Series.str.find fix for arrow strings when start < 0 (#56412)

Author: rohanjain101

doc/source/whatsnew/v2.2.0.rst

@@ -579,6 +579,7 @@ Strings
 - Bug in :meth:`DataFrame.apply` failing when ``engine="numba"`` and columns or index have ``StringDtype`` (:issue:`56189`)
 - Bug in :meth:`Index.str.cat` always casting result to object dtype (:issue:`56157`)
 - Bug in :meth:`Series.__mul__` for :class:`ArrowDtype` with ``pyarrow.string`` dtype and ``string[pyarrow]`` for the pyarrow backend (:issue:`51970`)
+- Bug in :meth:`Series.str.find` when ``start < 0`` for :class:`ArrowDtype` with ``pyarrow.string`` (:issue:`56411`)
 - Bug in :meth:`Series.str.replace` when ``n < 0`` for :class:`ArrowDtype` with ``pyarrow.string`` (:issue:`56404`)
 - Bug in :meth:`Series.str.startswith` and :meth:`Series.str.endswith` with arguments of type ``tuple[str, ...]`` for ``string[pyarrow]`` (:issue:`54942`)
 

pandas/core/arrays/arrow/array.py

@@ -2193,7 +2193,8 @@ def _str_find(self, sub: str, start: int = 0, end: int | None = None):
             slices = pc.utf8_slice_codeunits(self._pa_array, start, stop=end)
             result = pc.find_substring(slices, sub)
             not_found = pc.equal(result, -1)
-            offset_result = pc.add(result, end - start)
+            start_offset = max(0, start)
+            offset_result = pc.add(result, start_offset)
             result = pc.if_else(not_found, result, offset_result)
         elif start == 0 and end is None:
             slices = self._pa_array

pandas/tests/extension/test_arrow.py

@@ -1835,7 +1835,7 @@ def test_str_fullmatch(pat, case, na, exp):
 
 @pytest.mark.parametrize(
     "sub, start, end, exp, exp_typ",
-    [["ab", 0, None, [0, None], pa.int32()], ["bc", 1, 3, [2, None], pa.int64()]],
+    [["ab", 0, None, [0, None], pa.int32()], ["bc", 1, 3, [1, None], pa.int64()]],
 )
 def test_str_find(sub, start, end, exp, exp_typ):
     ser = pd.Series(["abc", None], dtype=ArrowDtype(pa.string()))
@@ -1844,6 +1844,14 @@ def test_str_find(sub, start, end, exp, exp_typ):
     tm.assert_series_equal(result, expected)
 
 
+def test_str_find_negative_start():
+    # GH 56411
+    ser = pd.Series(["abc", None], dtype=ArrowDtype(pa.string()))
+    result = ser.str.find(sub="b", start=-1000, end=3)
+    expected = pd.Series([1, None], dtype=ArrowDtype(pa.int64()))
+    tm.assert_series_equal(result, expected)
+
+
 def test_str_find_notimplemented():
     ser = pd.Series(["abc", None], dtype=ArrowDtype(pa.string()))
     with pytest.raises(NotImplementedError, match="find not implemented"):