[-] Try to make the datetime guesser better

Author: ben-cutler-tamarin

pandas/_libs/tslibs/parsing.pyx

@@ -1049,7 +1049,6 @@ def guess_datetime_format(dt_str: str, bint dayfirst=False) -> str | None:
     # rebuild string, capturing any inferred padding
     dt_str = "".join(tokens)
     if parsed_datetime.strftime(guessed_format) == dt_str:
-        _maybe_warn_about_dayfirst(guessed_format, dayfirst)
         return guessed_format
     else:
         return None
@@ -1072,29 +1071,6 @@ cdef str _fill_token(token: str, padding: int):
     return token_filled
 
 
-cdef void _maybe_warn_about_dayfirst(format: str, bint dayfirst) noexcept:
-    """Warn if guessed datetime format doesn't respect dayfirst argument."""
-    cdef:
-        int day_index = format.find("%d")
-        int month_index = format.find("%m")
-
-    if (day_index != -1) and (month_index != -1):
-        if (day_index > month_index) and dayfirst:
-            warnings.warn(
-                f"Parsing dates in {format} format when dayfirst=True was specified. "
-                "Pass `dayfirst=False` or specify a format to silence this warning.",
-                UserWarning,
-                stacklevel=find_stack_level(),
-            )
-        if (day_index < month_index) and not dayfirst:
-            warnings.warn(
-                f"Parsing dates in {format} format when dayfirst=False (the default) "
-                "was specified. "
-                "Pass `dayfirst=True` or specify a format to silence this warning.",
-                UserWarning,
-                stacklevel=find_stack_level(),
-            )
-
 
 cpdef str get_rule_month(str source):
     """

pandas/core/tools/datetimes.py

@@ -1,7 +1,7 @@
 from __future__ import annotations
 
 from collections import abc
-from datetime import date
+from datetime import date, datetime
 from functools import partial
 from itertools import islice
 from typing import (
@@ -131,24 +131,26 @@ class FulldatetimeDict(YearMonthDayDict, total=False):
 
 def _guess_datetime_format_for_array(arr, dayfirst: bool | None = False) -> str | None:
     # Try to guess the format based on the first non-NaN element, return None if can't
-    if (first_non_null := tslib.first_non_null(arr)) != -1:
-        if type(first_non_nan_element := arr[first_non_null]) is str:
-            # GH#32264 np.str_ object
-            guessed_format = guess_datetime_format(
-                first_non_nan_element, dayfirst=dayfirst
-            )
+    search_start = 0
+    allowed_formats = set()
+    while not search_start >= len(arr):
+        non_null_offset = tslib.first_non_null(arr[search_start:])
+        if non_null_offset == -1:
+            break
+        idx = search_start + non_null_offset
+        element = arr[idx]
+        if isinstance(element, str):
+            guessed_format = guess_datetime_format(str(element), dayfirst=dayfirst)
             if guessed_format is not None:
-                return guessed_format
-            # If there are multiple non-null elements, warn about
-            # how parsing might not be consistent
-            if tslib.first_non_null(arr[first_non_null + 1 :]) != -1:
-                warnings.warn(
-                    "Could not infer format, so each element will be parsed "
-                    "individually, falling back to `dateutil`. To ensure parsing is "
-                    "consistent and as-expected, please specify a format.",
-                    UserWarning,
-                    stacklevel=find_stack_level(),
-                )
+                allowed_formats.add(guessed_format)
+        search_start = idx + 1
+    # Look through the formats and see if one satisfies each item in the array
+    for fmt in list(allowed_formats):
+        try:
+            [datetime.strptime(date_string, fmt) for date_string in arr if date_string]
+            return fmt
+        except ValueError:
+            pass
     return None
 
 

pandas/tests/tslibs/test_parsing.py

@@ -424,3 +424,44 @@ def test_parse_datetime_string_with_reso_yearfirst(yearfirst, input):
         )
         assert except_out_dateutil == except_in_dateutil
         assert result[0] == expected
+
+
+from pandas.core.tools.datetimes import _guess_datetime_format_for_array
+
+
+@pytest.mark.parametrize(
+    "expected_format, array",
+    [
+        ("%d/%m/%Y", np.array(["01/02/2025", "30/07/2025"])),
+        ("%Y-%m-%d", np.array(["2025-08-09", "2025-08-13", None])),
+        ("%m/%d/%Y", np.array(["02/01/2025", "12/31/2025"])),
+        ("%d-%m-%Y", np.array(["01-02-2025", "30-07-2025"])),
+        ("%d.%m.%Y", np.array(["01.02.2025", "30.07.2025"])),
+        ("%Y/%m/%d", np.array(["2025/08/09", "2025/12/01"])),
+        ("%b %d, %Y", np.array(["Feb 01, 2025", "Jul 30, 2025"])),
+        ("%B %d, %Y", np.array(["February 01, 2025", "July 30, 2025"])),
+        ("%d %b %Y", np.array(["01 Feb 2025", "30 Jul 2025"])),
+        ("%d-%b-%Y", np.array(["01-Feb-2025", "30-Jul-2025"])),
+        ("%Y%m%d", np.array(["20250201", "20250730"])),
+        (None, np.array(["02/01/25", "12/31/25"])),
+        ("%Y-%m-%d %H:%M:%S", np.array(["2025-08-09 14:30:00", "2025-12-01 00:00:00"])),
+        ("%Y-%m-%dT%H:%M:%S", np.array(["2025-08-09T14:30:00", "2025-12-01T00:00:00"])),
+        (
+            "%Y-%m-%dT%H:%M:%S.%f",
+            np.array(["2025-08-09T14:30:00.123456", "2025-12-01T00:00:00.5"]),
+        ),
+        (
+            "%Y-%m-%d %H:%M:%S%z",
+            np.array(["2025-08-09 14:30:00+0000", "2025-12-01 09:15:00-0500"]),
+        ),
+        ("%Y-%m-%d", np.array(["2025-08-09", None, "2025-12-01"])),
+        (None, np.array(["2025/13/01", "not-a-date", ""])),
+        (
+            None,
+            np.array(["01/02/2025", "2025-02-01"]),
+        ),
+    ],
+)
+def test_guess_datetime_format_for_array(expected_format: str, array: np.array) -> None:
+    fmt = _guess_datetime_format_for_array(array, dayfirst=False)
+    assert fmt == expected_format, f"{fmt} does not match {expected_format}"