def import_data_from_sheet(
sheet,
sheet_header,
language_id: str,
implicit: t.Mapping[Literal["languageReference", "id", "value"], str] = {},
concept_column: t.Tuple[str, str] = ("Concept_ID", "Concept_ID"),
) -> t.Iterable[Form]:
row_iter = sheet.iter_rows()
# TODO?: compare header of this sheet to format of given dataset process
# row. Maybe unnecessary. In any case, do not complain about the unused
# variable.
header = next(row_iter) # noqa: F841
assert (
concept_column[1] in sheet_header
), f"Could not find concept column {concept_column[1]} in your excel sheet {sheet.title}."
for row in row_iter:
data = Form({k: clean_cell_value(cell) for k, cell in zip(sheet_header, row)})
if "?" in data.values():
continue
if "value" in implicit:
data[implicit["value"]] = "\t".join(map(str, data.values()))
concept_entry = data.pop(concept_column[1])
data[concept_column[0]] = concept_entry
if "id" in implicit:
data[implicit["id"]] = None
if "languageReference" in implicit:
data[implicit["languageReference"]] = language_id
yield data
def import_data_from_sheet(
sheet,
sheet_header,
implicit: t.Mapping[Literal["languageReference", "id", "value"], str] = {},
entries_to_concepts: t.Mapping[str, str] = KeyKeyDict(),
concept_column: t.Tuple[str, str] = ("Concept_ID", "Concept_ID"),
) -> t.Iterable[Form]:
row_iter = sheet.iter_rows()
# TODO?: compare header of this sheet to format of given data set process
# row. Maybe unnecessary. In any case, do not complain about the unused
# variable.
header = next(row_iter) # noqa: F841
assert (
concept_column[1] in sheet_header
), f"Could not find concept column {concept_column[0]} in your excel sheet {sheet.title}."
for row in row_iter:
data = Form({k: clean_cell_value(cell) for k, cell in zip(sheet_header, row)})
if "value" in implicit:
data[implicit["value"]] = "\t".join(map(str, data.values()))
try:
concept_entry = data.pop(concept_column[1])
data[concept_column[0]] = entries_to_concepts[concept_entry]
except KeyError:
logger.warning(
f"Concept {concept_entry} was not found. Please add it to the concepts table manually. The corresponding form was ignored and not added to the dataset."
)
data[concept_column[0]] = concept_entry
continue
if "id" in implicit:
data[implicit["id"]] = None
if "languageReference" in implicit:
data[implicit["languageReference"]] = sheet.title
yield data