def set_publication_details(apps, schema_editor):
Document = apps.get_model("indigo_api", "Document")
db_alias = schema_editor.connection.alias
for doc in Document.objects.using(db_alias).all():
a = Act(doc.document_xml)
doc.publication_number = a.publication_number
doc.publication_name = a.publication_name
doc.publication_date = a.publication_date
doc.save()
def migrate_xml(self, document_xml, frbr_uri, forward):
# Create a cobalt StructuredDocument from the document's existing XML
cobalt_doc = Act(document_xml)
# Update the document's FRBR URI in the XML (meta/identification block)
cobalt_doc.frbr_uri = frbr_uri
# Add `/akn` prefix (if forward=True; remove if forward=False)
# to hrefs in <ref> or <passiveRef> AKN elements, e.g.
# href="/za/act/2012/22" <--> href="/akn/za/act/2012/22"
for node in cobalt_doc.root.xpath(
"//a:*[self::a:ref or self::a:passiveRef][starts-with(@href, '/')]",
namespaces={'a': cobalt_doc.namespace}):
ref = node.get('href')
if FRBR_URI_RE.match(ref):
ref = self.new_frbr_uri(ref, forward)
node.set('href', ref)
return cobalt_doc.to_xml().decode('utf-8')
def test_migration_without_heading(self):
migration = ScheduleArticleToHcontainer()
xml = OLD_SCHEDULE % ("""
<paragraph id="schedule2.paragraph-0">
<content>
<p>None</p>
</content>
</paragraph>""")
act = Act(xml)
migration.migrate_act(act)
expected = NEW_SCHEDULE % """
<heading>Schedule</heading><paragraph id="schedule2.paragraph-0">
<content>
<p>None</p>
</content>
</paragraph>"""
self.assertMultiLineEqual(
expected,
etree.tostring(act.root, encoding='utf-8',
pretty_print=True).decode('utf-8'))
def validate_content(self, value):
try:
Act(value)
except LxmlError as e:
raise ValidationError("Invalid XML: %s" % str(e))
return value