def test_rewrite_conflicting_column_headers(self):
with _temp_csv("A,A,Column 4,\na,b,c,d") as path:
result = _internal_parse_csv(path, has_header=True)
assert_csv_result_equals(
result,
ParseCsvResult(
pa.table({
"A": ["a"],
"A 2": ["b"], # rewritten
"Column 4": ["c"],
"Column 5": ["d"], # rewritten
}),
[ParseCsvWarning.NumberedColumnNames(2, "A 2")],
),
)
def test_truncate_column_names(self):
with _temp_csv("ABC,ABCD,ABCDE,BCDEF\na,b,c,d") as path:
assert_csv_result_equals(
_internal_parse_csv(path, has_header=True),
ParseCsvResult(
pa.table({
"ABC": ["a"],
"ABCD": ["b"],
"AB 2": ["c"],
"BCDE": ["d"]
}),
[
ParseCsvWarning.TruncatedColumnNames(2, "AB 2"),
ParseCsvWarning.NumberedColumnNames(1, "AB 2"),
],
),
)