def get_document(doc_id):
"""
Retrieves a document
:param doc_id:
:param api_key:
:return:
"""
logger.info(f'Retrieving document {doc_id}')
api_key = key_manager.get_key('abstract')
client = elsclient.ElsClient(api_key)
doc = ElsAbstract(scopus_id=doc_id)
doc.read(client)
doc_data = doc.data
doc_data['_id'] = doc_data['coredata']['dc:identifier'].split(':')[1]
refs = get_doc_refs(doc_data['_id'])
if 'affiliation' in doc_data and isinstance(doc_data['affiliation'], dict):
doc_data['affiliation'] = [doc_data['affiliation']]
if refs:
refs_data = refs.get('references', {}).get('reference', [])
if isinstance(refs_data, list):
doc_data['references'] = refs_data
elif isinstance(refs_data, dict): # Fix the reference data to always make it an array
doc_data['references'] = [refs_data]
doc_data['last_modified'] = datetime.now()
return doc_data
def get_affiliation(affl_id):
logger.info(f'Retrieving affiliation {affl_id}')
api_key = key_manager.get_key('affiliation')
client = elsclient.ElsClient(api_key)
affl = ElsAffil(affil_id=affl_id)
if not affl.read(client):
logger.error("[!]Read affiliation failed: %s", affl_id)
affl.data['_id'] = affl.data['coredata']['dc:identifier'].split(':')[1]
yield affl.data
def get_doc_refs(doc_id):
"""
Retrieves references of a document
:param doc_id:
:param api_key:
:return:
"""
logger.info(f'Retrieving document references {doc_id}')
api_key = key_manager.get_key('abstract')
client = elsclient.ElsClient(api_key)
refs = ElsAbstract(scopus_id=doc_id, params={'view': 'REF'})
refs.read(client)
return refs.data
def get_docs_by_author(author_id):
"""
Get documents published by an author
:param author_id:
:param api_key:
:return:
"""
logger.info(f'Searching docs by author {author_id}')
api_key = key_manager.get_key('scopus_search')
client = elsclient.ElsClient(api_key)
search = ElsSearch(f'au-id({author_id})', 'scopus')
search.execute(client, True)
for doc in search.results:
yield doc
def get_doc_authors(doc_id):
"""
Retrieves document-author relationship in tuples
:param doc_id:
:param api_key:
:return:
"""
logger.info(f'Retrieving author affiliations of document {doc_id}')
api_key = key_manager.get_key('abstract')
client = elsclient.ElsClient(api_key)
abstract = ElsAbstract(scopus_id=doc_id, params={'field': 'author,affiliation'})
abstract.read(client)
for author in abstract.authors:
yield {'doc': doc_id, 'author': author.get('@auid', None), 'seq': author.get('@seq', None)}
def get_author_by_name(last_name, first_name, affiliation_id='60005248'):
"""
Search author by first name, last name and affiliation
:param last_name:
:param first_name:
:param affiliation_id:
:param api_key:
:return:
"""
logger.info('searching authors by name and affiliation')
api_key = key_manager.get_key('scopus_search')
query = f'authlast({last_name}) and authfirst({first_name}) and af-id({affiliation_id})'
client = elsclient.ElsClient(api_key)
search = ElsSearch(query, 'author')
search.execute(client)
yield {'first_name': first_name, 'last_name': last_name, 'results': search.results}
def get_serial_by_title(title):
"""
Retrieves serials by the title
:param title:
:param api_key:
:return:
"""
logger.info('Searching serial by title: ' + title)
api_key = key_manager.get_key('serial_title')
client = elsclient.ElsClient(api_key)
search = ElsSerialTitleSearch(title)
search.execute(client)
for serial in search.results:
serial['_id'] = serial['source-id']
yield serial
def get_author(author_id):
"""
Retrieves an author
:param author_id:
:param api_key:
:return:
"""
logger.info(f'Retrieving author {author_id}')
api_key = key_manager.get_key('author')
client = elsclient.ElsClient(api_key)
author = ElsAuthor(author_id=author_id)
if not author.read(client):
logger.error("[!]Read author failed: %s", author_id)
author.data['_id'] = author.data['coredata']['dc:identifier'].split(':')[1]
author.data['last_modified'] = datetime.now()
return author.data
def get_serial(serial_id):
"""
Retrieves a serial by a scopus ID
:param serial_id:
:param api_key:
:return:
"""
logger.info(f'Retrieving serial {serial_id}')
api_key = key_manager.get_key('serial')
client = elsclient.ElsClient(api_key)
serial = ElsSerial(scopus_id=serial_id, params={'view': 'STANDARD'})
if not serial.read(client):
logger.error("[!]Read serial failed: %s", serial_id)
if 'error' in serial.data:
logger.error("[!]Read serial with error: %s, %s", serial.data['error'], serial_id)
else:
data = serial.data['entry'][0]
data['_id'] = data['source-id']
return data
def get_docs_by_year(year, affl_id='60005248', get_all=False):
"""
Get documents by year
:param year:
:param affl_id:
:param get_all:
:param api_key:
:return:
"""
logger.info(f'Searching docs for year {year} and affiliation {affl_id}')
api_key = key_manager.get_key('scopus_search')
client = elsclient.ElsClient(api_key)
# Split the search since for recent years JHU has publications more than 5,000 each year.
search_one = ElsSearch(f'af-id({affl_id})', 'scopus', {'date': year})
search_one.execute(client, get_all)
all_results = search_one.results
#
# search_two = ElsSearch('af-id(60005248) AND NOT subjarea(MEDI)', 'scopus', {'date': year})
# search_two.execute(client, get_all)
# all_results += search_two.results
return all_results