diff --git a/Lib/test/support/_hypothesis_stubs/__init__.py b/Lib/test/support/_hypothesis_stubs/__init__.py new file mode 100644 index 0000000000..6ba5bb814b --- /dev/null +++ b/Lib/test/support/_hypothesis_stubs/__init__.py @@ -0,0 +1,111 @@ +from enum import Enum +import functools +import unittest + +__all__ = [ + "given", + "example", + "assume", + "reject", + "register_random", + "strategies", + "HealthCheck", + "settings", + "Verbosity", +] + +from . import strategies + + +def given(*_args, **_kwargs): + def decorator(f): + if examples := getattr(f, "_examples", []): + + @functools.wraps(f) + def test_function(self): + for example_args, example_kwargs in examples: + with self.subTest(*example_args, **example_kwargs): + f(self, *example_args, **example_kwargs) + + else: + # If we have found no examples, we must skip the test. If @example + # is applied after @given, it will re-wrap the test to remove the + # skip decorator. + test_function = unittest.skip( + "Hypothesis required for property test with no " + + "specified examples" + )(f) + + test_function._given = True + return test_function + + return decorator + + +def example(*args, **kwargs): + if bool(args) == bool(kwargs): + raise ValueError("Must specify exactly one of *args or **kwargs") + + def decorator(f): + base_func = getattr(f, "__wrapped__", f) + if not hasattr(base_func, "_examples"): + base_func._examples = [] + + base_func._examples.append((args, kwargs)) + + if getattr(f, "_given", False): + # If the given decorator is below all the example decorators, + # it would be erroneously skipped, so we need to re-wrap the new + # base function. + f = given()(base_func) + + return f + + return decorator + + +def assume(condition): + if not condition: + raise unittest.SkipTest("Unsatisfied assumption") + return True + + +def reject(): + assume(False) + + +def register_random(*args, **kwargs): + pass # pragma: no cover + + +def settings(*args, **kwargs): + return lambda f: f # pragma: nocover + + +class HealthCheck(Enum): + data_too_large = 1 + filter_too_much = 2 + too_slow = 3 + return_value = 5 + large_base_example = 7 + not_a_test_method = 8 + + @classmethod + def all(cls): + return list(cls) + + +class Verbosity(Enum): + quiet = 0 + normal = 1 + verbose = 2 + debug = 3 + + +class Phase(Enum): + explicit = 0 + reuse = 1 + generate = 2 + target = 3 + shrink = 4 + explain = 5 diff --git a/Lib/test/support/_hypothesis_stubs/_helpers.py b/Lib/test/support/_hypothesis_stubs/_helpers.py new file mode 100644 index 0000000000..3f6244e4db --- /dev/null +++ b/Lib/test/support/_hypothesis_stubs/_helpers.py @@ -0,0 +1,43 @@ +# Stub out only the subset of the interface that we actually use in our tests. +class StubClass: + def __init__(self, *args, **kwargs): + self.__stub_args = args + self.__stub_kwargs = kwargs + self.__repr = None + + def _with_repr(self, new_repr): + new_obj = self.__class__(*self.__stub_args, **self.__stub_kwargs) + new_obj.__repr = new_repr + return new_obj + + def __repr__(self): + if self.__repr is not None: + return self.__repr + + argstr = ", ".join(self.__stub_args) + kwargstr = ", ".join(f"{kw}={val}" for kw, val in self.__stub_kwargs.items()) + + in_parens = argstr + if kwargstr: + in_parens += ", " + kwargstr + + return f"{self.__class__.__qualname__}({in_parens})" + + +def stub_factory(klass, name, *, with_repr=None, _seen={}): + if (klass, name) not in _seen: + + class Stub(klass): + def __init__(self, *args, **kwargs): + super().__init__() + self.__stub_args = args + self.__stub_kwargs = kwargs + + Stub.__name__ = name + Stub.__qualname__ = name + if with_repr is not None: + Stub._repr = None + + _seen.setdefault((klass, name, with_repr), Stub) + + return _seen[(klass, name, with_repr)] diff --git a/Lib/test/support/_hypothesis_stubs/strategies.py b/Lib/test/support/_hypothesis_stubs/strategies.py new file mode 100644 index 0000000000..d2b885d41e --- /dev/null +++ b/Lib/test/support/_hypothesis_stubs/strategies.py @@ -0,0 +1,91 @@ +import functools + +from ._helpers import StubClass, stub_factory + + +class StubStrategy(StubClass): + def __make_trailing_repr(self, transformation_name, func): + func_name = func.__name__ or repr(func) + return f"{self!r}.{transformation_name}({func_name})" + + def map(self, pack): + return self._with_repr(self.__make_trailing_repr("map", pack)) + + def flatmap(self, expand): + return self._with_repr(self.__make_trailing_repr("flatmap", expand)) + + def filter(self, condition): + return self._with_repr(self.__make_trailing_repr("filter", condition)) + + def __or__(self, other): + new_repr = f"one_of({self!r}, {other!r})" + return self._with_repr(new_repr) + + +_STRATEGIES = { + "binary", + "booleans", + "builds", + "characters", + "complex_numbers", + "composite", + "data", + "dates", + "datetimes", + "decimals", + "deferred", + "dictionaries", + "emails", + "fixed_dictionaries", + "floats", + "fractions", + "from_regex", + "from_type", + "frozensets", + "functions", + "integers", + "iterables", + "just", + "lists", + "none", + "nothing", + "one_of", + "permutations", + "random_module", + "randoms", + "recursive", + "register_type_strategy", + "runner", + "sampled_from", + "sets", + "shared", + "slices", + "timedeltas", + "times", + "text", + "tuples", + "uuids", +} + +__all__ = sorted(_STRATEGIES) + + +def composite(f): + strategy = stub_factory(StubStrategy, f.__name__) + + @functools.wraps(f) + def inner(*args, **kwargs): + return strategy(*args, **kwargs) + + return inner + + +def __getattr__(name): + if name not in _STRATEGIES: + raise AttributeError(f"Unknown attribute {name}") + + return stub_factory(StubStrategy, f"hypothesis.strategies.{name}") + + +def __dir__(): + return __all__ diff --git a/Lib/test/support/hypothesis_helper.py b/Lib/test/support/hypothesis_helper.py new file mode 100644 index 0000000000..40f58a2f59 --- /dev/null +++ b/Lib/test/support/hypothesis_helper.py @@ -0,0 +1,45 @@ +import os + +try: + import hypothesis +except ImportError: + from . import _hypothesis_stubs as hypothesis +else: + # Regrtest changes to use a tempdir as the working directory, so we have + # to tell Hypothesis to use the original in order to persist the database. + from .os_helper import SAVEDCWD + from hypothesis.configuration import set_hypothesis_home_dir + + set_hypothesis_home_dir(os.path.join(SAVEDCWD, ".hypothesis")) + + # When using the real Hypothesis, we'll configure it to ignore occasional + # slow tests (avoiding flakiness from random VM slowness in CI). + hypothesis.settings.register_profile( + "slow-is-ok", + deadline=None, + suppress_health_check=[ + hypothesis.HealthCheck.too_slow, + hypothesis.HealthCheck.differing_executors, + ], + ) + hypothesis.settings.load_profile("slow-is-ok") + + # For local development, we'll write to the default on-local-disk database + # of failing examples, and also use a pull-through cache to automatically + # replay any failing examples discovered in CI. For details on how this + # works, see https://hypothesis.readthedocs.io/en/latest/database.html + if "CI" not in os.environ: + from hypothesis.database import ( + GitHubArtifactDatabase, + MultiplexedDatabase, + ReadOnlyDatabase, + ) + + hypothesis.settings.register_profile( + "cpython-local-dev", + database=MultiplexedDatabase( + hypothesis.settings.default.database, + ReadOnlyDatabase(GitHubArtifactDatabase("python", "cpython")), + ), + ) + hypothesis.settings.load_profile("cpython-local-dev") diff --git a/Lib/test/test_zoneinfo/__init__.py b/Lib/test/test_zoneinfo/__init__.py new file mode 100644 index 0000000000..4b16ecc311 --- /dev/null +++ b/Lib/test/test_zoneinfo/__init__.py @@ -0,0 +1,5 @@ +import os +from test.support import load_package_tests + +def load_tests(*args): + return load_package_tests(os.path.dirname(__file__), *args) diff --git a/Lib/test/test_zoneinfo/__main__.py b/Lib/test/test_zoneinfo/__main__.py new file mode 100644 index 0000000000..5cc4e055d5 --- /dev/null +++ b/Lib/test/test_zoneinfo/__main__.py @@ -0,0 +1,3 @@ +import unittest + +unittest.main('test.test_zoneinfo') diff --git a/Lib/test/test_zoneinfo/_support.py b/Lib/test/test_zoneinfo/_support.py new file mode 100644 index 0000000000..5a76c163fb --- /dev/null +++ b/Lib/test/test_zoneinfo/_support.py @@ -0,0 +1,100 @@ +import contextlib +import functools +import sys +import threading +import unittest +from test.support.import_helper import import_fresh_module + +OS_ENV_LOCK = threading.Lock() +TZPATH_LOCK = threading.Lock() +TZPATH_TEST_LOCK = threading.Lock() + + +def call_once(f): + """Decorator that ensures a function is only ever called once.""" + lock = threading.Lock() + cached = functools.lru_cache(None)(f) + + @functools.wraps(f) + def inner(): + with lock: + return cached() + + return inner + + +@call_once +def get_modules(): + """Retrieve two copies of zoneinfo: pure Python and C accelerated. + + Because this function manipulates the import system in a way that might + be fragile or do unexpected things if it is run many times, it uses a + `call_once` decorator to ensure that this is only ever called exactly + one time — in other words, when using this function you will only ever + get one copy of each module rather than a fresh import each time. + """ + import zoneinfo as c_module + + py_module = import_fresh_module("zoneinfo", blocked=["_zoneinfo"]) + + return py_module, c_module + + +@contextlib.contextmanager +def set_zoneinfo_module(module): + """Make sure sys.modules["zoneinfo"] refers to `module`. + + This is necessary because `pickle` will refuse to serialize + an type calling itself `zoneinfo.ZoneInfo` unless `zoneinfo.ZoneInfo` + refers to the same object. + """ + + NOT_PRESENT = object() + old_zoneinfo = sys.modules.get("zoneinfo", NOT_PRESENT) + sys.modules["zoneinfo"] = module + yield + if old_zoneinfo is not NOT_PRESENT: + sys.modules["zoneinfo"] = old_zoneinfo + else: # pragma: nocover + sys.modules.pop("zoneinfo") + + +class ZoneInfoTestBase(unittest.TestCase): + @classmethod + def setUpClass(cls): + cls.klass = cls.module.ZoneInfo + super().setUpClass() + + @contextlib.contextmanager + def tzpath_context(self, tzpath, block_tzdata=True, lock=TZPATH_LOCK): + def pop_tzdata_modules(): + tzdata_modules = {} + for modname in list(sys.modules): + if modname.split(".", 1)[0] != "tzdata": # pragma: nocover + continue + + tzdata_modules[modname] = sys.modules.pop(modname) + + return tzdata_modules + + with lock: + if block_tzdata: + # In order to fully exclude tzdata from the path, we need to + # clear the sys.modules cache of all its contents — setting the + # root package to None is not enough to block direct access of + # already-imported submodules (though it will prevent new + # imports of submodules). + tzdata_modules = pop_tzdata_modules() + sys.modules["tzdata"] = None + + old_path = self.module.TZPATH + try: + self.module.reset_tzpath(tzpath) + yield + finally: + if block_tzdata: + sys.modules.pop("tzdata") + for modname, module in tzdata_modules.items(): + sys.modules[modname] = module + + self.module.reset_tzpath(old_path) diff --git a/Lib/test/test_zoneinfo/data/update_test_data.py b/Lib/test/test_zoneinfo/data/update_test_data.py new file mode 100644 index 0000000000..e69de29bb2 diff --git a/Lib/test/test_zoneinfo/data/zoneinfo_data.json b/Lib/test/test_zoneinfo/data/zoneinfo_data.json new file mode 100644 index 0000000000..48e70ac6d3 --- /dev/null +++ b/Lib/test/test_zoneinfo/data/zoneinfo_data.json @@ -0,0 +1,190 @@ +{ + "data": { + "Africa/Abidjan": [ + "{Wp48S^xk9=GL@E0stWa761SMbT8$j-~f{VGF<>F7KxBg5R*{Ksocg8-YYVul=v7vZzaHN", + "uC=da5UI2rH18c!OnjV{y4u(+A!!VBKmY&$ORw>7UO^(500B;v0RR91bXh%WvBYQl0ssI2", + "00dcD" + ], + "Africa/Casablanca": [ + "{Wp48S^xk9=GL@E0stWa761SMbT8$j;0b&Kz+C_;7KxBg5R*{N&yjMUR~;C-fDaSOU;q-~", + "FqW+4{YBjbcw}`a!dW>b)R2-0a+uwf`P3{_Y@HuCz}S$J$ZJ>R_V<~|Fk>sgX4=%0vUrh-", + "lt@YP^Wrus;j?`Th#xRPzf<<~Hp4DH^gZX>d{+WOp~HNu8!{uWu}&XphAd{j1;rB4|9?R!", + "pqruAFUMt8#*WcrVS{;kLlY(cJRV$w?d2car%Rs>q9BgTU4", + "Ht-tQKZ7Z`9QqOb?R#b%z?rk>!CkH7jy3wja4NG2q)H}fNRKg8v{);Em;K3Cncf4C6&Oaj", + "V+DbX%o4+)CV3+e!Lm6dutu(0BQpH1T?W(~cQtKV*^_Pdx!LirjpTs?Bmt@vktjLq4;)O!", + "rrly=c*rwTwMJFd0I57`hgkc?=nyI4RZf9W$6DCWugmf&)wk^tWH17owj=#PGH7Xv-?9$j", + "njwDlkOE+BFNR9YXEmBpO;rqEw=e2IR-8^(W;8ma?M3JVd($2T>IW+0tk|Gm8>ftukRQ9J", + "8k3brzqMnVyjsLI-CKneFa)Lxvp_aq40f}0J3VVoWL5rox", + "`Kptivcp}o5xA^@>qNI%?zo=Yj4AMV?kbAA)j(1%)+Pp)bSn+7Yk`M{oE}L-Z!G6OMr5G+h", + "p)$3Lg{ono{4cN>Vr&>L4kXH;_VnBL5U!LgzqE%P7QQ*tue}O`3(TZ0`aKn&~8trOQ-rBXCp)f@P6RMO4l0+;b|5-pk9_ryNh}Zc*v%mvz_#", + "yd6fjB0g9{MmMnu8bG%#C~ugXK^S^k@?ab#", + "O|aE>dDTt4s4n69(~@t~!wniV%g7khFx~I*4>Y|V$4j5%KPF*-FyKIi@!Ho&", + "x8QQsksYt8)D+W)Ni!=G`ogSu^vLL-l#7A7=iIAKL2SuZk9F}NfNk86VI)9WZE?%2wC-ya", + "F~z#Qsq)LH0|_D8^5fU8X%GeQ4TB>R-dlziA&tZe&1ada208!$nk`7bOFO2S00G`Z@1A~t&lyL{p{eM{5)QGf7Mo5FW9==mlyXJt2", + "UwpntR7H0eSq!(aYq#aqUz&RM*tvuMI)AsM?K3-dV3-TT{t)!Iy#JTo=tXkzAM9~j2YbiO", + "ls3(H8Dc>Y|D1aqL51vjLbpYG;GvGTQB4bXuJ%mA;(B4eUpu$$@zv2vVcq-Y)VKbzp^tei", + "uzy}R{LuvDjpuVb`79O+CBmg{Wx!bvx$eu4zRE&", + "PehMb=&G<9$>iZ|bFE)0=4I?KLFGBC0I(0_svgw0%FiMsT%koo*!nEYc6GY@QnU}&4Isg;", + "l=|khi(!VaiSE2=Ny`&&tpi~~;{$uN}%f|7mBhAy;s3YT^sy!$eG~?`9mNJC9@4Bac_p^BZh)Yd_rWW5qh-?tKY(>5VHO", + "L*iT8P@wCavLj^yYbnDR+4ukhS+xPrpl)iqB?u)bj9a2aW==g6G3lCJd>(+Blfr)~^40F4f>cRZ^UF;RibfZ>0m73hR", + "C{$vTfC(STN`g7(B<=Z2556{}0`?p&|Akkst!4Xy4OT;A@c$XTUI3FRRjy*KA7uC56FD)z", + "^X{WV*sr(w!c$W357o!&eLO2wTDNOyw@gf(&R<t;=-Tu1TV{>%8ZVATC9tjD8|(&`$9YHvZ9bVe#>w", + "|8c;Tg|xE&)`*}LwM*E}q}q8^Qja%p`_U)*5DdLI9O@!e=3jFjOCrCq28b_bb;s>%D#iJB", + "CWJi{JH!Js;6nfayos$kq^OEX00HO-lokL0!mqm{vBYQl0ssI200dcD" + ], + "America/Santiago": [ + "{Wp48S^xk9=GL@E0stWa761SMbT8$j;0fRZ<6QtM7KxBg84(fsEAUJ$J{f-TXlPEUec5Ee", + "n+hsD4lC(QYax=JdSpoyje8%VM`GW}{bJ8@y$A8O&*$pw{(f~Os#}2w", + "eX6^Rgi$IT%n^V^85L>$_c7{cB^#ogV=rHBJGiz-RQNFGK?gdPi|q)j`&8)}KJ{qo6dixa", + "9@yYyVg+%lo0nO+Tw0-w2hJ%mafyWL)|", + ")?W6Bi%FWuGPA1Dru$XR4SZANsAthU2EoKHF6oEtKq`rwP", + "(VNegnI_NI%;ma$)wj{k!@KFB30Yo)IOrl>)$)D|+(5h&+%2vuwGuy^@S8FT^s21V5};>VA9Iu;?8bHz#r<;JtfZDI1(FT@edh0#", + "MYW$A1qkMGIwTZqqdYNE3gl#zp&NbL9Mp=voqN|;?gqR&4$)1`znddtEyuKS*^nMMD=0^>", + "7^z6-C4P67UWOXuMBubP>j6i~03aR@jD^-Y`JSYu#Yp0P8dLLJ0QOPE8=BoiuRX59YW7xg", + "WiexjHX%&0?`ZQCdxCdL^qd1v@kOjQKaWo2Y1++~LcA%FTq?5o%}fX1-RIvlB)1#iTNomGnUL=nM!>Ix|AGtON7!F1O?53kqlC2o-`ZGw*+s", + "NM$^9znsIJMwlgscE`|O3|;BRgsQMYm~`uv+nvuv`nigRa}X=BX=A5Sw$)WEklF7&c>_~$", + "zJ(m--bqXgiN^w-U=BJH9C0Qro(x90zo@rK;&TJ$nI@&k$ORgOb2s%gWbc}ok_27)Eoku~Fq|B-Ps+4J_", + "HPJMLJ2^_)cOU$p&3kNAlrV!)%~6r$BJ>OOi~=-<6byle{?zd4J{NG}o8tw|+#ZNLcpNwk", + "TuPE~sbJB8_RZb2DopStO+Wwux~F#S59zm%00I98;S&G=b(j+6vBYQl0ssI200dcD" + ], + "Asia/Tokyo": [ + "{Wp48S^xk9=GL@E0stWa761SMbT8$j-~luMgIxeB7KxBg5R*;y?l4Rl4neXH3cv!OtfK@h", + "KZzauI)S!FSDREPhhBS6Fb$&Vv#7%;?Te|>pF^0HBr&z_Tk<%vMW_QqjevRZOp8XVFgP<8", + "TkT#`9H&0Ua;gT1#rZLV0HqbAKK;_z@nO;6t0L}hOdk<>TdUa07R(LPI6@!GU$ty4=mwqHG-XVe*n(Yvgdlr+FqIU18!osi)48t~eWX8)&L", + "G)Ud^0zz@*AF+2r7E}Nf9Y72K~o-T%}D&z%}#7g2br?oH6ZiYH^%>J3D)TPKV(JY*bwjuw5=DsPB@~CrROZeN", + "x>A*H&CHrWt0`EP`m!F%waepl#|w#&`XgVc?~2M3uw$fGX~tf_Il!q#Aa<*8xlzQ2+7r6Z", + "^;Laa9F(WB_O&Dy2r>~@kSi16W{=6+i5GV=Uq~KX*~&HUN4oz7*O(gXIr}sDVcD`Ikgw#|", + "50ssal8s)Qy;?YGCf;*UKKKN!T4!Kqy_G;7PfQapugqvVBKy12v3TVH^L2", + "0?#5*VP~MOYfe$h`*L!7@tiW|_^X1N%<}`7YahiUYtMu5XwmOf3?dr+@zXHwW`z}ZDqZlT", + "<2Cs(<1%M!i6o&VK89BY0J7HPIo;O62s=|IbV^@y$N&#=>i^F00FcHoDl#3", + "Mdv&xvBYQl0ssI200dcD" + ], + "Europe/Dublin": [ + "{Wp48S^xk9=GL@E0stWa761SMbT8$j;0>b$_+0=h7KxBg5R*;&J77#T_U2R5sleVWFDmK~", + "Kzj5oh@`QKHvW^6V{jU-w>qg1tSt0c^vh;?qAqA0%t?;#S~6U8Qi", + "v&f1s9IH#g$m1k1a#3+lylw4mwT4QnEUUQdwg+xnEcBlgu31bAVabn41OMZVLGz6NDwG%X", + "uQar!b>GI{qSahE`AG}$kRWbuI~JCt;38)Xwbb~Qggs55t+MAHIxgDxzTJ;2xXx99+qCy4", + "45kC#v_l8fx|G&jlVvaciR<-wwf22l%4(t@S6tnX39#_K(4S0fu$FUs$isud9IKzCXB78NkARYq@9Dc0TGkhz);NtM_SSzEffN", + "l{2^*CKGdp52h!52A)6q9fUSltXF{T*Ehc9Q7u8!W7pE(Fv$D$cKUAt6wY=DA1mGgxC*VX", + "q_If3G#FY6-Voj`fIKk`0}Cc72_SD{v>468LV{pyBI33^p0E?}RwDA6Pkq--C~0jF&Z@Pv", + "!dx_1SN_)jwz@P$(oK%P!Tk9?fRjK88yxhxlcFtTjjZ$DYssSsa#ufYrR+}}nKS+r384o~", + "!Uw$nwTbF~qgRsgr0N#d@KIinx%hQB(SJyjJtDtIy(%mDm}ZBGN}dV6K~om|=U", + "VGkbciQ=^$_14|gT21!YQ)@y*Rd0i_lS6gtPBE9+ah%WIJPwzUTjIr+J1XckkmA!6WE16%", + "CVAl{Dn&-)=G$Bjh?bh0$Xt1UDcgXJjXzzojuw0>paV~?Sa`VN3FysqFxTzfKVAu*ucq#+m=|KSSMvp_#@-lwd+q*ue", + "FQ^5<|<0R-u4qYMbRqzSn&", + "Q7jSuvc%b+EZc%>nI(+&0Tl1Y>a6v4`uNFD-7$QrhHgS7Wnv~rDgfH;rQw3+m`LJxoM4v#", + "gK@?|B{RHJ*VxZgk#!p<_&-sjxOda0YaiJ1UnG41VPv(Et%ElzKRMcO$AfgU+Xnwg5p2_+", + "NrnZ1WfEj^fmHd^sx@%JWKkh#zaK0ox%rdP)zUmGZZnqmZ_9L=%6R8ibJH0bOT$AGhDo6{", + "fJ?;_U;D|^>5by2ul@i4Zf()InfFN}00EQ=q#FPL>RM>svBYQl0ssI200dcD" + ], + "Europe/Lisbon": [ + "{Wp48S^xk9=GL@E0stWa761SMbT8$j;0=rf*IfWA7KxBg5R*;*X|PN+G3LqthM?xgkNUN_", + ")gCt1Sc%YT6^TTomk4yVHXeyvQj8}l<;q&s7K}#Vnc8lII1?)AHh$*>OKUU4S;*h>v*ep0", + "xTi1cK2{aY*|2D*-~K<;-{_W+r@NvZ7-|NZv($ek_C%VfP0xjWeZP#CPXD`IKkakjh(kUd", + "&H)m;^Q(jGjIyiyrcUMtOP)u3A>sw6ux;Bmp3x$4QvQKMx5TrCx_!$srWQuXNs&`9=^IY1", + "yc&C31!sQh7P=Mk*#6x8Z@5^%ehR8UW$OWw0KMw}P1ycI^", + "4eh12oBUOV?S>n*d!+EM@>x#9PZD12iD=zaC;7`8dTfkU_6d}OZvSFSbGgXeKw}XyX@D=(", + ")D0!^DBGr8pXWBT$S-yhLP>Z3ys^VW3}RQ6{NGGVJG6vf*MH93vvNW6yLjie1;{4tVhg-KnSf|G`!", + "Z;j$7gJ1ows~RD=@n7I6aFd8rOR_7Y?E-$clI%1o5gA@O!KPa^(8^iFFeFykI-+z>E$mvp", + "E_h`vbHPjqkLs`Dn-0FV`R@z|h!S(Lb;M&|Exr!biY`%bfp$6`hK;GDhdP|^Q", + "*Ty*}1d41K>H2B{jrjE9aFK>yAQJBX9CD%-384S;0fw`PlprHGS`^b$oS-`I4VH7ji8ou-", + "g|060jfb1XcxiInT0oOoeR7#%e5Ug5#KW)nVSRvLHNe$SQHM@2)`S9L7>RL@Qx%fmm7?3u7P5TywFQ}C@S(pq}|", + "eLPT{C^{<0Q?uU&kSVd%!~8q3;Z0s3OqzF`$HRkePL5Ywgiwn{R(zi+jmOBFrVpW;)@UsU#%$8BcV#h@}m$#!Fglo&bwb78aYqOG_W7h{eb(+39&-mk4EIXq_", + "_`30=8sfA3=!3TO_TyS5X22~?6nKngZ|bq=grdq=9X)3xAkA42L!~rmS)n3w-~;lgz%Fhn", + "(?rXdp2ho~9?wmVs2JwVt~?@FVD%`tN69{(i3oQa;O0$E$lF&~Y#_H6bu6(BiwblJ>;-Fs", + "gA$Y$*?=X)n1pFkKn}F~`>=4)+LLQk?L*P!bhAm0;`N~z3QbUIyVrm%kOZ(n1JJsm0pyb8", + "!GV{d*C!9KXv;4vD4Q>-k#+x(!V5L@w5M>v2V5a`B>t(|B", + "|Fqr4^-{S*%Ep~ojUtx_CRbSQ(uFwu2=KH)Q@EBs@ZqRXn4mU;B!68;;IQs3Ub=n&UU%*m", + "k&zwD36&JSwsN(%k&x?H+tN^6)23c`I0=5^N_R0~1>tsFZ`^`3z~rXSXT&qcwa#n!%+Z#P", + "PG}(D^_CCILXnF|GKwabBh*xFS?4rwGo2vtJUwzrbv_$5PO+`?$l{H-jGB@X%S!OAhw;D4", + "XFycN3!XqQ&EorJOD3>~^U%Luw!jF<;6_q-f-S|6{cQDfZ2(4Xf1MMLr1=SA=MwVf2%Pp%VP;jn)|5Tf!-DbUGn%I-rkYaH7?$$O!t)wwClAisr3eUoeB^~T=U*_P~Y2*KdnO87>B!19sV=xZ5", + "yApq26RxgqA|*tmsvtL#OhcF(C<0EGWHP)BFl?h)_*7!{LoJiv%RsOs!q->n+DcV%9~B@RbC_1G_1g6`Yd~8|%-=2l~oGN!~TVv2Bnk>7wW8L@^?vX$f3AiT)(4nrCuTm9%(XC6Nai", + "E(;}7&=YZagjAN$O-cN;1u{dTkElmB0GT$|Wa)QMmKrx<|LCJ9qlUoFsUbD^H^6_8(w<0{", + "ftj&O1~p_%lh5z;zNV&sP+", + "NF2>iK{8KMUf+)<-)VxXbLxD(alL}N$AT-ogNbJSMMYeX+Z{jS)b8TK^PB=FxyBxzfmFto", + "eo0R`a(%NO?#aEH9|?Cv00000NIsFh6BW2800DjO0RR918Pu^`vBYQl0ssI200dcD" + ], + "UTC": [ + "{Wp48S^xk9=GL@E0stWa761SMbT8$j-~e#|9bEt_7KxBg5R*|3h1|xhHLji!C57qW6L*|H", + "pEErm00000ygu;I+>V)?00B92fhY-(AGY&-0RR9100dcD" + ] + }, + "metadata": { + "version": "2020a" + } + } \ No newline at end of file diff --git a/Lib/test/test_zoneinfo/test_zoneinfo.py b/Lib/test/test_zoneinfo/test_zoneinfo.py new file mode 100644 index 0000000000..8414721555 --- /dev/null +++ b/Lib/test/test_zoneinfo/test_zoneinfo.py @@ -0,0 +1,2259 @@ +from __future__ import annotations + +import base64 +import contextlib +import dataclasses +import importlib.metadata +import io +import json +import os +import pathlib +import pickle +import re +import shutil +import struct +import tempfile +import unittest +from datetime import date, datetime, time, timedelta, timezone +from functools import cached_property + +from test.support import MISSING_C_DOCSTRINGS +from test.test_zoneinfo import _support as test_support +from test.test_zoneinfo._support import OS_ENV_LOCK, TZPATH_TEST_LOCK, ZoneInfoTestBase +from test.support.import_helper import import_module, CleanImport + +lzma = import_module('lzma') +py_zoneinfo, c_zoneinfo = test_support.get_modules() + +try: + importlib.metadata.metadata("tzdata") + HAS_TZDATA_PKG = True +except importlib.metadata.PackageNotFoundError: + HAS_TZDATA_PKG = False + +ZONEINFO_DATA = None +ZONEINFO_DATA_V1 = None +TEMP_DIR = None +DATA_DIR = pathlib.Path(__file__).parent / "data" +ZONEINFO_JSON = DATA_DIR / "zoneinfo_data.json" +DRIVE = os.path.splitdrive('x:')[0] + +# Useful constants +ZERO = timedelta(0) +ONE_H = timedelta(hours=1) + + +def setUpModule(): + global TEMP_DIR + global ZONEINFO_DATA + global ZONEINFO_DATA_V1 + + TEMP_DIR = pathlib.Path(tempfile.mkdtemp(prefix="zoneinfo")) + ZONEINFO_DATA = ZoneInfoData(ZONEINFO_JSON, TEMP_DIR / "v2") + ZONEINFO_DATA_V1 = ZoneInfoData(ZONEINFO_JSON, TEMP_DIR / "v1", v1=True) + + +def tearDownModule(): + shutil.rmtree(TEMP_DIR) + + +class TzPathUserMixin: + """ + Adds a setUp() and tearDown() to make TZPATH manipulations thread-safe. + + Any tests that require manipulation of the TZPATH global are necessarily + thread unsafe, so we will acquire a lock and reset the TZPATH variable + to the default state before each test and release the lock after the test + is through. + """ + + @property + def tzpath(self): # pragma: nocover + return None + + @property + def block_tzdata(self): + return True + + def setUp(self): + with contextlib.ExitStack() as stack: + stack.enter_context( + self.tzpath_context( + self.tzpath, + block_tzdata=self.block_tzdata, + lock=TZPATH_TEST_LOCK, + ) + ) + self.addCleanup(stack.pop_all().close) + + super().setUp() + + +class DatetimeSubclassMixin: + """ + Replaces all ZoneTransition transition dates with a datetime subclass. + """ + + class DatetimeSubclass(datetime): + @classmethod + def from_datetime(cls, dt): + return cls( + dt.year, + dt.month, + dt.day, + dt.hour, + dt.minute, + dt.second, + dt.microsecond, + tzinfo=dt.tzinfo, + fold=dt.fold, + ) + + def load_transition_examples(self, key): + transition_examples = super().load_transition_examples(key) + for zt in transition_examples: + dt = zt.transition + new_dt = self.DatetimeSubclass.from_datetime(dt) + new_zt = dataclasses.replace(zt, transition=new_dt) + yield new_zt + + +class ZoneInfoTest(TzPathUserMixin, ZoneInfoTestBase): + module = py_zoneinfo + class_name = "ZoneInfo" + + def setUp(self): + super().setUp() + + # This is necessary because various subclasses pull from different + # data sources (e.g. tzdata, V1 files, etc). + self.klass.clear_cache() + + @property + def zoneinfo_data(self): + return ZONEINFO_DATA + + @property + def tzpath(self): + return [self.zoneinfo_data.tzpath] + + def zone_from_key(self, key): + return self.klass(key) + + def zones(self): + return ZoneDumpData.transition_keys() + + def fixed_offset_zones(self): + return ZoneDumpData.fixed_offset_zones() + + def load_transition_examples(self, key): + return ZoneDumpData.load_transition_examples(key) + + def test_str(self): + # Zones constructed with a key must have str(zone) == key + for key in self.zones(): + with self.subTest(key): + zi = self.zone_from_key(key) + + self.assertEqual(str(zi), key) + + # Zones with no key constructed should have str(zone) == repr(zone) + file_key = self.zoneinfo_data.keys[0] + file_path = self.zoneinfo_data.path_from_key(file_key) + + with open(file_path, "rb") as f: + with self.subTest(test_name="Repr test", path=file_path): + zi_ff = self.klass.from_file(f) + self.assertEqual(str(zi_ff), repr(zi_ff)) + + def test_repr(self): + # The repr is not guaranteed, but I think we can insist that it at + # least contain the name of the class. + key = next(iter(self.zones())) + + zi = self.klass(key) + class_name = self.class_name + with self.subTest(name="from key"): + self.assertRegex(repr(zi), class_name) + + file_key = self.zoneinfo_data.keys[0] + file_path = self.zoneinfo_data.path_from_key(file_key) + with open(file_path, "rb") as f: + zi_ff = self.klass.from_file(f, key=file_key) + + with self.subTest(name="from file with key"): + self.assertRegex(repr(zi_ff), class_name) + + with open(file_path, "rb") as f: + zi_ff_nk = self.klass.from_file(f) + + with self.subTest(name="from file without key"): + self.assertRegex(repr(zi_ff_nk), class_name) + + def test_key_attribute(self): + key = next(iter(self.zones())) + + def from_file_nokey(key): + with open(self.zoneinfo_data.path_from_key(key), "rb") as f: + return self.klass.from_file(f) + + constructors = ( + ("Primary constructor", self.klass, key), + ("no_cache", self.klass.no_cache, key), + ("from_file", from_file_nokey, None), + ) + + for msg, constructor, expected in constructors: + zi = constructor(key) + + # Ensure that the key attribute is set to the input to ``key`` + with self.subTest(msg): + self.assertEqual(zi.key, expected) + + # Ensure that the key attribute is read-only + with self.subTest(f"{msg}: readonly"): + with self.assertRaises(AttributeError): + zi.key = "Some/Value" + + def test_bad_keys(self): + bad_keys = [ + "Eurasia/Badzone", # Plausible but does not exist + "BZQ", + "America.Los_Angeles", + "🇨🇦", # Non-ascii + "America/New\ud800York", # Contains surrogate character + ] + + for bad_key in bad_keys: + with self.assertRaises(self.module.ZoneInfoNotFoundError): + self.klass(bad_key) + + def test_bad_keys_paths(self): + bad_keys = [ + "/America/Los_Angeles", # Absolute path + "America/Los_Angeles/", # Trailing slash - not normalized + "../zoneinfo/America/Los_Angeles", # Traverses above TZPATH + "America/../America/Los_Angeles", # Not normalized + "America/./Los_Angeles", + ] + + for bad_key in bad_keys: + with self.assertRaises(ValueError): + self.klass(bad_key) + + def test_bad_zones(self): + bad_zones = [ + b"", # Empty file + b"AAAA3" + b" " * 15, # Bad magic + ] + + for bad_zone in bad_zones: + fobj = io.BytesIO(bad_zone) + with self.assertRaises(ValueError): + self.klass.from_file(fobj) + + def test_fromutc_errors(self): + key = next(iter(self.zones())) + zone = self.zone_from_key(key) + + bad_values = [ + (datetime(2019, 1, 1, tzinfo=timezone.utc), ValueError), + (datetime(2019, 1, 1), ValueError), + (date(2019, 1, 1), TypeError), + (time(0), TypeError), + (0, TypeError), + ("2019-01-01", TypeError), + ] + + for val, exc_type in bad_values: + with self.subTest(val=val): + with self.assertRaises(exc_type): + zone.fromutc(val) + + def test_utc(self): + zi = self.klass("UTC") + dt = datetime(2020, 1, 1, tzinfo=zi) + + self.assertEqual(dt.utcoffset(), ZERO) + self.assertEqual(dt.dst(), ZERO) + self.assertEqual(dt.tzname(), "UTC") + + def test_unambiguous(self): + test_cases = [] + for key in self.zones(): + for zone_transition in self.load_transition_examples(key): + test_cases.append( + ( + key, + zone_transition.transition - timedelta(days=2), + zone_transition.offset_before, + ) + ) + + test_cases.append( + ( + key, + zone_transition.transition + timedelta(days=2), + zone_transition.offset_after, + ) + ) + + for key, dt, offset in test_cases: + with self.subTest(key=key, dt=dt, offset=offset): + tzi = self.zone_from_key(key) + dt = dt.replace(tzinfo=tzi) + + self.assertEqual(dt.tzname(), offset.tzname, dt) + self.assertEqual(dt.utcoffset(), offset.utcoffset, dt) + self.assertEqual(dt.dst(), offset.dst, dt) + + def test_folds_and_gaps(self): + test_cases = [] + for key in self.zones(): + tests = {"folds": [], "gaps": []} + for zt in self.load_transition_examples(key): + if zt.fold: + test_group = tests["folds"] + elif zt.gap: + test_group = tests["gaps"] + else: + # Assign a random variable here to disable the peephole + # optimizer so that coverage can see this line. + # See bpo-2506 for more information. + no_peephole_opt = None + continue + + # Cases are of the form key, dt, fold, offset + dt = zt.anomaly_start - timedelta(seconds=1) + test_group.append((dt, 0, zt.offset_before)) + test_group.append((dt, 1, zt.offset_before)) + + dt = zt.anomaly_start + test_group.append((dt, 0, zt.offset_before)) + test_group.append((dt, 1, zt.offset_after)) + + dt = zt.anomaly_start + timedelta(seconds=1) + test_group.append((dt, 0, zt.offset_before)) + test_group.append((dt, 1, zt.offset_after)) + + dt = zt.anomaly_end - timedelta(seconds=1) + test_group.append((dt, 0, zt.offset_before)) + test_group.append((dt, 1, zt.offset_after)) + + dt = zt.anomaly_end + test_group.append((dt, 0, zt.offset_after)) + test_group.append((dt, 1, zt.offset_after)) + + dt = zt.anomaly_end + timedelta(seconds=1) + test_group.append((dt, 0, zt.offset_after)) + test_group.append((dt, 1, zt.offset_after)) + + for grp, test_group in tests.items(): + test_cases.append(((key, grp), test_group)) + + for (key, grp), tests in test_cases: + with self.subTest(key=key, grp=grp): + tzi = self.zone_from_key(key) + + for dt, fold, offset in tests: + dt = dt.replace(fold=fold, tzinfo=tzi) + + self.assertEqual(dt.tzname(), offset.tzname, dt) + self.assertEqual(dt.utcoffset(), offset.utcoffset, dt) + self.assertEqual(dt.dst(), offset.dst, dt) + + def test_folds_from_utc(self): + for key in self.zones(): + zi = self.zone_from_key(key) + with self.subTest(key=key): + for zt in self.load_transition_examples(key): + if not zt.fold: + continue + + dt_utc = zt.transition_utc + dt_before_utc = dt_utc - timedelta(seconds=1) + dt_after_utc = dt_utc + timedelta(seconds=1) + + dt_before = dt_before_utc.astimezone(zi) + self.assertEqual(dt_before.fold, 0, (dt_before, dt_utc)) + + dt_after = dt_after_utc.astimezone(zi) + self.assertEqual(dt_after.fold, 1, (dt_after, dt_utc)) + + def test_time_variable_offset(self): + # self.zones() only ever returns variable-offset zones + for key in self.zones(): + zi = self.zone_from_key(key) + t = time(11, 15, 1, 34471, tzinfo=zi) + + with self.subTest(key=key): + self.assertIs(t.tzname(), None) + self.assertIs(t.utcoffset(), None) + self.assertIs(t.dst(), None) + + def test_time_fixed_offset(self): + for key, offset in self.fixed_offset_zones(): + zi = self.zone_from_key(key) + + t = time(11, 15, 1, 34471, tzinfo=zi) + + with self.subTest(key=key): + self.assertEqual(t.tzname(), offset.tzname) + self.assertEqual(t.utcoffset(), offset.utcoffset) + self.assertEqual(t.dst(), offset.dst) + + +class CZoneInfoTest(ZoneInfoTest): + module = c_zoneinfo + + @unittest.skipIf(MISSING_C_DOCSTRINGS, + "Signature information for builtins requires docstrings") + def test_signatures(self): + """Ensure that C module has valid method signatures.""" + import inspect + + must_have_signatures = ( + self.klass.clear_cache, + self.klass.no_cache, + self.klass.from_file, + ) + for method in must_have_signatures: + with self.subTest(method=method): + inspect.Signature.from_callable(method) + + def test_fold_mutate(self): + """Test that fold isn't mutated when no change is necessary. + + The underlying C API is capable of mutating datetime objects, and + may rely on the fact that addition of a datetime object returns a + new datetime; this test ensures that the input datetime to fromutc + is not mutated. + """ + + def to_subclass(dt): + class SameAddSubclass(type(dt)): + def __add__(self, other): + if other == timedelta(0): + return self + + return super().__add__(other) # pragma: nocover + + return SameAddSubclass( + dt.year, + dt.month, + dt.day, + dt.hour, + dt.minute, + dt.second, + dt.microsecond, + fold=dt.fold, + tzinfo=dt.tzinfo, + ) + + subclass = [False, True] + + key = "Europe/London" + zi = self.zone_from_key(key) + for zt in self.load_transition_examples(key): + if zt.fold and zt.offset_after.utcoffset == ZERO: + example = zt.transition_utc.replace(tzinfo=zi) + break + + for subclass in [False, True]: + if subclass: + dt = to_subclass(example) + else: + dt = example + + with self.subTest(subclass=subclass): + dt_fromutc = zi.fromutc(dt) + + self.assertEqual(dt_fromutc.fold, 1) + self.assertEqual(dt.fold, 0) + + +class ZoneInfoDatetimeSubclassTest(DatetimeSubclassMixin, ZoneInfoTest): + pass + + +class CZoneInfoDatetimeSubclassTest(DatetimeSubclassMixin, CZoneInfoTest): + pass + + +class ZoneInfoSubclassTest(ZoneInfoTest): + @classmethod + def setUpClass(cls): + super().setUpClass() + + class ZISubclass(cls.klass): + pass + + cls.class_name = "ZISubclass" + cls.parent_klass = cls.klass + cls.klass = ZISubclass + + def test_subclass_own_cache(self): + base_obj = self.parent_klass("Europe/London") + sub_obj = self.klass("Europe/London") + + self.assertIsNot(base_obj, sub_obj) + self.assertIsInstance(base_obj, self.parent_klass) + self.assertIsInstance(sub_obj, self.klass) + + +class CZoneInfoSubclassTest(ZoneInfoSubclassTest): + module = c_zoneinfo + + +class ZoneInfoV1Test(ZoneInfoTest): + @property + def zoneinfo_data(self): + return ZONEINFO_DATA_V1 + + def load_transition_examples(self, key): + # We will discard zdump examples outside the range epoch +/- 2**31, + # because they are not well-supported in Version 1 files. + epoch = datetime(1970, 1, 1) + max_offset_32 = timedelta(seconds=2 ** 31) + min_dt = epoch - max_offset_32 + max_dt = epoch + max_offset_32 + + for zt in ZoneDumpData.load_transition_examples(key): + if min_dt <= zt.transition <= max_dt: + yield zt + + +class CZoneInfoV1Test(ZoneInfoV1Test): + module = c_zoneinfo + + +@unittest.skipIf( + not HAS_TZDATA_PKG, "Skipping tzdata-specific tests: tzdata not installed" +) +class TZDataTests(ZoneInfoTest): + """ + Runs all the ZoneInfoTest tests, but against the tzdata package + + NOTE: The ZoneDumpData has frozen test data, but tzdata will update, so + some of the tests (particularly those related to the far future) may break + in the event that the time zone policies in the relevant time zones change. + """ + + @property + def tzpath(self): + return [] + + @property + def block_tzdata(self): + return False + + def zone_from_key(self, key): + return self.klass(key=key) + + +@unittest.skipIf( + not HAS_TZDATA_PKG, "Skipping tzdata-specific tests: tzdata not installed" +) +class CTZDataTests(TZDataTests): + module = c_zoneinfo + + +class WeirdZoneTest(ZoneInfoTestBase): + module = py_zoneinfo + + def test_one_transition(self): + LMT = ZoneOffset("LMT", -timedelta(hours=6, minutes=31, seconds=2)) + STD = ZoneOffset("STD", -timedelta(hours=6)) + + transitions = [ + ZoneTransition(datetime(1883, 6, 9, 14), LMT, STD), + ] + + after = "STD6" + + zf = self.construct_zone(transitions, after) + zi = self.klass.from_file(zf) + + dt0 = datetime(1883, 6, 9, 1, tzinfo=zi) + dt1 = datetime(1883, 6, 10, 1, tzinfo=zi) + + for dt, offset in [(dt0, LMT), (dt1, STD)]: + with self.subTest(name="local", dt=dt): + self.assertEqual(dt.tzname(), offset.tzname) + self.assertEqual(dt.utcoffset(), offset.utcoffset) + self.assertEqual(dt.dst(), offset.dst) + + dts = [ + ( + datetime(1883, 6, 9, 1, tzinfo=zi), + datetime(1883, 6, 9, 7, 31, 2, tzinfo=timezone.utc), + ), + ( + datetime(2010, 4, 1, 12, tzinfo=zi), + datetime(2010, 4, 1, 18, tzinfo=timezone.utc), + ), + ] + + for dt_local, dt_utc in dts: + with self.subTest(name="fromutc", dt=dt_local): + dt_actual = dt_utc.astimezone(zi) + self.assertEqual(dt_actual, dt_local) + + dt_utc_actual = dt_local.astimezone(timezone.utc) + self.assertEqual(dt_utc_actual, dt_utc) + + def test_one_zone_dst(self): + DST = ZoneOffset("DST", ONE_H, ONE_H) + transitions = [ + ZoneTransition(datetime(1970, 1, 1), DST, DST), + ] + + after = "STD0DST-1,0/0,J365/25" + + zf = self.construct_zone(transitions, after) + zi = self.klass.from_file(zf) + + dts = [ + datetime(1900, 3, 1), + datetime(1965, 9, 12), + datetime(1970, 1, 1), + datetime(2010, 11, 3), + datetime(2040, 1, 1), + ] + + for dt in dts: + dt = dt.replace(tzinfo=zi) + with self.subTest(dt=dt): + self.assertEqual(dt.tzname(), DST.tzname) + self.assertEqual(dt.utcoffset(), DST.utcoffset) + self.assertEqual(dt.dst(), DST.dst) + + def test_no_tz_str(self): + STD = ZoneOffset("STD", ONE_H, ZERO) + DST = ZoneOffset("DST", 2 * ONE_H, ONE_H) + + transitions = [] + for year in range(1996, 2000): + transitions.append( + ZoneTransition(datetime(year, 3, 1, 2), STD, DST) + ) + transitions.append( + ZoneTransition(datetime(year, 11, 1, 2), DST, STD) + ) + + after = "" + + zf = self.construct_zone(transitions, after) + + # According to RFC 8536, local times after the last transition time + # with an empty TZ string are unspecified. We will go with "hold the + # last transition", but the most we should promise is "doesn't crash." + zi = self.klass.from_file(zf) + + cases = [ + (datetime(1995, 1, 1), STD), + (datetime(1996, 4, 1), DST), + (datetime(1996, 11, 2), STD), + (datetime(2001, 1, 1), STD), + ] + + for dt, offset in cases: + dt = dt.replace(tzinfo=zi) + with self.subTest(dt=dt): + self.assertEqual(dt.tzname(), offset.tzname) + self.assertEqual(dt.utcoffset(), offset.utcoffset) + self.assertEqual(dt.dst(), offset.dst) + + # Test that offsets return None when using a datetime.time + t = time(0, tzinfo=zi) + with self.subTest("Testing datetime.time"): + self.assertIs(t.tzname(), None) + self.assertIs(t.utcoffset(), None) + self.assertIs(t.dst(), None) + + def test_tz_before_only(self): + # From RFC 8536 Section 3.2: + # + # If there are no transitions, local time for all timestamps is + # specified by the TZ string in the footer if present and nonempty; + # otherwise, it is specified by time type 0. + + offsets = [ + ZoneOffset("STD", ZERO, ZERO), + ZoneOffset("DST", ONE_H, ONE_H), + ] + + for offset in offsets: + # Phantom transition to set time type 0. + transitions = [ + ZoneTransition(None, offset, offset), + ] + + after = "" + + zf = self.construct_zone(transitions, after) + zi = self.klass.from_file(zf) + + dts = [ + datetime(1900, 1, 1), + datetime(1970, 1, 1), + datetime(2000, 1, 1), + ] + + for dt in dts: + dt = dt.replace(tzinfo=zi) + with self.subTest(offset=offset, dt=dt): + self.assertEqual(dt.tzname(), offset.tzname) + self.assertEqual(dt.utcoffset(), offset.utcoffset) + self.assertEqual(dt.dst(), offset.dst) + + def test_empty_zone(self): + zf = self.construct_zone([], "") + + with self.assertRaises(ValueError): + self.klass.from_file(zf) + + def test_zone_very_large_timestamp(self): + """Test when a transition is in the far past or future. + + Particularly, this is a concern if something: + + 1. Attempts to call ``datetime.timestamp`` for a datetime outside + of ``[datetime.min, datetime.max]``. + 2. Attempts to construct a timedelta outside of + ``[timedelta.min, timedelta.max]``. + + This actually occurs "in the wild", as some time zones on Ubuntu (at + least as of 2020) have an initial transition added at ``-2**58``. + """ + + LMT = ZoneOffset("LMT", timedelta(seconds=-968)) + GMT = ZoneOffset("GMT", ZERO) + + transitions = [ + (-(1 << 62), LMT, LMT), + ZoneTransition(datetime(1912, 1, 1), LMT, GMT), + ((1 << 62), GMT, GMT), + ] + + after = "GMT0" + + zf = self.construct_zone(transitions, after) + zi = self.klass.from_file(zf, key="Africa/Abidjan") + + offset_cases = [ + (datetime.min, LMT), + (datetime.max, GMT), + (datetime(1911, 12, 31), LMT), + (datetime(1912, 1, 2), GMT), + ] + + for dt_naive, offset in offset_cases: + dt = dt_naive.replace(tzinfo=zi) + with self.subTest(name="offset", dt=dt, offset=offset): + self.assertEqual(dt.tzname(), offset.tzname) + self.assertEqual(dt.utcoffset(), offset.utcoffset) + self.assertEqual(dt.dst(), offset.dst) + + utc_cases = [ + (datetime.min, datetime.min + timedelta(seconds=968)), + (datetime(1898, 12, 31, 23, 43, 52), datetime(1899, 1, 1)), + ( + datetime(1911, 12, 31, 23, 59, 59, 999999), + datetime(1912, 1, 1, 0, 16, 7, 999999), + ), + (datetime(1912, 1, 1, 0, 16, 8), datetime(1912, 1, 1, 0, 16, 8)), + (datetime(1970, 1, 1), datetime(1970, 1, 1)), + (datetime.max, datetime.max), + ] + + for naive_dt, naive_dt_utc in utc_cases: + dt = naive_dt.replace(tzinfo=zi) + dt_utc = naive_dt_utc.replace(tzinfo=timezone.utc) + + self.assertEqual(dt_utc.astimezone(zi), dt) + self.assertEqual(dt, dt_utc) + + def test_fixed_offset_phantom_transition(self): + UTC = ZoneOffset("UTC", ZERO, ZERO) + + transitions = [ZoneTransition(datetime(1970, 1, 1), UTC, UTC)] + + after = "UTC0" + zf = self.construct_zone(transitions, after) + zi = self.klass.from_file(zf, key="UTC") + + dt = datetime(2020, 1, 1, tzinfo=zi) + with self.subTest("datetime.datetime"): + self.assertEqual(dt.tzname(), UTC.tzname) + self.assertEqual(dt.utcoffset(), UTC.utcoffset) + self.assertEqual(dt.dst(), UTC.dst) + + t = time(0, tzinfo=zi) + with self.subTest("datetime.time"): + self.assertEqual(t.tzname(), UTC.tzname) + self.assertEqual(t.utcoffset(), UTC.utcoffset) + self.assertEqual(t.dst(), UTC.dst) + + def construct_zone(self, transitions, after=None, version=3): + # These are not used for anything, so we're not going to include + # them for now. + isutc = [] + isstd = [] + leap_seconds = [] + + offset_lists = [[], []] + trans_times_lists = [[], []] + trans_idx_lists = [[], []] + + v1_range = (-(2 ** 31), 2 ** 31) + v2_range = (-(2 ** 63), 2 ** 63) + ranges = [v1_range, v2_range] + + def zt_as_tuple(zt): + # zt may be a tuple (timestamp, offset_before, offset_after) or + # a ZoneTransition object — this is to allow the timestamp to be + # values that are outside the valid range for datetimes but still + # valid 64-bit timestamps. + if isinstance(zt, tuple): + return zt + + if zt.transition: + trans_time = int(zt.transition_utc.timestamp()) + else: + trans_time = None + + return (trans_time, zt.offset_before, zt.offset_after) + + transitions = sorted(map(zt_as_tuple, transitions), key=lambda x: x[0]) + + for zt in transitions: + trans_time, offset_before, offset_after = zt + + for v, (dt_min, dt_max) in enumerate(ranges): + offsets = offset_lists[v] + trans_times = trans_times_lists[v] + trans_idx = trans_idx_lists[v] + + if trans_time is not None and not ( + dt_min <= trans_time <= dt_max + ): + continue + + if offset_before not in offsets: + offsets.append(offset_before) + + if offset_after not in offsets: + offsets.append(offset_after) + + if trans_time is not None: + trans_times.append(trans_time) + trans_idx.append(offsets.index(offset_after)) + + isutcnt = len(isutc) + isstdcnt = len(isstd) + leapcnt = len(leap_seconds) + + zonefile = io.BytesIO() + + time_types = ("l", "q") + for v in range(min((version, 2))): + offsets = offset_lists[v] + trans_times = trans_times_lists[v] + trans_idx = trans_idx_lists[v] + time_type = time_types[v] + + # Translate the offsets into something closer to the C values + abbrstr = bytearray() + ttinfos = [] + + for offset in offsets: + utcoff = int(offset.utcoffset.total_seconds()) + isdst = bool(offset.dst) + abbrind = len(abbrstr) + + ttinfos.append((utcoff, isdst, abbrind)) + abbrstr += offset.tzname.encode("ascii") + b"\x00" + abbrstr = bytes(abbrstr) + + typecnt = len(offsets) + timecnt = len(trans_times) + charcnt = len(abbrstr) + + # Write the header + zonefile.write(b"TZif") + zonefile.write(b"%d" % version) + zonefile.write(b" " * 15) + zonefile.write( + struct.pack( + ">6l", isutcnt, isstdcnt, leapcnt, timecnt, typecnt, charcnt + ) + ) + + # Now the transition data + zonefile.write(struct.pack(f">{timecnt}{time_type}", *trans_times)) + zonefile.write(struct.pack(f">{timecnt}B", *trans_idx)) + + for ttinfo in ttinfos: + zonefile.write(struct.pack(">lbb", *ttinfo)) + + zonefile.write(bytes(abbrstr)) + + # Now the metadata and leap seconds + zonefile.write(struct.pack(f"{isutcnt}b", *isutc)) + zonefile.write(struct.pack(f"{isstdcnt}b", *isstd)) + zonefile.write(struct.pack(f">{leapcnt}l", *leap_seconds)) + + # Finally we write the TZ string if we're writing a Version 2+ file + if v > 0: + zonefile.write(b"\x0A") + zonefile.write(after.encode("ascii")) + zonefile.write(b"\x0A") + + zonefile.seek(0) + return zonefile + + +class CWeirdZoneTest(WeirdZoneTest): + module = c_zoneinfo + + +class TZStrTest(ZoneInfoTestBase): + module = py_zoneinfo + + NORMAL = 0 + FOLD = 1 + GAP = 2 + + @classmethod + def setUpClass(cls): + super().setUpClass() + + cls._populate_test_cases() + cls.populate_tzstr_header() + + @classmethod + def populate_tzstr_header(cls): + out = bytearray() + # The TZif format always starts with a Version 1 file followed by + # the Version 2+ file. In this case, we have no transitions, just + # the tzstr in the footer, so up to the footer, the files are + # identical and we can just write the same file twice in a row. + for _ in range(2): + out += b"TZif" # Magic value + out += b"3" # Version + out += b" " * 15 # Reserved + + # We will not write any of the manual transition parts + out += struct.pack(">6l", 0, 0, 0, 0, 0, 0) + + cls._tzif_header = bytes(out) + + def zone_from_tzstr(self, tzstr): + """Creates a zoneinfo file following a POSIX rule.""" + zonefile = io.BytesIO(self._tzif_header) + zonefile.seek(0, 2) + + # Write the footer + zonefile.write(b"\x0A") + zonefile.write(tzstr.encode("ascii")) + zonefile.write(b"\x0A") + + zonefile.seek(0) + + return self.klass.from_file(zonefile, key=tzstr) + + def test_tzstr_localized(self): + for tzstr, cases in self.test_cases.items(): + with self.subTest(tzstr=tzstr): + zi = self.zone_from_tzstr(tzstr) + + for dt_naive, offset, _ in cases: + dt = dt_naive.replace(tzinfo=zi) + + with self.subTest(tzstr=tzstr, dt=dt, offset=offset): + self.assertEqual(dt.tzname(), offset.tzname) + self.assertEqual(dt.utcoffset(), offset.utcoffset) + self.assertEqual(dt.dst(), offset.dst) + + def test_tzstr_from_utc(self): + for tzstr, cases in self.test_cases.items(): + with self.subTest(tzstr=tzstr): + zi = self.zone_from_tzstr(tzstr) + + for dt_naive, offset, dt_type in cases: + if dt_type == self.GAP: + continue # Cannot create a gap from UTC + + dt_utc = (dt_naive - offset.utcoffset).replace( + tzinfo=timezone.utc + ) + + # Check that we can go UTC -> Our zone + dt_act = dt_utc.astimezone(zi) + dt_exp = dt_naive.replace(tzinfo=zi) + + self.assertEqual(dt_act, dt_exp) + + if dt_type == self.FOLD: + self.assertEqual(dt_act.fold, dt_naive.fold, dt_naive) + else: + self.assertEqual(dt_act.fold, 0) + + # Now check that we can go our zone -> UTC + dt_act = dt_exp.astimezone(timezone.utc) + + self.assertEqual(dt_act, dt_utc) + + def test_extreme_tzstr(self): + tzstrs = [ + # Extreme offset hour + "AAA24", + "AAA+24", + "AAA-24", + "AAA24BBB,J60/2,J300/2", + "AAA+24BBB,J60/2,J300/2", + "AAA-24BBB,J60/2,J300/2", + "AAA4BBB24,J60/2,J300/2", + "AAA4BBB+24,J60/2,J300/2", + "AAA4BBB-24,J60/2,J300/2", + # Extreme offset minutes + "AAA4:00BBB,J60/2,J300/2", + "AAA4:59BBB,J60/2,J300/2", + "AAA4BBB5:00,J60/2,J300/2", + "AAA4BBB5:59,J60/2,J300/2", + # Extreme offset seconds + "AAA4:00:00BBB,J60/2,J300/2", + "AAA4:00:59BBB,J60/2,J300/2", + "AAA4BBB5:00:00,J60/2,J300/2", + "AAA4BBB5:00:59,J60/2,J300/2", + # Extreme total offset + "AAA24:59:59BBB5,J60/2,J300/2", + "AAA-24:59:59BBB5,J60/2,J300/2", + "AAA4BBB24:59:59,J60/2,J300/2", + "AAA4BBB-24:59:59,J60/2,J300/2", + # Extreme months + "AAA4BBB,M12.1.1/2,M1.1.1/2", + "AAA4BBB,M1.1.1/2,M12.1.1/2", + # Extreme weeks + "AAA4BBB,M1.5.1/2,M1.1.1/2", + "AAA4BBB,M1.1.1/2,M1.5.1/2", + # Extreme weekday + "AAA4BBB,M1.1.6/2,M2.1.1/2", + "AAA4BBB,M1.1.1/2,M2.1.6/2", + # Extreme numeric offset + "AAA4BBB,0/2,20/2", + "AAA4BBB,0/2,0/14", + "AAA4BBB,20/2,365/2", + "AAA4BBB,365/2,365/14", + # Extreme julian offset + "AAA4BBB,J1/2,J20/2", + "AAA4BBB,J1/2,J1/14", + "AAA4BBB,J20/2,J365/2", + "AAA4BBB,J365/2,J365/14", + # Extreme transition hour + "AAA4BBB,J60/167,J300/2", + "AAA4BBB,J60/+167,J300/2", + "AAA4BBB,J60/-167,J300/2", + "AAA4BBB,J60/2,J300/167", + "AAA4BBB,J60/2,J300/+167", + "AAA4BBB,J60/2,J300/-167", + # Extreme transition minutes + "AAA4BBB,J60/2:00,J300/2", + "AAA4BBB,J60/2:59,J300/2", + "AAA4BBB,J60/2,J300/2:00", + "AAA4BBB,J60/2,J300/2:59", + # Extreme transition seconds + "AAA4BBB,J60/2:00:00,J300/2", + "AAA4BBB,J60/2:00:59,J300/2", + "AAA4BBB,J60/2,J300/2:00:00", + "AAA4BBB,J60/2,J300/2:00:59", + # Extreme total transition time + "AAA4BBB,J60/167:59:59,J300/2", + "AAA4BBB,J60/-167:59:59,J300/2", + "AAA4BBB,J60/2,J300/167:59:59", + "AAA4BBB,J60/2,J300/-167:59:59", + ] + + for tzstr in tzstrs: + with self.subTest(tzstr=tzstr): + self.zone_from_tzstr(tzstr) + + def test_invalid_tzstr(self): + invalid_tzstrs = [ + "PST8PDT", # DST but no transition specified + "+11", # Unquoted alphanumeric + "GMT,M3.2.0/2,M11.1.0/3", # Transition rule but no DST + "GMT0+11,M3.2.0/2,M11.1.0/3", # Unquoted alphanumeric in DST + "PST8PDT,M3.2.0/2", # Only one transition rule + # Invalid offset hours + "AAA168", + "AAA+168", + "AAA-168", + "AAA168BBB,J60/2,J300/2", + "AAA+168BBB,J60/2,J300/2", + "AAA-168BBB,J60/2,J300/2", + "AAA4BBB168,J60/2,J300/2", + "AAA4BBB+168,J60/2,J300/2", + "AAA4BBB-168,J60/2,J300/2", + # Invalid offset minutes + "AAA4:0BBB,J60/2,J300/2", + "AAA4:100BBB,J60/2,J300/2", + "AAA4BBB5:0,J60/2,J300/2", + "AAA4BBB5:100,J60/2,J300/2", + # Invalid offset seconds + "AAA4:00:0BBB,J60/2,J300/2", + "AAA4:00:100BBB,J60/2,J300/2", + "AAA4BBB5:00:0,J60/2,J300/2", + "AAA4BBB5:00:100,J60/2,J300/2", + # Completely invalid dates + "AAA4BBB,M1443339,M11.1.0/3", + "AAA4BBB,M3.2.0/2,0349309483959c", + "AAA4BBB,,J300/2", + "AAA4BBB,z,J300/2", + "AAA4BBB,J60/2,", + "AAA4BBB,J60/2,z", + # Invalid months + "AAA4BBB,M13.1.1/2,M1.1.1/2", + "AAA4BBB,M1.1.1/2,M13.1.1/2", + "AAA4BBB,M0.1.1/2,M1.1.1/2", + "AAA4BBB,M1.1.1/2,M0.1.1/2", + # Invalid weeks + "AAA4BBB,M1.6.1/2,M1.1.1/2", + "AAA4BBB,M1.1.1/2,M1.6.1/2", + # Invalid weekday + "AAA4BBB,M1.1.7/2,M2.1.1/2", + "AAA4BBB,M1.1.1/2,M2.1.7/2", + # Invalid numeric offset + "AAA4BBB,-1/2,20/2", + "AAA4BBB,1/2,-1/2", + "AAA4BBB,367,20/2", + "AAA4BBB,1/2,367/2", + # Invalid julian offset + "AAA4BBB,J0/2,J20/2", + "AAA4BBB,J20/2,J366/2", + # Invalid transition time + "AAA4BBB,J60/2/3,J300/2", + "AAA4BBB,J60/2,J300/2/3", + # Invalid transition hour + "AAA4BBB,J60/168,J300/2", + "AAA4BBB,J60/+168,J300/2", + "AAA4BBB,J60/-168,J300/2", + "AAA4BBB,J60/2,J300/168", + "AAA4BBB,J60/2,J300/+168", + "AAA4BBB,J60/2,J300/-168", + # Invalid transition minutes + "AAA4BBB,J60/2:0,J300/2", + "AAA4BBB,J60/2:100,J300/2", + "AAA4BBB,J60/2,J300/2:0", + "AAA4BBB,J60/2,J300/2:100", + # Invalid transition seconds + "AAA4BBB,J60/2:00:0,J300/2", + "AAA4BBB,J60/2:00:100,J300/2", + "AAA4BBB,J60/2,J300/2:00:0", + "AAA4BBB,J60/2,J300/2:00:100", + ] + + for invalid_tzstr in invalid_tzstrs: + with self.subTest(tzstr=invalid_tzstr): + # Not necessarily a guaranteed property, but we should show + # the problematic TZ string if that's the cause of failure. + tzstr_regex = re.escape(invalid_tzstr) + with self.assertRaisesRegex(ValueError, tzstr_regex): + self.zone_from_tzstr(invalid_tzstr) + + @classmethod + def _populate_test_cases(cls): + # This method uses a somewhat unusual style in that it populates the + # test cases for each tzstr by using a decorator to automatically call + # a function that mutates the current dictionary of test cases. + # + # The population of the test cases is done in individual functions to + # give each set of test cases its own namespace in which to define + # its offsets (this way we don't have to worry about variable reuse + # causing problems if someone makes a typo). + # + # The decorator for calling is used to make it more obvious that each + # function is actually called (if it's not decorated, it's not called). + def call(f): + """Decorator to call the addition methods. + + This will call a function which adds at least one new entry into + the `cases` dictionary. The decorator will also assert that + something was added to the dictionary. + """ + prev_len = len(cases) + f() + assert len(cases) > prev_len, "Function did not add a test case!" + + NORMAL = cls.NORMAL + FOLD = cls.FOLD + GAP = cls.GAP + + cases = {} + + @call + def _add(): + # Transition to EDT on the 2nd Sunday in March at 4 AM, and + # transition back on the first Sunday in November at 3AM + tzstr = "EST5EDT,M3.2.0/4:00,M11.1.0/3:00" + + EST = ZoneOffset("EST", timedelta(hours=-5), ZERO) + EDT = ZoneOffset("EDT", timedelta(hours=-4), ONE_H) + + cases[tzstr] = ( + (datetime(2019, 3, 9), EST, NORMAL), + (datetime(2019, 3, 10, 3, 59), EST, NORMAL), + (datetime(2019, 3, 10, 4, 0, fold=0), EST, GAP), + (datetime(2019, 3, 10, 4, 0, fold=1), EDT, GAP), + (datetime(2019, 3, 10, 4, 1, fold=0), EST, GAP), + (datetime(2019, 3, 10, 4, 1, fold=1), EDT, GAP), + (datetime(2019, 11, 2), EDT, NORMAL), + (datetime(2019, 11, 3, 1, 59, fold=1), EDT, NORMAL), + (datetime(2019, 11, 3, 2, 0, fold=0), EDT, FOLD), + (datetime(2019, 11, 3, 2, 0, fold=1), EST, FOLD), + (datetime(2020, 3, 8, 3, 59), EST, NORMAL), + (datetime(2020, 3, 8, 4, 0, fold=0), EST, GAP), + (datetime(2020, 3, 8, 4, 0, fold=1), EDT, GAP), + (datetime(2020, 11, 1, 1, 59, fold=1), EDT, NORMAL), + (datetime(2020, 11, 1, 2, 0, fold=0), EDT, FOLD), + (datetime(2020, 11, 1, 2, 0, fold=1), EST, FOLD), + ) + + @call + def _add(): + # Transition to BST happens on the last Sunday in March at 1 AM GMT + # and the transition back happens the last Sunday in October at 2AM BST + tzstr = "GMT0BST-1,M3.5.0/1:00,M10.5.0/2:00" + + GMT = ZoneOffset("GMT", ZERO, ZERO) + BST = ZoneOffset("BST", ONE_H, ONE_H) + + cases[tzstr] = ( + (datetime(2019, 3, 30), GMT, NORMAL), + (datetime(2019, 3, 31, 0, 59), GMT, NORMAL), + (datetime(2019, 3, 31, 2, 0), BST, NORMAL), + (datetime(2019, 10, 26), BST, NORMAL), + (datetime(2019, 10, 27, 0, 59, fold=1), BST, NORMAL), + (datetime(2019, 10, 27, 1, 0, fold=0), BST, GAP), + (datetime(2019, 10, 27, 2, 0, fold=1), GMT, GAP), + (datetime(2020, 3, 29, 0, 59), GMT, NORMAL), + (datetime(2020, 3, 29, 2, 0), BST, NORMAL), + (datetime(2020, 10, 25, 0, 59, fold=1), BST, NORMAL), + (datetime(2020, 10, 25, 1, 0, fold=0), BST, FOLD), + (datetime(2020, 10, 25, 2, 0, fold=1), GMT, NORMAL), + ) + + @call + def _add(): + # Austrialian time zone - DST start is chronologically first + tzstr = "AEST-10AEDT,M10.1.0/2,M4.1.0/3" + + AEST = ZoneOffset("AEST", timedelta(hours=10), ZERO) + AEDT = ZoneOffset("AEDT", timedelta(hours=11), ONE_H) + + cases[tzstr] = ( + (datetime(2019, 4, 6), AEDT, NORMAL), + (datetime(2019, 4, 7, 1, 59), AEDT, NORMAL), + (datetime(2019, 4, 7, 1, 59, fold=1), AEDT, NORMAL), + (datetime(2019, 4, 7, 2, 0, fold=0), AEDT, FOLD), + (datetime(2019, 4, 7, 2, 1, fold=0), AEDT, FOLD), + (datetime(2019, 4, 7, 2, 0, fold=1), AEST, FOLD), + (datetime(2019, 4, 7, 2, 1, fold=1), AEST, FOLD), + (datetime(2019, 4, 7, 3, 0, fold=0), AEST, NORMAL), + (datetime(2019, 4, 7, 3, 0, fold=1), AEST, NORMAL), + (datetime(2019, 10, 5, 0), AEST, NORMAL), + (datetime(2019, 10, 6, 1, 59), AEST, NORMAL), + (datetime(2019, 10, 6, 2, 0, fold=0), AEST, GAP), + (datetime(2019, 10, 6, 2, 0, fold=1), AEDT, GAP), + (datetime(2019, 10, 6, 3, 0), AEDT, NORMAL), + ) + + @call + def _add(): + # Irish time zone - negative DST + tzstr = "IST-1GMT0,M10.5.0,M3.5.0/1" + + GMT = ZoneOffset("GMT", ZERO, -ONE_H) + IST = ZoneOffset("IST", ONE_H, ZERO) + + cases[tzstr] = ( + (datetime(2019, 3, 30), GMT, NORMAL), + (datetime(2019, 3, 31, 0, 59), GMT, NORMAL), + (datetime(2019, 3, 31, 2, 0), IST, NORMAL), + (datetime(2019, 10, 26), IST, NORMAL), + (datetime(2019, 10, 27, 0, 59, fold=1), IST, NORMAL), + (datetime(2019, 10, 27, 1, 0, fold=0), IST, FOLD), + (datetime(2019, 10, 27, 1, 0, fold=1), GMT, FOLD), + (datetime(2019, 10, 27, 2, 0, fold=1), GMT, NORMAL), + (datetime(2020, 3, 29, 0, 59), GMT, NORMAL), + (datetime(2020, 3, 29, 2, 0), IST, NORMAL), + (datetime(2020, 10, 25, 0, 59, fold=1), IST, NORMAL), + (datetime(2020, 10, 25, 1, 0, fold=0), IST, FOLD), + (datetime(2020, 10, 25, 2, 0, fold=1), GMT, NORMAL), + ) + + @call + def _add(): + # Pacific/Kosrae: Fixed offset zone with a quoted numerical tzname + tzstr = "<+11>-11" + + cases[tzstr] = ( + ( + datetime(2020, 1, 1), + ZoneOffset("+11", timedelta(hours=11)), + NORMAL, + ), + ) + + @call + def _add(): + # Quoted STD and DST, transitions at 24:00 + tzstr = "<-04>4<-03>,M9.1.6/24,M4.1.6/24" + + M04 = ZoneOffset("-04", timedelta(hours=-4)) + M03 = ZoneOffset("-03", timedelta(hours=-3), ONE_H) + + cases[tzstr] = ( + (datetime(2020, 5, 1), M04, NORMAL), + (datetime(2020, 11, 1), M03, NORMAL), + ) + + @call + def _add(): + # Permanent daylight saving time is modeled with transitions at 0/0 + # and J365/25, as mentioned in RFC 8536 Section 3.3.1 + tzstr = "EST5EDT,0/0,J365/25" + + EDT = ZoneOffset("EDT", timedelta(hours=-4), ONE_H) + + cases[tzstr] = ( + (datetime(2019, 1, 1), EDT, NORMAL), + (datetime(2019, 6, 1), EDT, NORMAL), + (datetime(2019, 12, 31, 23, 59, 59, 999999), EDT, NORMAL), + (datetime(2020, 1, 1), EDT, NORMAL), + (datetime(2020, 3, 1), EDT, NORMAL), + (datetime(2020, 6, 1), EDT, NORMAL), + (datetime(2020, 12, 31, 23, 59, 59, 999999), EDT, NORMAL), + (datetime(2400, 1, 1), EDT, NORMAL), + (datetime(2400, 3, 1), EDT, NORMAL), + (datetime(2400, 12, 31, 23, 59, 59, 999999), EDT, NORMAL), + ) + + @call + def _add(): + # Transitions on March 1st and November 1st of each year + tzstr = "AAA3BBB,J60/12,J305/12" + + AAA = ZoneOffset("AAA", timedelta(hours=-3)) + BBB = ZoneOffset("BBB", timedelta(hours=-2), ONE_H) + + cases[tzstr] = ( + (datetime(2019, 1, 1), AAA, NORMAL), + (datetime(2019, 2, 28), AAA, NORMAL), + (datetime(2019, 3, 1, 11, 59), AAA, NORMAL), + (datetime(2019, 3, 1, 12, fold=0), AAA, GAP), + (datetime(2019, 3, 1, 12, fold=1), BBB, GAP), + (datetime(2019, 3, 1, 13), BBB, NORMAL), + (datetime(2019, 11, 1, 10, 59), BBB, NORMAL), + (datetime(2019, 11, 1, 11, fold=0), BBB, FOLD), + (datetime(2019, 11, 1, 11, fold=1), AAA, FOLD), + (datetime(2019, 11, 1, 12), AAA, NORMAL), + (datetime(2019, 12, 31, 23, 59, 59, 999999), AAA, NORMAL), + (datetime(2020, 1, 1), AAA, NORMAL), + (datetime(2020, 2, 29), AAA, NORMAL), + (datetime(2020, 3, 1, 11, 59), AAA, NORMAL), + (datetime(2020, 3, 1, 12, fold=0), AAA, GAP), + (datetime(2020, 3, 1, 12, fold=1), BBB, GAP), + (datetime(2020, 3, 1, 13), BBB, NORMAL), + (datetime(2020, 11, 1, 10, 59), BBB, NORMAL), + (datetime(2020, 11, 1, 11, fold=0), BBB, FOLD), + (datetime(2020, 11, 1, 11, fold=1), AAA, FOLD), + (datetime(2020, 11, 1, 12), AAA, NORMAL), + (datetime(2020, 12, 31, 23, 59, 59, 999999), AAA, NORMAL), + ) + + @call + def _add(): + # Taken from America/Godthab, this rule has a transition on the + # Saturday before the last Sunday of March and October, at 22:00 + # and 23:00, respectively. This is encoded with negative start + # and end transition times. + tzstr = "<-03>3<-02>,M3.5.0/-2,M10.5.0/-1" + + N03 = ZoneOffset("-03", timedelta(hours=-3)) + N02 = ZoneOffset("-02", timedelta(hours=-2), ONE_H) + + cases[tzstr] = ( + (datetime(2020, 3, 27), N03, NORMAL), + (datetime(2020, 3, 28, 21, 59, 59), N03, NORMAL), + (datetime(2020, 3, 28, 22, fold=0), N03, GAP), + (datetime(2020, 3, 28, 22, fold=1), N02, GAP), + (datetime(2020, 3, 28, 23), N02, NORMAL), + (datetime(2020, 10, 24, 21), N02, NORMAL), + (datetime(2020, 10, 24, 22, fold=0), N02, FOLD), + (datetime(2020, 10, 24, 22, fold=1), N03, FOLD), + (datetime(2020, 10, 24, 23), N03, NORMAL), + ) + + @call + def _add(): + # Transition times with minutes and seconds + tzstr = "AAA3BBB,M3.2.0/01:30,M11.1.0/02:15:45" + + AAA = ZoneOffset("AAA", timedelta(hours=-3)) + BBB = ZoneOffset("BBB", timedelta(hours=-2), ONE_H) + + cases[tzstr] = ( + (datetime(2012, 3, 11, 1, 0), AAA, NORMAL), + (datetime(2012, 3, 11, 1, 30, fold=0), AAA, GAP), + (datetime(2012, 3, 11, 1, 30, fold=1), BBB, GAP), + (datetime(2012, 3, 11, 2, 30), BBB, NORMAL), + (datetime(2012, 11, 4, 1, 15, 44, 999999), BBB, NORMAL), + (datetime(2012, 11, 4, 1, 15, 45, fold=0), BBB, FOLD), + (datetime(2012, 11, 4, 1, 15, 45, fold=1), AAA, FOLD), + (datetime(2012, 11, 4, 2, 15, 45), AAA, NORMAL), + ) + + cls.test_cases = cases + + +class CTZStrTest(TZStrTest): + module = c_zoneinfo + + +class ZoneInfoCacheTest(TzPathUserMixin, ZoneInfoTestBase): + module = py_zoneinfo + + def setUp(self): + self.klass.clear_cache() + super().setUp() + + @property + def zoneinfo_data(self): + return ZONEINFO_DATA + + @property + def tzpath(self): + return [self.zoneinfo_data.tzpath] + + def test_ephemeral_zones(self): + self.assertIs( + self.klass("America/Los_Angeles"), self.klass("America/Los_Angeles") + ) + + def test_strong_refs(self): + tz0 = self.klass("Australia/Sydney") + tz1 = self.klass("Australia/Sydney") + + self.assertIs(tz0, tz1) + + def test_no_cache(self): + + tz0 = self.klass("Europe/Lisbon") + tz1 = self.klass.no_cache("Europe/Lisbon") + + self.assertIsNot(tz0, tz1) + + def test_cache_reset_tzpath(self): + """Test that the cache persists when tzpath has been changed. + + The PEP specifies that as long as a reference exists to one zone + with a given key, the primary constructor must continue to return + the same object. + """ + zi0 = self.klass("America/Los_Angeles") + with self.tzpath_context([]): + zi1 = self.klass("America/Los_Angeles") + + self.assertIs(zi0, zi1) + + def test_clear_cache_explicit_none(self): + la0 = self.klass("America/Los_Angeles") + self.klass.clear_cache(only_keys=None) + la1 = self.klass("America/Los_Angeles") + + self.assertIsNot(la0, la1) + + def test_clear_cache_one_key(self): + """Tests that you can clear a single key from the cache.""" + la0 = self.klass("America/Los_Angeles") + dub0 = self.klass("Europe/Dublin") + + self.klass.clear_cache(only_keys=["America/Los_Angeles"]) + + la1 = self.klass("America/Los_Angeles") + dub1 = self.klass("Europe/Dublin") + + self.assertIsNot(la0, la1) + self.assertIs(dub0, dub1) + + def test_clear_cache_two_keys(self): + la0 = self.klass("America/Los_Angeles") + dub0 = self.klass("Europe/Dublin") + tok0 = self.klass("Asia/Tokyo") + + self.klass.clear_cache( + only_keys=["America/Los_Angeles", "Europe/Dublin"] + ) + + la1 = self.klass("America/Los_Angeles") + dub1 = self.klass("Europe/Dublin") + tok1 = self.klass("Asia/Tokyo") + + self.assertIsNot(la0, la1) + self.assertIsNot(dub0, dub1) + self.assertIs(tok0, tok1) + + +class CZoneInfoCacheTest(ZoneInfoCacheTest): + module = c_zoneinfo + + +class ZoneInfoPickleTest(TzPathUserMixin, ZoneInfoTestBase): + module = py_zoneinfo + + def setUp(self): + self.klass.clear_cache() + + with contextlib.ExitStack() as stack: + stack.enter_context(test_support.set_zoneinfo_module(self.module)) + self.addCleanup(stack.pop_all().close) + + super().setUp() + + @property + def zoneinfo_data(self): + return ZONEINFO_DATA + + @property + def tzpath(self): + return [self.zoneinfo_data.tzpath] + + def test_cache_hit(self): + for proto in range(pickle.HIGHEST_PROTOCOL + 1): + with self.subTest(proto=proto): + zi_in = self.klass("Europe/Dublin") + pkl = pickle.dumps(zi_in, protocol=proto) + zi_rt = pickle.loads(pkl) + + with self.subTest(test="Is non-pickled ZoneInfo"): + self.assertIs(zi_in, zi_rt) + + zi_rt2 = pickle.loads(pkl) + with self.subTest(test="Is unpickled ZoneInfo"): + self.assertIs(zi_rt, zi_rt2) + + def test_cache_miss(self): + for proto in range(pickle.HIGHEST_PROTOCOL + 1): + with self.subTest(proto=proto): + zi_in = self.klass("Europe/Dublin") + pkl = pickle.dumps(zi_in, protocol=proto) + + del zi_in + self.klass.clear_cache() # Induce a cache miss + zi_rt = pickle.loads(pkl) + zi_rt2 = pickle.loads(pkl) + + self.assertIs(zi_rt, zi_rt2) + + def test_no_cache(self): + for proto in range(pickle.HIGHEST_PROTOCOL + 1): + with self.subTest(proto=proto): + zi_no_cache = self.klass.no_cache("Europe/Dublin") + + pkl = pickle.dumps(zi_no_cache, protocol=proto) + zi_rt = pickle.loads(pkl) + + with self.subTest(test="Not the pickled object"): + self.assertIsNot(zi_rt, zi_no_cache) + + zi_rt2 = pickle.loads(pkl) + with self.subTest(test="Not a second unpickled object"): + self.assertIsNot(zi_rt, zi_rt2) + + zi_cache = self.klass("Europe/Dublin") + with self.subTest(test="Not a cached object"): + self.assertIsNot(zi_rt, zi_cache) + + def test_from_file(self): + key = "Europe/Dublin" + with open(self.zoneinfo_data.path_from_key(key), "rb") as f: + zi_nokey = self.klass.from_file(f) + + f.seek(0) + zi_key = self.klass.from_file(f, key=key) + + test_cases = [ + (zi_key, "ZoneInfo with key"), + (zi_nokey, "ZoneInfo without key"), + ] + + for zi, test_name in test_cases: + for proto in range(pickle.HIGHEST_PROTOCOL + 1): + with self.subTest(test_name=test_name, proto=proto): + with self.assertRaises(pickle.PicklingError): + pickle.dumps(zi, protocol=proto) + + def test_pickle_after_from_file(self): + # This may be a bit of paranoia, but this test is to ensure that no + # global state is maintained in order to handle the pickle cache and + # from_file behavior, and that it is possible to interweave the + # constructors of each of these and pickling/unpickling without issues. + for proto in range(pickle.HIGHEST_PROTOCOL + 1): + with self.subTest(proto=proto): + key = "Europe/Dublin" + zi = self.klass(key) + + pkl_0 = pickle.dumps(zi, protocol=proto) + zi_rt_0 = pickle.loads(pkl_0) + self.assertIs(zi, zi_rt_0) + + with open(self.zoneinfo_data.path_from_key(key), "rb") as f: + zi_ff = self.klass.from_file(f, key=key) + + pkl_1 = pickle.dumps(zi, protocol=proto) + zi_rt_1 = pickle.loads(pkl_1) + self.assertIs(zi, zi_rt_1) + + with self.assertRaises(pickle.PicklingError): + pickle.dumps(zi_ff, protocol=proto) + + pkl_2 = pickle.dumps(zi, protocol=proto) + zi_rt_2 = pickle.loads(pkl_2) + self.assertIs(zi, zi_rt_2) + + +class CZoneInfoPickleTest(ZoneInfoPickleTest): + module = c_zoneinfo + + +class CallingConventionTest(ZoneInfoTestBase): + """Tests for functions with restricted calling conventions.""" + + module = py_zoneinfo + + @property + def zoneinfo_data(self): + return ZONEINFO_DATA + + def test_from_file(self): + with open(self.zoneinfo_data.path_from_key("UTC"), "rb") as f: + with self.assertRaises(TypeError): + self.klass.from_file(fobj=f) + + def test_clear_cache(self): + with self.assertRaises(TypeError): + self.klass.clear_cache(["UTC"]) + + +class CCallingConventionTest(CallingConventionTest): + module = c_zoneinfo + + +class TzPathTest(TzPathUserMixin, ZoneInfoTestBase): + module = py_zoneinfo + + @staticmethod + @contextlib.contextmanager + def python_tzpath_context(value): + path_var = "PYTHONTZPATH" + unset_env_sentinel = object() + old_env = unset_env_sentinel + try: + with OS_ENV_LOCK: + old_env = os.environ.get(path_var, None) + os.environ[path_var] = value + yield + finally: + if old_env is unset_env_sentinel: + # In this case, `old_env` was never retrieved from the + # environment for whatever reason, so there's no need to + # reset the environment TZPATH. + pass + elif old_env is None: + del os.environ[path_var] + else: + os.environ[path_var] = old_env # pragma: nocover + + def test_env_variable(self): + """Tests that the environment variable works with reset_tzpath.""" + new_paths = [ + ("", []), + (f"{DRIVE}/etc/zoneinfo", [f"{DRIVE}/etc/zoneinfo"]), + (f"{DRIVE}/a/b/c{os.pathsep}{DRIVE}/d/e/f", [f"{DRIVE}/a/b/c", f"{DRIVE}/d/e/f"]), + ] + + for new_path_var, expected_result in new_paths: + with self.python_tzpath_context(new_path_var): + with self.subTest(tzpath=new_path_var): + self.module.reset_tzpath() + tzpath = self.module.TZPATH + self.assertSequenceEqual(tzpath, expected_result) + + def test_env_variable_relative_paths(self): + test_cases = [ + [("path/to/somewhere",), ()], + [ + (f"{DRIVE}/usr/share/zoneinfo", "path/to/somewhere",), + (f"{DRIVE}/usr/share/zoneinfo",), + ], + [("../relative/path",), ()], + [ + (f"{DRIVE}/usr/share/zoneinfo", "../relative/path",), + (f"{DRIVE}/usr/share/zoneinfo",), + ], + [("path/to/somewhere", "../relative/path",), ()], + [ + ( + f"{DRIVE}/usr/share/zoneinfo", + "path/to/somewhere", + "../relative/path", + ), + (f"{DRIVE}/usr/share/zoneinfo",), + ], + ] + + for input_paths, expected_paths in test_cases: + path_var = os.pathsep.join(input_paths) + with self.python_tzpath_context(path_var): + with self.subTest("warning", path_var=path_var): + # Note: Per PEP 615 the warning is implementation-defined + # behavior, other implementations need not warn. + with self.assertWarns(self.module.InvalidTZPathWarning) as w: + self.module.reset_tzpath() + self.assertEqual(w.warnings[0].filename, __file__) + + tzpath = self.module.TZPATH + with self.subTest("filtered", path_var=path_var): + self.assertSequenceEqual(tzpath, expected_paths) + + def test_env_variable_relative_paths_warning_location(self): + path_var = "path/to/somewhere" + + with self.python_tzpath_context(path_var): + with CleanImport("zoneinfo", "zoneinfo._tzpath"): + with self.assertWarns(RuntimeWarning) as w: + import zoneinfo + InvalidTZPathWarning = zoneinfo.InvalidTZPathWarning + self.assertIsInstance(w.warnings[0].message, InvalidTZPathWarning) + # It should represent the current file: + self.assertEqual(w.warnings[0].filename, __file__) + + def test_reset_tzpath_kwarg(self): + self.module.reset_tzpath(to=[f"{DRIVE}/a/b/c"]) + + self.assertSequenceEqual(self.module.TZPATH, (f"{DRIVE}/a/b/c",)) + + def test_reset_tzpath_relative_paths(self): + bad_values = [ + ("path/to/somewhere",), + ("/usr/share/zoneinfo", "path/to/somewhere",), + ("../relative/path",), + ("/usr/share/zoneinfo", "../relative/path",), + ("path/to/somewhere", "../relative/path",), + ("/usr/share/zoneinfo", "path/to/somewhere", "../relative/path",), + ] + for input_paths in bad_values: + with self.subTest(input_paths=input_paths): + with self.assertRaises(ValueError): + self.module.reset_tzpath(to=input_paths) + + def test_tzpath_type_error(self): + bad_values = [ + "/etc/zoneinfo:/usr/share/zoneinfo", + b"/etc/zoneinfo:/usr/share/zoneinfo", + 0, + ] + + for bad_value in bad_values: + with self.subTest(value=bad_value): + with self.assertRaises(TypeError): + self.module.reset_tzpath(bad_value) + + def test_tzpath_attribute(self): + tzpath_0 = [f"{DRIVE}/one", f"{DRIVE}/two"] + tzpath_1 = [f"{DRIVE}/three"] + + with self.tzpath_context(tzpath_0): + query_0 = self.module.TZPATH + + with self.tzpath_context(tzpath_1): + query_1 = self.module.TZPATH + + self.assertSequenceEqual(tzpath_0, query_0) + self.assertSequenceEqual(tzpath_1, query_1) + + +class CTzPathTest(TzPathTest): + module = c_zoneinfo + + +class TestModule(ZoneInfoTestBase): + module = py_zoneinfo + + @property + def zoneinfo_data(self): + return ZONEINFO_DATA + + @cached_property + def _UTC_bytes(self): + zone_file = self.zoneinfo_data.path_from_key("UTC") + with open(zone_file, "rb") as f: + return f.read() + + def touch_zone(self, key, tz_root): + """Creates a valid TZif file at key under the zoneinfo root tz_root. + + tz_root must exist, but all folders below that will be created. + """ + if not os.path.exists(tz_root): + raise FileNotFoundError(f"{tz_root} does not exist.") + + root_dir, *tail = key.rsplit("/", 1) + if tail: # If there's no tail, then the first component isn't a dir + os.makedirs(os.path.join(tz_root, root_dir), exist_ok=True) + + zonefile_path = os.path.join(tz_root, key) + with open(zonefile_path, "wb") as f: + f.write(self._UTC_bytes) + + def test_getattr_error(self): + with self.assertRaises(AttributeError): + self.module.NOATTRIBUTE + + def test_dir_contains_all(self): + """dir(self.module) should at least contain everything in __all__.""" + module_all_set = set(self.module.__all__) + module_dir_set = set(dir(self.module)) + + difference = module_all_set - module_dir_set + + self.assertFalse(difference) + + def test_dir_unique(self): + """Test that there are no duplicates in dir(self.module)""" + module_dir = dir(self.module) + module_unique = set(module_dir) + + self.assertCountEqual(module_dir, module_unique) + + def test_available_timezones(self): + with self.tzpath_context([self.zoneinfo_data.tzpath]): + self.assertTrue(self.zoneinfo_data.keys) # Sanity check + + available_keys = self.module.available_timezones() + zoneinfo_keys = set(self.zoneinfo_data.keys) + + # If tzdata is not present, zoneinfo_keys == available_keys, + # otherwise it should be a subset. + union = zoneinfo_keys & available_keys + self.assertEqual(zoneinfo_keys, union) + + def test_available_timezones_weirdzone(self): + with tempfile.TemporaryDirectory() as td: + # Make a fictional zone at "Mars/Olympus_Mons" + self.touch_zone("Mars/Olympus_Mons", td) + + with self.tzpath_context([td]): + available_keys = self.module.available_timezones() + self.assertIn("Mars/Olympus_Mons", available_keys) + + def test_folder_exclusions(self): + expected = { + "America/Los_Angeles", + "America/Santiago", + "America/Indiana/Indianapolis", + "UTC", + "Europe/Paris", + "Europe/London", + "Asia/Tokyo", + "Australia/Sydney", + } + + base_tree = list(expected) + posix_tree = [f"posix/{x}" for x in base_tree] + right_tree = [f"right/{x}" for x in base_tree] + + cases = [ + ("base_tree", base_tree), + ("base_and_posix", base_tree + posix_tree), + ("base_and_right", base_tree + right_tree), + ("all_trees", base_tree + right_tree + posix_tree), + ] + + with tempfile.TemporaryDirectory() as td: + for case_name, tree in cases: + tz_root = os.path.join(td, case_name) + os.mkdir(tz_root) + + for key in tree: + self.touch_zone(key, tz_root) + + with self.tzpath_context([tz_root]): + with self.subTest(case_name): + actual = self.module.available_timezones() + self.assertEqual(actual, expected) + + def test_exclude_posixrules(self): + expected = { + "America/New_York", + "Europe/London", + } + + tree = list(expected) + ["posixrules"] + + with tempfile.TemporaryDirectory() as td: + for key in tree: + self.touch_zone(key, td) + + with self.tzpath_context([td]): + actual = self.module.available_timezones() + self.assertEqual(actual, expected) + + +class CTestModule(TestModule): + module = c_zoneinfo + + +class ExtensionBuiltTest(unittest.TestCase): + """Smoke test to ensure that the C and Python extensions are both tested. + + Because the intention is for the Python and C versions of ZoneInfo to + behave identically, these tests necessarily rely on implementation details, + so the tests may need to be adjusted if the implementations change. Do not + rely on these tests as an indication of stable properties of these classes. + """ + + def test_cache_location(self): + # The pure Python version stores caches on attributes, but the C + # extension stores them in C globals (at least for now) + self.assertFalse(hasattr(c_zoneinfo.ZoneInfo, "_weak_cache")) + self.assertTrue(hasattr(py_zoneinfo.ZoneInfo, "_weak_cache")) + + def test_gc_tracked(self): + import gc + + self.assertTrue(gc.is_tracked(py_zoneinfo.ZoneInfo)) + self.assertTrue(gc.is_tracked(c_zoneinfo.ZoneInfo)) + + +@dataclasses.dataclass(frozen=True) +class ZoneOffset: + tzname: str + utcoffset: timedelta + dst: timedelta = ZERO + + +@dataclasses.dataclass(frozen=True) +class ZoneTransition: + transition: datetime + offset_before: ZoneOffset + offset_after: ZoneOffset + + @property + def transition_utc(self): + return (self.transition - self.offset_before.utcoffset).replace( + tzinfo=timezone.utc + ) + + @property + def fold(self): + """Whether this introduces a fold""" + return self.offset_before.utcoffset > self.offset_after.utcoffset + + @property + def gap(self): + """Whether this introduces a gap""" + return self.offset_before.utcoffset < self.offset_after.utcoffset + + @property + def delta(self): + return self.offset_after.utcoffset - self.offset_before.utcoffset + + @property + def anomaly_start(self): + if self.fold: + return self.transition + self.delta + else: + return self.transition + + @property + def anomaly_end(self): + if not self.fold: + return self.transition + self.delta + else: + return self.transition + + +class ZoneInfoData: + def __init__(self, source_json, tzpath, v1=False): + self.tzpath = pathlib.Path(tzpath) + self.keys = [] + self.v1 = v1 + self._populate_tzpath(source_json) + + def path_from_key(self, key): + return self.tzpath / key + + def _populate_tzpath(self, source_json): + with open(source_json, "rb") as f: + zoneinfo_dict = json.load(f) + + zoneinfo_data = zoneinfo_dict["data"] + + for key, value in zoneinfo_data.items(): + self.keys.append(key) + raw_data = self._decode_text(value) + + if self.v1: + data = self._convert_to_v1(raw_data) + else: + data = raw_data + + destination = self.path_from_key(key) + destination.parent.mkdir(exist_ok=True, parents=True) + with open(destination, "wb") as f: + f.write(data) + + def _decode_text(self, contents): + raw_data = b"".join(map(str.encode, contents)) + decoded = base64.b85decode(raw_data) + + return lzma.decompress(decoded) + + def _convert_to_v1(self, contents): + assert contents[0:4] == b"TZif", "Invalid TZif data found!" + version = int(contents[4:5]) + + header_start = 4 + 16 + header_end = header_start + 24 # 6l == 24 bytes + assert version >= 2, "Version 1 file found: no conversion necessary" + isutcnt, isstdcnt, leapcnt, timecnt, typecnt, charcnt = struct.unpack( + ">6l", contents[header_start:header_end] + ) + + file_size = ( + timecnt * 5 + + typecnt * 6 + + charcnt + + leapcnt * 8 + + isstdcnt + + isutcnt + ) + file_size += header_end + out = b"TZif" + b"\x00" + contents[5:file_size] + + assert ( + contents[file_size : (file_size + 4)] == b"TZif" + ), "Version 2 file not truncated at Version 2 header" + + return out + + +class ZoneDumpData: + @classmethod + def transition_keys(cls): + return cls._get_zonedump().keys() + + @classmethod + def load_transition_examples(cls, key): + return cls._get_zonedump()[key] + + @classmethod + def fixed_offset_zones(cls): + if not cls._FIXED_OFFSET_ZONES: + cls._populate_fixed_offsets() + + return cls._FIXED_OFFSET_ZONES.items() + + @classmethod + def _get_zonedump(cls): + if not cls._ZONEDUMP_DATA: + cls._populate_zonedump_data() + return cls._ZONEDUMP_DATA + + @classmethod + def _populate_fixed_offsets(cls): + cls._FIXED_OFFSET_ZONES = { + "UTC": ZoneOffset("UTC", ZERO, ZERO), + } + + @classmethod + def _populate_zonedump_data(cls): + def _Africa_Abidjan(): + LMT = ZoneOffset("LMT", timedelta(seconds=-968)) + GMT = ZoneOffset("GMT", ZERO) + + return [ + ZoneTransition(datetime(1912, 1, 1), LMT, GMT), + ] + + def _Africa_Casablanca(): + P00_s = ZoneOffset("+00", ZERO, ZERO) + P01_d = ZoneOffset("+01", ONE_H, ONE_H) + P00_d = ZoneOffset("+00", ZERO, -ONE_H) + P01_s = ZoneOffset("+01", ONE_H, ZERO) + + return [ + # Morocco sometimes pauses DST during Ramadan + ZoneTransition(datetime(2018, 3, 25, 2), P00_s, P01_d), + ZoneTransition(datetime(2018, 5, 13, 3), P01_d, P00_s), + ZoneTransition(datetime(2018, 6, 17, 2), P00_s, P01_d), + # On October 28th Morocco set standard time to +01, + # with negative DST only during Ramadan + ZoneTransition(datetime(2018, 10, 28, 3), P01_d, P01_s), + ZoneTransition(datetime(2019, 5, 5, 3), P01_s, P00_d), + ZoneTransition(datetime(2019, 6, 9, 2), P00_d, P01_s), + ] + + def _America_Los_Angeles(): + LMT = ZoneOffset("LMT", timedelta(seconds=-28378), ZERO) + PST = ZoneOffset("PST", timedelta(hours=-8), ZERO) + PDT = ZoneOffset("PDT", timedelta(hours=-7), ONE_H) + PWT = ZoneOffset("PWT", timedelta(hours=-7), ONE_H) + PPT = ZoneOffset("PPT", timedelta(hours=-7), ONE_H) + + return [ + ZoneTransition(datetime(1883, 11, 18, 12, 7, 2), LMT, PST), + ZoneTransition(datetime(1918, 3, 31, 2), PST, PDT), + ZoneTransition(datetime(1918, 3, 31, 2), PST, PDT), + ZoneTransition(datetime(1918, 10, 27, 2), PDT, PST), + # Transition to Pacific War Time + ZoneTransition(datetime(1942, 2, 9, 2), PST, PWT), + # Transition from Pacific War Time to Pacific Peace Time + ZoneTransition(datetime(1945, 8, 14, 16), PWT, PPT), + ZoneTransition(datetime(1945, 9, 30, 2), PPT, PST), + ZoneTransition(datetime(2015, 3, 8, 2), PST, PDT), + ZoneTransition(datetime(2015, 11, 1, 2), PDT, PST), + # After 2038: Rules continue indefinitely + ZoneTransition(datetime(2450, 3, 13, 2), PST, PDT), + ZoneTransition(datetime(2450, 11, 6, 2), PDT, PST), + ] + + def _America_Santiago(): + LMT = ZoneOffset("LMT", timedelta(seconds=-16966), ZERO) + SMT = ZoneOffset("SMT", timedelta(seconds=-16966), ZERO) + N05 = ZoneOffset("-05", timedelta(seconds=-18000), ZERO) + N04 = ZoneOffset("-04", timedelta(seconds=-14400), ZERO) + N03 = ZoneOffset("-03", timedelta(seconds=-10800), ONE_H) + + return [ + ZoneTransition(datetime(1890, 1, 1), LMT, SMT), + ZoneTransition(datetime(1910, 1, 10), SMT, N05), + ZoneTransition(datetime(1916, 7, 1), N05, SMT), + ZoneTransition(datetime(2008, 3, 30), N03, N04), + ZoneTransition(datetime(2008, 10, 12), N04, N03), + ZoneTransition(datetime(2040, 4, 8), N03, N04), + ZoneTransition(datetime(2040, 9, 2), N04, N03), + ] + + def _Asia_Tokyo(): + JST = ZoneOffset("JST", timedelta(seconds=32400), ZERO) + JDT = ZoneOffset("JDT", timedelta(seconds=36000), ONE_H) + + # Japan had DST from 1948 to 1951, and it was unusual in that + # the transition from DST to STD occurred at 25:00, and is + # denominated as such in the time zone database + return [ + ZoneTransition(datetime(1948, 5, 2), JST, JDT), + ZoneTransition(datetime(1948, 9, 12, 1), JDT, JST), + ZoneTransition(datetime(1951, 9, 9, 1), JDT, JST), + ] + + def _Australia_Sydney(): + LMT = ZoneOffset("LMT", timedelta(seconds=36292), ZERO) + AEST = ZoneOffset("AEST", timedelta(seconds=36000), ZERO) + AEDT = ZoneOffset("AEDT", timedelta(seconds=39600), ONE_H) + + return [ + ZoneTransition(datetime(1895, 2, 1), LMT, AEST), + ZoneTransition(datetime(1917, 1, 1, 0, 1), AEST, AEDT), + ZoneTransition(datetime(1917, 3, 25, 2), AEDT, AEST), + ZoneTransition(datetime(2012, 4, 1, 3), AEDT, AEST), + ZoneTransition(datetime(2012, 10, 7, 2), AEST, AEDT), + ZoneTransition(datetime(2040, 4, 1, 3), AEDT, AEST), + ZoneTransition(datetime(2040, 10, 7, 2), AEST, AEDT), + ] + + def _Europe_Dublin(): + LMT = ZoneOffset("LMT", timedelta(seconds=-1500), ZERO) + DMT = ZoneOffset("DMT", timedelta(seconds=-1521), ZERO) + IST_0 = ZoneOffset("IST", timedelta(seconds=2079), ONE_H) + GMT_0 = ZoneOffset("GMT", ZERO, ZERO) + BST = ZoneOffset("BST", ONE_H, ONE_H) + GMT_1 = ZoneOffset("GMT", ZERO, -ONE_H) + IST_1 = ZoneOffset("IST", ONE_H, ZERO) + + return [ + ZoneTransition(datetime(1880, 8, 2, 0), LMT, DMT), + ZoneTransition(datetime(1916, 5, 21, 2), DMT, IST_0), + ZoneTransition(datetime(1916, 10, 1, 3), IST_0, GMT_0), + ZoneTransition(datetime(1917, 4, 8, 2), GMT_0, BST), + ZoneTransition(datetime(2016, 3, 27, 1), GMT_1, IST_1), + ZoneTransition(datetime(2016, 10, 30, 2), IST_1, GMT_1), + ZoneTransition(datetime(2487, 3, 30, 1), GMT_1, IST_1), + ZoneTransition(datetime(2487, 10, 26, 2), IST_1, GMT_1), + ] + + def _Europe_Lisbon(): + WET = ZoneOffset("WET", ZERO, ZERO) + WEST = ZoneOffset("WEST", ONE_H, ONE_H) + CET = ZoneOffset("CET", ONE_H, ZERO) + CEST = ZoneOffset("CEST", timedelta(seconds=7200), ONE_H) + + return [ + ZoneTransition(datetime(1992, 3, 29, 1), WET, WEST), + ZoneTransition(datetime(1992, 9, 27, 2), WEST, CET), + ZoneTransition(datetime(1993, 3, 28, 2), CET, CEST), + ZoneTransition(datetime(1993, 9, 26, 3), CEST, CET), + ZoneTransition(datetime(1996, 3, 31, 2), CET, WEST), + ZoneTransition(datetime(1996, 10, 27, 2), WEST, WET), + ] + + def _Europe_London(): + LMT = ZoneOffset("LMT", timedelta(seconds=-75), ZERO) + GMT = ZoneOffset("GMT", ZERO, ZERO) + BST = ZoneOffset("BST", ONE_H, ONE_H) + + return [ + ZoneTransition(datetime(1847, 12, 1), LMT, GMT), + ZoneTransition(datetime(2005, 3, 27, 1), GMT, BST), + ZoneTransition(datetime(2005, 10, 30, 2), BST, GMT), + ZoneTransition(datetime(2043, 3, 29, 1), GMT, BST), + ZoneTransition(datetime(2043, 10, 25, 2), BST, GMT), + ] + + def _Pacific_Kiritimati(): + LMT = ZoneOffset("LMT", timedelta(seconds=-37760), ZERO) + N1040 = ZoneOffset("-1040", timedelta(seconds=-38400), ZERO) + N10 = ZoneOffset("-10", timedelta(seconds=-36000), ZERO) + P14 = ZoneOffset("+14", timedelta(seconds=50400), ZERO) + + # This is literally every transition in Christmas Island history + return [ + ZoneTransition(datetime(1901, 1, 1), LMT, N1040), + ZoneTransition(datetime(1979, 10, 1), N1040, N10), + # They skipped December 31, 1994 + ZoneTransition(datetime(1994, 12, 31), N10, P14), + ] + + cls._ZONEDUMP_DATA = { + "Africa/Abidjan": _Africa_Abidjan(), + "Africa/Casablanca": _Africa_Casablanca(), + "America/Los_Angeles": _America_Los_Angeles(), + "America/Santiago": _America_Santiago(), + "Australia/Sydney": _Australia_Sydney(), + "Asia/Tokyo": _Asia_Tokyo(), + "Europe/Dublin": _Europe_Dublin(), + "Europe/Lisbon": _Europe_Lisbon(), + "Europe/London": _Europe_London(), + "Pacific/Kiritimati": _Pacific_Kiritimati(), + } + + _ZONEDUMP_DATA = None + _FIXED_OFFSET_ZONES = None + + +if __name__ == '__main__': + unittest.main() diff --git a/Lib/test/test_zoneinfo/test_zoneinfo_property.py b/Lib/test/test_zoneinfo/test_zoneinfo_property.py new file mode 100644 index 0000000000..feaa77f3e7 --- /dev/null +++ b/Lib/test/test_zoneinfo/test_zoneinfo_property.py @@ -0,0 +1,368 @@ +import contextlib +import datetime +import os +import pickle +import unittest +import zoneinfo + +from test.support.hypothesis_helper import hypothesis + +import test.test_zoneinfo._support as test_support + +ZoneInfoTestBase = test_support.ZoneInfoTestBase + +py_zoneinfo, c_zoneinfo = test_support.get_modules() + +UTC = datetime.timezone.utc +MIN_UTC = datetime.datetime.min.replace(tzinfo=UTC) +MAX_UTC = datetime.datetime.max.replace(tzinfo=UTC) +ZERO = datetime.timedelta(0) + + +def _valid_keys(): + """Get available time zones, including posix/ and right/ directories.""" + from importlib import resources + + available_zones = sorted(zoneinfo.available_timezones()) + TZPATH = zoneinfo.TZPATH + + def valid_key(key): + for root in TZPATH: + key_file = os.path.join(root, key) + if os.path.exists(key_file): + return True + + components = key.split("/") + package_name = ".".join(["tzdata.zoneinfo"] + components[:-1]) + resource_name = components[-1] + + try: + return resources.files(package_name).joinpath(resource_name).is_file() + except ModuleNotFoundError: + return False + + # This relies on the fact that dictionaries maintain insertion order — for + # shrinking purposes, it is preferable to start with the standard version, + # then move to the posix/ version, then to the right/ version. + out_zones = {"": available_zones} + for prefix in ["posix", "right"]: + prefix_out = [] + for key in available_zones: + prefix_key = f"{prefix}/{key}" + if valid_key(prefix_key): + prefix_out.append(prefix_key) + + out_zones[prefix] = prefix_out + + output = [] + for keys in out_zones.values(): + output.extend(keys) + + return output + + +VALID_KEYS = _valid_keys() +if not VALID_KEYS: + raise unittest.SkipTest("No time zone data available") + + +def valid_keys(): + return hypothesis.strategies.sampled_from(VALID_KEYS) + + +KEY_EXAMPLES = [ + "Africa/Abidjan", + "Africa/Casablanca", + "America/Los_Angeles", + "America/Santiago", + "Asia/Tokyo", + "Australia/Sydney", + "Europe/Dublin", + "Europe/Lisbon", + "Europe/London", + "Pacific/Kiritimati", + "UTC", +] + + +def add_key_examples(f): + for key in KEY_EXAMPLES: + f = hypothesis.example(key)(f) + return f + + +class ZoneInfoTest(ZoneInfoTestBase): + module = py_zoneinfo + + @hypothesis.given(key=valid_keys()) + @add_key_examples + def test_str(self, key): + zi = self.klass(key) + self.assertEqual(str(zi), key) + + @hypothesis.given(key=valid_keys()) + @add_key_examples + def test_key(self, key): + zi = self.klass(key) + + self.assertEqual(zi.key, key) + + @hypothesis.given( + dt=hypothesis.strategies.one_of( + hypothesis.strategies.datetimes(), hypothesis.strategies.times() + ) + ) + @hypothesis.example(dt=datetime.datetime.min) + @hypothesis.example(dt=datetime.datetime.max) + @hypothesis.example(dt=datetime.datetime(1970, 1, 1)) + @hypothesis.example(dt=datetime.datetime(2039, 1, 1)) + @hypothesis.example(dt=datetime.time(0)) + @hypothesis.example(dt=datetime.time(12, 0)) + @hypothesis.example(dt=datetime.time(23, 59, 59, 999999)) + def test_utc(self, dt): + zi = self.klass("UTC") + dt_zi = dt.replace(tzinfo=zi) + + self.assertEqual(dt_zi.utcoffset(), ZERO) + self.assertEqual(dt_zi.dst(), ZERO) + self.assertEqual(dt_zi.tzname(), "UTC") + + +class CZoneInfoTest(ZoneInfoTest): + module = c_zoneinfo + + +class ZoneInfoPickleTest(ZoneInfoTestBase): + module = py_zoneinfo + + def setUp(self): + with contextlib.ExitStack() as stack: + stack.enter_context(test_support.set_zoneinfo_module(self.module)) + self.addCleanup(stack.pop_all().close) + + super().setUp() + + @hypothesis.given(key=valid_keys()) + @add_key_examples + def test_pickle_unpickle_cache(self, key): + zi = self.klass(key) + pkl_str = pickle.dumps(zi) + zi_rt = pickle.loads(pkl_str) + + self.assertIs(zi, zi_rt) + + @hypothesis.given(key=valid_keys()) + @add_key_examples + def test_pickle_unpickle_no_cache(self, key): + zi = self.klass.no_cache(key) + pkl_str = pickle.dumps(zi) + zi_rt = pickle.loads(pkl_str) + + self.assertIsNot(zi, zi_rt) + self.assertEqual(str(zi), str(zi_rt)) + + @hypothesis.given(key=valid_keys()) + @add_key_examples + def test_pickle_unpickle_cache_multiple_rounds(self, key): + """Test that pickle/unpickle is idempotent.""" + zi_0 = self.klass(key) + pkl_str_0 = pickle.dumps(zi_0) + zi_1 = pickle.loads(pkl_str_0) + pkl_str_1 = pickle.dumps(zi_1) + zi_2 = pickle.loads(pkl_str_1) + pkl_str_2 = pickle.dumps(zi_2) + + self.assertEqual(pkl_str_0, pkl_str_1) + self.assertEqual(pkl_str_1, pkl_str_2) + + self.assertIs(zi_0, zi_1) + self.assertIs(zi_0, zi_2) + self.assertIs(zi_1, zi_2) + + @hypothesis.given(key=valid_keys()) + @add_key_examples + def test_pickle_unpickle_no_cache_multiple_rounds(self, key): + """Test that pickle/unpickle is idempotent.""" + zi_cache = self.klass(key) + + zi_0 = self.klass.no_cache(key) + pkl_str_0 = pickle.dumps(zi_0) + zi_1 = pickle.loads(pkl_str_0) + pkl_str_1 = pickle.dumps(zi_1) + zi_2 = pickle.loads(pkl_str_1) + pkl_str_2 = pickle.dumps(zi_2) + + self.assertEqual(pkl_str_0, pkl_str_1) + self.assertEqual(pkl_str_1, pkl_str_2) + + self.assertIsNot(zi_0, zi_1) + self.assertIsNot(zi_0, zi_2) + self.assertIsNot(zi_1, zi_2) + + self.assertIsNot(zi_0, zi_cache) + self.assertIsNot(zi_1, zi_cache) + self.assertIsNot(zi_2, zi_cache) + + +class CZoneInfoPickleTest(ZoneInfoPickleTest): + module = c_zoneinfo + + +class ZoneInfoCacheTest(ZoneInfoTestBase): + module = py_zoneinfo + + @hypothesis.given(key=valid_keys()) + @add_key_examples + def test_cache(self, key): + zi_0 = self.klass(key) + zi_1 = self.klass(key) + + self.assertIs(zi_0, zi_1) + + @hypothesis.given(key=valid_keys()) + @add_key_examples + def test_no_cache(self, key): + zi_0 = self.klass.no_cache(key) + zi_1 = self.klass.no_cache(key) + + self.assertIsNot(zi_0, zi_1) + + +class CZoneInfoCacheTest(ZoneInfoCacheTest): + klass = c_zoneinfo.ZoneInfo + + +class PythonCConsistencyTest(unittest.TestCase): + """Tests that the C and Python versions do the same thing.""" + + def _is_ambiguous(self, dt): + return dt.replace(fold=not dt.fold).utcoffset() == dt.utcoffset() + + @hypothesis.given(dt=hypothesis.strategies.datetimes(), key=valid_keys()) + @hypothesis.example(dt=datetime.datetime.min, key="America/New_York") + @hypothesis.example(dt=datetime.datetime.max, key="America/New_York") + @hypothesis.example(dt=datetime.datetime(1970, 1, 1), key="America/New_York") + @hypothesis.example(dt=datetime.datetime(2020, 1, 1), key="Europe/Paris") + @hypothesis.example(dt=datetime.datetime(2020, 6, 1), key="Europe/Paris") + def test_same_str(self, dt, key): + py_dt = dt.replace(tzinfo=py_zoneinfo.ZoneInfo(key)) + c_dt = dt.replace(tzinfo=c_zoneinfo.ZoneInfo(key)) + + self.assertEqual(str(py_dt), str(c_dt)) + + @hypothesis.given(dt=hypothesis.strategies.datetimes(), key=valid_keys()) + @hypothesis.example(dt=datetime.datetime(1970, 1, 1), key="America/New_York") + @hypothesis.example(dt=datetime.datetime(2020, 2, 5), key="America/New_York") + @hypothesis.example(dt=datetime.datetime(2020, 8, 12), key="America/New_York") + @hypothesis.example(dt=datetime.datetime(2040, 1, 1), key="Africa/Casablanca") + @hypothesis.example(dt=datetime.datetime(1970, 1, 1), key="Europe/Paris") + @hypothesis.example(dt=datetime.datetime(2040, 1, 1), key="Europe/Paris") + @hypothesis.example(dt=datetime.datetime.min, key="Asia/Tokyo") + @hypothesis.example(dt=datetime.datetime.max, key="Asia/Tokyo") + def test_same_offsets_and_names(self, dt, key): + py_dt = dt.replace(tzinfo=py_zoneinfo.ZoneInfo(key)) + c_dt = dt.replace(tzinfo=c_zoneinfo.ZoneInfo(key)) + + self.assertEqual(py_dt.tzname(), c_dt.tzname()) + self.assertEqual(py_dt.utcoffset(), c_dt.utcoffset()) + self.assertEqual(py_dt.dst(), c_dt.dst()) + + @hypothesis.given( + dt=hypothesis.strategies.datetimes(timezones=hypothesis.strategies.just(UTC)), + key=valid_keys(), + ) + @hypothesis.example(dt=MIN_UTC, key="Asia/Tokyo") + @hypothesis.example(dt=MAX_UTC, key="Asia/Tokyo") + @hypothesis.example(dt=MIN_UTC, key="America/New_York") + @hypothesis.example(dt=MAX_UTC, key="America/New_York") + @hypothesis.example( + dt=datetime.datetime(2006, 10, 29, 5, 15, tzinfo=UTC), + key="America/New_York", + ) + def test_same_from_utc(self, dt, key): + py_zi = py_zoneinfo.ZoneInfo(key) + c_zi = c_zoneinfo.ZoneInfo(key) + + # Convert to UTC: This can overflow, but we just care about consistency + py_overflow_exc = None + c_overflow_exc = None + try: + py_dt = dt.astimezone(py_zi) + except OverflowError as e: + py_overflow_exc = e + + try: + c_dt = dt.astimezone(c_zi) + except OverflowError as e: + c_overflow_exc = e + + if (py_overflow_exc is not None) != (c_overflow_exc is not None): + raise py_overflow_exc or c_overflow_exc # pragma: nocover + + if py_overflow_exc is not None: + return # Consistently raises the same exception + + # PEP 495 says that an inter-zone comparison between ambiguous + # datetimes is always False. + if py_dt != c_dt: + self.assertEqual( + self._is_ambiguous(py_dt), + self._is_ambiguous(c_dt), + (py_dt, c_dt), + ) + + self.assertEqual(py_dt.tzname(), c_dt.tzname()) + self.assertEqual(py_dt.utcoffset(), c_dt.utcoffset()) + self.assertEqual(py_dt.dst(), c_dt.dst()) + + @hypothesis.given(dt=hypothesis.strategies.datetimes(), key=valid_keys()) + @hypothesis.example(dt=datetime.datetime.max, key="America/New_York") + @hypothesis.example(dt=datetime.datetime.min, key="America/New_York") + @hypothesis.example(dt=datetime.datetime.min, key="Asia/Tokyo") + @hypothesis.example(dt=datetime.datetime.max, key="Asia/Tokyo") + def test_same_to_utc(self, dt, key): + py_dt = dt.replace(tzinfo=py_zoneinfo.ZoneInfo(key)) + c_dt = dt.replace(tzinfo=c_zoneinfo.ZoneInfo(key)) + + # Convert from UTC: Overflow OK if it happens in both implementations + py_overflow_exc = None + c_overflow_exc = None + try: + py_utc = py_dt.astimezone(UTC) + except OverflowError as e: + py_overflow_exc = e + + try: + c_utc = c_dt.astimezone(UTC) + except OverflowError as e: + c_overflow_exc = e + + if (py_overflow_exc is not None) != (c_overflow_exc is not None): + raise py_overflow_exc or c_overflow_exc # pragma: nocover + + if py_overflow_exc is not None: + return # Consistently raises the same exception + + self.assertEqual(py_utc, c_utc) + + @hypothesis.given(key=valid_keys()) + @add_key_examples + def test_cross_module_pickle(self, key): + py_zi = py_zoneinfo.ZoneInfo(key) + c_zi = c_zoneinfo.ZoneInfo(key) + + with test_support.set_zoneinfo_module(py_zoneinfo): + py_pkl = pickle.dumps(py_zi) + + with test_support.set_zoneinfo_module(c_zoneinfo): + c_pkl = pickle.dumps(c_zi) + + with test_support.set_zoneinfo_module(c_zoneinfo): + # Python → C + py_to_c_zi = pickle.loads(py_pkl) + self.assertIs(py_to_c_zi, c_zi) + + with test_support.set_zoneinfo_module(py_zoneinfo): + # C → Python + c_to_py_zi = pickle.loads(c_pkl) + self.assertIs(c_to_py_zi, py_zi) diff --git a/Lib/zoneinfo/__init__.py b/Lib/zoneinfo/__init__.py new file mode 100644 index 0000000000..f5510ee049 --- /dev/null +++ b/Lib/zoneinfo/__init__.py @@ -0,0 +1,31 @@ +__all__ = [ + "ZoneInfo", + "reset_tzpath", + "available_timezones", + "TZPATH", + "ZoneInfoNotFoundError", + "InvalidTZPathWarning", +] + +from . import _tzpath +from ._common import ZoneInfoNotFoundError + +try: + from _zoneinfo import ZoneInfo +except ImportError: # pragma: nocover + from ._zoneinfo import ZoneInfo + +reset_tzpath = _tzpath.reset_tzpath +available_timezones = _tzpath.available_timezones +InvalidTZPathWarning = _tzpath.InvalidTZPathWarning + + +def __getattr__(name): + if name == "TZPATH": + return _tzpath.TZPATH + else: + raise AttributeError(f"module {__name__!r} has no attribute {name!r}") + + +def __dir__(): + return sorted(list(globals()) + ["TZPATH"]) diff --git a/Lib/zoneinfo/_common.py b/Lib/zoneinfo/_common.py new file mode 100644 index 0000000000..98cdfe37ca --- /dev/null +++ b/Lib/zoneinfo/_common.py @@ -0,0 +1,164 @@ +import struct + + +def load_tzdata(key): + from importlib import resources + + components = key.split("/") + package_name = ".".join(["tzdata.zoneinfo"] + components[:-1]) + resource_name = components[-1] + + try: + return resources.files(package_name).joinpath(resource_name).open("rb") + except (ImportError, FileNotFoundError, UnicodeEncodeError): + # There are three types of exception that can be raised that all amount + # to "we cannot find this key": + # + # ImportError: If package_name doesn't exist (e.g. if tzdata is not + # installed, or if there's an error in the folder name like + # Amrica/New_York) + # FileNotFoundError: If resource_name doesn't exist in the package + # (e.g. Europe/Krasnoy) + # UnicodeEncodeError: If package_name or resource_name are not UTF-8, + # such as keys containing a surrogate character. + raise ZoneInfoNotFoundError(f"No time zone found with key {key}") + + +def load_data(fobj): + header = _TZifHeader.from_file(fobj) + + if header.version == 1: + time_size = 4 + time_type = "l" + else: + # Version 2+ has 64-bit integer transition times + time_size = 8 + time_type = "q" + + # Version 2+ also starts with a Version 1 header and data, which + # we need to skip now + skip_bytes = ( + header.timecnt * 5 # Transition times and types + + header.typecnt * 6 # Local time type records + + header.charcnt # Time zone designations + + header.leapcnt * 8 # Leap second records + + header.isstdcnt # Standard/wall indicators + + header.isutcnt # UT/local indicators + ) + + fobj.seek(skip_bytes, 1) + + # Now we need to read the second header, which is not the same + # as the first + header = _TZifHeader.from_file(fobj) + + typecnt = header.typecnt + timecnt = header.timecnt + charcnt = header.charcnt + + # The data portion starts with timecnt transitions and indices + if timecnt: + trans_list_utc = struct.unpack( + f">{timecnt}{time_type}", fobj.read(timecnt * time_size) + ) + trans_idx = struct.unpack(f">{timecnt}B", fobj.read(timecnt)) + else: + trans_list_utc = () + trans_idx = () + + # Read the ttinfo struct, (utoff, isdst, abbrind) + if typecnt: + utcoff, isdst, abbrind = zip( + *(struct.unpack(">lbb", fobj.read(6)) for i in range(typecnt)) + ) + else: + utcoff = () + isdst = () + abbrind = () + + # Now read the abbreviations. They are null-terminated strings, indexed + # not by position in the array but by position in the unsplit + # abbreviation string. I suppose this makes more sense in C, which uses + # null to terminate the strings, but it's inconvenient here... + abbr_vals = {} + abbr_chars = fobj.read(charcnt) + + def get_abbr(idx): + # Gets a string starting at idx and running until the next \x00 + # + # We cannot pre-populate abbr_vals by splitting on \x00 because there + # are some zones that use subsets of longer abbreviations, like so: + # + # LMT\x00AHST\x00HDT\x00 + # + # Where the idx to abbr mapping should be: + # + # {0: "LMT", 4: "AHST", 5: "HST", 9: "HDT"} + if idx not in abbr_vals: + span_end = abbr_chars.find(b"\x00", idx) + abbr_vals[idx] = abbr_chars[idx:span_end].decode() + + return abbr_vals[idx] + + abbr = tuple(get_abbr(idx) for idx in abbrind) + + # The remainder of the file consists of leap seconds (currently unused) and + # the standard/wall and ut/local indicators, which are metadata we don't need. + # In version 2 files, we need to skip the unnecessary data to get at the TZ string: + if header.version >= 2: + # Each leap second record has size (time_size + 4) + skip_bytes = header.isutcnt + header.isstdcnt + header.leapcnt * 12 + fobj.seek(skip_bytes, 1) + + c = fobj.read(1) # Should be \n + assert c == b"\n", c + + tz_bytes = b"" + while (c := fobj.read(1)) != b"\n": + tz_bytes += c + + tz_str = tz_bytes + else: + tz_str = None + + return trans_idx, trans_list_utc, utcoff, isdst, abbr, tz_str + + +class _TZifHeader: + __slots__ = [ + "version", + "isutcnt", + "isstdcnt", + "leapcnt", + "timecnt", + "typecnt", + "charcnt", + ] + + def __init__(self, *args): + for attr, val in zip(self.__slots__, args, strict=True): + setattr(self, attr, val) + + @classmethod + def from_file(cls, stream): + # The header starts with a 4-byte "magic" value + if stream.read(4) != b"TZif": + raise ValueError("Invalid TZif file: magic not found") + + _version = stream.read(1) + if _version == b"\x00": + version = 1 + else: + version = int(_version) + stream.read(15) + + args = (version,) + + # Slots are defined in the order that the bytes are arranged + args = args + struct.unpack(">6l", stream.read(24)) + + return cls(*args) + + +class ZoneInfoNotFoundError(KeyError): + """Exception raised when a ZoneInfo key is not found.""" diff --git a/Lib/zoneinfo/_tzpath.py b/Lib/zoneinfo/_tzpath.py new file mode 100644 index 0000000000..5db17bea04 --- /dev/null +++ b/Lib/zoneinfo/_tzpath.py @@ -0,0 +1,181 @@ +import os +import sysconfig + + +def _reset_tzpath(to=None, stacklevel=4): + global TZPATH + + tzpaths = to + if tzpaths is not None: + if isinstance(tzpaths, (str, bytes)): + raise TypeError( + f"tzpaths must be a list or tuple, " + + f"not {type(tzpaths)}: {tzpaths!r}" + ) + + if not all(map(os.path.isabs, tzpaths)): + raise ValueError(_get_invalid_paths_message(tzpaths)) + base_tzpath = tzpaths + else: + env_var = os.environ.get("PYTHONTZPATH", None) + if env_var is None: + env_var = sysconfig.get_config_var("TZPATH") + base_tzpath = _parse_python_tzpath(env_var, stacklevel) + + TZPATH = tuple(base_tzpath) + + +def reset_tzpath(to=None): + """Reset global TZPATH.""" + # We need `_reset_tzpath` helper function because it produces a warning, + # it is used as both a module-level call and a public API. + # This is how we equalize the stacklevel for both calls. + _reset_tzpath(to) + + +def _parse_python_tzpath(env_var, stacklevel): + if not env_var: + return () + + raw_tzpath = env_var.split(os.pathsep) + new_tzpath = tuple(filter(os.path.isabs, raw_tzpath)) + + # If anything has been filtered out, we will warn about it + if len(new_tzpath) != len(raw_tzpath): + import warnings + + msg = _get_invalid_paths_message(raw_tzpath) + + warnings.warn( + "Invalid paths specified in PYTHONTZPATH environment variable. " + + msg, + InvalidTZPathWarning, + stacklevel=stacklevel, + ) + + return new_tzpath + + +def _get_invalid_paths_message(tzpaths): + invalid_paths = (path for path in tzpaths if not os.path.isabs(path)) + + prefix = "\n " + indented_str = prefix + prefix.join(invalid_paths) + + return ( + "Paths should be absolute but found the following relative paths:" + + indented_str + ) + + +def find_tzfile(key): + """Retrieve the path to a TZif file from a key.""" + _validate_tzfile_path(key) + for search_path in TZPATH: + filepath = os.path.join(search_path, key) + if os.path.isfile(filepath): + return filepath + + return None + + +_TEST_PATH = os.path.normpath(os.path.join("_", "_"))[:-1] + + +def _validate_tzfile_path(path, _base=_TEST_PATH): + if os.path.isabs(path): + raise ValueError( + f"ZoneInfo keys may not be absolute paths, got: {path}" + ) + + # We only care about the kinds of path normalizations that would change the + # length of the key - e.g. a/../b -> a/b, or a/b/ -> a/b. On Windows, + # normpath will also change from a/b to a\b, but that would still preserve + # the length. + new_path = os.path.normpath(path) + if len(new_path) != len(path): + raise ValueError( + f"ZoneInfo keys must be normalized relative paths, got: {path}" + ) + + resolved = os.path.normpath(os.path.join(_base, new_path)) + if not resolved.startswith(_base): + raise ValueError( + f"ZoneInfo keys must refer to subdirectories of TZPATH, got: {path}" + ) + + +del _TEST_PATH + + +def available_timezones(): + """Returns a set containing all available time zones. + + .. caution:: + + This may attempt to open a large number of files, since the best way to + determine if a given file on the time zone search path is to open it + and check for the "magic string" at the beginning. + """ + from importlib import resources + + valid_zones = set() + + # Start with loading from the tzdata package if it exists: this has a + # pre-assembled list of zones that only requires opening one file. + try: + with resources.files("tzdata").joinpath("zones").open("r") as f: + for zone in f: + zone = zone.strip() + if zone: + valid_zones.add(zone) + except (ImportError, FileNotFoundError): + pass + + def valid_key(fpath): + try: + with open(fpath, "rb") as f: + return f.read(4) == b"TZif" + except Exception: # pragma: nocover + return False + + for tz_root in TZPATH: + if not os.path.exists(tz_root): + continue + + for root, dirnames, files in os.walk(tz_root): + if root == tz_root: + # right/ and posix/ are special directories and shouldn't be + # included in the output of available zones + if "right" in dirnames: + dirnames.remove("right") + if "posix" in dirnames: + dirnames.remove("posix") + + for file in files: + fpath = os.path.join(root, file) + + key = os.path.relpath(fpath, start=tz_root) + if os.sep != "/": # pragma: nocover + key = key.replace(os.sep, "/") + + if not key or key in valid_zones: + continue + + if valid_key(fpath): + valid_zones.add(key) + + if "posixrules" in valid_zones: + # posixrules is a special symlink-only time zone where it exists, it + # should not be included in the output + valid_zones.remove("posixrules") + + return valid_zones + + +class InvalidTZPathWarning(RuntimeWarning): + """Warning raised if an invalid path is specified in PYTHONTZPATH.""" + + +TZPATH = () +_reset_tzpath(stacklevel=5) diff --git a/Lib/zoneinfo/_zoneinfo.py b/Lib/zoneinfo/_zoneinfo.py new file mode 100644 index 0000000000..b77dc0ed39 --- /dev/null +++ b/Lib/zoneinfo/_zoneinfo.py @@ -0,0 +1,772 @@ +import bisect +import calendar +import collections +import functools +import re +import weakref +from datetime import datetime, timedelta, tzinfo + +from . import _common, _tzpath + +EPOCH = datetime(1970, 1, 1) +EPOCHORDINAL = datetime(1970, 1, 1).toordinal() + +# It is relatively expensive to construct new timedelta objects, and in most +# cases we're looking at the same deltas, like integer numbers of hours, etc. +# To improve speed and memory use, we'll keep a dictionary with references +# to the ones we've already used so far. +# +# Loading every time zone in the 2020a version of the time zone database +# requires 447 timedeltas, which requires approximately the amount of space +# that ZoneInfo("America/New_York") with 236 transitions takes up, so we will +# set the cache size to 512 so that in the common case we always get cache +# hits, but specifically crafted ZoneInfo objects don't leak arbitrary amounts +# of memory. +@functools.lru_cache(maxsize=512) +def _load_timedelta(seconds): + return timedelta(seconds=seconds) + + +class ZoneInfo(tzinfo): + _strong_cache_size = 8 + _strong_cache = collections.OrderedDict() + _weak_cache = weakref.WeakValueDictionary() + __module__ = "zoneinfo" + + def __init_subclass__(cls): + cls._strong_cache = collections.OrderedDict() + cls._weak_cache = weakref.WeakValueDictionary() + + def __new__(cls, key): + instance = cls._weak_cache.get(key, None) + if instance is None: + instance = cls._weak_cache.setdefault(key, cls._new_instance(key)) + instance._from_cache = True + + # Update the "strong" cache + cls._strong_cache[key] = cls._strong_cache.pop(key, instance) + + if len(cls._strong_cache) > cls._strong_cache_size: + cls._strong_cache.popitem(last=False) + + return instance + + @classmethod + def no_cache(cls, key): + obj = cls._new_instance(key) + obj._from_cache = False + + return obj + + @classmethod + def _new_instance(cls, key): + obj = super().__new__(cls) + obj._key = key + obj._file_path = obj._find_tzfile(key) + + if obj._file_path is not None: + file_obj = open(obj._file_path, "rb") + else: + file_obj = _common.load_tzdata(key) + + with file_obj as f: + obj._load_file(f) + + return obj + + @classmethod + def from_file(cls, fobj, /, key=None): + obj = super().__new__(cls) + obj._key = key + obj._file_path = None + obj._load_file(fobj) + obj._file_repr = repr(fobj) + + # Disable pickling for objects created from files + obj.__reduce__ = obj._file_reduce + + return obj + + @classmethod + def clear_cache(cls, *, only_keys=None): + if only_keys is not None: + for key in only_keys: + cls._weak_cache.pop(key, None) + cls._strong_cache.pop(key, None) + + else: + cls._weak_cache.clear() + cls._strong_cache.clear() + + @property + def key(self): + return self._key + + def utcoffset(self, dt): + return self._find_trans(dt).utcoff + + def dst(self, dt): + return self._find_trans(dt).dstoff + + def tzname(self, dt): + return self._find_trans(dt).tzname + + def fromutc(self, dt): + """Convert from datetime in UTC to datetime in local time""" + + if not isinstance(dt, datetime): + raise TypeError("fromutc() requires a datetime argument") + if dt.tzinfo is not self: + raise ValueError("dt.tzinfo is not self") + + timestamp = self._get_local_timestamp(dt) + num_trans = len(self._trans_utc) + + if num_trans >= 1 and timestamp < self._trans_utc[0]: + tti = self._tti_before + fold = 0 + elif ( + num_trans == 0 or timestamp > self._trans_utc[-1] + ) and not isinstance(self._tz_after, _ttinfo): + tti, fold = self._tz_after.get_trans_info_fromutc( + timestamp, dt.year + ) + elif num_trans == 0: + tti = self._tz_after + fold = 0 + else: + idx = bisect.bisect_right(self._trans_utc, timestamp) + + if num_trans > 1 and timestamp >= self._trans_utc[1]: + tti_prev, tti = self._ttinfos[idx - 2 : idx] + elif timestamp > self._trans_utc[-1]: + tti_prev = self._ttinfos[-1] + tti = self._tz_after + else: + tti_prev = self._tti_before + tti = self._ttinfos[0] + + # Detect fold + shift = tti_prev.utcoff - tti.utcoff + fold = shift.total_seconds() > timestamp - self._trans_utc[idx - 1] + dt += tti.utcoff + if fold: + return dt.replace(fold=1) + else: + return dt + + def _find_trans(self, dt): + if dt is None: + if self._fixed_offset: + return self._tz_after + else: + return _NO_TTINFO + + ts = self._get_local_timestamp(dt) + + lt = self._trans_local[dt.fold] + + num_trans = len(lt) + + if num_trans and ts < lt[0]: + return self._tti_before + elif not num_trans or ts > lt[-1]: + if isinstance(self._tz_after, _TZStr): + return self._tz_after.get_trans_info(ts, dt.year, dt.fold) + else: + return self._tz_after + else: + # idx is the transition that occurs after this timestamp, so we + # subtract off 1 to get the current ttinfo + idx = bisect.bisect_right(lt, ts) - 1 + assert idx >= 0 + return self._ttinfos[idx] + + def _get_local_timestamp(self, dt): + return ( + (dt.toordinal() - EPOCHORDINAL) * 86400 + + dt.hour * 3600 + + dt.minute * 60 + + dt.second + ) + + def __str__(self): + if self._key is not None: + return f"{self._key}" + else: + return repr(self) + + def __repr__(self): + if self._key is not None: + return f"{self.__class__.__name__}(key={self._key!r})" + else: + return f"{self.__class__.__name__}.from_file({self._file_repr})" + + def __reduce__(self): + return (self.__class__._unpickle, (self._key, self._from_cache)) + + def _file_reduce(self): + import pickle + + raise pickle.PicklingError( + "Cannot pickle a ZoneInfo file created from a file stream." + ) + + @classmethod + def _unpickle(cls, key, from_cache, /): + if from_cache: + return cls(key) + else: + return cls.no_cache(key) + + def _find_tzfile(self, key): + return _tzpath.find_tzfile(key) + + def _load_file(self, fobj): + # Retrieve all the data as it exists in the zoneinfo file + trans_idx, trans_utc, utcoff, isdst, abbr, tz_str = _common.load_data( + fobj + ) + + # Infer the DST offsets (needed for .dst()) from the data + dstoff = self._utcoff_to_dstoff(trans_idx, utcoff, isdst) + + # Convert all the transition times (UTC) into "seconds since 1970-01-01 local time" + trans_local = self._ts_to_local(trans_idx, trans_utc, utcoff) + + # Construct `_ttinfo` objects for each transition in the file + _ttinfo_list = [ + _ttinfo( + _load_timedelta(utcoffset), _load_timedelta(dstoffset), tzname + ) + for utcoffset, dstoffset, tzname in zip(utcoff, dstoff, abbr) + ] + + self._trans_utc = trans_utc + self._trans_local = trans_local + self._ttinfos = [_ttinfo_list[idx] for idx in trans_idx] + + # Find the first non-DST transition + for i in range(len(isdst)): + if not isdst[i]: + self._tti_before = _ttinfo_list[i] + break + else: + if self._ttinfos: + self._tti_before = self._ttinfos[0] + else: + self._tti_before = None + + # Set the "fallback" time zone + if tz_str is not None and tz_str != b"": + self._tz_after = _parse_tz_str(tz_str.decode()) + else: + if not self._ttinfos and not _ttinfo_list: + raise ValueError("No time zone information found.") + + if self._ttinfos: + self._tz_after = self._ttinfos[-1] + else: + self._tz_after = _ttinfo_list[-1] + + # Determine if this is a "fixed offset" zone, meaning that the output + # of the utcoffset, dst and tzname functions does not depend on the + # specific datetime passed. + # + # We make three simplifying assumptions here: + # + # 1. If _tz_after is not a _ttinfo, it has transitions that might + # actually occur (it is possible to construct TZ strings that + # specify STD and DST but no transitions ever occur, such as + # AAA0BBB,0/0,J365/25). + # 2. If _ttinfo_list contains more than one _ttinfo object, the objects + # represent different offsets. + # 3. _ttinfo_list contains no unused _ttinfos (in which case an + # otherwise fixed-offset zone with extra _ttinfos defined may + # appear to *not* be a fixed offset zone). + # + # Violations to these assumptions would be fairly exotic, and exotic + # zones should almost certainly not be used with datetime.time (the + # only thing that would be affected by this). + if len(_ttinfo_list) > 1 or not isinstance(self._tz_after, _ttinfo): + self._fixed_offset = False + elif not _ttinfo_list: + self._fixed_offset = True + else: + self._fixed_offset = _ttinfo_list[0] == self._tz_after + + @staticmethod + def _utcoff_to_dstoff(trans_idx, utcoffsets, isdsts): + # Now we must transform our ttis and abbrs into `_ttinfo` objects, + # but there is an issue: .dst() must return a timedelta with the + # difference between utcoffset() and the "standard" offset, but + # the "base offset" and "DST offset" are not encoded in the file; + # we can infer what they are from the isdst flag, but it is not + # sufficient to just look at the last standard offset, because + # occasionally countries will shift both DST offset and base offset. + + typecnt = len(isdsts) + dstoffs = [0] * typecnt # Provisionally assign all to 0. + dst_cnt = sum(isdsts) + dst_found = 0 + + for i in range(1, len(trans_idx)): + if dst_cnt == dst_found: + break + + idx = trans_idx[i] + + dst = isdsts[idx] + + # We're only going to look at daylight saving time + if not dst: + continue + + # Skip any offsets that have already been assigned + if dstoffs[idx] != 0: + continue + + dstoff = 0 + utcoff = utcoffsets[idx] + + comp_idx = trans_idx[i - 1] + + if not isdsts[comp_idx]: + dstoff = utcoff - utcoffsets[comp_idx] + + if not dstoff and idx < (typecnt - 1): + comp_idx = trans_idx[i + 1] + + # If the following transition is also DST and we couldn't + # find the DST offset by this point, we're going to have to + # skip it and hope this transition gets assigned later + if isdsts[comp_idx]: + continue + + dstoff = utcoff - utcoffsets[comp_idx] + + if dstoff: + dst_found += 1 + dstoffs[idx] = dstoff + else: + # If we didn't find a valid value for a given index, we'll end up + # with dstoff = 0 for something where `isdst=1`. This is obviously + # wrong - one hour will be a much better guess than 0 + for idx in range(typecnt): + if not dstoffs[idx] and isdsts[idx]: + dstoffs[idx] = 3600 + + return dstoffs + + @staticmethod + def _ts_to_local(trans_idx, trans_list_utc, utcoffsets): + """Generate number of seconds since 1970 *in the local time*. + + This is necessary to easily find the transition times in local time""" + if not trans_list_utc: + return [[], []] + + # Start with the timestamps and modify in-place + trans_list_wall = [list(trans_list_utc), list(trans_list_utc)] + + if len(utcoffsets) > 1: + offset_0 = utcoffsets[0] + offset_1 = utcoffsets[trans_idx[0]] + if offset_1 > offset_0: + offset_1, offset_0 = offset_0, offset_1 + else: + offset_0 = offset_1 = utcoffsets[0] + + trans_list_wall[0][0] += offset_0 + trans_list_wall[1][0] += offset_1 + + for i in range(1, len(trans_idx)): + offset_0 = utcoffsets[trans_idx[i - 1]] + offset_1 = utcoffsets[trans_idx[i]] + + if offset_1 > offset_0: + offset_1, offset_0 = offset_0, offset_1 + + trans_list_wall[0][i] += offset_0 + trans_list_wall[1][i] += offset_1 + + return trans_list_wall + + +class _ttinfo: + __slots__ = ["utcoff", "dstoff", "tzname"] + + def __init__(self, utcoff, dstoff, tzname): + self.utcoff = utcoff + self.dstoff = dstoff + self.tzname = tzname + + def __eq__(self, other): + return ( + self.utcoff == other.utcoff + and self.dstoff == other.dstoff + and self.tzname == other.tzname + ) + + def __repr__(self): # pragma: nocover + return ( + f"{self.__class__.__name__}" + + f"({self.utcoff}, {self.dstoff}, {self.tzname})" + ) + + +_NO_TTINFO = _ttinfo(None, None, None) + + +class _TZStr: + __slots__ = ( + "std", + "dst", + "start", + "end", + "get_trans_info", + "get_trans_info_fromutc", + "dst_diff", + ) + + def __init__( + self, std_abbr, std_offset, dst_abbr, dst_offset, start=None, end=None + ): + self.dst_diff = dst_offset - std_offset + std_offset = _load_timedelta(std_offset) + self.std = _ttinfo( + utcoff=std_offset, dstoff=_load_timedelta(0), tzname=std_abbr + ) + + self.start = start + self.end = end + + dst_offset = _load_timedelta(dst_offset) + delta = _load_timedelta(self.dst_diff) + self.dst = _ttinfo(utcoff=dst_offset, dstoff=delta, tzname=dst_abbr) + + # These are assertions because the constructor should only be called + # by functions that would fail before passing start or end + assert start is not None, "No transition start specified" + assert end is not None, "No transition end specified" + + self.get_trans_info = self._get_trans_info + self.get_trans_info_fromutc = self._get_trans_info_fromutc + + def transitions(self, year): + start = self.start.year_to_epoch(year) + end = self.end.year_to_epoch(year) + return start, end + + def _get_trans_info(self, ts, year, fold): + """Get the information about the current transition - tti""" + start, end = self.transitions(year) + + # With fold = 0, the period (denominated in local time) with the + # smaller offset starts at the end of the gap and ends at the end of + # the fold; with fold = 1, it runs from the start of the gap to the + # beginning of the fold. + # + # So in order to determine the DST boundaries we need to know both + # the fold and whether DST is positive or negative (rare), and it + # turns out that this boils down to fold XOR is_positive. + if fold == (self.dst_diff >= 0): + end -= self.dst_diff + else: + start += self.dst_diff + + if start < end: + isdst = start <= ts < end + else: + isdst = not (end <= ts < start) + + return self.dst if isdst else self.std + + def _get_trans_info_fromutc(self, ts, year): + start, end = self.transitions(year) + start -= self.std.utcoff.total_seconds() + end -= self.dst.utcoff.total_seconds() + + if start < end: + isdst = start <= ts < end + else: + isdst = not (end <= ts < start) + + # For positive DST, the ambiguous period is one dst_diff after the end + # of DST; for negative DST, the ambiguous period is one dst_diff before + # the start of DST. + if self.dst_diff > 0: + ambig_start = end + ambig_end = end + self.dst_diff + else: + ambig_start = start + ambig_end = start - self.dst_diff + + fold = ambig_start <= ts < ambig_end + + return (self.dst if isdst else self.std, fold) + + +def _post_epoch_days_before_year(year): + """Get the number of days between 1970-01-01 and YEAR-01-01""" + y = year - 1 + return y * 365 + y // 4 - y // 100 + y // 400 - EPOCHORDINAL + + +class _DayOffset: + __slots__ = ["d", "julian", "hour", "minute", "second"] + + def __init__(self, d, julian, hour=2, minute=0, second=0): + min_day = 0 + julian # convert bool to int + if not min_day <= d <= 365: + raise ValueError(f"d must be in [{min_day}, 365], not: {d}") + + self.d = d + self.julian = julian + self.hour = hour + self.minute = minute + self.second = second + + def year_to_epoch(self, year): + days_before_year = _post_epoch_days_before_year(year) + + d = self.d + if self.julian and d >= 59 and calendar.isleap(year): + d += 1 + + epoch = (days_before_year + d) * 86400 + epoch += self.hour * 3600 + self.minute * 60 + self.second + + return epoch + + +class _CalendarOffset: + __slots__ = ["m", "w", "d", "hour", "minute", "second"] + + _DAYS_BEFORE_MONTH = ( + -1, + 0, + 31, + 59, + 90, + 120, + 151, + 181, + 212, + 243, + 273, + 304, + 334, + ) + + def __init__(self, m, w, d, hour=2, minute=0, second=0): + if not 1 <= m <= 12: + raise ValueError("m must be in [1, 12]") + + if not 1 <= w <= 5: + raise ValueError("w must be in [1, 5]") + + if not 0 <= d <= 6: + raise ValueError("d must be in [0, 6]") + + self.m = m + self.w = w + self.d = d + self.hour = hour + self.minute = minute + self.second = second + + @classmethod + def _ymd2ord(cls, year, month, day): + return ( + _post_epoch_days_before_year(year) + + cls._DAYS_BEFORE_MONTH[month] + + (month > 2 and calendar.isleap(year)) + + day + ) + + # TODO: These are not actually epoch dates as they are expressed in local time + def year_to_epoch(self, year): + """Calculates the datetime of the occurrence from the year""" + # We know year and month, we need to convert w, d into day of month + # + # Week 1 is the first week in which day `d` (where 0 = Sunday) appears. + # Week 5 represents the last occurrence of day `d`, so we need to know + # the range of the month. + first_day, days_in_month = calendar.monthrange(year, self.m) + + # This equation seems magical, so I'll break it down: + # 1. calendar says 0 = Monday, POSIX says 0 = Sunday + # so we need first_day + 1 to get 1 = Monday -> 7 = Sunday, + # which is still equivalent because this math is mod 7 + # 2. Get first day - desired day mod 7: -1 % 7 = 6, so we don't need + # to do anything to adjust negative numbers. + # 3. Add 1 because month days are a 1-based index. + month_day = (self.d - (first_day + 1)) % 7 + 1 + + # Now use a 0-based index version of `w` to calculate the w-th + # occurrence of `d` + month_day += (self.w - 1) * 7 + + # month_day will only be > days_in_month if w was 5, and `w` means + # "last occurrence of `d`", so now we just check if we over-shot the + # end of the month and if so knock off 1 week. + if month_day > days_in_month: + month_day -= 7 + + ordinal = self._ymd2ord(year, self.m, month_day) + epoch = ordinal * 86400 + epoch += self.hour * 3600 + self.minute * 60 + self.second + return epoch + + +def _parse_tz_str(tz_str): + # The tz string has the format: + # + # std[offset[dst[offset],start[/time],end[/time]]] + # + # std and dst must be 3 or more characters long and must not contain + # a leading colon, embedded digits, commas, nor a plus or minus signs; + # The spaces between "std" and "offset" are only for display and are + # not actually present in the string. + # + # The format of the offset is ``[+|-]hh[:mm[:ss]]`` + + offset_str, *start_end_str = tz_str.split(",", 1) + + parser_re = re.compile( + r""" + (?P[^<0-9:.+-]+|<[a-zA-Z0-9+-]+>) + (?: + (?P[+-]?\d{1,3}(?::\d{2}(?::\d{2})?)?) + (?: + (?P[^0-9:.+-]+|<[a-zA-Z0-9+-]+>) + (?P[+-]?\d{1,3}(?::\d{2}(?::\d{2})?)?)? + )? # dst + )? # stdoff + """, + re.ASCII|re.VERBOSE + ) + + m = parser_re.fullmatch(offset_str) + + if m is None: + raise ValueError(f"{tz_str} is not a valid TZ string") + + std_abbr = m.group("std") + dst_abbr = m.group("dst") + dst_offset = None + + std_abbr = std_abbr.strip("<>") + + if dst_abbr: + dst_abbr = dst_abbr.strip("<>") + + if std_offset := m.group("stdoff"): + try: + std_offset = _parse_tz_delta(std_offset) + except ValueError as e: + raise ValueError(f"Invalid STD offset in {tz_str}") from e + else: + std_offset = 0 + + if dst_abbr is not None: + if dst_offset := m.group("dstoff"): + try: + dst_offset = _parse_tz_delta(dst_offset) + except ValueError as e: + raise ValueError(f"Invalid DST offset in {tz_str}") from e + else: + dst_offset = std_offset + 3600 + + if not start_end_str: + raise ValueError(f"Missing transition rules: {tz_str}") + + start_end_strs = start_end_str[0].split(",", 1) + try: + start, end = (_parse_dst_start_end(x) for x in start_end_strs) + except ValueError as e: + raise ValueError(f"Invalid TZ string: {tz_str}") from e + + return _TZStr(std_abbr, std_offset, dst_abbr, dst_offset, start, end) + elif start_end_str: + raise ValueError(f"Transition rule present without DST: {tz_str}") + else: + # This is a static ttinfo, don't return _TZStr + return _ttinfo( + _load_timedelta(std_offset), _load_timedelta(0), std_abbr + ) + + +def _parse_dst_start_end(dststr): + date, *time = dststr.split("/", 1) + type = date[:1] + if type == "M": + n_is_julian = False + m = re.fullmatch(r"M(\d{1,2})\.(\d).(\d)", date, re.ASCII) + if m is None: + raise ValueError(f"Invalid dst start/end date: {dststr}") + date_offset = tuple(map(int, m.groups())) + offset = _CalendarOffset(*date_offset) + else: + if type == "J": + n_is_julian = True + date = date[1:] + else: + n_is_julian = False + + doy = int(date) + offset = _DayOffset(doy, n_is_julian) + + if time: + offset.hour, offset.minute, offset.second = _parse_transition_time(time[0]) + + return offset + + +def _parse_transition_time(time_str): + match = re.fullmatch( + r"(?P[+-])?(?P\d{1,3})(:(?P\d{2})(:(?P\d{2}))?)?", + time_str, + re.ASCII + ) + if match is None: + raise ValueError(f"Invalid time: {time_str}") + + h, m, s = (int(v or 0) for v in match.group("h", "m", "s")) + + if h > 167: + raise ValueError( + f"Hour must be in [0, 167]: {time_str}" + ) + + if match.group("sign") == "-": + h, m, s = -h, -m, -s + + return h, m, s + + +def _parse_tz_delta(tz_delta): + match = re.fullmatch( + r"(?P[+-])?(?P\d{1,3})(:(?P\d{2})(:(?P\d{2}))?)?", + tz_delta, + re.ASCII + ) + # Anything passed to this function should already have hit an equivalent + # regular expression to find the section to parse. + assert match is not None, tz_delta + + h, m, s = (int(v or 0) for v in match.group("h", "m", "s")) + + total = h * 3600 + m * 60 + s + + if h > 24: + raise ValueError( + f"Offset hours must be in [0, 24]: {tz_delta}" + ) + + # Yes, +5 maps to an offset of -5h + if match.group("sign") != "-": + total = -total + + return total