gh-158252: Parse twelve-hour clocks in C

Use the current locale for AM/PM indicators and preserve the default
without an indicator. Include the equivalent %k, %l and %P directives.
Locales with empty or non-ASCII indicators retain the Python parser.

Author: pganssle

Lib/test/test_strptime.py

@@ -16,6 +16,26 @@
 
 
 class PublicStrptimeTests(unittest.TestCase):
+    @run_with_locales('LC_TIME', 'C', 'en_US', 'fr_FR', 'de_DE', 'ja_JP', '')
+    def test_twelve_hour_clock(self):
+        for hour in range(24):
+            value = datetime.datetime(2024, 1, 1, hour, 24, 35)
+            for fmt in ('%Y-%m-%d %I:%M:%S %p', '%p %I:%M:%S', '%I', '%p'):
+                self.assert_matches_python(value.strftime(fmt), fmt)
+
+    @support.run_with_locale('LC_TIME', 'C')
+    def test_hour_aliases_and_precedence(self):
+        for text, fmt in (
+            ('12 AM', '%I %p'), ('12 PM', '%I %p'),
+            (' 1 pm', '%l %P'), (' 0', '%k'),
+            ('PM 12', '%p %I'), ('23 AM', '%H %p'),
+            ('12 13 PM', '%I %H %p'), ('13 12 PM', '%H %I %p'),
+            ('00 AM', '%I %p'), ('13 PM', '%I %p'),
+            ('1 pM', '%I %p'), ('1 PX', '%I %p'),
+            ('AM PM', '%p %P'),
+        ):
+            self.assert_matches_python(text, fmt)
+
     @run_with_locales('LC_TIME', 'C', 'en_US', 'fr_FR', 'de_DE', 'ja_JP', '')
     def test_locale_names(self):
         for month in range(1, 13):

Python/_strptime.c

@@ -10,6 +10,7 @@
  * Supported inputs are exact str objects containing only ASCII, with:
  * - numeric %Y, %y, %m, %d, %H, %M, %S, and %f fields;
  * - ASCII locale month/weekday names and numeric %w/%u weekdays;
+ * - %I with ASCII %p indicators, and the %k, %l, %P aliases;
  * - a terminal %z: empty, Z, or +/-HHMM or +/-HH:MM with hours below 24;
  * - literal characters, %%, and ASCII whitespace.
  * Duplicate directives, mixed %Y/%y, and day-of-month without a year use the
@@ -44,13 +45,14 @@ strptime_name(const unsigned char *data, Py_ssize_t length, Py_ssize_t pos,
      * As in TimeRE, prefer the longest match, then the first equal name.
      */
     int month = directive == 'b' || directive == 'B';
-    int count = month ? 12 : 7;
+    int count = month ? 12 : directive == 'p' ? 2 : 7;
     char format[] = {'%', directive, '\0'};
     int matched = 0;
     for (int i = 0; i < count; i++) {
         struct tm tm = {.tm_year = 101, .tm_mday = 1,
                         .tm_mon = month ? i : 0,
-                        .tm_wday = month ? 0 : (i + 1) % 7};
+                        .tm_wday = month ? 0 : (i + 1) % 7,
+                        .tm_hour = i == 0 ? 1 : 22};
         char name[256];
         size_t size = strftime(name, sizeof(name), format, &tm);
         if (size == 0) {
@@ -61,12 +63,31 @@ strptime_name(const unsigned char *data, Py_ssize_t length, Py_ssize_t pos,
                 return 0;
             }
         }
-        if (size <= (size_t)matched || size > (size_t)(length - pos)) {
+        char *start = name;
+        if (directive == 'p') {
+            while (size && Py_ISSPACE(start[size - 1])) {
+                size--;
+            }
+            while (size && Py_ISSPACE(*start)) {
+                start++;
+                size--;
+            }
+            /* Empty indicators have special defaulting rules in Python. */
+            if (size == 0) {
+                return 0;
+            }
+        }
+        if (data == NULL) {
+            continue;
+        }
+        if (size < (size_t)matched ||
+            (size == (size_t)matched && directive != 'p') ||
+            size > (size_t)(length - pos)) {
             continue;
         }
         size_t j;
         for (j = 0; j < size; j++) {
-            if (Py_TOLOWER(data[pos + j]) != Py_TOLOWER(name[j])) {
+            if (Py_TOLOWER(data[pos + j]) != Py_TOLOWER(start[j])) {
                 break;
             }
         }
@@ -75,7 +96,7 @@ strptime_name(const unsigned char *data, Py_ssize_t length, Py_ssize_t pos,
             *value = month ? i + 1 : i;
         }
     }
-    return matched;
+    return data == NULL ? 1 : matched;
 #else
     return 0;
 #endif
@@ -134,6 +155,7 @@ _PyTime_Strptime(PyObject *string, PyObject *format, _PyTime_StrptimeFields *fie
         FRACTION = 1U << 7,
         OFFSET = 1U << 8,
         WEEKDAY = 1U << 9,
+        AMPM = 1U << 10,
         ANY_YEAR = YEAR | SHORT_YEAR,
     };
     static const unsigned short directive_bits[128] = {
@@ -147,6 +169,7 @@ _PyTime_Strptime(PyObject *string, PyObject *format, _PyTime_StrptimeFields *fie
         ['f'] = FRACTION,
         ['z'] = OFFSET,
         ['w'] = WEEKDAY,
+        ['p'] = AMPM,
     };
     /* String subclasses and non-ASCII input retain Python's behavior. */
     if (!PyUnicode_CheckExact(string) || !PyUnicode_CheckExact(format) ||
@@ -160,6 +183,8 @@ _PyTime_Strptime(PyObject *string, PyObject *format, _PyTime_StrptimeFields *fie
     *fields = (_PyTime_StrptimeFields){.year = 1900, .month = 1, .day = 1,
                                      .gmtoff = INT_MIN, .weekday = -1};
     unsigned int seen = 0;
+    int hour12 = 0;
+    int ampm = -1;
     Py_ssize_t pos = 0;
 
     for (Py_ssize_t i = 0; i < fmt_length; i++) {
@@ -186,6 +211,15 @@ _PyTime_Strptime(PyObject *string, PyObject *format, _PyTime_StrptimeFields *fie
             return 0;
         }
         c = fmt[i];
+        if (c == 'k') {
+            c = 'H';
+        }
+        else if (c == 'l') {
+            c = 'I';
+        }
+        else if (c == 'P') {
+            c = 'p';
+        }
         if (c == '%') {
             if (pos == length || data[pos++] != '%') {
                 return 0;
@@ -201,6 +235,9 @@ _PyTime_Strptime(PyObject *string, PyObject *format, _PyTime_StrptimeFields *fie
         else if (c == 'a' || c == 'A' || c == 'u') {
             field = 'w';
         }
+        else if (c == 'I') {
+            field = 'H';
+        }
         unsigned int bit = directive_bits[field];
         if (bit == 0) {
             return 0;
@@ -210,7 +247,7 @@ _PyTime_Strptime(PyObject *string, PyObject *format, _PyTime_StrptimeFields *fie
         }
         seen |= bit;
 
-        if (c == 'b' || c == 'B' || c == 'a' || c == 'A') {
+        if (c == 'b' || c == 'B' || c == 'a' || c == 'A' || c == 'p') {
             int value;
             int count = strptime_name(data, length, pos, c, &value);
             if (count == 0) {
@@ -220,6 +257,9 @@ _PyTime_Strptime(PyObject *string, PyObject *format, _PyTime_StrptimeFields *fie
             if (field == 'm') {
                 fields->month = value;
             }
+            else if (c == 'p') {
+                ampm = value;
+            }
             else {
                 fields->weekday = value;
             }
@@ -249,7 +289,8 @@ _PyTime_Strptime(PyObject *string, PyObject *format, _PyTime_StrptimeFields *fie
         else if (field == 'w') {
             maximum = 1;
         }
-        else if ((c == 'd' || c == 'H') && pos < length && data[pos] == ' ') {
+        else if ((c == 'd' || c == 'H' || c == 'I') &&
+                 pos < length && data[pos] == ' ') {
             pos++;
             maximum = 1;
         }
@@ -260,6 +301,13 @@ _PyTime_Strptime(PyObject *string, PyObject *format, _PyTime_StrptimeFields *fie
         }
         pos += count;
         switch (c) {
+            case 'I':
+                if (value < 1 || value > 12) {
+                    return 0;
+                }
+                hour12 = 1;
+                fields->hour = value;
+                break;
             case 'w':
                 if (value > 6) {
                     return 0;
@@ -316,6 +364,12 @@ _PyTime_Strptime(PyObject *string, PyObject *format, _PyTime_StrptimeFields *fie
                 break;
         }
     }
+    if (hour12) {
+        if (ampm < 0 && !strptime_name(NULL, 0, 0, 'p', &ampm)) {
+            return 0;
+        }
+        fields->hour = fields->hour % 12 + (ampm == 1 ? 12 : 0);
+    }
     static const int days_before_month[] = {
         0, 0, 31, 59, 90, 120, 151, 181, 212, 243, 273, 304, 334, 365
     };