Skip to content

Commit aca32d5

Browse files
MarcoGorelliMarcoGorelli
committed
keep fixing
1 parent 005759c commit aca32d5

File tree

2 files changed

+50
-156
lines changed

2 files changed

+50
-156
lines changed

‎tests/test_series.py

-156
Original file line numberDiff line numberDiff line change
@@ -1571,162 +1571,6 @@ def test_categorical_codes():
15711571
assert_type(cat.codes, "np_ndarray_int")
15721572

15731573

1574-
def test_string_accessors():
1575-
s = pd.Series(
1576-
["applep", "bananap", "Cherryp", "DATEp", "eGGpLANTp", "123p", "23.45p"]
1577-
)
1578-
s2 = pd.Series([["apple", "banana"], ["cherry", "date"], [1, "eggplant"]])
1579-
s3 = pd.Series(["a1", "b2", "c3"])
1580-
s4 = pd.Series([b"a1", b"b2", b"c3"])
1581-
check(assert_type(s.str.capitalize(), "pd.Series[str]"), pd.Series, str)
1582-
check(assert_type(s.str.casefold(), "pd.Series[str]"), pd.Series, str)
1583-
check(assert_type(s.str.cat(sep="X"), str), str)
1584-
check(assert_type(s.str.center(10), "pd.Series[str]"), pd.Series, str)
1585-
check(assert_type(s.str.contains("a"), "pd.Series[bool]"), pd.Series, np.bool_)
1586-
check(
1587-
assert_type(s.str.contains(re.compile(r"a")), "pd.Series[bool]"),
1588-
pd.Series,
1589-
np.bool_,
1590-
)
1591-
check(assert_type(s.str.count("pp"), "pd.Series[int]"), pd.Series, np.integer)
1592-
check(assert_type(s4.str.decode("utf-8"), "pd.Series[str]"), pd.Series, str)
1593-
check(assert_type(s.str.encode("latin-1"), "pd.Series[bytes]"), pd.Series, bytes)
1594-
check(assert_type(s.str.endswith("e"), "pd.Series[bool]"), pd.Series, np.bool_)
1595-
check(
1596-
assert_type(s.str.endswith(("e", "f")), "pd.Series[bool]"), pd.Series, np.bool_
1597-
)
1598-
check(assert_type(s3.str.extract(r"([ab])?(\d)"), pd.DataFrame), pd.DataFrame)
1599-
check(assert_type(s3.str.extractall(r"([ab])?(\d)"), pd.DataFrame), pd.DataFrame)
1600-
check(assert_type(s.str.find("p"), "pd.Series[int]"), pd.Series, np.int64)
1601-
check(assert_type(s.str.findall("pp"), "pd.Series[list[str]]"), pd.Series, list)
1602-
check(assert_type(s.str.fullmatch("apple"), "pd.Series[bool]"), pd.Series, np.bool_)
1603-
check(assert_type(s.str.get(2), "pd.Series[str]"), pd.Series, str)
1604-
check(assert_type(s.str.get_dummies(), pd.DataFrame), pd.DataFrame)
1605-
check(assert_type(s.str.index("p"), "pd.Series[int]"), pd.Series, np.int64)
1606-
check(assert_type(s.str.isalnum(), "pd.Series[bool]"), pd.Series, np.bool_)
1607-
check(assert_type(s.str.isalpha(), "pd.Series[bool]"), pd.Series, np.bool_)
1608-
check(assert_type(s.str.isdecimal(), "pd.Series[bool]"), pd.Series, np.bool_)
1609-
check(assert_type(s.str.isdigit(), "pd.Series[bool]"), pd.Series, np.bool_)
1610-
check(assert_type(s.str.isnumeric(), "pd.Series[bool]"), pd.Series, np.bool_)
1611-
check(assert_type(s.str.islower(), "pd.Series[bool]"), pd.Series, np.bool_)
1612-
check(assert_type(s.str.isspace(), "pd.Series[bool]"), pd.Series, np.bool_)
1613-
check(assert_type(s.str.istitle(), "pd.Series[bool]"), pd.Series, np.bool_)
1614-
check(assert_type(s.str.isupper(), "pd.Series[bool]"), pd.Series, np.bool_)
1615-
check(assert_type(s2.str.join("-"), pd.Series), pd.Series)
1616-
check(assert_type(s.str.len(), "pd.Series[int]"), pd.Series, np.integer)
1617-
check(assert_type(s.str.ljust(80), "pd.Series[str]"), pd.Series, str)
1618-
check(assert_type(s.str.lower(), "pd.Series[str]"), pd.Series, str)
1619-
check(assert_type(s.str.lstrip("a"), "pd.Series[str]"), pd.Series, str)
1620-
check(assert_type(s.str.match("pp"), "pd.Series[bool]"), pd.Series, np.bool_)
1621-
check(assert_type(s.str.normalize("NFD"), "pd.Series[str]"), pd.Series, str)
1622-
check(assert_type(s.str.pad(80, "right"), "pd.Series[str]"), pd.Series, str)
1623-
check(assert_type(s.str.partition("p"), pd.DataFrame), pd.DataFrame)
1624-
check(assert_type(s.str.removeprefix("a"), "pd.Series[str]"), pd.Series, str)
1625-
check(assert_type(s.str.removesuffix("e"), "pd.Series[str]"), pd.Series, str)
1626-
check(assert_type(s.str.repeat(2), "pd.Series[str]"), pd.Series, str)
1627-
check(assert_type(s.str.replace("a", "X"), "pd.Series[str]"), pd.Series, str)
1628-
check(assert_type(s.str.rfind("e"), "pd.Series[int]"), pd.Series, np.int64)
1629-
check(assert_type(s.str.rindex("p"), "pd.Series[int]"), pd.Series, np.int64)
1630-
check(assert_type(s.str.rjust(80), "pd.Series[str]"), pd.Series, str)
1631-
check(assert_type(s.str.rpartition("p"), pd.DataFrame), pd.DataFrame)
1632-
check(assert_type(s.str.rsplit("a"), "pd.Series[list[str]]"), pd.Series, list)
1633-
check(assert_type(s.str.rsplit("a", expand=True), pd.DataFrame), pd.DataFrame)
1634-
check(
1635-
assert_type(s.str.rsplit("a", expand=False), "pd.Series[list[str]]"),
1636-
pd.Series,
1637-
list,
1638-
)
1639-
check(assert_type(s.str.rstrip(), "pd.Series[str]"), pd.Series, str)
1640-
check(assert_type(s.str.slice(0, 4, 2), "pd.Series[str]"), pd.Series, str)
1641-
check(
1642-
assert_type(s.str.slice_replace(0, 2, "XX"), "pd.Series[str]"), pd.Series, str
1643-
)
1644-
check(assert_type(s.str.split("a"), "pd.Series[list[str]]"), pd.Series, list)
1645-
# GH 194
1646-
check(assert_type(s.str.split("a", expand=True), pd.DataFrame), pd.DataFrame)
1647-
check(
1648-
assert_type(s.str.split("a", expand=False), "pd.Series[list[str]]"),
1649-
pd.Series,
1650-
list,
1651-
)
1652-
check(assert_type(s.str.startswith("a"), "pd.Series[bool]"), pd.Series, np.bool_)
1653-
check(
1654-
assert_type(s.str.startswith(("a", "b")), "pd.Series[bool]"),
1655-
pd.Series,
1656-
np.bool_,
1657-
)
1658-
check(assert_type(s.str.strip(), "pd.Series[str]"), pd.Series, str)
1659-
check(assert_type(s.str.swapcase(), "pd.Series[str]"), pd.Series, str)
1660-
check(assert_type(s.str.title(), "pd.Series[str]"), pd.Series, str)
1661-
check(
1662-
assert_type(s.str.translate({241: "n"}), "pd.Series[str]"),
1663-
pd.Series,
1664-
str,
1665-
)
1666-
check(assert_type(s.str.upper(), "pd.Series[str]"), pd.Series, str)
1667-
check(assert_type(s.str.wrap(80), "pd.Series[str]"), pd.Series, str)
1668-
check(assert_type(s.str.zfill(10), "pd.Series[str]"), pd.Series, str)
1669-
1670-
1671-
def test_series_overloads_cat():
1672-
s = pd.Series(
1673-
["applep", "bananap", "Cherryp", "DATEp", "eGGpLANTp", "123p", "23.45p"]
1674-
)
1675-
check(assert_type(s.str.cat(sep=";"), str), str)
1676-
check(assert_type(s.str.cat(None, sep=";"), str), str)
1677-
check(
1678-
assert_type(
1679-
s.str.cat(["A", "B", "C", "D", "E", "F", "G"], sep=";"),
1680-
"pd.Series[str]",
1681-
),
1682-
pd.Series,
1683-
str,
1684-
)
1685-
check(
1686-
assert_type(
1687-
s.str.cat(pd.Series(["A", "B", "C", "D", "E", "F", "G"]), sep=";"),
1688-
"pd.Series[str]",
1689-
),
1690-
pd.Series,
1691-
str,
1692-
)
1693-
unknown_s: UnknownSeries = pd.DataFrame({"a": ["a", "b"]})["a"]
1694-
check(assert_type(s.str.cat(unknown_s, sep=";"), "pd.Series[str]"), pd.Series, str)
1695-
1696-
1697-
def test_series_overloads_partition():
1698-
s = pd.Series(
1699-
[
1700-
"ap;pl;ep",
1701-
"ban;an;ap",
1702-
"Che;rr;yp",
1703-
"DA;TEp",
1704-
"eGGp;LANT;p",
1705-
"12;3p",
1706-
"23.45p",
1707-
]
1708-
)
1709-
check(assert_type(s.str.partition(sep=";"), pd.DataFrame), pd.DataFrame)
1710-
check(
1711-
assert_type(s.str.partition(sep=";", expand=True), pd.DataFrame), pd.DataFrame
1712-
)
1713-
check(
1714-
assert_type(s.str.partition(sep=";", expand=False), "pd.Series[type[object]]"),
1715-
pd.Series,
1716-
object,
1717-
)
1718-
1719-
check(assert_type(s.str.rpartition(sep=";"), pd.DataFrame), pd.DataFrame)
1720-
check(
1721-
assert_type(s.str.rpartition(sep=";", expand=True), pd.DataFrame), pd.DataFrame
1722-
)
1723-
check(
1724-
assert_type(s.str.rpartition(sep=";", expand=False), "pd.Series[type[object]]"),
1725-
pd.Series,
1726-
object,
1727-
)
1728-
1729-
17301574
def test_series_overloads_extract():
17311575
s = pd.Series(
17321576
["appl;ep", "ban;anap", "Cherr;yp", "DATEp", "eGGp;LANTp", "12;3p", "23.45p"]

‎tests/test_string_accessors.py

+50
Original file line numberDiff line numberDiff line change
@@ -299,3 +299,53 @@ def test_index_overloads_partition():
299299
pd.Index,
300300
object,
301301
)
302+
303+
304+
def test_series_overloads_cat():
305+
s = pd.Series(DATA)
306+
check(assert_type(s.str.cat(sep=";"), str), str)
307+
check(assert_type(s.str.cat(None, sep=";"), str), str)
308+
check(
309+
assert_type(
310+
s.str.cat(["A", "B", "C", "D", "E", "F", "G"], sep=";"),
311+
"pd.Series[str]",
312+
),
313+
pd.Series,
314+
str,
315+
)
316+
check(
317+
assert_type(
318+
s.str.cat(pd.Series(["A", "B", "C", "D", "E", "F", "G"]), sep=";"),
319+
"pd.Series[str]",
320+
),
321+
pd.Series,
322+
str,
323+
)
324+
unknown_s = pd.DataFrame({"a": ["a", "b"]})["a"]
325+
check(assert_type(s.str.cat(unknown_s, sep=";"), "pd.Series[str]"), pd.Series, str)
326+
327+
328+
def test_index_overloads_cat():
329+
idx = pd.Index(DATA)
330+
check(assert_type(idx.str.cat(sep=";"), str), str)
331+
check(assert_type(idx.str.cat(None, sep=";"), str), str)
332+
check(
333+
assert_type(
334+
idx.str.cat(["A", "B", "C", "D", "E", "F", "G"], sep=";"),
335+
"pd.Index[str]",
336+
),
337+
pd.Index,
338+
str,
339+
)
340+
check(
341+
assert_type(
342+
idx.str.cat(pd.Index(["A", "B", "C", "D", "E", "F", "G"]), sep=";"),
343+
"pd.Index[str]",
344+
),
345+
pd.Index,
346+
str,
347+
)
348+
unknown_idx = pd.DataFrame({"a": ["a", "b"]}).set_index("a").index
349+
check(
350+
assert_type(idx.str.cat(unknown_idx, sep=";"), "pd.Index[str]"), pd.Index, str
351+
)

0 commit comments

Comments
 (0)