def upload_to_amcat(self, articles):
"""Create articles in Amcat using AmcatAPI
Args:
articles: List of articles in Amcat's format
"""
try:
self.amcat_api.create_articles(project=amcat.project, articleset=self.set_id, json_data=articles)
except:
log.exception('Could not upload articles to amcat. url: {amcat.project}; set: {self.set_id}; '
'data: {articles}'.format(**locals()))
def article_ocr(identifier):
"""Retrieve ocr full text for article with identifier
"""
url = DelpherAPI.ocr_url(identifier)
try:
response = request.get(url)
except:
log.exception('Could not get OCR data for url {url}.'.format(**locals()))
return '<failed to load>'
if response is None:
log.error('Did not get OCR data for url {url}.'.format(**locals()))
return '<failed to load>'
else:
# Each paragraph is one item in the list
return "\n\n".join([response[key] for key in sorted(response.keys()) if key != 'title'])
def list_next_articles(self):
"""Retrieve next page of search results
"""
url = self.results_url()
try:
response = request.get(url)
except:
log.exception('Could not get results for url {url}'.format(**locals()))
return []
self.number_of_records = response['numberOfRecords']
log.info('Page {self.page} of article list retrieved. '
'{self.records_processed} of {self.number_of_records} articles processed.'.format(**locals()))
return response['records']
def setup_amcat(self, from_date, until_date):
"""Create Amcat API object and save configuration data for later use.
"""
amcat_api = AmcatAPI(amcat.host, amcat.username, amcat.password)
log.info('Setup Amcat API with host {0}, username {1}. Use project {2}'.format(amcat.host, amcat.username,
amcat.project))
now = datetime.now().replace(microsecond=0)
set_name = amcat.set_name_template.format(**locals())
try:
aset = amcat_api.create_set(project=amcat.project, name=set_name, provenance=amcat.data_provenance)
except:
log.exception('Could not create article set')
raise
log.info('Created article set in Amcat. ID: {0}'.format(aset['id']))
self.set_id = aset['id']
return amcat_api
