..4890f764 100644
--- a/mygpo/core/slugs.py
+++ b/mygpo/core/slugs.py
@@ -151,41 +151,3 @@ class SlugMixin(DocumentSchema):
 
         # remove from merged slugs
         self.merged_slugs = list(set(self.merged_slugs) - set([slug]))
-
-
-def get_duplicate_slugs(episodes):
-    """ Finds duplicate slugs and yields (slug, duplicates) pairs for each slug
-
-    Such a pair is only yielded for each slug that actually has a duplicate.
-    The "duplicates" list does not contain the selected "winner" of a set of
-    duplicates. """
-
-    # we build a dict of {slug: [episode1, episode2, ...], ...}
-    # for each slug all episodes are given that use this slug
-    slugs = defaultdict(list)
-
-    for episode in episodes:
-        all_slugs = filter(None, [episode.slug] + episode.merged_slugs)
-        for slug in all_slugs:
-            slugs[slug].append(episode)
-
-    # filter out unique slugs
-    dups = {s: eps for (s, eps) in slugs.items() if len(eps) > 1}
-
-    for slug, episodes in dups.items():
-        merged, main = partition(episodes, lambda e: e.slug == slug)
-
-        main, merged = list(main), list(merged)
-
-        # we want to determine exactly one winner, the rest is in "merged"
-        if len(main) == 1:
-            winner = main[0]
-
-        if len(main) < 1:
-            winner = merged.pop()
-
-        if len(main) > 1:
-            winner, merged = main[0], main[1:] + merged
-
-        # for every loser, remove the slug
-        yield slug, merged
diff --git a/mygpo/core/tests.py b/mygpo/core/tests.py
index 95cd946e..ac8ac7a1 100644
--- a/mygpo/core/tests.py
+++ b/mygpo/core/tests.py
@@ -18,38 +18,10 @@
 import unittest
 import doctest
 
-from django.test import TestCase
-
 import mygpo.utils
-from mygpo.core.slugs import get_duplicate_slugs
-from mygpo.core.models import Episode
-
-
-class UnifySlugTests(unittest.TestCase):
-
-    def test_unify(self):
-
-        from mygpo.core.models import Episode
-        a = Episode(_id='a', slug='1')
-        b = Episode(_id='b', merged_slugs=['1'])
-        c = Episode(_id='c', merged_slugs=['1'])
-
-        dups_list = list(get_duplicate_slugs([a, b, c]))
-
-        # only one duplicate slug is reported
-        self.assertEquals(len(dups_list), 1)
-
-        slug, dups = dups_list[0]
-
-        self.assertEquals(slug, '1')
-        self.assertEquals(len(dups), 2)
-        self.assertEquals(dups[0], b)
-        self.assertEquals(dups[1], c)
-        self.assertEquals(dups, [b, c])
 
 
 def suite():
     suite = unittest.TestSuite()
     suite.addTest(doctest.DocTestSuite(mygpo.utils))
-    suite.addTest(unittest.TestLoader().loadTestsFromTestCase(UnifySlugTests))
     return suite
-- 
2.11.4.GIT