def test_reserve_dtype(self):
t = TypedDfBuilder("a").reserve("x", dtype=np.float32).build()
df = t.convert(pd.DataFrame([pd.Series(dict(x="0.5"))]))
assert df.column_names() == ["x"]
assert df.to_numpy().tolist() == [[0.5]]
with pytest.raises(ValueError):
t.convert(pd.DataFrame([pd.Series(dict(x="kitten"))]))
def test_drop(self):
t = TypedDfBuilder("a").reserve("column").drop("trash").build()
typ: DfTyping = t.get_typing()
assert typ.columns_to_drop == {"trash"}
df = t.convert(pd.DataFrame([pd.Series(dict(x="x", zz="y"))]))
assert df.column_names() == ["x", "zz"]
df = t.convert(pd.DataFrame([pd.Series(dict(x="x", trash="y"))]))
assert df.column_names() == ["x"]
def test_condition(self):
t = TypedDfBuilder("a").verify(always_ok).build()
typ: DfTyping = t.get_typing()
assert typ.required_columns == []
assert typ.required_index_names == []
assert typ.verifications == [always_ok]
TypedDf(pd.DataFrame())
t = TypedDfBuilder("a").verify(always_fail).build()
with pytest.raises(VerificationFailedError):
t.convert(pd.DataFrame())
def test_no_overwrite(self):
t = TypedDfBuilder("a").reserve("x", "y").build()
df = t.convert(pd.DataFrame([pd.Series(dict(x="cat", y="dog"))]))
with tmpfile(".csv") as path:
df.write_file(path, overwrite=False)
with pytest.raises(FileExistsError):
df.write_file(path, overwrite=False)
def test_mkdir(self):
t = TypedDfBuilder("a").reserve("x", "y").build()
df = t.convert(pd.DataFrame([pd.Series(dict(x="cat", y="dog"))]))
with tmpdir() as path:
df.write_file(path / "a.csv", mkdirs=True)
with tmpdir() as path:
with pytest.raises(FileNotFoundError):
df.write_file(path / "b.csv")
def test_pass_io_options(self):
t = TypedDfBuilder("a").reserve("x",
"y").add_write_kwargs(FileFormat.csv,
sep="&").build()
df = t.convert(pd.DataFrame([pd.Series(dict(x="cat", y="dog"))]))
with tmpfile(".csv") as path:
df.write_file(path)
lines = path.read_text(encoding="utf8").splitlines()
assert lines == ["x&y", "cat&dog"]
def test_dir_hash(self):
t = TypedDfBuilder("a").reserve("x", "y").build()
df = t.convert(pd.DataFrame([pd.Series(dict(x="cat", y="kitten"))]))
with tmpfile(".csv") as path:
hash_dir = Checksums().get_dirsum_of_file(path)
hash_dir.unlink(missing_ok=True)
df.write_file(path, dir_hash=True)
assert hash_dir.exists()
got = Checksums().load_dirsum_exact(hash_dir)
assert list(got.keys()) == [path]
hit = got[path]
assert len(hit) == 64
t.read_file(path, dir_hash=True)
t.read_file(path, hex_hash=hit)
def test_file_hash(self):
t = TypedDfBuilder("a").reserve("x", "y").build()
df = t.convert(pd.DataFrame([pd.Series(dict(x="cat", y="dog"))]))
# unfortunately, the file that gets output is os-dependent
# \n vs \r\n is an issue, so we can't check the exact hash
with tmpfile(".csv") as path:
df.write_file(path, file_hash=True)
hash_file = Checksums().get_filesum_of_file(path)
assert hash_file.exists()
got = Checksums().load_filesum_of_file(path)
assert got.file_path == path
hit = got.hash_value
assert len(hit) == 64
t.read_file(path, file_hash=True)
t.read_file(path, hex_hash=hit)
def test_strict(self):
# strict columns but not index
t = TypedDfBuilder("a").strict(index=False, cols=True).build()
typ: DfTyping = t.get_typing()
assert typ.more_indices_allowed
assert not typ.more_columns_allowed
t.convert(pd.DataFrame([pd.Series(dict(x="x"))]).set_index("x"))
with pytest.raises(UnexpectedColumnError):
t.convert(pd.DataFrame([pd.Series(dict(x="x"))]))
# strict index but not columns
t = TypedDfBuilder("a").strict(True, False).build()
typ: DfTyping = t.get_typing()
assert typ.more_columns_allowed
assert not typ.more_indices_allowed
t.convert(pd.DataFrame([pd.Series(dict(x="x"))]))
with pytest.raises(UnexpectedIndexNameError):
df = PrettyDf(
pd.DataFrame([pd.Series(dict(x="x"))]).set_index("x"))
assert df.index_names() == ["x"]
assert df.column_names() == []
t.convert(df)
# neither strict
t = TypedDfBuilder("a").strict(False, False).build()
t.convert(pd.DataFrame([pd.Series(dict(x="x"))]))
def test_attrs_hard(self):
meta = None
try:
t = TypedDfBuilder("a").reserve("x", "y").build()
df = t.convert(pd.DataFrame([pd.Series(dict(x="cat",
y="kitten"))]))
df.attrs["matrix"] = np.zeros((2, 2))
with tmpfile(".csv") as path:
df.write_file(path, attrs=True)
meta = Path(str(path) + ".attrs.json")
assert meta.exists()
df = t.read_file(path, attrs=True)
assert df.attrs == {"matrix": [["0.0", "0.0"], ["0.0", "0.0"]]}
finally:
if meta is not None:
meta.unlink(missing_ok=True)
def test_attrs(self):
meta = None
try:
t = TypedDfBuilder("a").reserve("x", "y").build()
df = t.convert(pd.DataFrame([pd.Series(dict(x="cat",
y="kitten"))]))
df.attrs["fruit"] = "apple"
with tmpfile(".csv") as path:
df.write_file(path, attrs=True)
meta = Path(str(path) + ".attrs.json")
assert meta.exists()
data = meta.read_text(encoding="utf-8").replace("\n",
"").replace(
" ", "")
assert data == '{"fruit": "apple"}'
df = t.read_file(path, attrs=True)
assert df.attrs == {"fruit": "apple"}
finally:
if meta is not None:
meta.unlink(missing_ok=True)
