gh-158252: Match Unicode whitespace in the C strptime parser

Py_ISSPACE excludes ASCII separators U+001C through U+001F, which the
Python parser treats as whitespace. Consuming one as a literal can leave
a space for the next numeric field, changing how adjacent fields split.

Use Py_UNICODE_ISSPACE and add regression tests for all four public
entry points. These cases returned different dates or times before the
fix.

Author: pganssle

Lib/test/datetimetester.py

@@ -1281,6 +1281,13 @@ def test_strptime_whitespace(self):
                     fmt = format_space.join(('%Y', '%m', '%d'))
                     self.assertEqual(date.strptime(text, fmt), date(2024, 2, 29))
 
+    def test_strptime_whitespace_before_adjacent_fields(self):
+        for separator in ('\x1c', '\x1d', '\x1e', '\x1f'):
+            with self.subTest(separator=separator):
+                text = f'2024{separator} 112'
+                fmt = f'%Y{separator}%d%m'
+                self.assertEqual(date.strptime(text, fmt), date(2024, 2, 11))
+
     def test_strptime_single_digit(self):
         # bpo-34903: Check that single digit dates are allowed.
         strptime = date.strptime
@@ -3202,6 +3209,14 @@ def test_strptime_whitespace(self):
                     self.assertEqual(self.theclass.strptime(text, fmt),
                                      self.theclass(2024, 2, 29, 12, 34, 56))
 
+    def test_strptime_whitespace_before_adjacent_fields(self):
+        for separator in ('\x1c', '\x1d', '\x1e', '\x1f'):
+            with self.subTest(separator=separator):
+                text = f'2024{separator} 112 34{separator} 123'
+                fmt = f'%Y{separator}%d%m %M{separator}%H%S'
+                self.assertEqual(self.theclass.strptime(text, fmt),
+                                 self.theclass(2024, 2, 11, 12, 34, 3))
+
     def test_strptime_numeric_fallback(self):
         cases = (
             ('2024\0-02-29', '%Y\0-%m-%d', (2024, 2, 29)),
@@ -4435,6 +4450,14 @@ def test_strptime_whitespace(self):
                     self.assertEqual(self.theclass.strptime(text, fmt),
                                      self.theclass(12, 34, 56))
 
+    def test_strptime_whitespace_before_adjacent_fields(self):
+        for separator in ('\x1c', '\x1d', '\x1e', '\x1f'):
+            with self.subTest(separator=separator):
+                text = f'34{separator} 123'
+                fmt = f'%M{separator}%H%S'
+                self.assertEqual(self.theclass.strptime(text, fmt),
+                                 self.theclass(12, 34, 3))
+
     def test_strptime_tz(self):
         strptime = self.theclass.strptime
         self.assertEqual(strptime("+0002", "%z").utcoffset(), 2 * MINUTE)

Lib/test/test_time.py

@@ -382,6 +382,14 @@ def test_strptime_whitespace(self):
                     self.assertEqual(time.strptime(text, fmt),
                                      (2024, 2, 29, 12, 34, 56, 3, 60, -1))
 
+    def test_strptime_whitespace_before_adjacent_fields(self):
+        for separator in ('\x1c', '\x1d', '\x1e', '\x1f'):
+            with self.subTest(separator=separator):
+                text = f'2024{separator} 112 34{separator} 123'
+                fmt = f'%Y{separator}%d%m %M{separator}%H%S'
+                self.assertEqual(time.strptime(text, fmt),
+                                 (2024, 2, 11, 12, 34, 3, 6, 42, -1))
+
     def test_strptime_numeric_calendar(self):
         cases = (
             ('0001-01-01', 1, 1, 1, 0, 1),

Python/_strptime.c

@@ -111,16 +111,17 @@ _PyTime_Strptime(PyObject *string, PyObject *format, _PyTime_StrptimeFields *fie
 
     for (Py_ssize_t i = 0; i < fmt_length; i++) {
         unsigned char c = fmt[i];
-        if (Py_ISSPACE(c)) {
-            if (pos == length || !Py_ISSPACE(data[pos])) {
+        /* Match regex \s, including ASCII separators U+001C through U+001F. */
+        if (Py_UNICODE_ISSPACE(c)) {
+            if (pos == length || !Py_UNICODE_ISSPACE(data[pos])) {
                 return 0;
             }
-            while (i + 1 < fmt_length && Py_ISSPACE(fmt[i + 1])) {
+            while (i + 1 < fmt_length && Py_UNICODE_ISSPACE(fmt[i + 1])) {
                 i++;
             }
             do {
                 pos++;
-            } while (pos < length && Py_ISSPACE(data[pos]));
+            } while (pos < length && Py_UNICODE_ISSPACE(data[pos]));
             continue;
         }
         if (c != '%') {