apps/lesmianator/models.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 import cPickle
   6 from datetime import datetime
   7 from random import randint
   8 from StringIO import StringIO
   9
  10 from django.core.files.base import ContentFile
  11 from django.db import models
  12 from django.utils.timezone import utc
  13 from django.utils.translation import ugettext_lazy as _
  14 from django.core.urlresolvers import reverse
  15 from django.contrib.auth.models import User
  16 from django.contrib.contenttypes.models import ContentType
  17 from django.contrib.contenttypes.fields import GenericForeignKey
  18 from django.conf import settings
  19
  20 from jsonfield import JSONField
  21 from catalogue.models import Book, Tag
  22
  23
  24 class Poem(models.Model):
  25     slug = models.SlugField(_('slug'), max_length=120, db_index=True)
  26     text = models.TextField(_('text'))
  27     created_by = models.ForeignKey(User, null=True)
  28     created_from = JSONField(_('extra information'), null=True, blank=True)
  29     created_at = models.DateTimeField(_('creation date'), auto_now_add=True, editable=False)
  30     seen_at = models.DateTimeField(_('last view date'), auto_now_add=True, editable=False)
  31     view_count = models.IntegerField(_('view count'), default=1)
  32
  33     try:
  34         f = open(settings.LESMIANATOR_PICKLE)
  35         global_dictionary = cPickle.load(f)
  36         f.close()
  37     except:
  38         global_dictionary = {}
  39
  40     def visit(self):
  41         self.view_count += 1
  42         self.seen_at = datetime.utcnow().replace(tzinfo=utc)
  43         self.save()
  44
  45     def __unicode__(self):
  46         return "%s (%s...)" % (self.slug, self.text[:20])
  47
  48     @staticmethod
  49     def choose_letter(word, continuations):
  50         if word not in continuations:
  51             return u'\n'
  52
  53         choices = sum((continuations[word][letter]
  54                        for letter in continuations[word]))
  55         r = randint(0, choices - 1)
  56
  57         for letter in continuations[word]:
  58             r -= continuations[word][letter]
  59             if r < 0:
  60                 return letter
  61
  62     @classmethod
  63     def write(cls, continuations=None, length=3, min_lines=2, maxlen=1000):
  64         if continuations is None:
  65             continuations = cls.global_dictionary
  66         if not continuations:
  67             return ''
  68
  69         letters = []
  70         word = u''
  71
  72         finished_stanza_verses = 0
  73         current_stanza_verses = 0
  74         verse_start = True
  75
  76         char_count = 0
  77
  78         # do `min_lines' non-empty verses and then stop,
  79         # but let Lesmianator finish his last stanza.
  80         while finished_stanza_verses < min_lines and char_count < maxlen:
  81             letter = cls.choose_letter(word, continuations)
  82             letters.append(letter)
  83             word = word[-length + 1:] + letter
  84             char_count += 1
  85
  86             if letter == u'\n':
  87                 if verse_start:
  88                     finished_stanza_verses += current_stanza_verses
  89                     current_stanza_verses = 0
  90                 else:
  91                     current_stanza_verses += 1
  92                     verse_start = True
  93             else:
  94                 verse_start = False
  95
  96         return ''.join(letters).strip()
  97
  98     def get_absolute_url(self):
  99         return reverse('get_poem', kwargs={'poem': self.slug})
 100
 101
 102 class Continuations(models.Model):
 103     pickle = models.FileField(_('Continuations file'), upload_to='lesmianator')
 104     content_type = models.ForeignKey(ContentType)
 105     object_id = models.PositiveIntegerField()
 106     content_object = GenericForeignKey('content_type', 'object_id')
 107
 108     class Meta:
 109         unique_together = (('content_type', 'object_id'), )
 110
 111     def __unicode__(self):
 112         return "Continuations for: %s" % unicode(self.content_object)
 113
 114     @staticmethod
 115     def join_conts(a, b):
 116         for pre in b:
 117             a.setdefault(pre, {})
 118             for post in b[pre]:
 119                 a[pre].setdefault(post, 0)
 120                 a[pre][post] += b[pre][post]
 121         return a
 122
 123     @classmethod
 124     def for_book(cls, book, length=3):
 125         # count from this book only
 126         output = StringIO()
 127         wldoc = book.wldocument(parse_dublincore=False)
 128         output = wldoc.as_text(('raw-text',)).get_string()
 129         del wldoc
 130
 131         conts = {}
 132         last_word = ''
 133         for letter in output.decode('utf-8').strip().lower():
 134             mydict = conts.setdefault(last_word, {})
 135             mydict.setdefault(letter, 0)
 136             mydict[letter] += 1
 137             last_word = last_word[-length+1:] + letter
 138         # add children
 139         return reduce(cls.join_conts,
 140                       (cls.get(child) for child in book.children.all().iterator()),
 141                       conts)
 142
 143     @classmethod
 144     def for_set(cls, tag):
 145         # book contains its descendants, we don't want them twice
 146         books = Book.tagged.with_any((tag,))
 147         l_tags = Tag.objects.filter(category='book', slug__in=[book.book_tag_slug() for book in books.iterator()])
 148         descendants_keys = [book.pk for book in Book.tagged.with_any(l_tags).iterator()]
 149         if descendants_keys:
 150             books = books.exclude(pk__in=descendants_keys)
 151
 152         cont_tabs = (cls.get(b) for b in books.iterator())
 153         return reduce(cls.join_conts, cont_tabs)
 154
 155     @classmethod
 156     def get(cls, sth):
 157         object_type = ContentType.objects.get_for_model(sth)
 158         should_keys = set([sth.id])
 159         if isinstance(sth, Tag):
 160             should_keys = set(b.pk for b in Book.tagged.with_any((sth,)).iterator())
 161         try:
 162             obj = cls.objects.get(content_type=object_type, object_id=sth.id)
 163             if not obj.pickle:
 164                 raise cls.DoesNotExist
 165             f = open(obj.pickle.path)
 166             keys, conts = cPickle.load(f)
 167             f.close()
 168             if set(keys) != should_keys:
 169                 raise cls.DoesNotExist
 170             return conts
 171         except cls.DoesNotExist:
 172             if isinstance(sth, Book):
 173                 conts = cls.for_book(sth)
 174             elif isinstance(sth, Tag):
 175                 conts = cls.for_set(sth)
 176             else:
 177                 raise NotImplemented('Lesmianator continuations: only Book and Tag supported')
 178
 179             c, created = cls.objects.get_or_create(content_type=object_type, object_id=sth.id)
 180             c.pickle.save(sth.slug+'.p', ContentFile(cPickle.dumps((should_keys, conts))))
 181             c.save()
 182             return conts
 183
 184