def check(self, metadata):
"""Enriquece e normaliza itens do dicionário ``metadata``, que representa
metadados de um periódico.
A estrutura de ``metadata`` é a mesma retornada pelo formato JSON, do
``articlemeta.scielo.org``, conforme exemplo:
https://gist.github.com/gustavofonseca/92638fe6e1f85dd84bcebce72e83b76e
"""
metadata_copy = metadata.copy()
journal = Journal(metadata_copy)
issns = set([
journal.any_issn(priority=u'electronic'),
journal.any_issn(priority=u'print'),
journal.scielo_issn
])
metadata_copy['code'] = journal.scielo_issn
metadata_copy['issns'] = list(issns)
metadata_copy['collection'] = journal.collection_acronym
if not isinstance(journal.data['processing_date'], datetime):
try:
metadata_copy['processing_date'] = datetime.strptime(journal.data['processing_date'], '%Y-%m-%d')
except:
metadata_copy['processing_date'] = datetime.now()
return metadata_copy
def check(self, metadata):
"""Enriquece e normaliza itens do dicionário ``metadata``, que representa
metadados de um periódico.
A estrutura de ``metadata`` é a mesma retornada pelo formato JSON, do
``articlemeta.scielo.org``, conforme exemplo:
https://gist.github.com/gustavofonseca/92638fe6e1f85dd84bcebce72e83b76e
"""
metadata_copy = metadata.copy()
journal = Journal(metadata_copy)
issns = set([
journal.any_issn(priority=u'electronic'),
journal.any_issn(priority=u'print'), journal.scielo_issn
])
metadata_copy['code'] = journal.scielo_issn
metadata_copy['issns'] = list(issns)
metadata_copy['collection'] = journal.collection_acronym
if not isinstance(journal.data['processing_date'], datetime):
try:
metadata_copy['processing_date'] = datetime.strptime(
journal.data['processing_date'], '%Y-%m-%d')
except:
metadata_copy['processing_date'] = datetime.now()
return metadata_copy
def _check_journal_meta(self, metadata):
"""
This method will check the given metadata and retrieve
a new dictionary with some new fields.
"""
journal = Journal(metadata)
issns = set([journal.any_issn(priority=u"electronic"), journal.any_issn(priority=u"print")])
metadata["code"] = list(issns)
metadata["collection"] = journal.collection_acronym
return metadata
def _check_journal_meta(self, metadata):
"""
This method will check the given metadata and retrieve
a new dictionary with some new fields.
"""
journal = Journal(metadata)
issns = set([
journal.any_issn(priority=u'electronic'),
journal.any_issn(priority=u'print'), journal.scielo_issn
])
metadata['code'] = journal.scielo_issn
metadata['issns'] = list(issns)
metadata['collection'] = journal.collection_acronym
return metadata
def _check_journal_meta(self, metadata):
"""
This method will check the given metadata and retrieve
a new dictionary with some new fields.
"""
journal = Journal(metadata)
issns = set([
journal.any_issn(priority=u'electronic'),
journal.any_issn(priority=u'print'),
journal.scielo_issn
])
metadata['code'] = journal.scielo_issn
metadata['issns'] = list(issns)
metadata['collection'] = journal.collection_acronym
return metadata
def _check_journal_meta(self, metadata):
"""
This method will check the given metadata and retrieve
a new dictionary with some new fields.
"""
journal = Journal(metadata)
issns = set([
journal.any_issn(priority=u'electronic'),
journal.any_issn(priority=u'print'),
journal.scielo_issn
])
metadata['code'] = journal.scielo_issn
metadata['issns'] = list(issns)
metadata['collection'] = journal.collection_acronym
if not isinstance(journal.processing_date, datetime):
try:
metadata['processing_date'] = datetime.strptime(journal.processing_date, '%Y-%m-%d')
except:
metadata['processing_date'] = datetime.now()
return metadata
def journal_as_kernel(journal: Journal) -> dict:
"""Gera um dicionário com a estrutura esperada pela API do Kernel a
partir da estrutura gerada pelo isis2json"""
_payload = {}
_id = journal.any_issn()
if not _id:
_id = journal.scielo_issn
_payload["_id"] = _id
if journal.mission:
_payload["mission"] = [{
"language": lang,
"value": value
} for lang, value in journal.mission.items()]
else:
_payload["mission"] = []
_payload["title"] = journal.title or ""
_payload["title_iso"] = journal.abbreviated_iso_title or ""
_payload["short_title"] = journal.abbreviated_title or ""
_payload["acronym"] = journal.acronym or ""
_payload["scielo_issn"] = journal.scielo_issn or ""
_payload["print_issn"] = journal.print_issn or ""
_payload["electronic_issn"] = journal.electronic_issn or ""
_payload["status"] = {}
if journal.status_history:
_status = journal.status_history[-1]
_payload["status"]["status"] = _status[1]
if _status[2]:
_payload["status"]["reason"] = _status[2]
_payload["subject_areas"] = []
if journal.subject_areas:
for subject_area in journal.subject_areas:
# TODO: Algumas áreas estão em caixa baixa, o que devemos fazer?
# A Base MST possui uma grande área que é considerada errada
# é preciso normalizar o valor
if subject_area.upper() == "LINGUISTICS, LETTERS AND ARTS":
subject_area = "LINGUISTIC, LITERATURE AND ARTS"
_payload["subject_areas"].append(subject_area.upper())
_payload["sponsors"] = []
if journal.sponsors:
_payload["sponsors"] = [{
"name": sponsor
} for sponsor in journal.sponsors]
_payload["subject_categories"] = journal.wos_subject_areas or []
_payload["online_submission_url"] = journal.submission_url or ""
_payload["next_journal"] = {}
if journal.next_title:
_payload["next_journal"]["name"] = journal.next_title
_payload["previous_journal"] = {}
if journal.previous_title:
_payload["previous_journal"]["name"] = journal.previous_title
_payload["contact"] = {}
if journal.editor_email:
_payload["contact"]["email"] = journal.editor_email
if journal.editor_address:
_payload["contact"]["address"] = journal.editor_address
return _payload