def import_from_upload(self, upload, frbr_uri, request):
""" Create a new Document by importing it from a
:class:`django.core.files.uploadedfile.UploadedFile` instance.
"""
self.reformat = True
if upload.content_type in ['text/xml', 'application/xml']:
# just assume it's valid AKN xml
doc = Document.randomized(frbr_uri)
doc.content = upload.read().decode('utf-8')
return doc
if upload.content_type == 'application/vnd.openxmlformats-officedocument.wordprocessingml.document':
# pre-process docx to HTML and then import html
html = self.docx_to_html(upload)
doc = self.import_from_text(html, frbr_uri, '.html')
elif upload.content_type == 'application/pdf':
doc = self.import_from_pdf(upload, frbr_uri)
else:
# slaw will do its best
with self.tempfile_for_upload(upload) as f:
doc = self.import_from_file(f.name, frbr_uri)
self.analyse_after_import(doc)
return doc
def import_from_file(self, fname, frbr_uri):
cmd = ['bundle', 'exec', 'slaw', 'parse']
if self.fragment:
cmd.extend(['--fragment', self.fragment])
if self.fragment_id_prefix:
cmd.extend(['--id-prefix', self.fragment_id_prefix])
if self.section_number_position:
cmd.extend(
['--section-number-position', self.section_number_position])
cmd.extend(['--grammar', self.slaw_grammar])
cmd.append(fname)
code, stdout, stderr = self.shell(cmd)
if code > 0:
raise ValueError(stderr)
if not stdout:
raise ValueError("We couldn't get any useful text out of the file")
if self.fragment:
doc = Fragment(stdout.decode('utf-8'))
else:
doc = Document.randomized(frbr_uri)
doc.content = stdout.decode('utf-8')
doc.frbr_uri = frbr_uri # reset it
doc.title = None
doc.copy_attributes()
self.log.info("Successfully imported from %s" % fname)
return doc
def document(request, doc_id=None):
if doc_id:
doc = get_object_or_404(Document, pk=doc_id)
doc_json = json.dumps(None)
else:
# it's new!
doc = Document.randomized(request.user, title='(untitled)')
doc.tags = None
doc_json = json.dumps(
DocumentSerializer(instance=doc, context={
'request': request
}).data)
form = DocumentForm(instance=doc)
countries = Country.objects.select_related('country').prefetch_related(
'locality_set', 'publication_set', 'country').all()
countries = {c.code: c.as_json() for c in countries}
countries_json = json.dumps(countries)
serializer = DocumentListSerializer(context={'request': request})
documents_json = json.dumps(
serializer.to_representation(DocumentViewSet.queryset.all()))
return render(
request, 'document/show.html', {
'document': doc,
'document_json': doc_json,
'document_content_json': json.dumps(doc.document_xml),
'documents_json': documents_json,
'form': form,
'subtypes': Subtype.objects.order_by('name').all(),
'languages': Language.objects.select_related('language').all(),
'countries': Country.objects.select_related('country').all(),
'countries_json': countries_json,
'view': 'DocumentView',
})
