def get_feed_parser(self, provider, article=None):
"""
Returns instance of configured feed parser for the given provider.
:param provider: Ingest Provider Details.
:type provider: dict :py:class: `superdesk.io.ingest_provider_model.IngestProviderResource`
:param article: article which needs to parsed by the parser, defaults to None. When None, the registered parser
will be returned without checking if the parser can parse the article.
:return: Feed Parser instance.
:rtype: :py:class:`superdesk.io.feed_parsers.FeedParser`
:raises: SuperdeskIngestError.parserNotFoundError()
if either feed_parser value is empty or Feed Parser not found.
"""
parser = registered_feed_parsers.get(provider.get('feed_parser', ''))
if not parser:
raise SuperdeskIngestError.parserNotFoundError(provider=provider)
if article is not None and not parser.can_parse(article):
raise SuperdeskIngestError.parserNotFoundError(provider=provider)
if article is not None:
parser = parser.__class__()
return parser
def get_feed_parser(self, provider, article=None):
"""
Returns instance of configured feed parser for the given provider.
:param provider: Ingest Provider Details.
:type provider: dict :py:class: `superdesk.io.ingest_provider_model.IngestProviderResource`
:param article: article which needs to parsed by the parser, defaults to None. When None, the registered parser
will be returned without checking if the parser can parse the article.
:return: Feed Parser instance.
:rtype: :py:class:`superdesk.io.feed_parsers.FeedParser`
:raises: SuperdeskIngestError.parserNotFoundError()
if either feed_parser value is empty or Feed Parser not found.
"""
parser = registered_feed_parsers.get(provider.get('feed_parser', ''))
if not parser:
raise SuperdeskIngestError.parserNotFoundError(provider=provider)
if article is not None and not parser.can_parse(article):
raise SuperdeskIngestError.parserNotFoundError(provider=provider)
if article is not None:
parser = parser.__class__()
return parser
def test_ingest_update_same_event(self):
xml = ET.fromstring("""<?xml version="1.0" encoding="ISO-8859-1" standalone="yes"?>
<document>
<guid>NTB-123456</guid>
<time>2016-08-10T15:02:02</time>
<publiseres>True</publiseres>
<ntbId>NBRP160810_144545_ja_00</ntbId>
<service>newscalendar</service>
<title>Original Content</title>
<location>Fr. Nansens plass 17, Tromsø, Troms</location>
<timeStart>2016-09-05T09:00:00</timeStart>
<timeEnd>2016-09-05T16:00:00</timeEnd>
<alldayevent>False</alldayevent>
<priority>5</priority>
<regions>
<region>Norge</region>
</regions>
<districts>
<district parent="Norge">Troms</district>
</districts>
<category>Innenriks</category>
<subcategory>Redplan element</subcategory>
<subjects>
<subject>Kriminalitet og rettsvesen</subject>
<subject parent="Kriminalitet">Drap;Rettssaker</subject>
</subjects>
<emailwriter>[email protected]</emailwriter>
<messagetype>Redplan redaksjon</messagetype>
<geo>
<latitude>69.65482639999999</latitude>
<longitude>18.96509590000005</longitude>
</geo>
<content>Original Content</content>
<mediaList>
<media id="" mediaType="" mimeType="ukjent">
<caption></caption>
</media>
</mediaList>
</document>""")
with self.app.test_request_context(self.app.config['URL_PREFIX']):
# ingest event
events = self.get_parsed_documents(registered_feed_parsers.get('ntb_event_xml'), xml)
provider = get_resource_service('ingest_providers').find_one(req=None, _id=self.providers.get('ntbevent'))
self.ingest_items(events, provider, registered_feeding_services.get('event_file'))
ingested_event = get_resource_service('events').find_one(req=None, _id='NTB-123456')
self.assertTrue(ingested_event['_id'], 'NTB-123456')
self.assertTrue(ingested_event['name'], 'Original Content')
self.assertTrue(ingested_event['dates']['start'], '2016-09-05T09:00:00')
self.assertTrue(ingested_event['dates']['end'], '2016-09-05T16:00:00')
self.assertTrue(ingested_event['_planning_schedule'][0]['scheduled'], '2016-09-05T09:00:00')
# ingest updated event
events = self.get_parsed_documents(registered_feed_parsers.get('ntb_event_xml'), xml)
events[0]['dates']['start'] = '2016-09-06T10:00:00'
events[0]['dates']['end'] = '2016-09-06T14:00:00'
events[0]['name'] = 'Updated Content'
self.ingest_items(events, provider, registered_feeding_services.get('event_file'))
ingested_event = get_resource_service('events').find_one(req=None, _id='NTB-123456')
self.assertTrue(ingested_event['_id'], 'NTB-123456')
self.assertTrue(ingested_event['name'], 'Updated Content')
self.assertTrue(ingested_event['dates']['start'], '2016-09-05T09:00:00')
self.assertTrue(ingested_event['dates']['end'], '2016-09-05T16:00:00')
self.assertTrue(ingested_event['_planning_schedule'][0]['scheduled'], '2016-09-16T16:00:00')