Fix: manage.py importbooks -P means: don't build pdf.
[wolnelektury.git] / apps / catalogue / models.py
index bf724fd..930f98d 100644 (file)
@@ -6,6 +6,7 @@ from datetime import datetime
 
 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
@@ -21,14 +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 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 os import unlink
 
 TAG_CATEGORIES = (
     ('author', _('author')),
@@ -47,6 +48,10 @@ MEDIA_FORMATS = (
     ('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__()
@@ -291,14 +296,16 @@ 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)
-    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)
 
+    html_built = django.dispatch.Signal()
+
     class AlreadyExists(Exception):
         pass
 
@@ -313,10 +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)
 
+        ret = super(Book, self).save(force_insert, force_update)
+
         if reset_short_html:
             self.reset_short_html()
 
-        return super(Book, self).save(force_insert, force_update)
+        return ret
 
     @permalink
     def get_absolute_url(self):
@@ -397,6 +406,9 @@ class Book(models.Model):
         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))
@@ -405,14 +417,15 @@ class Book(models.Model):
             fragm.reset_short_html()
 
     def short_html(self):
-        cache_key = "Book.short_html/%d/%s" % (self.id, get_language())
-        short_html = cache.get(cache_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 is not None:
-            print 'b.s from cache'
             return mark_safe(short_html)
         else:
-            print 'b.s manual'
             tags = self.tags.filter(~Q(category__in=('set', 'theme', 'book')))
             tags = [mark_safe(u'<a href="%s">%s</a>' % (tag.get_absolute_url(), tag.name)) for tag in tags]
 
@@ -434,7 +447,9 @@ class Book(models.Model):
 
             short_html = unicode(render_to_string('catalogue/book_short.html',
                 {'book': self, 'tags': tags, 'formats': formats}))
-            cache.set(cache_key, short_html)
+
+            if self.id:
+                cache.set(cache_key, short_html, CACHE_FOREVER)
             return mark_safe(short_html)
 
     @property
@@ -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
-    
+
     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.
         """
-    
         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
@@ -600,6 +630,7 @@ class Book(models.Model):
                 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
 
@@ -618,7 +649,7 @@ class Book(models.Model):
             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
 
         # check for parts before we do anything
@@ -691,6 +722,9 @@ class Book(models.Model):
         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
@@ -702,14 +736,18 @@ class Book(models.Model):
                 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()
 
-        book.save()
         return book
 
     def reset_tag_counter(self):
+        if self.id is None:
+            return
+
         cache_key = "Book.tag_counter/%d" % self.id
         cache.delete(cache_key)
         if self.parent:
@@ -717,11 +755,13 @@ class Book(models.Model):
 
     @property
     def tag_counter(self):
-        cache_key = "Book.tag_counter/%d" % self.id
-        tags = cache.get(cache_key)
-        print 'tag'
+        if self.id:
+            cache_key = "Book.tag_counter/%d" % self.id
+            tags = cache.get(cache_key)
+        else:
+            tags = None
+
         if tags is None:
-            print 'tag manual'
             tags = {}
             for child in self.children.all().order_by():
                 for tag_pk, value in child.tag_counter.iteritems():
@@ -729,10 +769,14 @@ class Book(models.Model):
             for tag in self.tags.exclude(category__in=('book', 'theme', 'set')).order_by():
                 tags[tag.pk] = 1
 
-            cache.set(cache_key, tags)
+            if self.id:
+                cache.set(cache_key, tags, CACHE_FOREVER)
         return tags
 
     def reset_theme_counter(self):
+        if self.id is None:
+            return
+
         cache_key = "Book.theme_counter/%d" % self.id
         cache.delete(cache_key)
         if self.parent:
@@ -740,17 +784,20 @@ class Book(models.Model):
 
     @property
     def theme_counter(self):
-        cache_key = "Book.theme_counter/%d" % self.id
-        tags = cache.get(cache_key)
-        print 'theme'
+        if self.id:
+            cache_key = "Book.theme_counter/%d" % self.id
+            tags = cache.get(cache_key)
+        else:
+            tags = None
+
         if tags is None:
-            print 'theme manual'
             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)
+            if self.id:
+                cache.set(cache_key, tags, CACHE_FOREVER)
         return tags
 
     def pretty_title(self, html_links=False):
@@ -808,22 +855,27 @@ class Fragment(models.Model):
         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):
-        cache_key = "Fragment.short_html/%d/%s" % (self.id, get_language())
-        short_html = cache.get(cache_key)
+        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:
-            print 'f.s from cache'
             return mark_safe(short_html)
         else:
-            print 'f.s manual'
             short_html = unicode(render_to_string('catalogue/fragment_short.html',
                 {'fragment': self}))
-            cache.set(cache_key, short_html)
+            if self.id:
+                cache.set(cache_key, short_html, CACHE_FOREVER)
             return mark_safe(short_html)