def test_json_syntax_error(self):
assert_json_result_equals(
_parse_json_with_defaults("not JSON"),
ParseJsonResult(
pyarrow.table({}),
[
ParseJsonWarning.TODO_i18n(
"JSON parse error at byte 1: Invalid value.")
],
),
)
def test_json_not_array(self):
assert_json_result_equals(
_parse_json_with_defaults('"foo"'),
ParseJsonResult(
pyarrow.table({}),
[
ParseJsonWarning.TODO_i18n(
'JSON is not an Array or Object containing an Array; got: "foo"'
)
],
),
)
def test_json_not_records(self):
assert_json_result_equals(
_parse_json_with_defaults(["foo", "bar"]),
ParseJsonResult(
pyarrow.table({}),
[
ParseJsonWarning.TODO_i18n(
'skipped 2 non-Object records; example Array item 0: "foo"'
)
],
),
)
def test_json_replace_badly_encoded_characters(self):
assert_json_result_equals(
_parse_json_with_defaults('[{"x": "café"}]'.encode("windows-1252"),
encoding="utf-8"),
ParseJsonResult(
pyarrow.table({"x": ["caf�"]}),
[
ParseJsonWarning.RepairedEncoding(
encoding="utf-8",
first_invalid_byte=233,
first_invalid_byte_position=11,
)
],
),
)
def test_max_bytes_per_column_name(self):
assert_json_result_equals(
_parse_json_with_defaults([{
"ABCD": "x",
"BCDEFG": "y"
}]),
ParseJsonResult(
pyarrow.table({
"AB": ["x"],
"BC": ["y"]
}),
[
ParseJsonWarning.TODO_i18n(
"truncated 2 column names; example AB")
],
),
)
def test_max_rows(self):
assert_json_result_equals(
_parse_json_with_defaults([{
"A": "a"
}, {
"A": "b"
}, {
"A": "c"
}]),
ParseJsonResult(
pyarrow.table({"A": ["a", "b"]}),
[
ParseJsonWarning.TODO_i18n(
"skipped 1 rows (after row limit of 2)")
],
),
)
def test_max_bytes_per_value(self):
assert_json_result_equals(
_parse_json_with_defaults([{
"A": ["abc", "def"],
"B": "ghij"
}]),
ParseJsonResult(
pyarrow.table({
"A": ['["a'],
"B": ["ghi"]
}),
[
ParseJsonWarning.TODO_i18n(
"truncated 2 values (value byte limit is 3; see row 0 column A)"
)
],
),
)
def test_max_bytes_text(self):
assert_json_result_equals(
_parse_json_with_defaults([{
"A": "abcd",
"B": "bcde"
}, {
"A": "c",
"B": "def"
}]),
ParseJsonResult(
pyarrow.table({
"A": ["abcd"],
"B": ["bcde"]
}),
[
ParseJsonWarning.TODO_i18n(
"stopped at limit of 8 bytes of data")
],
),
)
def test_max_columns(self):
assert_json_result_equals(
_parse_json_with_defaults([{
"A": "a",
"B": "b",
"C": "c"
}, {
"A": "aa",
"B": "bb"
}]),
ParseJsonResult(
pyarrow.table({
"A": ["a", "aa"],
"B": ["b", "bb"]
}),
[
ParseJsonWarning.TODO_i18n(
"skipped column C (after column limit of 2)")
],
),
)