Update to new librarian api for html, txt.

[wolnelektury.git] / src / catalogue / models / book.py

diff --git a/src/catalogue/models/book.py b/src/catalogue/models/book.py
index ac8c02b..7d36662 100644 (file)
--- a/src/catalogue/models/book.py
+++ b/src/catalogue/models/book.py
@@ -19,7 +19,7 @@ from django.utils.translation import gettext_lazy as _, get_language
 from fnpdjango.storage import BofhFileSystemStorage
 from lxml import html
 from librarian.cover import WLCover
 from fnpdjango.storage import BofhFileSystemStorage
 from lxml import html
 from librarian.cover import WLCover

-from librarian.html import transform_abstrakt
+from librarian.builders.html import AbstraktHtmlBuilder

 from librarian.builders import builders
 from newtagging import managers
 from catalogue import constants
 from librarian.builders import builders
 from newtagging import managers
 from catalogue import constants


@@ -327,7 +327,10 @@ class Book(models.Model):
         return int(total)
 
     def get_time(self):
         return int(total)
 
     def get_time(self):

-        return round(self.xml_file.size / 1000 * 40)
+        try:
+            return round(self.xml_file.size / 1000 * 40)
+        except ValueError:
+            return 0

     
     def has_media(self, type_):
         if type_ in Book.formats:
     
     def has_media(self, type_):
         if type_ in Book.formats:


@@ -554,11 +557,8 @@ class Book(models.Model):
                 urlretrieve('%s/%s' % (remote_gallery_url, ilustr_src), ilustr_path)
 
     def load_abstract(self):
                 urlretrieve('%s/%s' % (remote_gallery_url, ilustr_src), ilustr_path)
 
     def load_abstract(self):

-        abstract = self.wldocument(parse_dublincore=False).edoc.getroot().find('.//abstrakt')
-        if abstract is not None:
-            self.abstract = transform_abstrakt(abstract)
-        else:
-            self.abstract = ''
+        self.abstract = AbstraktHtmlBuilder().build(
+            self.wldocument2()).get_bytes().decode('utf-8')

 
     def load_toc(self):
         self.toc = ''
 
     def load_toc(self):
         self.toc = ''


@@ -653,11 +653,6 @@ class Book(models.Model):
 
         meta_tags = Tag.tags_from_info(book_info)
 
 
         meta_tags = Tag.tags_from_info(book_info)
 

-        for tag, relationship in meta_tags:
-            if not tag.for_books:
-                tag.for_books = True
-                tag.save()
-

         just_tags = [t for (t, rel) in meta_tags if not rel]
         book.tags = set(just_tags + book_shelves)
         book.save()  # update sort_key_author
         just_tags = [t for (t, rel) in meta_tags if not rel]
         book.tags = set(just_tags + book_shelves)
         book.save()  # update sort_key_author


@@ -722,13 +717,14 @@ class Book(models.Model):
         cls.published.send(sender=cls, instance=book)
         return book
 
         cls.published.send(sender=cls, instance=book)
         return book
 

+    # TODO TEST

     def update_references(self):
         Entity = apps.get_model('references', 'Entity')
         doc = self.wldocument2()
     def update_references(self):
         Entity = apps.get_model('references', 'Entity')
         doc = self.wldocument2()

-        doc._compat_assign_section_ids()
-        doc._compat_assign_ordered_ids()
+        doc.assign_ids()
+

         refs = {}
         refs = {}

-        for ref_elem in doc.references():
+        for i, ref_elem in enumerate(doc.references()):

             uri = ref_elem.attrib.get('href', '')
             if not uri:
                 continue
             uri = ref_elem.attrib.get('href', '')
             if not uri:
                 continue


@@ -747,10 +743,8 @@ class Book(models.Model):
                 refs[uri] = ref
                 if not ref_created:
                     ref.occurence_set.all().delete()
                 refs[uri] = ref
                 if not ref_created:
                     ref.occurence_set.all().delete()

-            sec = ref_elem.get_link()
-            m = re.match(r'sec(\d+)', sec)
-            assert m is not None
-            sec = int(m.group(1))
+            anchor = ref_elem.get_link()
+

             snippet = ref_elem.get_snippet()
             b = builders['html-snippet']()
             for s in snippet:
             snippet = ref_elem.get_snippet()
             b = builders['html-snippet']()
             for s in snippet:


@@ -758,7 +752,8 @@ class Book(models.Model):
             html = b.output().get_bytes().decode('utf-8')
 
             ref.occurence_set.create(
             html = b.output().get_bytes().decode('utf-8')
 
             ref.occurence_set.create(

-                section=sec,
+                section=i,
+                anchor=anchor,

                 html=html
             )
         self.reference_set.exclude(entity__uri__in=refs).delete()
                 html=html
             )
         self.reference_set.exclude(entity__uri__in=refs).delete()