/Lib/test/test_strptime.py

  1"""PyUnit testing against strptime"""
  2
  3import unittest
  4import time
  5import locale
  6import re
  7import sys
  8from test import test_support
  9from datetime import date as datetime_date
 10
 11import _strptime
 12
 13class getlang_Tests(unittest.TestCase):
 14    """Test _getlang"""
 15    def test_basic(self):
 16        self.failUnlessEqual(_strptime._getlang(), locale.getlocale(locale.LC_TIME))
 17
 18class LocaleTime_Tests(unittest.TestCase):
 19    """Tests for _strptime.LocaleTime.
 20
 21    All values are lower-cased when stored in LocaleTime, so make sure to
 22    compare values after running ``lower`` on them.
 23
 24    """
 25
 26    def setUp(self):
 27        """Create time tuple based on current time."""
 28        self.time_tuple = time.localtime()
 29        self.LT_ins = _strptime.LocaleTime()
 30
 31    def compare_against_time(self, testing, directive, tuple_position,
 32                             error_msg):
 33        """Helper method that tests testing against directive based on the
 34        tuple_position of time_tuple.  Uses error_msg as error message.
 35
 36        """
 37        strftime_output = time.strftime(directive, self.time_tuple).lower()
 38        comparison = testing[self.time_tuple[tuple_position]]
 39        self.failUnless(strftime_output in testing, "%s: not found in tuple" %
 40                                                    error_msg)
 41        self.failUnless(comparison == strftime_output,
 42                        "%s: position within tuple incorrect; %s != %s" %
 43                        (error_msg, comparison, strftime_output))
 44
 45    def test_weekday(self):
 46        # Make sure that full and abbreviated weekday names are correct in
 47        # both string and position with tuple
 48        self.compare_against_time(self.LT_ins.f_weekday, '%A', 6,
 49                                  "Testing of full weekday name failed")
 50        self.compare_against_time(self.LT_ins.a_weekday, '%a', 6,
 51                                  "Testing of abbreviated weekday name failed")
 52
 53    def test_month(self):
 54        # Test full and abbreviated month names; both string and position
 55        # within the tuple
 56        self.compare_against_time(self.LT_ins.f_month, '%B', 1,
 57                                  "Testing against full month name failed")
 58        self.compare_against_time(self.LT_ins.a_month, '%b', 1,
 59                                  "Testing against abbreviated month name failed")
 60
 61    def test_am_pm(self):
 62        # Make sure AM/PM representation done properly
 63        strftime_output = time.strftime("%p", self.time_tuple).lower()
 64        self.failUnless(strftime_output in self.LT_ins.am_pm,
 65                        "AM/PM representation not in tuple")
 66        if self.time_tuple[3] < 12: position = 0
 67        else: position = 1
 68        self.failUnless(strftime_output == self.LT_ins.am_pm[position],
 69                        "AM/PM representation in the wrong position within the tuple")
 70
 71    def test_timezone(self):
 72        # Make sure timezone is correct
 73        timezone = time.strftime("%Z", self.time_tuple).lower()
 74        if timezone:
 75            self.failUnless(timezone in self.LT_ins.timezone[0] or \
 76                            timezone in self.LT_ins.timezone[1],
 77                            "timezone %s not found in %s" %
 78                            (timezone, self.LT_ins.timezone))
 79
 80    def test_date_time(self):
 81        # Check that LC_date_time, LC_date, and LC_time are correct
 82        # the magic date is used so as to not have issues with %c when day of
 83        #  the month is a single digit and has a leading space.  This is not an
 84        #  issue since strptime still parses it correctly.  The problem is
 85        #  testing these directives for correctness by comparing strftime
 86        #  output.
 87        magic_date = (1999, 3, 17, 22, 44, 55, 2, 76, 0)
 88        strftime_output = time.strftime("%c", magic_date)
 89        self.failUnless(strftime_output == time.strftime(self.LT_ins.LC_date_time,
 90                                                         magic_date),
 91                        "LC_date_time incorrect")
 92        strftime_output = time.strftime("%x", magic_date)
 93        self.failUnless(strftime_output == time.strftime(self.LT_ins.LC_date,
 94                                                         magic_date),
 95                        "LC_date incorrect")
 96        strftime_output = time.strftime("%X", magic_date)
 97        self.failUnless(strftime_output == time.strftime(self.LT_ins.LC_time,
 98                                                         magic_date),
 99                        "LC_time incorrect")
100        LT = _strptime.LocaleTime()
101        LT.am_pm = ('', '')
102        self.failUnless(LT.LC_time, "LocaleTime's LC directives cannot handle "
103                                    "empty strings")
104
105    def test_lang(self):
106        # Make sure lang is set to what _getlang() returns
107        # Assuming locale has not changed between now and when self.LT_ins was created
108        self.failUnlessEqual(self.LT_ins.lang, _strptime._getlang())
109
110
111class TimeRETests(unittest.TestCase):
112    """Tests for TimeRE."""
113
114    def setUp(self):
115        """Construct generic TimeRE object."""
116        self.time_re = _strptime.TimeRE()
117        self.locale_time = _strptime.LocaleTime()
118
119    def test_pattern(self):
120        # Test TimeRE.pattern
121        pattern_string = self.time_re.pattern(r"%a %A %d")
122        self.failUnless(pattern_string.find(self.locale_time.a_weekday[2]) != -1,
123                        "did not find abbreviated weekday in pattern string '%s'" %
124                         pattern_string)
125        self.failUnless(pattern_string.find(self.locale_time.f_weekday[4]) != -1,
126                        "did not find full weekday in pattern string '%s'" %
127                         pattern_string)
128        self.failUnless(pattern_string.find(self.time_re['d']) != -1,
129                        "did not find 'd' directive pattern string '%s'" %
130                         pattern_string)
131
132    def test_pattern_escaping(self):
133        # Make sure any characters in the format string that might be taken as
134        # regex syntax is escaped.
135        pattern_string = self.time_re.pattern("\d+")
136        self.failUnless(r"\\d\+" in pattern_string,
137                        "%s does not have re characters escaped properly" %
138                        pattern_string)
139
140    def test_compile(self):
141        # Check that compiled regex is correct
142        found = self.time_re.compile(r"%A").match(self.locale_time.f_weekday[6])
143        self.failUnless(found and found.group('A') == self.locale_time.f_weekday[6],
144                        "re object for '%A' failed")
145        compiled = self.time_re.compile(r"%a %b")
146        found = compiled.match("%s %s" % (self.locale_time.a_weekday[4],
147                               self.locale_time.a_month[4]))
148        self.failUnless(found,
149            "Match failed with '%s' regex and '%s' string" %
150             (compiled.pattern, "%s %s" % (self.locale_time.a_weekday[4],
151                                           self.locale_time.a_month[4])))
152        self.failUnless(found.group('a') == self.locale_time.a_weekday[4] and
153                         found.group('b') == self.locale_time.a_month[4],
154                        "re object couldn't find the abbreviated weekday month in "
155                         "'%s' using '%s'; group 'a' = '%s', group 'b' = %s'" %
156                         (found.string, found.re.pattern, found.group('a'),
157                          found.group('b')))
158        for directive in ('a','A','b','B','c','d','H','I','j','m','M','p','S',
159                          'U','w','W','x','X','y','Y','Z','%'):
160            compiled = self.time_re.compile("%" + directive)
161            found = compiled.match(time.strftime("%" + directive))
162            self.failUnless(found, "Matching failed on '%s' using '%s' regex" %
163                                    (time.strftime("%" + directive),
164                                     compiled.pattern))
165
166    def test_blankpattern(self):
167        # Make sure when tuple or something has no values no regex is generated.
168        # Fixes bug #661354
169        test_locale = _strptime.LocaleTime()
170        test_locale.timezone = (frozenset(), frozenset())
171        self.failUnless(_strptime.TimeRE(test_locale).pattern("%Z") == '',
172                        "with timezone == ('',''), TimeRE().pattern('%Z') != ''")
173
174    def test_matching_with_escapes(self):
175        # Make sure a format that requires escaping of characters works
176        compiled_re = self.time_re.compile("\w+ %m")
177        found = compiled_re.match("\w+ 10")
178        self.failUnless(found, "Escaping failed of format '\w+ 10'")
179
180    def test_locale_data_w_regex_metacharacters(self):
181        # Check that if locale data contains regex metacharacters they are
182        # escaped properly.
183        # Discovered by bug #1039270 .
184        locale_time = _strptime.LocaleTime()
185        locale_time.timezone = (frozenset(("utc", "gmt",
186                                            "Tokyo (standard time)")),
187                                frozenset("Tokyo (daylight time)"))
188        time_re = _strptime.TimeRE(locale_time)
189        self.failUnless(time_re.compile("%Z").match("Tokyo (standard time)"),
190                        "locale data that contains regex metacharacters is not"
191                        " properly escaped")
192
193    def test_whitespace_substitution(self):
194        # When pattern contains whitespace, make sure it is taken into account
195        # so as to not allow to subpatterns to end up next to each other and
196        # "steal" characters from each other.
197        pattern = self.time_re.pattern('%j %H')
198        self.failUnless(not re.match(pattern, "180"))
199        self.failUnless(re.match(pattern, "18 0"))
200
201
202class StrptimeTests(unittest.TestCase):
203    """Tests for _strptime.strptime."""
204
205    def setUp(self):
206        """Create testing time tuple."""
207        self.time_tuple = time.gmtime()
208
209    def test_ValueError(self):
210        # Make sure ValueError is raised when match fails or format is bad
211        self.assertRaises(ValueError, _strptime._strptime_time, data_string="%d",
212                          format="%A")
213        for bad_format in ("%", "% ", "%e"):
214            try:
215                _strptime._strptime_time("2005", bad_format)
216            except ValueError:
217                continue
218            except Exception, err:
219                self.fail("'%s' raised %s, not ValueError" %
220                            (bad_format, err.__class__.__name__))
221            else:
222                self.fail("'%s' did not raise ValueError" % bad_format)
223
224    def test_unconverteddata(self):
225        # Check ValueError is raised when there is unconverted data
226        self.assertRaises(ValueError, _strptime._strptime_time, "10 12", "%m")
227
228    def helper(self, directive, position):
229        """Helper fxn in testing."""
230        strf_output = time.strftime("%" + directive, self.time_tuple)
231        strp_output = _strptime._strptime_time(strf_output, "%" + directive)
232        self.failUnless(strp_output[position] == self.time_tuple[position],
233                        "testing of '%s' directive failed; '%s' -> %s != %s" %
234                         (directive, strf_output, strp_output[position],
235                          self.time_tuple[position]))
236
237    def test_year(self):
238        # Test that the year is handled properly
239        for directive in ('y', 'Y'):
240            self.helper(directive, 0)
241        # Must also make sure %y values are correct for bounds set by Open Group
242        for century, bounds in ((1900, ('69', '99')), (2000, ('00', '68'))):
243            for bound in bounds:
244                strp_output = _strptime._strptime_time(bound, '%y')
245                expected_result = century + int(bound)
246                self.failUnless(strp_output[0] == expected_result,
247                                "'y' test failed; passed in '%s' "
248                                "and returned '%s'" % (bound, strp_output[0]))
249
250    def test_month(self):
251        # Test for month directives
252        for directive in ('B', 'b', 'm'):
253            self.helper(directive, 1)
254
255    def test_day(self):
256        # Test for day directives
257        self.helper('d', 2)
258
259    def test_hour(self):
260        # Test hour directives
261        self.helper('H', 3)
262        strf_output = time.strftime("%I %p", self.time_tuple)
263        strp_output = _strptime._strptime_time(strf_output, "%I %p")
264        self.failUnless(strp_output[3] == self.time_tuple[3],
265                        "testing of '%%I %%p' directive failed; '%s' -> %s != %s" %
266                         (strf_output, strp_output[3], self.time_tuple[3]))
267
268    def test_minute(self):
269        # Test minute directives
270        self.helper('M', 4)
271
272    def test_second(self):
273        # Test second directives
274        self.helper('S', 5)
275
276    def test_fraction(self):
277        import datetime
278        now = datetime.datetime.now()
279        tup, frac = _strptime._strptime(str(now), format="%Y-%m-%d %H:%M:%S.%f")
280        self.assertEqual(frac, now.microsecond)
281
282    def test_weekday(self):
283        # Test weekday directives
284        for directive in ('A', 'a', 'w'):
285            self.helper(directive,6)
286
287    def test_julian(self):
288        # Test julian directives
289        self.helper('j', 7)
290
291    def test_timezone(self):
292        # Test timezone directives.
293        # When gmtime() is used with %Z, entire result of strftime() is empty.
294        # Check for equal timezone names deals with bad locale info when this
295        # occurs; first found in FreeBSD 4.4.
296        strp_output = _strptime._strptime_time("UTC", "%Z")
297        self.failUnlessEqual(strp_output.tm_isdst, 0)
298        strp_output = _strptime._strptime_time("GMT", "%Z")
299        self.failUnlessEqual(strp_output.tm_isdst, 0)
300        if sys.platform == "mac":
301            # Timezones don't really work on MacOS9
302            return
303        time_tuple = time.localtime()
304        strf_output = time.strftime("%Z")  #UTC does not have a timezone
305        strp_output = _strptime._strptime_time(strf_output, "%Z")
306        locale_time = _strptime.LocaleTime()
307        if time.tzname[0] != time.tzname[1] or not time.daylight:
308            self.failUnless(strp_output[8] == time_tuple[8],
309                            "timezone check failed; '%s' -> %s != %s" %
310                             (strf_output, strp_output[8], time_tuple[8]))
311        else:
312            self.failUnless(strp_output[8] == -1,
313                            "LocaleTime().timezone has duplicate values and "
314                             "time.daylight but timezone value not set to -1")
315
316    def test_bad_timezone(self):
317        # Explicitly test possibility of bad timezone;
318        # when time.tzname[0] == time.tzname[1] and time.daylight
319        if sys.platform == "mac":
320            return #MacOS9 has severely broken timezone support.
321        tz_name = time.tzname[0]
322        if tz_name.upper() in ("UTC", "GMT"):
323            return
324        try:
325            original_tzname = time.tzname
326            original_daylight = time.daylight
327            time.tzname = (tz_name, tz_name)
328            time.daylight = 1
329            tz_value = _strptime._strptime_time(tz_name, "%Z")[8]
330            self.failUnlessEqual(tz_value, -1,
331                    "%s lead to a timezone value of %s instead of -1 when "
332                    "time.daylight set to %s and passing in %s" %
333                    (time.tzname, tz_value, time.daylight, tz_name))
334        finally:
335            time.tzname = original_tzname
336            time.daylight = original_daylight
337
338    def test_date_time(self):
339        # Test %c directive
340        for position in range(6):
341            self.helper('c', position)
342
343    def test_date(self):
344        # Test %x directive
345        for position in range(0,3):
346            self.helper('x', position)
347
348    def test_time(self):
349        # Test %X directive
350        for position in range(3,6):
351            self.helper('X', position)
352
353    def test_percent(self):
354        # Make sure % signs are handled properly
355        strf_output = time.strftime("%m %% %Y", self.time_tuple)
356        strp_output = _strptime._strptime_time(strf_output, "%m %% %Y")
357        self.failUnless(strp_output[0] == self.time_tuple[0] and
358                         strp_output[1] == self.time_tuple[1],
359                        "handling of percent sign failed")
360
361    def test_caseinsensitive(self):
362        # Should handle names case-insensitively.
363        strf_output = time.strftime("%B", self.time_tuple)
364        self.failUnless(_strptime._strptime_time(strf_output.upper(), "%B"),
365                        "strptime does not handle ALL-CAPS names properly")
366        self.failUnless(_strptime._strptime_time(strf_output.lower(), "%B"),
367                        "strptime does not handle lowercase names properly")
368        self.failUnless(_strptime._strptime_time(strf_output.capitalize(), "%B"),
369                        "strptime does not handle capword names properly")
370
371    def test_defaults(self):
372        # Default return value should be (1900, 1, 1, 0, 0, 0, 0, 1, 0)
373        defaults = (1900, 1, 1, 0, 0, 0, 0, 1, -1)
374        strp_output = _strptime._strptime_time('1', '%m')
375        self.failUnless(strp_output == defaults,
376                        "Default values for strptime() are incorrect;"
377                        " %s != %s" % (strp_output, defaults))
378
379    def test_escaping(self):
380        # Make sure all characters that have regex significance are escaped.
381        # Parentheses are in a purposeful order; will cause an error of
382        # unbalanced parentheses when the regex is compiled if they are not
383        # escaped.
384        # Test instigated by bug #796149 .
385        need_escaping = ".^$*+?{}\[]|)("
386        self.failUnless(_strptime._strptime_time(need_escaping, need_escaping))
387
388class Strptime12AMPMTests(unittest.TestCase):
389    """Test a _strptime regression in '%I %p' at 12 noon (12 PM)"""
390
391    def test_twelve_noon_midnight(self):
392        eq = self.assertEqual
393        eq(time.strptime('12 PM', '%I %p')[3], 12)
394        eq(time.strptime('12 AM', '%I %p')[3], 0)
395        eq(_strptime._strptime_time('12 PM', '%I %p')[3], 12)
396        eq(_strptime._strptime_time('12 AM', '%I %p')[3], 0)
397
398
399class JulianTests(unittest.TestCase):
400    """Test a _strptime regression that all julian (1-366) are accepted"""
401
402    def test_all_julian_days(self):
403        eq = self.assertEqual
404        for i in range(1, 367):
405            # use 2004, since it is a leap year, we have 366 days
406            eq(_strptime._strptime_time('%d 2004' % i, '%j %Y')[7], i)
407
408class CalculationTests(unittest.TestCase):
409    """Test that strptime() fills in missing info correctly"""
410
411    def setUp(self):
412        self.time_tuple = time.gmtime()
413
414    def test_julian_calculation(self):
415        # Make sure that when Julian is missing that it is calculated
416        format_string = "%Y %m %d %H %M %S %w %Z"
417        result = _strptime._strptime_time(time.strftime(format_string, self.time_tuple),
418                                    format_string)
419        self.failUnless(result.tm_yday == self.time_tuple.tm_yday,
420                        "Calculation of tm_yday failed; %s != %s" %
421                         (result.tm_yday, self.time_tuple.tm_yday))
422
423    def test_gregorian_calculation(self):
424        # Test that Gregorian date can be calculated from Julian day
425        format_string = "%Y %H %M %S %w %j %Z"
426        result = _strptime._strptime_time(time.strftime(format_string, self.time_tuple),
427                                    format_string)
428        self.failUnless(result.tm_year == self.time_tuple.tm_year and
429                         result.tm_mon == self.time_tuple.tm_mon and
430                         result.tm_mday == self.time_tuple.tm_mday,
431                        "Calculation of Gregorian date failed;"
432                         "%s-%s-%s != %s-%s-%s" %
433                         (result.tm_year, result.tm_mon, result.tm_mday,
434                          self.time_tuple.tm_year, self.time_tuple.tm_mon,
435                          self.time_tuple.tm_mday))
436
437    def test_day_of_week_calculation(self):
438        # Test that the day of the week is calculated as needed
439        format_string = "%Y %m %d %H %S %j %Z"
440        result = _strptime._strptime_time(time.strftime(format_string, self.time_tuple),
441                                    format_string)
442        self.failUnless(result.tm_wday == self.time_tuple.tm_wday,
443                        "Calculation of day of the week failed;"
444                         "%s != %s" % (result.tm_wday, self.time_tuple.tm_wday))
445
446    def test_week_of_year_and_day_of_week_calculation(self):
447        # Should be able to infer date if given year, week of year (%U or %W)
448        # and day of the week
449        def test_helper(ymd_tuple, test_reason):
450            for directive in ('W', 'U'):
451                format_string = "%%Y %%%s %%w" % directive
452                dt_date = datetime_date(*ymd_tuple)
453                strp_input = dt_date.strftime(format_string)
454                strp_output = _strptime._strptime_time(strp_input, format_string)
455                self.failUnless(strp_output[:3] == ymd_tuple,
456                        "%s(%s) test failed w/ '%s': %s != %s (%s != %s)" %
457                            (test_reason, directive, strp_input,
458                                strp_output[:3], ymd_tuple,
459                                strp_output[7], dt_date.timetuple()[7]))
460        test_helper((1901, 1, 3), "week 0")
461        test_helper((1901, 1, 8), "common case")
462        test_helper((1901, 1, 13), "day on Sunday")
463        test_helper((1901, 1, 14), "day on Monday")
464        test_helper((1905, 1, 1), "Jan 1 on Sunday")
465        test_helper((1906, 1, 1), "Jan 1 on Monday")
466        test_helper((1906, 1, 7), "first Sunday in a year starting on Monday")
467        test_helper((1905, 12, 31), "Dec 31 on Sunday")
468        test_helper((1906, 12, 31), "Dec 31 on Monday")
469        test_helper((2008, 12, 29), "Monday in the last week of the year")
470        test_helper((2008, 12, 22), "Monday in the second-to-last week of the "
471                                    "year")
472        test_helper((1978, 10, 23), "randomly chosen date")
473        test_helper((2004, 12, 18), "randomly chosen date")
474        test_helper((1978, 10, 23), "year starting and ending on Monday while "
475                                        "date not on Sunday or Monday")
476        test_helper((1917, 12, 17), "year starting and ending on Monday with "
477                                        "a Monday not at the beginning or end "
478                                        "of the year")
479        test_helper((1917, 12, 31), "Dec 31 on Monday with year starting and "
480                                        "ending on Monday")
481        test_helper((2007, 01, 07), "First Sunday of 2007")
482        test_helper((2007, 01, 14), "Second Sunday of 2007")
483        test_helper((2006, 12, 31), "Last Sunday of 2006")
484        test_helper((2006, 12, 24), "Second to last Sunday of 2006")
485
486
487class CacheTests(unittest.TestCase):
488    """Test that caching works properly."""
489
490    def test_time_re_recreation(self):
491        # Make sure cache is recreated when current locale does not match what
492        # cached object was created with.
493        _strptime._strptime_time("10", "%d")
494        _strptime._strptime_time("2005", "%Y")
495        _strptime._TimeRE_cache.locale_time.lang = "Ni"
496        original_time_re = id(_strptime._TimeRE_cache)
497        _strptime._strptime_time("10", "%d")
498        self.failIfEqual(original_time_re, id(_strptime._TimeRE_cache))
499        self.failUnlessEqual(len(_strptime._regex_cache), 1)
500
501    def test_regex_cleanup(self):
502        # Make sure cached regexes are discarded when cache becomes "full".
503        try:
504            del _strptime._regex_cache['%d']
505        except KeyError:
506            pass
507        bogus_key = 0
508        while len(_strptime._regex_cache) <= _strptime._CACHE_MAX_SIZE:
509            _strptime._regex_cache[bogus_key] = None
510            bogus_key += 1
511        _strptime._strptime_time("10", "%d")
512        self.failUnlessEqual(len(_strptime._regex_cache), 1)
513
514    def test_new_localetime(self):
515        # A new LocaleTime instance should be created when a new TimeRE object
516        # is created.
517        locale_time_id = id(_strptime._TimeRE_cache.locale_time)
518        _strptime._TimeRE_cache.locale_time.lang = "Ni"
519        _strptime._strptime_time("10", "%d")
520        self.failIfEqual(locale_time_id,
521                         id(_strptime._TimeRE_cache.locale_time))
522
523    def test_TimeRE_recreation(self):
524        # The TimeRE instance should be recreated upon changing the locale.
525        locale_info = locale.getlocale(locale.LC_TIME)
526        try:
527            locale.setlocale(locale.LC_TIME, ('en_US', 'UTF8'))
528        except locale.Error:
529            return
530        try:
531            _strptime._strptime_time('10', '%d')
532            # Get id of current cache object.
533            first_time_re_id = id(_strptime._TimeRE_cache)
534            try:
535                # Change the locale and force a recreation of the cache.
536                locale.setlocale(locale.LC_TIME, ('de_DE', 'UTF8'))
537                _strptime._strptime_time('10', '%d')
538                # Get the new cache object's id.
539                second_time_re_id = id(_strptime._TimeRE_cache)
540                # They should not be equal.
541                self.failIfEqual(first_time_re_id, second_time_re_id)
542            # Possible test locale is not supported while initial locale is.
543            # If this is the case just suppress the exception and fall-through
544            # to the reseting to the original locale.
545            except locale.Error:
546                pass
547        # Make sure we don't trample on the locale setting once we leave the
548        # test.
549        finally:
550            locale.setlocale(locale.LC_TIME, locale_info)
551
552
553def test_main():
554    test_support.run_unittest(
555        getlang_Tests,
556        LocaleTime_Tests,
557        TimeRETests,
558        StrptimeTests,
559        Strptime12AMPMTests,
560        JulianTests,
561        CalculationTests,
562        CacheTests
563    )
564
565
566if __name__ == '__main__':
567    test_main()