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.db.models import permalink
  13 from django.utils.translation import ugettext_lazy as _
  14 from django.core.urlresolvers import reverse
  15 from django.db.models.signals import m2m_changed
  16 from django.contrib.auth.models import User
  17 from django.contrib.contenttypes.models import ContentType
  18 from django.contrib.contenttypes import generic
  19 from django.conf import settings
  20
  21 from catalogue.fields import JSONField
  22 from catalogue.models import Book, Tag
  23
  24
  25 class Poem(models.Model):
  26     slug = models.SlugField(_('slug'), max_length=120, db_index=True)
  27     text = models.TextField(_('text'))
  28     created_by = models.ForeignKey(User, null=True)
  29     created_from = JSONField(_('extra information'), null=True, blank=True)
  30     created_at = models.DateTimeField(_('creation date'), auto_now_add=True, editable=False)
  31     seen_at = models.DateTimeField(_('last view date'), auto_now_add=True, editable=False)
  32     view_count = models.IntegerField(_('view count'), default=1)
  33
  34     try:
  35         f = open(settings.LESMIANATOR_PICKLE)
  36         global_dictionary = cPickle.load(f)
  37         f.close()
  38     except:
  39         global_dictionary = {}
  40
  41     def visit(self):
  42         self.view_count += 1
  43         self.seen_at = datetime.now()
  44         self.save()
  45
  46     def __unicode__(self):
  47         return "%s (%s...)" % (self.slug, self.text[:20])
  48
  49     @staticmethod
  50     def choose_letter(word, continuations):
  51         if word not in continuations:
  52             return u'\n'
  53
  54         choices = sum((continuations[word][letter]
  55                        for letter in continuations[word]))
  56         r = randint(0, choices - 1)
  57
  58         for letter in continuations[word]:
  59             r -= continuations[word][letter]
  60             if r < 0:
  61                 return letter
  62
  63     @classmethod
  64     def write(cls, continuations=None, length=3, min_lines=2, maxlen=1000):
  65         if continuations is None:
  66             continuations = cls.global_dictionary
  67         if not continuations:
  68             return ''
  69
  70         letters = []
  71         word = u''
  72
  73         finished_stanza_verses = 0
  74         current_stanza_verses = 0
  75         verse_start = True
  76
  77         char_count = 0
  78
  79         # do `min_lines' non-empty verses and then stop,
  80         # but let Lesmianator finish his last stanza.
  81         while finished_stanza_verses < min_lines and char_count < maxlen:
  82             letter = cls.choose_letter(word, continuations)
  83             letters.append(letter)
  84             word = word[-length + 1:] + letter
  85             char_count += 1
  86
  87             if letter == u'\n':
  88                 if verse_start:
  89                     finished_stanza_verses += current_stanza_verses
  90                     current_stanza_verses = 0
  91                 else:
  92                     current_stanza_verses += 1
  93                     verse_start = True
  94             else:
  95                 verse_start = False
  96
  97         return ''.join(letters).strip()
  98
  99     def get_absolute_url(self):
 100         return reverse('get_poem', kwargs={'poem': self.slug})
 101
 102
 103 class Continuations(models.Model):
 104     pickle = models.FileField(_('Continuations file'), upload_to='lesmianator')
 105     content_type = models.ForeignKey(ContentType)
 106     object_id = models.PositiveIntegerField()
 107     content_object = generic.GenericForeignKey('content_type', 'object_id')
 108
 109     class Meta:
 110         unique_together = (('content_type', 'object_id'), )
 111
 112     def __unicode__(self):
 113         return "Continuations for: %s" % unicode(self.content_object)
 114
 115     @staticmethod
 116     def join_conts(a, b):
 117         for pre in b:
 118             a.setdefault(pre, {})
 119             for post in b[pre]:
 120                 a[pre].setdefault(post, 0)
 121                 a[pre][post] += b[pre][post]
 122         return a
 123
 124     @classmethod
 125     def for_book(cls, book, length=3):
 126         from librarian import text
 127
 128         # count from this book only
 129         output = StringIO()
 130         f = open(book.xml_file.path)
 131         text.transform(f, output, False, ('raw-text',))
 132         f.close()
 133         conts = {}
 134         last_word = ''
 135         for letter in output.getvalue().decode('utf-8').strip().lower():
 136             mydict = conts.setdefault(last_word, {})
 137             mydict.setdefault(letter, 0)
 138             mydict[letter] += 1
 139             last_word = last_word[-length+1:] + letter
 140         # add children
 141         return reduce(cls.join_conts,
 142                       (cls.get(child) for child in book.children.all()),
 143                       conts)
 144
 145     @classmethod
 146     def for_set(cls, tag):
 147         # book contains its descendants, we don't want them twice
 148         books = Book.tagged.with_any((tag,))
 149         l_tags = Tag.objects.filter(category='book', slug__in=[book.book_tag_slug() for book in books])
 150         descendants_keys = [book.pk for book in Book.tagged.with_any(l_tags)]
 151         if descendants_keys:
 152             books = books.exclude(pk__in=descendants_keys)
 153
 154         cont_tabs = (cls.get(b) for b in books)
 155         return reduce(cls.join_conts, cont_tabs)
 156
 157     @classmethod
 158     def get(cls, sth):
 159         object_type = ContentType.objects.get_for_model(sth)
 160         should_keys = set([sth.id])
 161         if isinstance(sth, Tag):
 162             should_keys = set(b.pk for b in Book.tagged.with_any((sth,)))
 163         try:
 164             obj = cls.objects.get(content_type=object_type, object_id=sth.id)
 165             if not obj.pickle:
 166                 raise cls.DoesNotExist
 167             f = open(obj.pickle.path)
 168             keys, conts = cPickle.load(f)
 169             f.close()
 170             if set(keys) != should_keys:
 171                 raise cls.DoesNotExist
 172             return conts
 173         except cls.DoesNotExist:
 174             if isinstance(sth, Book):
 175                 conts = cls.for_book(sth)
 176             elif isinstance(sth, Tag):
 177                 conts = cls.for_set(sth)
 178             else:
 179                 raise NotImplemented('Lesmianator continuations: only Book and Tag supported')
 180
 181             c, created = cls.objects.get_or_create(content_type=object_type, object_id=sth.id)
 182             c.pickle.save(sth.slug+'.p', ContentFile(cPickle.dumps((should_keys, conts))))
 183             c.save()
 184             return conts
 185
 186