src/lesmianator/models.py

   1 # This file is part of Wolnelektury, licensed under GNU Affero GPLv3 or later.
   2 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
   3 #
   4 from functools import reduce
   5 import pickle
   6 from pickle import PickleError
   7 from datetime import datetime
   8 from random import randint
   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.contrib.auth.models import User
  15 from django.contrib.contenttypes.models import ContentType
  16 from django.contrib.contenttypes.fields import GenericForeignKey
  17 from django.conf import settings
  18 from django.urls import reverse
  19
  20 from catalogue.models import Book, Tag
  21
  22
  23 class Poem(models.Model):
  24     slug = models.SlugField(_('slug'), max_length=120, db_index=True)
  25     text = models.TextField(_('text'))
  26     created_by = models.ForeignKey(User, models.SET_NULL, null=True)
  27     created_from = models.TextField(_('extra information'), null=True, blank=True)
  28     created_at = models.DateTimeField(_('creation date'), auto_now_add=True, editable=False)
  29     seen_at = models.DateTimeField(_('last view date'), auto_now_add=True, editable=False)
  30     view_count = models.IntegerField(_('view count'), default=1)
  31
  32     try:
  33         f = open(settings.LESMIANATOR_PICKLE, 'rb')
  34         global_dictionary = pickle.load(f)
  35         f.close()
  36     except (IOError, AttributeError, PickleError):
  37         global_dictionary = {}
  38
  39     def visit(self):
  40         self.view_count += 1
  41         self.seen_at = datetime.utcnow().replace(tzinfo=utc)
  42         self.save()
  43
  44     def __str__(self):
  45         return "%s (%s...)" % (self.slug, self.text[:20])
  46
  47     @staticmethod
  48     def choose_letter(word, continuations):
  49         if word not in continuations:
  50             return '\n'
  51
  52         choices = sum((continuations[word][letter]
  53                        for letter in continuations[word]))
  54         r = randint(0, choices - 1)
  55
  56         for letter in continuations[word]:
  57             r -= continuations[word][letter]
  58             if r < 0:
  59                 return letter
  60
  61     @classmethod
  62     def write(cls, continuations=None, length=3, min_lines=2, maxlen=1000):
  63         if continuations is None:
  64             continuations = cls.global_dictionary
  65         if not continuations:
  66             return ''
  67
  68         letters = []
  69         word = ''
  70
  71         finished_stanza_verses = 0
  72         current_stanza_verses = 0
  73         verse_start = True
  74
  75         char_count = 0
  76
  77         # do `min_lines' non-empty verses and then stop,
  78         # but let Lesmianator finish his last stanza.
  79         while finished_stanza_verses < min_lines and char_count < maxlen:
  80             letter = cls.choose_letter(word, continuations)
  81             letters.append(letter)
  82             word = word[-length + 1:] + letter
  83             char_count += 1
  84
  85             if letter == '\n':
  86                 if verse_start:
  87                     finished_stanza_verses += current_stanza_verses
  88                     current_stanza_verses = 0
  89                 else:
  90                     current_stanza_verses += 1
  91                     verse_start = True
  92             else:
  93                 verse_start = False
  94
  95         return ''.join(letters).strip()
  96
  97     def get_absolute_url(self):
  98         return reverse('get_poem', kwargs={'poem': self.slug})
  99
 100
 101 class Continuations(models.Model):
 102     pickle = models.FileField(_('Continuations file'), upload_to='lesmianator')
 103     content_type = models.ForeignKey(ContentType, models.CASCADE)
 104     object_id = models.PositiveIntegerField()
 105     content_object = GenericForeignKey('content_type', 'object_id')
 106
 107     class Meta:
 108         unique_together = (('content_type', 'object_id'), )
 109
 110     def __str__(self):
 111         return "Continuations for: %s" % str(self.content_object)
 112
 113     @staticmethod
 114     def join_conts(a, b):
 115         for pre in b:
 116             a.setdefault(pre, {})
 117             for post in b[pre]:
 118                 a[pre].setdefault(post, 0)
 119                 a[pre][post] += b[pre][post]
 120         return a
 121
 122     @classmethod
 123     def for_book(cls, book, length=3):
 124         # count from this book only
 125         wldoc = book.wldocument(parse_dublincore=False)
 126         output = wldoc.as_text(('raw-text',)).get_bytes()
 127         del wldoc
 128
 129         conts = {}
 130         last_word = ''
 131         for letter in output.decode('utf-8').strip().lower():
 132             mydict = conts.setdefault(last_word, {})
 133             mydict.setdefault(letter, 0)
 134             mydict[letter] += 1
 135             last_word = last_word[-length+1:] + letter
 136         # add children
 137         return reduce(cls.join_conts,
 138                       (cls.get(child) for child in book.children.all().iterator()),
 139                       conts)
 140
 141     @classmethod
 142     def for_set(cls, tag):
 143         books = Book.tagged_top_level([tag])
 144         cont_tabs = (cls.get(b) for b in books.iterator())
 145         return reduce(cls.join_conts, cont_tabs)
 146
 147     @classmethod
 148     def get(cls, sth):
 149         object_type = ContentType.objects.get_for_model(sth)
 150         should_keys = {sth.id}
 151         if isinstance(sth, Tag):
 152             should_keys = set(b.pk for b in Book.tagged.with_any((sth,)).iterator())
 153         try:
 154             obj = cls.objects.get(content_type=object_type, object_id=sth.id)
 155             if not obj.pickle:
 156                 raise cls.DoesNotExist
 157             f = open(obj.pickle.path, 'rb')
 158             keys, conts = pickle.load(f)
 159             f.close()
 160             if set(keys) != should_keys:
 161                 raise cls.DoesNotExist
 162             return conts
 163         except cls.DoesNotExist:
 164             if isinstance(sth, Book):
 165                 conts = cls.for_book(sth)
 166             elif isinstance(sth, Tag):
 167                 conts = cls.for_set(sth)
 168             else:
 169                 raise NotImplementedError('Lesmianator continuations: only Book and Tag supported')
 170
 171             c, created = cls.objects.get_or_create(content_type=object_type, object_id=sth.id)
 172             c.pickle.save(sth.slug+'.p', ContentFile(pickle.dumps((should_keys, conts))))
 173             c.save()
 174             return conts