From 009618f1125838af3c4afc772f2593637766fd45 Mon Sep 17 00:00:00 2001 From: "Miss Islington (bot)" <31488909+miss-islington@users.noreply.github.com> Date: Sat, 29 Jun 2024 08:57:33 +0200 Subject: [PATCH] [3.13] gh-120713: Normalize year with century for datetime.strftime (GH-120820) (GH-121144) (cherry picked from commit 6d34938dc8163f4a4bcc68069a1645a7ab76e935) Co-authored-by: blhsing --- Lib/_pydatetime.py | 19 +++++++ Lib/test/datetimetester.py | 32 ++++++----- ...-06-21-06-37-46.gh-issue-120713.WBbQx4.rst | 2 + Modules/_datetimemodule.c | 54 +++++++++++++++++-- configure | 52 ++++++++++++++++++ configure.ac | 28 ++++++++++ pyconfig.h.in | 3 ++ 7 files changed, 174 insertions(+), 16 deletions(-) create mode 100644 Misc/NEWS.d/next/Library/2024-06-21-06-37-46.gh-issue-120713.WBbQx4.rst diff --git a/Lib/_pydatetime.py b/Lib/_pydatetime.py index 34ccb2da13d..27cacb8e01f 100644 --- a/Lib/_pydatetime.py +++ b/Lib/_pydatetime.py @@ -204,6 +204,17 @@ def _format_offset(off, sep=':'): s += '.%06d' % ss.microseconds return s +_normalize_century = None +def _need_normalize_century(): + global _normalize_century + if _normalize_century is None: + try: + _normalize_century = ( + _time.strftime("%Y", (99, 1, 1, 0, 0, 0, 0, 1, 0)) != "0099") + except ValueError: + _normalize_century = True + return _normalize_century + # Correctly substitute for %z and %Z escapes in strftime formats. def _wrap_strftime(object, format, timetuple): # Don't call utcoffset() or tzname() unless actually needed. @@ -261,6 +272,14 @@ def _wrap_strftime(object, format, timetuple): # strftime is going to have at this: escape % Zreplace = s.replace('%', '%%') newformat.append(Zreplace) + elif ch in 'YG' and object.year < 1000 and _need_normalize_century(): + # Note that datetime(1000, 1, 1).strftime('%G') == '1000' so + # year 1000 for %G can go on the fast path. + if ch == 'G': + year = int(_time.strftime("%G", timetuple)) + else: + year = object.year + push('{:04}'.format(year)) else: push('%') push(ch) diff --git a/Lib/test/datetimetester.py b/Lib/test/datetimetester.py index 77472d84e60..3f67c6990b6 100644 --- a/Lib/test/datetimetester.py +++ b/Lib/test/datetimetester.py @@ -1716,18 +1716,26 @@ class TestDate(HarmlessMixedComparison, unittest.TestCase): self.assertTrue(self.theclass.max) def test_strftime_y2k(self): - for y in (1, 49, 70, 99, 100, 999, 1000, 1970): - d = self.theclass(y, 1, 1) - # Issue 13305: For years < 1000, the value is not always - # padded to 4 digits across platforms. The C standard - # assumes year >= 1900, so it does not specify the number - # of digits. - if d.strftime("%Y") != '%04d' % y: - # Year 42 returns '42', not padded - self.assertEqual(d.strftime("%Y"), '%d' % y) - # '0042' is obtained anyway - if support.has_strftime_extensions: - self.assertEqual(d.strftime("%4Y"), '%04d' % y) + # Test that years less than 1000 are 0-padded; note that the beginning + # of an ISO 8601 year may fall in an ISO week of the year before, and + # therefore needs an offset of -1 when formatting with '%G'. + dataset = ( + (1, 0), + (49, -1), + (70, 0), + (99, 0), + (100, -1), + (999, 0), + (1000, 0), + (1970, 0), + ) + for year, offset in dataset: + for specifier in 'YG': + with self.subTest(year=year, specifier=specifier): + d = self.theclass(year, 1, 1) + if specifier == 'G': + year += offset + self.assertEqual(d.strftime(f"%{specifier}"), f"{year:04d}") def test_replace(self): cls = self.theclass diff --git a/Misc/NEWS.d/next/Library/2024-06-21-06-37-46.gh-issue-120713.WBbQx4.rst b/Misc/NEWS.d/next/Library/2024-06-21-06-37-46.gh-issue-120713.WBbQx4.rst new file mode 100644 index 00000000000..18386a43edd --- /dev/null +++ b/Misc/NEWS.d/next/Library/2024-06-21-06-37-46.gh-issue-120713.WBbQx4.rst @@ -0,0 +1,2 @@ +:meth:`datetime.datetime.strftime` now 0-pads years with less than four digits for the format specifiers ``%Y`` and ``%G`` on Linux. +Patch by Ben Hsing diff --git a/Modules/_datetimemodule.c b/Modules/_datetimemodule.c index 135d6cb683f..6c1ae29636e 100644 --- a/Modules/_datetimemodule.c +++ b/Modules/_datetimemodule.c @@ -1848,6 +1848,11 @@ wrap_strftime(PyObject *object, PyObject *format, PyObject *timetuple, const char *ptoappend; /* ptr to string to append to output buffer */ Py_ssize_t ntoappend; /* # of bytes to append to output buffer */ +#ifdef Py_NORMALIZE_CENTURY + /* Buffer of maximum size of formatted year permitted by long. */ + char buf[SIZEOF_LONG*5/2+2]; +#endif + assert(object && format && timetuple); assert(PyUnicode_Check(format)); /* Convert the input format to a C string and size */ @@ -1855,6 +1860,11 @@ wrap_strftime(PyObject *object, PyObject *format, PyObject *timetuple, if (!pin) return NULL; + PyObject *strftime = _PyImport_GetModuleAttrString("time", "strftime"); + if (strftime == NULL) { + goto Done; + } + /* Scan the input format, looking for %z/%Z/%f escapes, building * a new format. Since computing the replacements for those codes * is expensive, don't unless they're actually used. @@ -1936,8 +1946,47 @@ wrap_strftime(PyObject *object, PyObject *format, PyObject *timetuple, ptoappend = PyBytes_AS_STRING(freplacement); ntoappend = PyBytes_GET_SIZE(freplacement); } +#ifdef Py_NORMALIZE_CENTURY + else if (ch == 'Y' || ch == 'G') { + /* 0-pad year with century as necessary */ + PyObject *item = PyTuple_GET_ITEM(timetuple, 0); + long year_long = PyLong_AsLong(item); + + if (year_long == -1 && PyErr_Occurred()) { + goto Done; + } + /* Note that datetime(1000, 1, 1).strftime('%G') == '1000' so year + 1000 for %G can go on the fast path. */ + if (year_long >= 1000) { + goto PassThrough; + } + if (ch == 'G') { + PyObject *year_str = PyObject_CallFunction(strftime, "sO", + "%G", timetuple); + if (year_str == NULL) { + goto Done; + } + PyObject *year = PyNumber_Long(year_str); + Py_DECREF(year_str); + if (year == NULL) { + goto Done; + } + year_long = PyLong_AsLong(year); + Py_DECREF(year); + if (year_long == -1 && PyErr_Occurred()) { + goto Done; + } + } + + ntoappend = PyOS_snprintf(buf, sizeof(buf), "%04ld", year_long); + ptoappend = buf; + } +#endif else { /* percent followed by something else */ +#ifdef Py_NORMALIZE_CENTURY + PassThrough: +#endif ptoappend = pin - 2; ntoappend = 2; } @@ -1969,17 +2018,13 @@ wrap_strftime(PyObject *object, PyObject *format, PyObject *timetuple, goto Done; { PyObject *format; - PyObject *strftime = _PyImport_GetModuleAttrString("time", "strftime"); - if (strftime == NULL) - goto Done; format = PyUnicode_FromString(PyBytes_AS_STRING(newfmt)); if (format != NULL) { result = PyObject_CallFunctionObjArgs(strftime, format, timetuple, NULL); Py_DECREF(format); } - Py_DECREF(strftime); } Done: Py_XDECREF(freplacement); @@ -1987,6 +2032,7 @@ wrap_strftime(PyObject *object, PyObject *format, PyObject *timetuple, Py_XDECREF(colonzreplacement); Py_XDECREF(Zreplacement); Py_XDECREF(newfmt); + Py_XDECREF(strftime); return result; } diff --git a/configure b/configure index e90f4759d3f..c92ec305333 100755 --- a/configure +++ b/configure @@ -25866,6 +25866,58 @@ printf "%s\n" "#define HAVE_STAT_TV_NSEC2 1" >>confdefs.h fi +{ printf "%s\n" "$as_me:${as_lineno-$LINENO}: checking whether year with century should be normalized for strftime" >&5 +printf %s "checking whether year with century should be normalized for strftime... " >&6; } +if test ${ac_cv_normalize_century+y} +then : + printf %s "(cached) " >&6 +else $as_nop + +if test "$cross_compiling" = yes +then : + ac_cv_normalize_century=yes +else $as_nop + cat confdefs.h - <<_ACEOF >conftest.$ac_ext +/* end confdefs.h. */ + +#include +#include + +int main(void) +{ + char year[5]; + struct tm date = { + .tm_year = -1801, + .tm_mon = 0, + .tm_mday = 1 + }; + if (strftime(year, sizeof(year), "%Y", &date) && !strcmp(year, "0099")) { + return 1; + } + return 0; +} + +_ACEOF +if ac_fn_c_try_run "$LINENO" +then : + ac_cv_normalize_century=yes +else $as_nop + ac_cv_normalize_century=no +fi +rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \ + conftest.$ac_objext conftest.beam conftest.$ac_ext +fi + +fi +{ printf "%s\n" "$as_me:${as_lineno-$LINENO}: result: $ac_cv_normalize_century" >&5 +printf "%s\n" "$ac_cv_normalize_century" >&6; } +if test "$ac_cv_normalize_century" = yes +then + +printf "%s\n" "#define Py_NORMALIZE_CENTURY 1" >>confdefs.h + +fi + have_curses=no have_panel=no diff --git a/configure.ac b/configure.ac index 93dd489d14b..5fd9b181734 100644 --- a/configure.ac +++ b/configure.ac @@ -6567,6 +6567,34 @@ then [Define if you have struct stat.st_mtimensec]) fi +AC_CACHE_CHECK([whether year with century should be normalized for strftime], [ac_cv_normalize_century], [ +AC_RUN_IFELSE([AC_LANG_SOURCE([[ +#include +#include + +int main(void) +{ + char year[5]; + struct tm date = { + .tm_year = -1801, + .tm_mon = 0, + .tm_mday = 1 + }; + if (strftime(year, sizeof(year), "%Y", &date) && !strcmp(year, "0099")) { + return 1; + } + return 0; +} +]])], +[ac_cv_normalize_century=yes], +[ac_cv_normalize_century=no], +[ac_cv_normalize_century=yes])]) +if test "$ac_cv_normalize_century" = yes +then + AC_DEFINE([Py_NORMALIZE_CENTURY], [1], + [Define if year with century should be normalized for strftime.]) +fi + dnl check for ncurses/ncursesw and panel/panelw dnl NOTE: old curses is not detected. dnl have_curses=[no, ncursesw, ncurses] diff --git a/pyconfig.h.in b/pyconfig.h.in index c279b147db3..a90f9363f0c 100644 --- a/pyconfig.h.in +++ b/pyconfig.h.in @@ -1659,6 +1659,9 @@ SipHash13: 3, externally defined: 0 */ #undef Py_HASH_ALGORITHM +/* Define if year with century should be normalized for strftime. */ +#undef Py_NORMALIZE_CENTURY + /* Define if rl_startup_hook takes arguments */ #undef Py_RL_STARTUP_HOOK_TAKES_ARGS