Made dcparser a submodule of librarian.
[wolnelektury.git] / catalogue / models.py
1 # -*- coding: utf-8 -*-
2 from django.db import models
3 from django.db.models import permalink, Q
4 from django.utils.translation import ugettext_lazy as _
5 from django.contrib.auth.models import User
6 from django.core.files import File
7 from django.template.loader import render_to_string
8 from django.utils.safestring import mark_safe
9
10 from newtagging.models import TagBase
11 from newtagging import managers
12
13 from librarian import html, dcparser
14
15
16 TAG_CATEGORIES = (
17     ('author', _('author')),
18     ('epoch', _('epoch')),
19     ('kind', _('kind')),
20     ('genre', _('genre')),
21     ('theme', _('theme')),
22     ('set', _('set')),
23 )
24
25
26 class TagSubcategoryManager(models.Manager):
27     def __init__(self, subcategory):
28         super(TagSubcategoryManager, self).__init__()
29         self.subcategory = subcategory
30         
31     def get_query_set(self):
32         return super(TagSubcategoryManager, self).get_query_set().filter(category=self.subcategory)
33
34
35 class Tag(TagBase):
36     name = models.CharField(_('name'), max_length=50, unique=True, db_index=True)
37     slug = models.SlugField(_('slug'), unique=True, db_index=True)
38     sort_key = models.SlugField(_('sort key'), db_index=True)
39     category = models.CharField(_('category'), max_length=50, blank=False, null=False, 
40         db_index=True, choices=TAG_CATEGORIES)
41     description = models.TextField(blank=True)
42     
43     user = models.ForeignKey(User, blank=True, null=True)
44     
45     def has_description(self):
46         return len(self.description) > 0
47     has_description.short_description = _('description')
48     has_description.boolean = True
49
50     @permalink
51     def get_absolute_url(self):
52         return ('catalogue.views.tagged_object_list', [self.slug])
53     
54     class Meta:
55         ordering = ('sort_key',)
56         verbose_name = _('tag')
57         verbose_name_plural = _('tags')
58     
59     def __unicode__(self):
60         return self.name
61
62     @staticmethod
63     def get_tag_list(tags):
64         if isinstance(tags, basestring):
65             tag_slugs = tags.split('/')
66             return [Tag.objects.get(slug=slug) for slug in tag_slugs]
67         else:
68             return TagBase.get_tag_list(tags)
69
70
71 class Book(models.Model):
72     title = models.CharField(_('title'), max_length=120)
73     slug = models.SlugField(_('slug'), unique=True, db_index=True)
74     description = models.TextField(_('description'), blank=True)
75     created_at = models.DateTimeField(_('creation date'), auto_now=True)
76     _short_html = models.TextField(_('short HTML'), editable=False)
77     
78     # Formats
79     xml_file = models.FileField(_('XML file'), upload_to='books/xml', blank=True)
80     pdf_file = models.FileField(_('PDF file'), upload_to='books/pdf', blank=True)
81     odt_file = models.FileField(_('ODT file'), upload_to='books/odt', blank=True)
82     html_file = models.FileField(_('HTML file'), upload_to='books/html', blank=True)
83     
84     objects = managers.ModelTaggedItemManager(Tag)
85     tags = managers.TagDescriptor(Tag)
86     
87     def short_html(self):
88         if len(self._short_html):
89             return mark_safe(self._short_html)
90         else:
91             tags = self.tags.filter(~Q(category__in=('set', 'theme')))
92             tags = [u'<a href="%s">%s</a>' % (tag.get_absolute_url(), tag.name) for tag in tags]
93
94             formats = []
95             if self.html_file:
96                 formats.append(u'<a href="%s">Czytaj online</a>' % self.html_file.url)
97             if self.pdf_file:
98                 formats.append(u'<a href="%s">Plik PDF</a>' % self.pdf_file.url)
99             if self.odt_file:
100                 formats.append(u'<a href="%s">Plik ODT</a>' % self.odt_file.url)
101             
102             self._short_html = render_to_string('catalogue/book_short.html',
103                 {'book': self, 'tags': tags, 'formats': formats})
104             self.save()
105             return mark_safe(self._short_html)
106     
107     def has_description(self):
108         return len(self.description) > 0
109     has_description.short_description = _('description')
110     has_description.boolean = True
111     
112     def has_pdf_file(self):
113         return bool(self.pdf_file)
114     has_pdf_file.short_description = 'PDF'
115     has_pdf_file.boolean = True
116     
117     def has_odt_file(self):
118         return bool(self.odt_file)
119     has_odt_file.short_description = 'ODT'
120     has_odt_file.boolean = True
121     
122     def has_html_file(self):
123         return bool(self.html_file)
124     has_html_file.short_description = 'HTML'
125     has_html_file.boolean = True
126
127     @staticmethod
128     def from_xml_file(xml_file):
129         from tempfile import NamedTemporaryFile
130         from slughifi import slughifi
131         from markupstring import MarkupString
132         
133         # Read book metadata
134         book_info = dcparser.parse(xml_file)
135         book = Book(title=book_info.title, slug=slughifi(book_info.title))
136         book.save()
137         
138         book_tags = []
139         for category in ('kind', 'genre', 'author', 'epoch'):    
140             tag_name = getattr(book_info, category)
141             tag_sort_key = tag_name
142             if category == 'author':
143                 tag_sort_key = tag_name.last_name
144                 tag_name = ' '.join(tag_name.first_names) + ' ' + tag_name.last_name
145             tag, created = Tag.objects.get_or_create(name=tag_name,
146                 slug=slughifi(tag_name), sort_key=slughifi(tag_sort_key), category=category)
147             tag.save()
148             book_tags.append(tag)
149         book.tags = book_tags
150         
151         # Save XML and HTML files
152         book.xml_file.save('%s.xml' % book.slug, File(file(xml_file)), save=False)
153         
154         html_file = NamedTemporaryFile()
155         html.transform(book.xml_file.path, html_file)
156         book.html_file.save('%s.html' % book.slug, File(html_file), save=False)
157         
158         # Extract fragments
159         closed_fragments, open_fragments = html.extract_fragments(book.html_file.path)
160         book_themes = []
161         for fragment in closed_fragments.values():
162             text = fragment.to_string()
163             short_text = ''
164             if (len(MarkupString(text)) > 240):
165                 short_text = MarkupString(text)[:160]
166             new_fragment = Fragment(text=text, short_text=short_text, anchor=fragment.id, book=book)
167                 
168             theme_names = [s.strip() for s in fragment.themes.split(',')]
169             themes = []
170             for theme_name in theme_names:
171                 tag, created = Tag.objects.get_or_create(name=theme_name,
172                     slug=slughifi(theme_name), sort_key=slughifi(theme_name), category='theme')
173                 tag.save()
174                 themes.append(tag)
175             new_fragment.save()
176             new_fragment.tags = list(book.tags) + themes
177             book_themes += themes
178         
179         book_themes = set(book_themes)
180         book.tags = list(book.tags) + list(book_themes)
181         return book.save()
182     
183     @permalink
184     def get_absolute_url(self):
185         return ('catalogue.views.book_detail', [self.slug])
186         
187     class Meta:
188         ordering = ('title',)
189         verbose_name = _('book')
190         verbose_name_plural = _('books')
191
192     def __unicode__(self):
193         return self.title
194
195
196 class Fragment(models.Model):
197     text = models.TextField()
198     short_text = models.TextField(editable=False)
199     _short_html = models.TextField(editable=False)
200     anchor = models.IntegerField()
201     book = models.ForeignKey(Book, related_name='fragments')
202
203     objects = managers.ModelTaggedItemManager(Tag)
204     tags = managers.TagDescriptor(Tag)
205     
206     def short_html(self):
207         if len(self._short_html):
208             return mark_safe(self._short_html)
209         else:
210             book_authors = [u'<a href="%s">%s</a>' % (tag.get_absolute_url(), tag.name) 
211                 for tag in self.book.tags if tag.category == 'author']
212             
213             self._short_html = render_to_string('catalogue/fragment_short.html',
214                 {'fragment': self, 'book': self.book, 'book_authors': book_authors})
215             self.save()
216             return mark_safe(self._short_html)
217         
218     class Meta:
219         ordering = ('book', 'anchor',)
220         verbose_name = _('fragment')
221         verbose_name_plural = _('fragments')
222