def export_matches(collection, authz):
"""Export the top N matches of cross-referencing for the given collection
to an Excel formatted export."""
excel = ExcelWriter()
headers = [
"Score",
"Entity Name",
"Entity Date",
"Entity Countries",
"Candidate Collection",
"Candidate Name",
"Candidate Date",
"Candidate Countries",
"Entity Link",
"Candidate Link",
]
sheet = excel.make_sheet("Cross-reference", headers)
batch = []
for match in index.iter_matches(collection, authz):
batch.append(match)
if len(batch) >= BULK_PAGE:
_iter_match_batch(excel, sheet, batch)
batch = []
if len(batch):
_iter_match_batch(excel, sheet, batch)
return excel.get_bytesio()
def export_matches(export_id):
"""Export the top N matches of cross-referencing for the given collection
to an Excel formatted export."""
export = Export.by_id(export_id)
export_dir = ensure_path(mkdtemp(prefix="aleph.export."))
try:
role = Role.by_id(export.creator_id)
authz = Authz.from_role(role)
collection = Collection.by_id(export.collection_id)
file_name = "%s - Crossreference.xlsx" % collection.label
file_path = export_dir.joinpath(file_name)
excel = ExcelWriter()
headers = [
"Score",
"Entity Name",
"Entity Date",
"Entity Countries",
"Candidate Collection",
"Candidate Name",
"Candidate Date",
"Candidate Countries",
"Entity Link",
"Candidate Link",
]
sheet = excel.make_sheet("Cross-reference", headers)
batch = []
for match in iter_matches(collection, authz):
batch.append(match)
if len(batch) >= BULK_PAGE:
_iter_match_batch(excel, sheet, batch)
batch = []
if len(batch):
_iter_match_batch(excel, sheet, batch)
with open(file_path, "wb") as fp:
buffer = excel.get_bytesio()
for data in buffer:
fp.write(data)
complete_export(export_id, file_path)
except Exception:
log.exception("Failed to process export [%s]", export_id)
export = Export.by_id(export_id)
export.set_status(status=Status.FAILED)
db.session.commit()
finally:
shutil.rmtree(export_dir)
def test_xref(self):
matches = list(iter_matches(self.coll_a, self.authz))
assert 0 == len(matches), len(matches)
xref_collection(self.stage, self.coll_a)
matches = list(iter_matches(self.coll_a, self.authz))
assert 3 == len(matches), len(matches)