def parseAuthor(self, authorStr):
"""Hathi data files include an author column that combines author name
with their birth and death dates (sometimes). This method parses
those dates from the name and assigns them as Date objects to the
constructed agent record. This record is then assigned to the work.
"""
logger.info('Storing author {} for work {}'.format(
authorStr, self.work))
authorDateGroup = re.search(r'([0-9\-c?\'.]{4,})', authorStr)
authorDates = None
if authorDateGroup is not None:
authorDates = authorDateGroup.group(1)
authorName = authorStr.replace(authorDates, '').strip(' ,.')
logger.debug('Found lifespan dates {}'.format(authorDates))
else:
authorName = authorStr
logger.debug('Found no lifespan dates')
authorRec = Agent(name=authorName, role='author')
if authorDates is not None:
logger.info('Creating date objects for author lifespan')
lifespan = authorDates.strip(' ,.').split('-')
if len(lifespan) == 1:
logger.debug('Found single date, default to death_date')
dateType = 'death_date'
datePrefix = re.search(r' b(?: |\.)', authorStr)
if datePrefix is not None:
authorRec.name = re.sub(r' b(?: |\.|$)', '',
authorName).strip(' ,.')
logger.debug('Detected single birth_date (living author)')
dateType = 'birth_date'
logger.debug('Storing single date {} of type {}'.format(
lifespan[0], dateType))
authorRec.addClassItem(
'dates', Date, **{
'display_date': lifespan[0],
'date_range': lifespan[0],
'date_type': dateType
})
else:
logger.debug('Storing lifespan {}-{} as dates'.format(
lifespan[0], lifespan[1]))
authorRec.addClassItem(
'dates', Date, **{
'display_date': lifespan[0],
'date_range': lifespan[0],
'date_type': 'birth_date'
})
authorRec.addClassItem(
'dates', Date, **{
'display_date': lifespan[1],
'date_range': lifespan[1],
'date_type': 'death_date'
})
logger.debug('Appending agent record {} to work'.format(authorRec))
self.work.agents.append(authorRec)
def parseAuthor(author):
"""Parse a supplied author into an agent record."""
authorDict = {
'name': author.text,
'viaf': author.get('viaf'),
'lcnaf': author.get('lc')
}
if authorDict['viaf'] is None or authorDict['lcnaf'] is None:
logger.info('Querying VIAF for {}'.format(authorDict['name']))
viafResp = requests.get('{}{}'.format(
'https://dev-platform.nypl.org/api/v0.1/research-now/viaf-lookup?queryName=',
quote_plus(authorDict['name'])
))
responseJSON = viafResp.json()
logger.debug(responseJSON)
if 'viaf' in responseJSON:
logger.debug('Found VIAF {} for agent'.format(responseJSON.get('viaf', None)))
if responseJSON['name'] != authorDict['name']:
authorDict['aliases'] = [authorDict['name']]
authorDict['name'] = responseJSON.get('name', '')
authorDict['viaf'] = responseJSON.get('viaf', None)
authorDict['lcnaf'] = responseJSON.get('lcnaf', None)
return Agent.createFromDict(**authorDict)
def buildAgent(name, role, agentType='personal'):
newAgent = Agent(name=name, role=role)
viafResp = requests.get('{}{}&queryType={}'.format(
'https://dev-platform.nypl.org/api/v0.1/research-now/viaf-lookup?queryName=',
quote_plus(name), agentType))
responseJSON = viafResp.json()
logger.debug(responseJSON)
if 'viaf' in responseJSON:
if responseJSON['name'] != name:
newAgent.aliases.append(name)
newAgent.name = responseJSON.get('name', '')
newAgent.viaf = responseJSON.get('viaf', None)
newAgent.lcnaf = responseJSON.get('lcnaf', None)
return newAgent
def buildAgent(name, role):
newAgent = Agent(name=name, role=role)
queryStr = '{}{}'.format(
'https://dev-platform.nypl.org/api/v0.1/research-now/viaf-lookup?queryName=',
quote_plus(name)
)
if role in ['publisher', 'manufacturer']:
queryStr = '{}&{}'.format(queryStr, 'queryType=corporate')
viafResp = requests.get(queryStr)
responseJSON = viafResp.json()
logger.debug(responseJSON)
if 'viaf' in responseJSON:
if responseJSON['name'] != name:
newAgent.aliases.append(name)
newAgent.name = responseJSON.get('name', '')
newAgent.viaf = responseJSON.get('viaf', None)
newAgent.lcnaf = responseJSON.get('lcnaf', None)
return newAgent
def test_agent_merging(self):
new = [Agent('Test, Tester', 'tester', ['Bad, Tester'])]
existing = [Agent('Testing, Tester', 'user', ['Tester, Bad'])]
merged = Agent.checkForMatches(new, existing)
self.assertEqual(len(merged), 1)
def test_agent_non_matching(self):
new = [Agent('Test, Tester', 'tester')]
existing = [Agent('Other, Agent', 'user')]
merged = Agent.checkForMatches(new, existing)
self.assertEqual(len(merged), 2)
def test_agent_repr(self):
agentTest = Agent(name='Test, Tester', role='tester')
self.assertEqual(
str(agentTest),
'<Agent(name=Test, Tester, roles=tester)>'
)
def test_agent_create_role_list(self):
agentTest = Agent(role=['tester'])
self.assertEqual(agentTest.roles, ['tester'])
def test_agent_create(self):
agentTest = Agent()
self.assertIsInstance(agentTest, Agent)