def issue_to_kernel(issue):
"""Transforma um objeto Issue (xylose) para o formato
de dados equivalente ao persistido pelo Kernel em um banco
mongodb"""
issn_id = issue.data["issue"]["v35"][0]["_"]
_creation_date = parse_date(issue.publication_date)
_metadata = {}
_bundle = {
"created": _creation_date,
"updated": _creation_date,
"items": [],
"metadata": _metadata,
}
_year = str(date_to_datetime(_creation_date).year)
_month = str(date_to_datetime(_creation_date).month)
_metadata["publication_year"] = set_metadata(_creation_date, _year)
if issue.volume:
_metadata["volume"] = set_metadata(_creation_date, issue.volume)
if issue.number:
_metadata["number"] = set_metadata(_creation_date, issue.number)
_supplement = None
if issue.type is "supplement":
_supplement = "0"
if issue.supplement_volume:
_supplement = issue.supplement_volume
elif issue.supplement_number:
_supplement = issue.supplement_number
_metadata["supplement"] = set_metadata(_creation_date, _supplement)
if issue.titles:
_titles = [{
"language": lang,
"value": value
} for lang, value in issue.titles.items()]
_metadata["titles"] = set_metadata(_creation_date, _titles)
publication_months = {}
if issue.start_month and issue.end_month:
publication_months["range"] = (int(issue.start_month),
int(issue.end_month))
elif _month:
publication_months["month"] = int(_month)
_metadata["publication_months"] = set_metadata(_creation_date,
publication_months)
_id = scielo_ids_generator.issue_id(issn_id, _year, issue.volume,
issue.number, _supplement)
_bundle["_id"] = _id
_bundle["id"] = _id
return _bundle
def test_issn_year_volume_suppl(self):
self.assertEqual(
scielo_ids_generator.issue_id("ISSN",
"YEAR",
"VOLUME",
supplement="SUPPL"),
"ISSN-YEAR-vVOLUME-sSUPPL",
)
def test_issue_id_returns_ISSN_YEAR_NUM_and_no_vol_because_it_is_00(self):
self.assertEqual(
scielo_ids_generator.issue_id("ISSN",
"YEAR",
volume="00",
number="19"),
"ISSN-YEAR-n19",
)
def get_documents_bundle(session_db, data):
issns = list(
set([
data[issn_type] for issn_type in ("eissn", "pissn", "issn")
if data.get(issn_type)
]))
is_issue = data.get("volume") or data.get("number")
bad_bundle_id = []
for issn in issns:
if is_issue:
bundle_id = scielo_ids_generator.issue_id(
issn,
data.get("year"),
data.get("volume"),
data.get("number"),
data.get("supplement"),
)
else:
bundle_id = scielo_ids_generator.aops_bundle_id(issn)
logger.debug("Fetch documents bundle {}".format(bundle_id))
try:
documents_bundle = session_db.documents_bundles.fetch(bundle_id)
except DoesNotExist:
bad_bundle_id.append(bundle_id)
else:
return documents_bundle
if is_issue:
raise ValueError("Nenhum documents_bundle encontrado %s" %
", ".join(bad_bundle_id))
else:
bad_issn = []
for issn in issns:
try:
documents_bundle = create_aop_bundle(session_db, issn)
except DoesNotExist:
bad_issn.append(issn)
else:
return documents_bundle
raise ValueError("Nenhum periódico encontrado para criação do AOP %s" %
", ".join(bad_issn))
def test_issn_year_number(self):
self.assertEqual(
scielo_ids_generator.issue_id("ISSN", "YEAR", number="03"),
"ISSN-YEAR-n3",
)
def test_issn_year_volume(self):
self.assertEqual(
scielo_ids_generator.issue_id("ISSN", "YEAR", "VOLUME"),
"ISSN-YEAR-vVOLUME",
)
def register_documents_in_documents_bundle(session_db, file_documents: str,
file_journals: str) -> None:
err_filename = os.path.join(config.get("ERRORS_PATH"),
"insert_documents_in_bundle.err")
not_registered = []
journals = reading.read_json_file(file_journals)
documents = reading.read_json_file(file_documents)
data_journal = {}
for journal in journals:
o_journal = Journal(journal)
if o_journal.print_issn:
data_journal[o_journal.print_issn] = o_journal.scielo_issn
if o_journal.electronic_issn:
data_journal[o_journal.electronic_issn] = o_journal.scielo_issn
if o_journal.scielo_issn:
data_journal[o_journal.scielo_issn] = o_journal.scielo_issn
documents_bundles = {}
for scielo_id, document in documents.items():
is_issue = bool(document.get("volume") or document.get("number"))
issn = ""
for issn_type in ("eissn", "pissn", "issn"):
issn = document.get(issn_type)
if issn: break
if is_issue:
bundle_id = scielo_ids_generator.issue_id(
data_journal[issn],
document.get("year"),
document.get("volume"),
document.get("number"),
document.get("supplement"),
)
else:
bundle_id = scielo_ids_generator.aops_bundle_id(data_journal[issn])
documents_bundles.setdefault(bundle_id, {})
documents_bundles[bundle_id].setdefault("items", [])
documents_bundles[bundle_id]["items"].append({
"id":
scielo_id,
"order":
document.get("order", ""),
})
documents_bundles[bundle_id]["data"] = {
"is_issue": is_issue,
"bundle_id": bundle_id,
"issn": data_journal[document.get("issn")],
}
for documents_bundle in documents_bundles.values():
data = documents_bundle["data"]
items = documents_bundle["items"]
try:
documents_bundle = get_documents_bundle(session_db,
data["bundle_id"],
data["is_issue"],
data["issn"])
except ValueError as error:
files.write_file(err_filename, data["bundle_id"] + "\n", "a")
not_registered.append(data["bundle_id"])
else:
link_documents_bundles_with_documents(documents_bundle, items,
session_db)
def test_issue_id_raises_value_error_because_vol_00_and_num_00(self):
with self.assertRaises(scielo_ids_generator.NotAnIssueError):
scielo_ids_generator.issue_id("ISSN",
"YEAR",
volume="00",
number="00")
