Unlisted collections.
[wolnelektury.git] / src / catalogue / helpers.py
index bb7fa68..c4b9267 100644 (file)
@@ -1,4 +1,3 @@
-# -*- coding: utf-8 -*-
 # This file is part of Wolnelektury, licensed under GNU Affero GPLv3 or later.
 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
 #
@@ -8,14 +7,14 @@ from django.core.cache import cache
 
 from .models import Tag, Book
 from os.path import getmtime
-import cPickle
+import pickle
 from collections import defaultdict
 
 
 BOOK_CATEGORIES = ('author', 'epoch', 'genre', 'kind')
 
 _COUNTERS = None
-_COUNTER_TIME = None
+_COUNTER_TIME = 0
 
 
 def get_top_level_related_tags(tags, categories=None):
@@ -28,8 +27,17 @@ def get_top_level_related_tags(tags, categories=None):
     global _COUNTERS, _COUNTER_TIME
     # First, check that we have a valid and recent version of the counters.
     if getmtime(settings.CATALOGUE_COUNTERS_FILE) > _COUNTER_TIME:
-        with open(settings.CATALOGUE_COUNTERS_FILE) as f:
-            _COUNTERS = cPickle.load(f)
+        for i in range(10):
+            try:
+                with open(settings.CATALOGUE_COUNTERS_FILE, 'rb') as f:
+                    _COUNTERS = pickle.load(f)
+            except (EOFError, ValueError):
+                if i < 9:
+                    continue
+                else:
+                    raise
+            else:
+                break
 
     tagids = tuple(sorted(t.pk for t in tags))
     try:
@@ -39,7 +47,6 @@ def get_top_level_related_tags(tags, categories=None):
 
     related = Tag.objects.filter(pk__in=related_ids)
 
-    # TODO: do we really need that?
     if categories is not None:
         related = related.filter(category__in=categories)
 
@@ -60,7 +67,7 @@ def update_counters():
     def count_for_book(book, count_by_combination=None, parent_combinations=None):
         if not parent_combinations:
             parent_combinations = set()
-        tags = sorted(tuple(t.pk for t in book.tags.filter(category__in=('author', 'genre', 'epoch', 'kind'))))
+        tags = sorted(book.tags.filter(category__in=('author', 'genre', 'epoch', 'kind')).values_list('pk', flat=True))
         combs = list(combinations(tags))
         for c in combs:
             if c not in parent_combinations:
@@ -70,7 +77,7 @@ def update_counters():
             count_for_book(child, count_by_combination, combs_for_child)
 
     count_by_combination = defaultdict(lambda: 0)
-    for b in Book.objects.filter(parent=None):
+    for b in Book.objects.filter(findable=True, parent=None):
         count_for_book(b, count_by_combination)
 
     next_combinations = defaultdict(set)
@@ -87,14 +94,14 @@ def update_counters():
         "next": dict(next_combinations),
     }
 
-    with open(settings.CATALOGUE_COUNTERS_FILE, 'w') as f:
-        cPickle.dump(counters, f)
+    with open(settings.CATALOGUE_COUNTERS_FILE, 'wb') as f:
+        pickle.dump(counters, f)
 
 
 def get_audiobook_tags():
     audiobook_tag_ids = cache.get('audiobook_tags')
     if audiobook_tag_ids is None:
-        books_with_audiobook = Book.objects.filter(media__type__in=('mp3', 'ogg'))\
+        books_with_audiobook = Book.objects.filter(findable=True, media__type__in=('mp3', 'ogg'))\
             .distinct().values_list('pk', flat=True)
         audiobook_tag_ids = Tag.objects.filter(
             items__content_type=ContentType.objects.get_for_model(Book),