Fix: manage.py importbooks -P means: don't build pdf.

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

diff --git a/apps/catalogue/models.py b/apps/catalogue/models.py
index 2722286..930f98d 100644 (file)
--- a/apps/catalogue/models.py
+++ b/apps/catalogue/models.py
@@ -6,6 +6,8 @@ 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

+import django.dispatch
+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


@@ -20,14 +22,14 @@ from django.conf import settings
 from newtagging.models import TagBase, tags_updated
 from newtagging import managers
 from catalogue.fields import JSONField, OverwritingFileField
 from newtagging.models import TagBase, tags_updated
 from newtagging import managers
 from catalogue.fields import JSONField, OverwritingFileField

-from catalogue.utils import ExistingFile
+from catalogue.utils import ExistingFile, BookImportDocProvider

 
 from librarian import dcparser, html, epub, NoDublinCore
 import mutagen
 from mutagen import id3
 from slughifi import slughifi
 from sortify import sortify
 
 from librarian import dcparser, html, epub, NoDublinCore
 import mutagen
 from mutagen import id3
 from slughifi import slughifi
 from sortify import sortify

-
+from os import unlink

 
 TAG_CATEGORIES = (
     ('author', _('author')),
 
 TAG_CATEGORIES = (
     ('author', _('author')),


@@ -46,6 +48,10 @@ MEDIA_FORMATS = (
     ('daisy', _('DAISY file')), 
 )
 
     ('daisy', _('DAISY file')), 
 )
 

+# not quite, but Django wants you to set a timeout
+CACHE_FOREVER = 2419200  # 28 days
+
+

 class TagSubcategoryManager(models.Manager):
     def __init__(self, subcategory):
         super(TagSubcategoryManager, self).__init__()
 class TagSubcategoryManager(models.Manager):
     def __init__(self, subcategory):
         super(TagSubcategoryManager, self).__init__()


@@ -282,7 +288,6 @@ class Book(models.Model):
     description   = models.TextField(_('description'), blank=True)
     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)
     description   = models.TextField(_('description'), blank=True)
     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)

-    _short_html   = models.TextField(_('short HTML'), editable=False)

     parent_number = models.IntegerField(_('parent number'), default=0)
     extra_info    = JSONField(_('extra information'), default='{}')
     gazeta_link   = models.CharField(blank=True, max_length=240)
     parent_number = models.IntegerField(_('parent number'), default=0)
     extra_info    = JSONField(_('extra information'), default='{}')
     gazeta_link   = models.CharField(blank=True, max_length=240)


@@ -291,16 +296,15 @@ class Book(models.Model):
     xml_file      = models.FileField(_('XML file'), upload_to=book_upload_path('xml'), blank=True)
     html_file     = models.FileField(_('HTML file'), upload_to=book_upload_path('html'), blank=True)
     pdf_file      = models.FileField(_('PDF file'), upload_to=book_upload_path('pdf'), blank=True)
     xml_file      = models.FileField(_('XML file'), upload_to=book_upload_path('xml'), blank=True)
     html_file     = models.FileField(_('HTML file'), upload_to=book_upload_path('html'), blank=True)
     pdf_file      = models.FileField(_('PDF file'), upload_to=book_upload_path('pdf'), blank=True)

-    epub_file     = models.FileField(_('EPUB file'), upload_to=book_upload_path('epub'), blank=True)    
-    txt_file      = models.FileField(_('TXT file'), upload_to=book_upload_path('txt'), blank=True)        
-
+    epub_file     = models.FileField(_('EPUB file'), upload_to=book_upload_path('epub'), blank=True)
+    txt_file      = models.FileField(_('TXT file'), upload_to=book_upload_path('txt'), blank=True)
+    

     parent        = models.ForeignKey('self', blank=True, null=True, related_name='children')
     objects  = models.Manager()
     tagged   = managers.ModelTaggedItemManager(Tag)
     tags     = managers.TagDescriptor(Tag)
 
     parent        = models.ForeignKey('self', blank=True, null=True, related_name='children')
     objects  = models.Manager()
     tagged   = managers.ModelTaggedItemManager(Tag)
     tags     = managers.TagDescriptor(Tag)
 

-    _tag_counter = JSONField(null=True, editable=False)
-    _theme_counter = JSONField(null=True, editable=False)
+    html_built = django.dispatch.Signal()

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


@@ -316,16 +320,12 @@ class Book(models.Model):
     def save(self, force_insert=False, force_update=False, reset_short_html=True, **kwargs):
         self.sort_key = sortify(self.title)
 
     def save(self, force_insert=False, force_update=False, reset_short_html=True, **kwargs):
         self.sort_key = sortify(self.title)
 

+        ret = super(Book, self).save(force_insert, force_update)
+

         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 ret

 
     @permalink
     def get_absolute_url(self):
 
     @permalink
     def get_absolute_url(self):


@@ -405,11 +405,25 @@ 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):
+        if self.id is None:
+            return
+
+        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)
+        if self.id:
+            cache_key = "Book.short_html/%d/%s" % (self.id, get_language())
+            short_html = cache.get(cache_key)
+        else:
+            short_html = None

 
 

-        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')))


@@ -431,11 +445,12 @@ 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}))

 
 

+            if self.id:
+                cache.set(cache_key, short_html, CACHE_FOREVER)
+            return mark_safe(short_html)

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


@@ -489,34 +504,49 @@ class Book(models.Model):
         return bool(self.has_media("ogg"))
     has_ogg_file.short_description = 'OGG'
     has_ogg_file.boolean = True
         return bool(self.has_media("ogg"))
     has_ogg_file.short_description = 'OGG'
     has_ogg_file.boolean = True

-    
+

     def has_daisy_file(self):
         return bool(self.has_media("daisy"))
     has_daisy_file.short_description = 'DAISY'
     def has_daisy_file(self):
         return bool(self.has_media("daisy"))
     has_daisy_file.short_description = 'DAISY'

-    has_daisy_file.boolean = True    
-    
+    has_daisy_file.boolean = True
+
+    def build_pdf(self):
+        """ (Re)builds the pdf file.
+
+        """
+        from librarian import pdf, ParseError
+        from tempfile import NamedTemporaryFile
+        import os
+
+        try:
+            path, fname = os.path.realpath(self.xml_file.path).rsplit('/', 1)
+            try:
+                pdf_file = NamedTemporaryFile(delete=False)
+
+                pdf.transform(BookImportDocProvider(self),
+                          file_path=str(self.xml_file.path),
+                          output_file=pdf_file,
+                          )
+
+                self.pdf_file.save('%s.pdf' % self.slug, File(open(pdf_file.name)))
+            finally:
+                unlink(pdf_file.name)
+
+        except ParseError, e:
+            print '%(file)s:%(name)s:%(message)s; use -v to see more output' % {
+                'file': self.xml_file.path,
+                'name': e.__class__.__name__,
+                'message': e
+                }
+

     def build_epub(self, remove_descendants=True):
         """ (Re)builds the epub file.
             If book has a parent, does nothing.
             Unless remove_descendants is False, descendants' epubs are removed.
         """
     def build_epub(self, remove_descendants=True):
         """ (Re)builds the epub file.
             If book has a parent, does nothing.
             Unless remove_descendants is False, descendants' epubs are removed.
         """

-    

         from StringIO import StringIO
         from hashlib import sha1
         from django.core.files.base import ContentFile
         from StringIO import StringIO
         from hashlib import sha1
         from django.core.files.base import ContentFile

-        from librarian import DocProvider
-
-        class BookImportDocProvider(DocProvider):
-            """ used for joined EPUBs """
-
-            def __init__(self, book):
-                self.book = book
-
-            def by_slug(self, slug):
-                if slug == self.book.slug:
-                    return self.book.xml_file
-                else:
-                    return Book.objects.get(slug=slug).xml_file

 
         if self.parent:
             # don't need an epub
 
         if self.parent:
             # don't need an epub


@@ -547,7 +577,62 @@ 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()
+            self.html_built.send(sender=self)
+            return True
+        return False

 
 
     @classmethod
 
 
     @classmethod


@@ -564,11 +649,8 @@ class Book(models.Model):
             xml_file.close()
 
     @classmethod
             xml_file.close()
 
     @classmethod

-    def from_text_and_meta(cls, raw_file, book_info, overwrite=False, build_epub=True, build_txt=True):
+    def from_text_and_meta(cls, raw_file, book_info, overwrite=False, build_epub=True, build_txt=True, build_pdf=True):

         import re
         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 = []


@@ -597,10 +679,9 @@ 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:


@@ -617,9 +698,9 @@ class Book(models.Model):
                     tag.name = tag_name
                     tag.sort_key = sortify(tag_sort_key.lower())
                     tag.save()
                     tag.name = tag_name
                     tag.sort_key = sortify(tag_sort_key.lower())
                     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()
 


@@ -634,53 +715,16 @@ 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_EPUB and build_epub:
             book.root_ancestor.build_epub()
 
             if not settings.NO_BUILD_TXT and build_txt:
                 book.build_txt()
 
         if not settings.NO_BUILD_EPUB and build_epub:
             book.root_ancestor.build_epub()
 

+        if not settings.NO_BUILD_PDF and build_pdf:
+            book.root_ancestor.build_pdf()
+

         book_descendants = list(book.children.all())
         # add l-tag to descendants and their fragments
         # delete unnecessary EPUB files
         book_descendants = list(book.children.all())
         # add l-tag to descendants and their fragments
         # delete unnecessary EPUB files


@@ -692,57 +736,69 @@ class Book(models.Model):
                 fragment.tags = set(list(fragment.tags) + [book_tag])
             book_descendants += list(child_book.children.all())
 
                 fragment.tags = set(list(fragment.tags) + [book_tag])
             book_descendants += list(child_book.children.all())
 

+        book.save()
+

         # refresh cache
         book.reset_tag_counter()
         book.reset_theme_counter()
 
         # refresh cache
         book.reset_tag_counter()
         book.reset_theme_counter()
 

-        book.save()

         return book
 
         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)
+        if self.id is None:
+            return
+
+        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)
+        if self.id:
+            cache_key = "Book.tag_counter/%d" % self.id
+            tags = cache.get(cache_key)
+        else:
+            tags = None
+
+        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
+
+            if self.id:
+                cache.set(cache_key, tags, CACHE_FOREVER)

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

-        self._theme_counter = None
-        self.save(reset_short_html=False)
+        if self.id is None:
+            return
+
+        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())
+        if self.id:
+            cache_key = "Book.theme_counter/%d" % self.id
+            tags = cache.get(cache_key)
+        else:
+            tags = None
+
+        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
+
+            if self.id:
+                cache.set(cache_key, tags, CACHE_FOREVER)
+        return tags

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


@@ -783,7 +839,6 @@ class Book(models.Model):
 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')
 


@@ -799,16 +854,29 @@ 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):
+        if self.id is None:
+            return
+
+        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):
+        if self.id:
+            cache_key = "Fragment.short_html/%d/%s" % (self.id, get_language())
+            short_html = cache.get(cache_key)
+        else:
+            short_html = None
+
+        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}))
+            if self.id:
+                cache.set(cache_key, short_html, CACHE_FOREVER)
+            return mark_safe(short_html)

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