def test_build_table_schema(self):
result = build_table_schema(self.df, version=False)
expected = {
"fields": [
{
"name": "idx",
"type": "integer"
},
{
"name": "A",
"type": "integer"
},
{
"name": "B",
"type": "string"
},
{
"name": "C",
"type": "datetime"
},
{
"name": "D",
"type": "duration"
},
],
"primaryKey": ["idx"],
}
assert result == expected
result = build_table_schema(self.df)
assert "pandas_version" in result
def test_build_table_schema(self):
result = build_table_schema(self.df, version=False)
expected = {
"fields": [
{
"name": "index",
"type": "integer"
},
{
"name": "A",
"type": "any",
"extDtype": "DateDtype"
},
{
"name": "B",
"type": "any",
"extDtype": "decimal"
},
{
"name": "C",
"type": "any",
"extDtype": "string"
},
{
"name": "D",
"type": "integer",
"extDtype": "Int64"
},
],
"primaryKey": ["index"],
}
assert result == expected
result = build_table_schema(self.df)
assert "pandas_version" in result
def test_series(self):
s = pd.Series([1, 2, 3], name="foo")
result = build_table_schema(s, version=False)
expected = {
"fields": [
{"name": "index", "type": "integer"},
{"name": "foo", "type": "integer"},
],
"primaryKey": ["index"],
}
assert result == expected
result = build_table_schema(s)
assert "pandas_version" in result
def test_mi_falsey_name(self):
# GH 16203
df = pd.DataFrame(
np.random.randn(4, 4),
index=pd.MultiIndex.from_product([("A", "B"), ("a", "b")]),
)
result = [x["name"] for x in build_table_schema(df)["fields"]]
assert result == ["level_0", "level_1", 0, 1, 2, 3]
def test_series_unnamed(self):
result = build_table_schema(pd.Series([1, 2, 3]), version=False)
expected = {
"fields": [
{"name": "index", "type": "integer"},
{"name": "values", "type": "integer"},
],
"primaryKey": ["index"],
}
assert result == expected
def test_multiindex(self):
df = self.df.copy()
idx = pd.MultiIndex.from_product([("a", "b"), (1, 2)])
df.index = idx
result = build_table_schema(df, version=False)
expected = {
"fields": [
{
"name": "level_0",
"type": "string"
},
{
"name": "level_1",
"type": "integer"
},
{
"name": "A",
"type": "integer"
},
{
"name": "B",
"type": "string"
},
{
"name": "C",
"type": "datetime"
},
{
"name": "D",
"type": "duration"
},
],
"primaryKey": ["level_0", "level_1"],
}
assert result == expected
df.index.names = ["idx0", None]
expected["fields"][0]["name"] = "idx0"
expected["primaryKey"] = ["idx0", "level_1"]
result = build_table_schema(df, version=False)
assert result == expected
def test_build_table_schema(self):
df = DataFrame({
"A": DateArray([dt.date(2021, 10, 10)]),
"B": DecimalArray([decimal.Decimal(10)]),
"C": array(["pandas"], dtype="string"),
"D": array([10], dtype="Int64"),
})
result = build_table_schema(df, version=False)
expected = {
"fields": [
{
"name": "index",
"type": "integer"
},
{
"name": "A",
"type": "any",
"extDtype": "DateDtype"
},
{
"name": "B",
"type": "any",
"extDtype": "decimal"
},
{
"name": "C",
"type": "any",
"extDtype": "string"
},
{
"name": "D",
"type": "integer",
"extDtype": "Int64"
},
],
"primaryKey": ["index"],
}
assert result == expected
result = build_table_schema(df)
assert "pandas_version" in result
def __init__(
self,
obj,
orient: str | None,
date_format: str,
double_precision: int,
ensure_ascii: bool,
date_unit: str,
index: bool,
default_handler: Callable[[Any], JSONSerializable] | None = None,
indent: int = 0,
):
"""
Adds a `schema` attribute with the Table Schema, resets
the index (can't do in caller, because the schema inference needs
to know what the index is, forces orient to records, and forces
date_format to 'iso'.
"""
super().__init__(
obj,
orient,
date_format,
double_precision,
ensure_ascii,
date_unit,
index,
default_handler=default_handler,
indent=indent,
)
if date_format != "iso":
msg = (
"Trying to write with `orient='table'` and "
f"`date_format='{date_format}'`. Table Schema requires dates "
"to be formatted with `date_format='iso'`")
raise ValueError(msg)
self.schema = build_table_schema(obj, index=self.index)
# NotImplemented on a column MultiIndex
if obj.ndim == 2 and isinstance(obj.columns, MultiIndex):
raise NotImplementedError(
"orient='table' is not supported for MultiIndex columns")
# TODO: Do this timedelta properly in objToJSON.c See GH #15137
if ((obj.ndim == 1) and (obj.name in set(obj.index.names))
or len(obj.columns.intersection(obj.index.names))):
msg = "Overlapping names between the index and columns"
raise ValueError(msg)
obj = obj.copy()
timedeltas = obj.select_dtypes(include=["timedelta"]).columns
if len(timedeltas):
obj[timedeltas] = obj[timedeltas].applymap(lambda x: x.isoformat())
# Convert PeriodIndex to datetimes before serializing
if is_period_dtype(obj.index.dtype):
obj.index = obj.index.to_timestamp()
# exclude index from obj if index=False
if not self.index:
self.obj = obj.reset_index(drop=True)
else:
self.obj = obj.reset_index(drop=False)
self.date_format = "iso"
self.orient = "records"
self.index = index
