don't justify

[wolnelektury.git] / apps / catalogue / models.py

diff --git a/apps/catalogue/models.py b/apps/catalogue/models.py
index 6291d97..7eefce7 100644 (file)
--- a/apps/catalogue/models.py
+++ b/apps/catalogue/models.py
@@ -2,8 +2,11 @@
 # This file is part of Wolnelektury, licensed under GNU Affero GPLv3 or later.
 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
 #
 # This file is part of Wolnelektury, licensed under GNU Affero GPLv3 or later.
 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
 #

+from datetime import datetime
+

 from django.db import models
 from django.db.models import permalink, Q
 from django.db import models
 from django.db.models import permalink, Q

+from django.core.cache import cache

 from django.utils.translation import ugettext_lazy as _
 from django.contrib.auth.models import User
 from django.core.files import File
 from django.utils.translation import ugettext_lazy as _
 from django.contrib.auth.models import User
 from django.core.files import File


@@ -17,13 +20,14 @@ from django.conf import settings
 
 from newtagging.models import TagBase, tags_updated
 from newtagging import managers
 
 from newtagging.models import TagBase, tags_updated
 from newtagging import managers

-from catalogue.fields import JSONField
+from catalogue.fields import JSONField, OverwritingFileField

 from catalogue.utils import ExistingFile
 
 from librarian import dcparser, html, epub, NoDublinCore
 import mutagen
 from mutagen import id3
 from slughifi import slughifi
 from catalogue.utils import ExistingFile
 
 from librarian import dcparser, html, epub, NoDublinCore
 import mutagen
 from mutagen import id3
 from slughifi import slughifi

+from sortify import sortify

 
 
 TAG_CATEGORIES = (
 
 
 TAG_CATEGORIES = (


@@ -66,6 +70,9 @@ class Tag(TagBase):
     gazeta_link = models.CharField(blank=True, max_length=240)
     wiki_link = models.CharField(blank=True, max_length=240)
 
     gazeta_link = models.CharField(blank=True, max_length=240)
     wiki_link = models.CharField(blank=True, max_length=240)
 

+    created_at    = models.DateTimeField(_('creation date'), auto_now_add=True, db_index=True)
+    changed_at    = models.DateTimeField(_('creation date'), auto_now=True, db_index=True)
+

     class UrlDeprecationWarning(DeprecationWarning):
         pass
 
     class UrlDeprecationWarning(DeprecationWarning):
         pass
 


@@ -184,7 +191,7 @@ def book_upload_path(ext=None, maxlen=100):
 class BookMedia(models.Model):
     type        = models.CharField(_('type'), choices=MEDIA_FORMATS, max_length="100")
     name        = models.CharField(_('name'), max_length="100")
 class BookMedia(models.Model):
     type        = models.CharField(_('type'), choices=MEDIA_FORMATS, max_length="100")
     name        = models.CharField(_('name'), max_length="100")

-    file        = models.FileField(_('file'), upload_to=book_upload_path())
+    file        = OverwritingFileField(_('file'), upload_to=book_upload_path())

     uploaded_at = models.DateTimeField(_('creation date'), auto_now_add=True, editable=False)
     extra_info  = JSONField(_('extra information'), default='{}', editable=False)
     book = models.ForeignKey('Book', related_name='media')
     uploaded_at = models.DateTimeField(_('creation date'), auto_now_add=True, editable=False)
     extra_info  = JSONField(_('extra information'), default='{}', editable=False)
     book = models.ForeignKey('Book', related_name='media')


@@ -200,16 +207,13 @@ class BookMedia(models.Model):
 
     def save(self, *args, **kwargs):
         try:
 
     def save(self, *args, **kwargs):
         try:

-            b = BookMedia.objects.get(pk=self.pk)
+            old = BookMedia.objects.get(pk=self.pk)

         except BookMedia.DoesNotExist, e:
             pass
         else:
         except BookMedia.DoesNotExist, e:
             pass
         else:

-            # if file is replaced, delete the old one
-            if self.file.path != b.file.path:
-                b.file.delete(save=False)

             # if name changed, change the file name, too
             # if name changed, change the file name, too

-            elif self.name != b.name:
-                self.file.save(None, ExistingFile(self.file.path))
+            if slughifi(self.name) != slughifi(old.name):
+                self.file.save(None, ExistingFile(self.file.path), save=False, leave=True)

 
         super(BookMedia, self).save(*args, **kwargs)
         extra_info = self.get_extra_info_value()
 
         super(BookMedia, self).save(*args, **kwargs)
         extra_info = self.get_extra_info_value()


@@ -274,12 +278,13 @@ class BookMedia(models.Model):
 
 class Book(models.Model):
     title         = models.CharField(_('title'), max_length=120)
 
 class Book(models.Model):
     title         = models.CharField(_('title'), max_length=120)

+    sort_key = models.CharField(_('sort_key'), max_length=120, db_index=True, editable=False)

     slug          = models.SlugField(_('slug'), max_length=120, unique=True, db_index=True)
     description   = models.TextField(_('description'), blank=True)
     slug          = models.SlugField(_('slug'), max_length=120, unique=True, db_index=True)
     description   = models.TextField(_('description'), blank=True)

-    created_at    = models.DateTimeField(_('creation date'), auto_now_add=True)
-    _short_html   = models.TextField(_('short HTML'), editable=False)
+    created_at    = models.DateTimeField(_('creation date'), auto_now_add=True, db_index=True)
+    changed_at    = models.DateTimeField(_('creation date'), auto_now=True, db_index=True)

     parent_number = models.IntegerField(_('parent number'), default=0)
     parent_number = models.IntegerField(_('parent number'), default=0)

-    extra_info    = JSONField(_('extra information'))
+    extra_info    = JSONField(_('extra information'), default='{}')

     gazeta_link   = models.CharField(blank=True, max_length=240)
     wiki_link     = models.CharField(blank=True, max_length=240)
     # files generated during publication
     gazeta_link   = models.CharField(blank=True, max_length=240)
     wiki_link     = models.CharField(blank=True, max_length=240)
     # files generated during publication


@@ -294,14 +299,11 @@ class Book(models.Model):
     tagged   = managers.ModelTaggedItemManager(Tag)
     tags     = managers.TagDescriptor(Tag)
 
     tagged   = managers.ModelTaggedItemManager(Tag)
     tags     = managers.TagDescriptor(Tag)
 

-    _tag_counter = JSONField(null=True, editable=False)
-    _theme_counter = JSONField(null=True, editable=False)
-

     class AlreadyExists(Exception):
         pass
 
     class Meta:
     class AlreadyExists(Exception):
         pass
 
     class Meta:

-        ordering = ('title',)
+        ordering = ('sort_key',)

         verbose_name = _('book')
         verbose_name_plural = _('books')
 
         verbose_name = _('book')
         verbose_name_plural = _('books')
 


@@ -309,14 +311,10 @@ class Book(models.Model):
         return self.title
 
     def save(self, force_insert=False, force_update=False, reset_short_html=True, **kwargs):
         return self.title
 
     def save(self, force_insert=False, force_update=False, reset_short_html=True, **kwargs):

+        self.sort_key = sortify(self.title)
+

         if reset_short_html:
         if reset_short_html:

-            # Reset _short_html during save
-            update = {}
-            for key in filter(lambda x: x.startswith('_short_html'), self.__dict__):
-                update[key] = ''
-                self.__setattr__(key, '')
-            # Fragment.short_html relies on book's tags, so reset it here too
-            self.fragments.all().update(**update)
+            self.reset_short_html()

 
         return super(Book, self).save(force_insert, force_update)
 
 
         return super(Book, self).save(force_insert, force_update)
 


@@ -398,11 +396,19 @@ class Book(models.Model):
     def get_daisy(self):
         return self.get_media("daisy")                       
 
     def get_daisy(self):
         return self.get_media("daisy")                       
 

+    def reset_short_html(self):
+        cache_key = "Book.short_html/%d/%s"
+        for lang, langname in settings.LANGUAGES:
+            cache.delete(cache_key % (self.id, lang))
+        # Fragment.short_html relies on book's tags, so reset it here too
+        for fragm in self.fragments.all():
+            fragm.reset_short_html()
+

     def short_html(self):
     def short_html(self):

-        key = '_short_html_%s' % get_language()
-        short_html = getattr(self, key)
+        cache_key = "Book.short_html/%d/%s" % (self.id, get_language())
+        short_html = cache.get(cache_key)

 
 

-        if short_html and len(short_html):
+        if short_html is not None:

             return mark_safe(short_html)
         else:
             tags = self.tags.filter(~Q(category__in=('set', 'theme', 'book')))
             return mark_safe(short_html)
         else:
             tags = self.tags.filter(~Q(category__in=('set', 'theme', 'book')))


@@ -424,11 +430,10 @@ class Book(models.Model):
 
             formats = [mark_safe(format) for format in formats]
 
 
             formats = [mark_safe(format) for format in formats]
 

-            setattr(self, key, unicode(render_to_string('catalogue/book_short.html',
-                {'book': self, 'tags': tags, 'formats': formats})))
-            self.save(reset_short_html=False)
-            return mark_safe(getattr(self, key))
-
+            short_html = unicode(render_to_string('catalogue/book_short.html',
+                {'book': self, 'tags': tags, 'formats': formats}))
+            cache.set(cache_key, short_html)
+            return mark_safe(short_html)

 
     @property
     def root_ancestor(self):
 
     @property
     def root_ancestor(self):


@@ -540,7 +545,61 @@ class Book(models.Model):
         out = StringIO()
         text.transform(open(self.xml_file.path), out)
         self.txt_file.save('%s.txt' % self.slug, ContentFile(out.getvalue()))
         out = StringIO()
         text.transform(open(self.xml_file.path), out)
         self.txt_file.save('%s.txt' % self.slug, ContentFile(out.getvalue()))

-        self.save()
+
+
+    def build_html(self):
+        from tempfile import NamedTemporaryFile
+        from markupstring import MarkupString
+
+        meta_tags = list(self.tags.filter(
+            category__in=('author', 'epoch', 'genre', 'kind')))
+        book_tag = self.book_tag()
+
+        html_file = NamedTemporaryFile()
+        if html.transform(self.xml_file.path, html_file, parse_dublincore=False):
+            self.html_file.save('%s.html' % self.slug, File(html_file))
+
+            # get ancestor l-tags for adding to new fragments
+            ancestor_tags = []
+            p = self.parent
+            while p:
+                ancestor_tags.append(p.book_tag())
+                p = p.parent
+
+            # Delete old fragments and create them from scratch
+            self.fragments.all().delete()
+            # Extract fragments
+            closed_fragments, open_fragments = html.extract_fragments(self.html_file.path)
+            for fragment in closed_fragments.values():
+                try:
+                    theme_names = [s.strip() for s in fragment.themes.split(',')]
+                except AttributeError:
+                    continue
+                themes = []
+                for theme_name in theme_names:
+                    if not theme_name:
+                        continue
+                    tag, created = Tag.objects.get_or_create(slug=slughifi(theme_name), category='theme')
+                    if created:
+                        tag.name = theme_name
+                        tag.sort_key = theme_name.lower()
+                        tag.save()
+                    themes.append(tag)
+                if not themes:
+                    continue
+
+                text = fragment.to_string()
+                short_text = ''
+                if (len(MarkupString(text)) > 240):
+                    short_text = unicode(MarkupString(text)[:160])
+                new_fragment = Fragment.objects.create(anchor=fragment.id, book=self,
+                    text=text, short_text=short_text)
+
+                new_fragment.save()
+                new_fragment.tags = set(meta_tags + themes + [book_tag] + ancestor_tags)
+            self.save()
+            return True
+        return False

 
 
     @classmethod
 
 
     @classmethod


@@ -559,9 +618,6 @@ class Book(models.Model):
     @classmethod
     def from_text_and_meta(cls, raw_file, book_info, overwrite=False, build_epub=True, build_txt=True):
         import re
     @classmethod
     def from_text_and_meta(cls, raw_file, book_info, overwrite=False, build_epub=True, build_txt=True):
         import re

-        from tempfile import NamedTemporaryFile
-        from markupstring import MarkupString
-        from django.core.files.storage import default_storage

 
         # check for parts before we do anything
         children = []
 
         # check for parts before we do anything
         children = []


@@ -590,10 +646,9 @@ class Book(models.Model):
 
         book.title = book_info.title
         book.set_extra_info_value(book_info.to_dict())
 
         book.title = book_info.title
         book.set_extra_info_value(book_info.to_dict())

-        book._short_html = ''

         book.save()
 
         book.save()
 

-        book_tags = []
+        meta_tags = []

         categories = (('kinds', 'kind'), ('genres', 'genre'), ('authors', 'author'), ('epochs', 'epoch'))
         for field_name, category in categories:
             try:
         categories = (('kinds', 'kind'), ('genres', 'genre'), ('authors', 'author'), ('epochs', 'epoch'))
         for field_name, category in categories:
             try:


@@ -608,11 +663,11 @@ class Book(models.Model):
                 tag, created = Tag.objects.get_or_create(slug=slughifi(tag_name), category=category)
                 if created:
                     tag.name = tag_name
                 tag, created = Tag.objects.get_or_create(slug=slughifi(tag_name), category=category)
                 if created:
                     tag.name = tag_name

-                    tag.sort_key = tag_sort_key.lower()
+                    tag.sort_key = sortify(tag_sort_key.lower())

                     tag.save()
                     tag.save()

-                book_tags.append(tag)
+                meta_tags.append(tag)

 
 

-        book.tags = set(book_tags + book_shelves)
+        book.tags = set(meta_tags + book_shelves)

 
         book_tag = book.book_tag()
 
 
         book_tag = book.book_tag()
 


@@ -627,47 +682,7 @@ class Book(models.Model):
         # delete old fragments when overwriting
         book.fragments.all().delete()
 
         # delete old fragments when overwriting
         book.fragments.all().delete()
 

-        html_file = NamedTemporaryFile()
-        if html.transform(book.xml_file.path, html_file, parse_dublincore=False):
-            book.html_file.save('%s.html' % book.slug, File(html_file), save=False)
-
-            # get ancestor l-tags for adding to new fragments
-            ancestor_tags = []
-            p = book.parent
-            while p:
-                ancestor_tags.append(p.book_tag())
-                p = p.parent
-
-            # Extract fragments
-            closed_fragments, open_fragments = html.extract_fragments(book.html_file.path)
-            for fragment in closed_fragments.values():
-                try:
-                    theme_names = [s.strip() for s in fragment.themes.split(',')]
-                except AttributeError:
-                    continue
-                themes = []
-                for theme_name in theme_names:
-                    if not theme_name:
-                        continue
-                    tag, created = Tag.objects.get_or_create(slug=slughifi(theme_name), category='theme')
-                    if created:
-                        tag.name = theme_name
-                        tag.sort_key = theme_name.lower()
-                        tag.save()
-                    themes.append(tag)
-                if not themes:
-                    continue
-
-                text = fragment.to_string()
-                short_text = ''
-                if (len(MarkupString(text)) > 240):
-                    short_text = unicode(MarkupString(text)[:160])
-                new_fragment, created = Fragment.objects.get_or_create(anchor=fragment.id, book=book,
-                    defaults={'text': text, 'short_text': short_text})
-
-                new_fragment.save()
-                new_fragment.tags = set(book_tags + themes + [book_tag] + ancestor_tags)
-
+        if book.build_html():

             if not settings.NO_BUILD_TXT and build_txt:
                 book.build_txt()
 
             if not settings.NO_BUILD_TXT and build_txt:
                 book.build_txt()
 


@@ -692,50 +707,45 @@ class Book(models.Model):
         book.save()
         return book
 
         book.save()
         return book
 

-
-    def refresh_tag_counter(self):
-        tags = {}
-        for child in self.children.all().order_by():
-            for tag_pk, value in child.tag_counter.iteritems():
-                tags[tag_pk] = tags.get(tag_pk, 0) + value
-        for tag in self.tags.exclude(category__in=('book', 'theme', 'set')).order_by():
-            tags[tag.pk] = 1
-        self.set__tag_counter_value(tags)
-        self.save(reset_short_html=False)
-        return tags
-

     def reset_tag_counter(self):
     def reset_tag_counter(self):

-        self._tag_counter = None
-        self.save(reset_short_html=False)
+        cache_key = "Book.tag_counter/%d" % self.id
+        cache.delete(cache_key)

         if self.parent:
             self.parent.reset_tag_counter()
 
     @property
     def tag_counter(self):
         if self.parent:
             self.parent.reset_tag_counter()
 
     @property
     def tag_counter(self):

-        if self._tag_counter is None:
-            return self.refresh_tag_counter()
-        return dict((int(k), v) for k, v in self.get__tag_counter_value().iteritems())
-
-    def refresh_theme_counter(self):
-        tags = {}
-        for fragment in Fragment.tagged.with_any([self.book_tag()]).order_by():
-            for tag in fragment.tags.filter(category='theme').order_by():
-                tags[tag.pk] = tags.get(tag.pk, 0) + 1
-        self.set__theme_counter_value(tags)
-        self.save(reset_short_html=False)
+        cache_key = "Book.tag_counter/%d" % self.id
+        tags = cache.get(cache_key)
+        if tags is None:
+            tags = {}
+            for child in self.children.all().order_by():
+                for tag_pk, value in child.tag_counter.iteritems():
+                    tags[tag_pk] = tags.get(tag_pk, 0) + value
+            for tag in self.tags.exclude(category__in=('book', 'theme', 'set')).order_by():
+                tags[tag.pk] = 1
+
+            cache.set(cache_key, tags)

         return tags
 
     def reset_theme_counter(self):
         return tags
 
     def reset_theme_counter(self):

-        self._theme_counter = None
-        self.save(reset_short_html=False)
+        cache_key = "Book.theme_counter/%d" % self.id
+        cache.delete(cache_key)

         if self.parent:
             self.parent.reset_theme_counter()
 
     @property
     def theme_counter(self):
         if self.parent:
             self.parent.reset_theme_counter()
 
     @property
     def theme_counter(self):

-        if self._theme_counter is None:
-            return self.refresh_theme_counter()
-        return dict((int(k), v) for k, v in self.get__theme_counter_value().iteritems())
+        cache_key = "Book.theme_counter/%d" % self.id
+        tags = cache.get(cache_key)
+        if tags is None:
+            tags = {}
+            for fragment in Fragment.tagged.with_any([self.book_tag()]).order_by():
+                for tag in fragment.tags.filter(category='theme').order_by():
+                    tags[tag.pk] = tags.get(tag.pk, 0) + 1
+
+            cache.set(cache_key, tags)
+        return tags

 
     def pretty_title(self, html_links=False):
         book = self
 
     def pretty_title(self, html_links=False):
         book = self


@@ -754,11 +764,28 @@ class Book(models.Model):
 
         return ', '.join(names)
 
 
         return ', '.join(names)
 

+    @classmethod
+    def tagged_top_level(cls, tags):
+        """ Returns top-level books tagged with `tags'.
+
+        It only returns those books which don't have ancestors which are
+        also tagged with those tags.
+
+        """
+        # get relevant books and their tags
+        objects = cls.tagged.with_all(tags)
+        # eliminate descendants
+        l_tags = Tag.objects.filter(category='book', slug__in=[book.book_tag_slug() for book in objects])
+        descendants_keys = [book.pk for book in cls.tagged.with_any(l_tags)]
+        if descendants_keys:
+            objects = objects.exclude(pk__in=descendants_keys)
+
+        return objects
+

 
 class Fragment(models.Model):
     text = models.TextField()
     short_text = models.TextField(editable=False)
 
 class Fragment(models.Model):
     text = models.TextField()
     short_text = models.TextField(editable=False)

-    _short_html = models.TextField(editable=False)

     anchor = models.CharField(max_length=120)
     book = models.ForeignKey(Book, related_name='fragments')
 
     anchor = models.CharField(max_length=120)
     book = models.ForeignKey(Book, related_name='fragments')
 


@@ -774,16 +801,22 @@ class Fragment(models.Model):
     def get_absolute_url(self):
         return '%s#m%s' % (reverse('book_text', kwargs={'slug': self.book.slug}), self.anchor)
 
     def get_absolute_url(self):
         return '%s#m%s' % (reverse('book_text', kwargs={'slug': self.book.slug}), self.anchor)
 

+    def reset_short_html(self):
+        cache_key = "Fragment.short_html/%d/%s"
+        for lang, langname in settings.LANGUAGES:
+            cache.delete(cache_key % (self.id, lang))
+

     def short_html(self):
     def short_html(self):

-        key = '_short_html_%s' % get_language()
-        short_html = getattr(self, key)
-        if short_html and len(short_html):
+        cache_key = "Fragment.short_html/%d/%s" % (self.id, get_language())
+        short_html = cache.get(cache_key)
+
+        if short_html is not None:

             return mark_safe(short_html)
         else:
             return mark_safe(short_html)
         else:

-            setattr(self, key, unicode(render_to_string('catalogue/fragment_short.html',
-                {'fragment': self})))
-            self.save()
-            return mark_safe(getattr(self, key))
+            short_html = unicode(render_to_string('catalogue/fragment_short.html',
+                {'fragment': self}))
+            cache.set(cache_key, short_html)
+            return mark_safe(short_html)

 
 
 class FileRecord(models.Model):
 
 
 class FileRecord(models.Model):


@@ -809,7 +842,8 @@ class FileRecord(models.Model):
 
 def _tags_updated_handler(sender, affected_tags, **kwargs):
     # reset tag global counter
 
 def _tags_updated_handler(sender, affected_tags, **kwargs):
     # reset tag global counter

-    Tag.objects.filter(pk__in=[tag.pk for tag in affected_tags]).update(book_count=None)
+    # we want Tag.changed_at updated for API to know the tag was touched
+    Tag.objects.filter(pk__in=[tag.pk for tag in affected_tags]).update(book_count=None, changed_at=datetime.now())

 
     # if book tags changed, reset book tag counter
     if isinstance(sender, Book) and \
 
     # if book tags changed, reset book tag counter
     if isinstance(sender, Book) and \