src/catalogue/helpers.py

   1 # -*- coding: utf-8 -*-
   2 # This file is part of Wolnelektury, licensed under GNU Affero GPLv3 or later.
   3 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
   4 #
   5 from django.conf import settings
   6 from .models import Tag, Book
   7 from os.path import getmtime
   8 import cPickle
   9 from collections import defaultdict
  10
  11
  12 BOOK_CATEGORIES = ('author', 'epoch', 'genre', 'kind')
  13
  14 _COUNTERS = None
  15 _COUNTER_TIME = None
  16
  17
  18 def get_top_level_related_tags(tags, categories=None):
  19     """
  20     Finds tags related to given tags through books, and counts their usage.
  21
  22     Takes ancestry into account: if a tag is applied to a book, its
  23     usage on the book's descendants is ignored.
  24     """
  25     global _COUNTERS, _COUNTER_TIME
  26     # First, check that we have a valid and recent version of the counters.
  27     if getmtime(settings.CATALOGUE_COUNTERS_FILE) > _COUNTER_TIME:
  28         with open(settings.CATALOGUE_COUNTERS_FILE) as f:
  29             _COUNTERS = cPickle.load(f)
  30
  31     tagids = tuple(sorted(t.pk for t in tags))
  32     try:
  33         related_ids = _COUNTERS['next'][tagids]
  34     except KeyError:
  35         return
  36
  37     related = Tag.objects.filter(pk__in=related_ids)
  38
  39     # TODO: do we really need that?
  40     if categories is not None:
  41         related = related.filter(category__in=categories)
  42
  43     for tag in related:
  44         tag.count = _COUNTERS['count'][tuple(sorted(tagids + (tag.pk,)))]
  45         yield tag
  46
  47
  48 def update_counters():
  49     def combinations(things):
  50         if len(things):
  51             for c in combinations(things[1:]):
  52                 yield c
  53                 yield (things[0],) + c
  54         else:
  55             yield ()
  56
  57     def count_for_book(book, count_by_combination=None, parent_combinations=None):
  58         if not parent_combinations:
  59             parent_combinations = set()
  60         tags = sorted(tuple(t.pk for t in book.tags.filter(category__in=('author', 'genre', 'epoch', 'kind'))))
  61         combs = list(combinations(tags))
  62         for c in combs:
  63             if c not in parent_combinations:
  64                 count_by_combination[c] += 1
  65         combs_for_child = set(list(parent_combinations) + combs)
  66         for child in book.children.all():
  67             count_for_book(child, count_by_combination, combs_for_child)
  68
  69     count_by_combination = defaultdict(lambda: 0)
  70     for b in Book.objects.filter(parent=None):
  71         count_for_book(b, count_by_combination)
  72
  73     next_combinations = defaultdict(set)
  74     # Now build an index of all combinations.
  75     for c in count_by_combination.keys():
  76         if not c:
  77             continue
  78         for n in c:
  79             rest = tuple(x for x in c if x != n)
  80             next_combinations[rest].add(n)
  81
  82     counters = {
  83         "count": dict(count_by_combination),
  84         "next": dict(next_combinations),
  85     }
  86
  87     with open(settings.CATALOGUE_COUNTERS_FILE, 'w') as f:
  88         cPickle.dump(counters, f)