+ def __init__(self, *args, **kwargs):
+ user = kwargs.pop('user')
+ super(DocumentTextSaveForm, self).__init__(*args, **kwargs)
+ if user and user.is_authenticated():
+ self.fields['author_name'].required = False
+ self.fields['author_email'].required = False
+
+ def clean_text(self):
+ text = self.cleaned_data['text']
+ try:
+ doc = Document.from_string(text)
+ except ValueError as e:
+ raise ValidationError(e.message)
+
+ from librarian import SSTNS, DCNS
+ root_elem = doc.edoc.getroot()
+ if len(root_elem) < 1 or root_elem[0].tag != SSTNS('metadata'):
+ raise ValidationError("The first tag in section should be metadata")
+ if len(root_elem) < 2 or root_elem[1].tag != SSTNS('header'):
+ raise ValidationError("The first tag after metadata should be header")
+ header = root_elem[1]
+ if not getattr(header, 'text', None) or not header.text.strip():
+ raise ValidationError(
+ "The first header should contain the title in plain text (no links, emphasis etc.) and cannot be empty")
+
+ cover_url = doc.meta.get_one(DCNS('relation.coverimage.url'))
+ if cover_url:
+ ext = cover_url.rsplit('.', 1)[-1].lower()
+ if ext not in ('jpg', 'jpeg', 'png', 'gif', 'svg'):
+ raise ValidationError('Invalid cover image format, should be an image file (jpg, png, gif, svg). '
+ 'Change it in Metadata.')
+ metadata = metadata_from_text(text)
+ if 'multiple_values' in metadata:
+ raise ValidationError('Category "%s" does not allow multiple values.' % metadata['multiple_values'])
+ for category in Category.objects.all():
+ values = metadata.get(category.dc_tag)
+ if values is None:
+ continue
+ if not category.multiple:
+ values = [values]
+ if not values:
+ values = []
+ for value in values:
+ if not category.tag_set.filter(dc_value=value):
+ raise ValidationError('Invalid value for dc:%s: %s' % (category.dc_tag, value))
+ return text
+