diff options
Diffstat (limited to 'third_party/python/setuptools/pkg_resources/_vendor/packaging')
17 files changed, 2770 insertions, 1619 deletions
diff --git a/third_party/python/setuptools/pkg_resources/_vendor/packaging/__about__.py b/third_party/python/setuptools/pkg_resources/_vendor/packaging/__about__.py deleted file mode 100644 index 4d998578d7..0000000000 --- a/third_party/python/setuptools/pkg_resources/_vendor/packaging/__about__.py +++ /dev/null @@ -1,27 +0,0 @@ -# This file is dual licensed under the terms of the Apache License, Version -# 2.0, and the BSD License. See the LICENSE file in the root of this repository -# for complete details. -from __future__ import absolute_import, division, print_function - -__all__ = [ - "__title__", - "__summary__", - "__uri__", - "__version__", - "__author__", - "__email__", - "__license__", - "__copyright__", -] - -__title__ = "packaging" -__summary__ = "Core utilities for Python packages" -__uri__ = "https://github.com/pypa/packaging" - -__version__ = "20.4" - -__author__ = "Donald Stufft and individual contributors" -__email__ = "donald@stufft.io" - -__license__ = "BSD-2-Clause or Apache-2.0" -__copyright__ = "Copyright 2014-2019 %s" % __author__ diff --git a/third_party/python/setuptools/pkg_resources/_vendor/packaging/__init__.py b/third_party/python/setuptools/pkg_resources/_vendor/packaging/__init__.py index a0cf67df52..13cadc7f04 100644 --- a/third_party/python/setuptools/pkg_resources/_vendor/packaging/__init__.py +++ b/third_party/python/setuptools/pkg_resources/_vendor/packaging/__init__.py @@ -1,26 +1,15 @@ # This file is dual licensed under the terms of the Apache License, Version # 2.0, and the BSD License. See the LICENSE file in the root of this repository # for complete details. -from __future__ import absolute_import, division, print_function -from .__about__ import ( - __author__, - __copyright__, - __email__, - __license__, - __summary__, - __title__, - __uri__, - __version__, -) +__title__ = "packaging" +__summary__ = "Core utilities for Python packages" +__uri__ = "https://github.com/pypa/packaging" -__all__ = [ - "__title__", - "__summary__", - "__uri__", - "__version__", - "__author__", - "__email__", - "__license__", - "__copyright__", -] +__version__ = "23.1" + +__author__ = "Donald Stufft and individual contributors" +__email__ = "donald@stufft.io" + +__license__ = "BSD-2-Clause or Apache-2.0" +__copyright__ = "2014-2019 %s" % __author__ diff --git a/third_party/python/setuptools/pkg_resources/_vendor/packaging/_compat.py b/third_party/python/setuptools/pkg_resources/_vendor/packaging/_compat.py deleted file mode 100644 index e54bd4ede8..0000000000 --- a/third_party/python/setuptools/pkg_resources/_vendor/packaging/_compat.py +++ /dev/null @@ -1,38 +0,0 @@ -# This file is dual licensed under the terms of the Apache License, Version -# 2.0, and the BSD License. See the LICENSE file in the root of this repository -# for complete details. -from __future__ import absolute_import, division, print_function - -import sys - -from ._typing import TYPE_CHECKING - -if TYPE_CHECKING: # pragma: no cover - from typing import Any, Dict, Tuple, Type - - -PY2 = sys.version_info[0] == 2 -PY3 = sys.version_info[0] == 3 - -# flake8: noqa - -if PY3: - string_types = (str,) -else: - string_types = (basestring,) - - -def with_metaclass(meta, *bases): - # type: (Type[Any], Tuple[Type[Any], ...]) -> Any - """ - Create a base class with a metaclass. - """ - # This requires a bit of explanation: the basic idea is to make a dummy - # metaclass for one level of class instantiation that replaces itself with - # the actual metaclass. - class metaclass(meta): # type: ignore - def __new__(cls, name, this_bases, d): - # type: (Type[Any], str, Tuple[Any], Dict[Any, Any]) -> Any - return meta(name, bases, d) - - return type.__new__(metaclass, "temporary_class", (), {}) diff --git a/third_party/python/setuptools/pkg_resources/_vendor/packaging/_elffile.py b/third_party/python/setuptools/pkg_resources/_vendor/packaging/_elffile.py new file mode 100644 index 0000000000..6fb19b30bb --- /dev/null +++ b/third_party/python/setuptools/pkg_resources/_vendor/packaging/_elffile.py @@ -0,0 +1,108 @@ +""" +ELF file parser. + +This provides a class ``ELFFile`` that parses an ELF executable in a similar +interface to ``ZipFile``. Only the read interface is implemented. + +Based on: https://gist.github.com/lyssdod/f51579ae8d93c8657a5564aefc2ffbca +ELF header: https://refspecs.linuxfoundation.org/elf/gabi4+/ch4.eheader.html +""" + +import enum +import os +import struct +from typing import IO, Optional, Tuple + + +class ELFInvalid(ValueError): + pass + + +class EIClass(enum.IntEnum): + C32 = 1 + C64 = 2 + + +class EIData(enum.IntEnum): + Lsb = 1 + Msb = 2 + + +class EMachine(enum.IntEnum): + I386 = 3 + S390 = 22 + Arm = 40 + X8664 = 62 + AArc64 = 183 + + +class ELFFile: + """ + Representation of an ELF executable. + """ + + def __init__(self, f: IO[bytes]) -> None: + self._f = f + + try: + ident = self._read("16B") + except struct.error: + raise ELFInvalid("unable to parse identification") + magic = bytes(ident[:4]) + if magic != b"\x7fELF": + raise ELFInvalid(f"invalid magic: {magic!r}") + + self.capacity = ident[4] # Format for program header (bitness). + self.encoding = ident[5] # Data structure encoding (endianness). + + try: + # e_fmt: Format for program header. + # p_fmt: Format for section header. + # p_idx: Indexes to find p_type, p_offset, and p_filesz. + e_fmt, self._p_fmt, self._p_idx = { + (1, 1): ("<HHIIIIIHHH", "<IIIIIIII", (0, 1, 4)), # 32-bit LSB. + (1, 2): (">HHIIIIIHHH", ">IIIIIIII", (0, 1, 4)), # 32-bit MSB. + (2, 1): ("<HHIQQQIHHH", "<IIQQQQQQ", (0, 2, 5)), # 64-bit LSB. + (2, 2): (">HHIQQQIHHH", ">IIQQQQQQ", (0, 2, 5)), # 64-bit MSB. + }[(self.capacity, self.encoding)] + except KeyError: + raise ELFInvalid( + f"unrecognized capacity ({self.capacity}) or " + f"encoding ({self.encoding})" + ) + + try: + ( + _, + self.machine, # Architecture type. + _, + _, + self._e_phoff, # Offset of program header. + _, + self.flags, # Processor-specific flags. + _, + self._e_phentsize, # Size of section. + self._e_phnum, # Number of sections. + ) = self._read(e_fmt) + except struct.error as e: + raise ELFInvalid("unable to parse machine and section information") from e + + def _read(self, fmt: str) -> Tuple[int, ...]: + return struct.unpack(fmt, self._f.read(struct.calcsize(fmt))) + + @property + def interpreter(self) -> Optional[str]: + """ + The path recorded in the ``PT_INTERP`` section header. + """ + for index in range(self._e_phnum): + self._f.seek(self._e_phoff + self._e_phentsize * index) + try: + data = self._read(self._p_fmt) + except struct.error: + continue + if data[self._p_idx[0]] != 3: # Not PT_INTERP. + continue + self._f.seek(data[self._p_idx[1]]) + return os.fsdecode(self._f.read(data[self._p_idx[2]])).strip("\0") + return None diff --git a/third_party/python/setuptools/pkg_resources/_vendor/packaging/_manylinux.py b/third_party/python/setuptools/pkg_resources/_vendor/packaging/_manylinux.py new file mode 100644 index 0000000000..449c655be6 --- /dev/null +++ b/third_party/python/setuptools/pkg_resources/_vendor/packaging/_manylinux.py @@ -0,0 +1,240 @@ +import collections +import contextlib +import functools +import os +import re +import sys +import warnings +from typing import Dict, Generator, Iterator, NamedTuple, Optional, Tuple + +from ._elffile import EIClass, EIData, ELFFile, EMachine + +EF_ARM_ABIMASK = 0xFF000000 +EF_ARM_ABI_VER5 = 0x05000000 +EF_ARM_ABI_FLOAT_HARD = 0x00000400 + + +# `os.PathLike` not a generic type until Python 3.9, so sticking with `str` +# as the type for `path` until then. +@contextlib.contextmanager +def _parse_elf(path: str) -> Generator[Optional[ELFFile], None, None]: + try: + with open(path, "rb") as f: + yield ELFFile(f) + except (OSError, TypeError, ValueError): + yield None + + +def _is_linux_armhf(executable: str) -> bool: + # hard-float ABI can be detected from the ELF header of the running + # process + # https://static.docs.arm.com/ihi0044/g/aaelf32.pdf + with _parse_elf(executable) as f: + return ( + f is not None + and f.capacity == EIClass.C32 + and f.encoding == EIData.Lsb + and f.machine == EMachine.Arm + and f.flags & EF_ARM_ABIMASK == EF_ARM_ABI_VER5 + and f.flags & EF_ARM_ABI_FLOAT_HARD == EF_ARM_ABI_FLOAT_HARD + ) + + +def _is_linux_i686(executable: str) -> bool: + with _parse_elf(executable) as f: + return ( + f is not None + and f.capacity == EIClass.C32 + and f.encoding == EIData.Lsb + and f.machine == EMachine.I386 + ) + + +def _have_compatible_abi(executable: str, arch: str) -> bool: + if arch == "armv7l": + return _is_linux_armhf(executable) + if arch == "i686": + return _is_linux_i686(executable) + return arch in {"x86_64", "aarch64", "ppc64", "ppc64le", "s390x"} + + +# If glibc ever changes its major version, we need to know what the last +# minor version was, so we can build the complete list of all versions. +# For now, guess what the highest minor version might be, assume it will +# be 50 for testing. Once this actually happens, update the dictionary +# with the actual value. +_LAST_GLIBC_MINOR: Dict[int, int] = collections.defaultdict(lambda: 50) + + +class _GLibCVersion(NamedTuple): + major: int + minor: int + + +def _glibc_version_string_confstr() -> Optional[str]: + """ + Primary implementation of glibc_version_string using os.confstr. + """ + # os.confstr is quite a bit faster than ctypes.DLL. It's also less likely + # to be broken or missing. This strategy is used in the standard library + # platform module. + # https://github.com/python/cpython/blob/fcf1d003bf4f0100c/Lib/platform.py#L175-L183 + try: + # Should be a string like "glibc 2.17". + version_string: str = getattr(os, "confstr")("CS_GNU_LIBC_VERSION") + assert version_string is not None + _, version = version_string.rsplit() + except (AssertionError, AttributeError, OSError, ValueError): + # os.confstr() or CS_GNU_LIBC_VERSION not available (or a bad value)... + return None + return version + + +def _glibc_version_string_ctypes() -> Optional[str]: + """ + Fallback implementation of glibc_version_string using ctypes. + """ + try: + import ctypes + except ImportError: + return None + + # ctypes.CDLL(None) internally calls dlopen(NULL), and as the dlopen + # manpage says, "If filename is NULL, then the returned handle is for the + # main program". This way we can let the linker do the work to figure out + # which libc our process is actually using. + # + # We must also handle the special case where the executable is not a + # dynamically linked executable. This can occur when using musl libc, + # for example. In this situation, dlopen() will error, leading to an + # OSError. Interestingly, at least in the case of musl, there is no + # errno set on the OSError. The single string argument used to construct + # OSError comes from libc itself and is therefore not portable to + # hard code here. In any case, failure to call dlopen() means we + # can proceed, so we bail on our attempt. + try: + process_namespace = ctypes.CDLL(None) + except OSError: + return None + + try: + gnu_get_libc_version = process_namespace.gnu_get_libc_version + except AttributeError: + # Symbol doesn't exist -> therefore, we are not linked to + # glibc. + return None + + # Call gnu_get_libc_version, which returns a string like "2.5" + gnu_get_libc_version.restype = ctypes.c_char_p + version_str: str = gnu_get_libc_version() + # py2 / py3 compatibility: + if not isinstance(version_str, str): + version_str = version_str.decode("ascii") + + return version_str + + +def _glibc_version_string() -> Optional[str]: + """Returns glibc version string, or None if not using glibc.""" + return _glibc_version_string_confstr() or _glibc_version_string_ctypes() + + +def _parse_glibc_version(version_str: str) -> Tuple[int, int]: + """Parse glibc version. + + We use a regexp instead of str.split because we want to discard any + random junk that might come after the minor version -- this might happen + in patched/forked versions of glibc (e.g. Linaro's version of glibc + uses version strings like "2.20-2014.11"). See gh-3588. + """ + m = re.match(r"(?P<major>[0-9]+)\.(?P<minor>[0-9]+)", version_str) + if not m: + warnings.warn( + f"Expected glibc version with 2 components major.minor," + f" got: {version_str}", + RuntimeWarning, + ) + return -1, -1 + return int(m.group("major")), int(m.group("minor")) + + +@functools.lru_cache() +def _get_glibc_version() -> Tuple[int, int]: + version_str = _glibc_version_string() + if version_str is None: + return (-1, -1) + return _parse_glibc_version(version_str) + + +# From PEP 513, PEP 600 +def _is_compatible(name: str, arch: str, version: _GLibCVersion) -> bool: + sys_glibc = _get_glibc_version() + if sys_glibc < version: + return False + # Check for presence of _manylinux module. + try: + import _manylinux # noqa + except ImportError: + return True + if hasattr(_manylinux, "manylinux_compatible"): + result = _manylinux.manylinux_compatible(version[0], version[1], arch) + if result is not None: + return bool(result) + return True + if version == _GLibCVersion(2, 5): + if hasattr(_manylinux, "manylinux1_compatible"): + return bool(_manylinux.manylinux1_compatible) + if version == _GLibCVersion(2, 12): + if hasattr(_manylinux, "manylinux2010_compatible"): + return bool(_manylinux.manylinux2010_compatible) + if version == _GLibCVersion(2, 17): + if hasattr(_manylinux, "manylinux2014_compatible"): + return bool(_manylinux.manylinux2014_compatible) + return True + + +_LEGACY_MANYLINUX_MAP = { + # CentOS 7 w/ glibc 2.17 (PEP 599) + (2, 17): "manylinux2014", + # CentOS 6 w/ glibc 2.12 (PEP 571) + (2, 12): "manylinux2010", + # CentOS 5 w/ glibc 2.5 (PEP 513) + (2, 5): "manylinux1", +} + + +def platform_tags(linux: str, arch: str) -> Iterator[str]: + if not _have_compatible_abi(sys.executable, arch): + return + # Oldest glibc to be supported regardless of architecture is (2, 17). + too_old_glibc2 = _GLibCVersion(2, 16) + if arch in {"x86_64", "i686"}: + # On x86/i686 also oldest glibc to be supported is (2, 5). + too_old_glibc2 = _GLibCVersion(2, 4) + current_glibc = _GLibCVersion(*_get_glibc_version()) + glibc_max_list = [current_glibc] + # We can assume compatibility across glibc major versions. + # https://sourceware.org/bugzilla/show_bug.cgi?id=24636 + # + # Build a list of maximum glibc versions so that we can + # output the canonical list of all glibc from current_glibc + # down to too_old_glibc2, including all intermediary versions. + for glibc_major in range(current_glibc.major - 1, 1, -1): + glibc_minor = _LAST_GLIBC_MINOR[glibc_major] + glibc_max_list.append(_GLibCVersion(glibc_major, glibc_minor)) + for glibc_max in glibc_max_list: + if glibc_max.major == too_old_glibc2.major: + min_minor = too_old_glibc2.minor + else: + # For other glibc major versions oldest supported is (x, 0). + min_minor = -1 + for glibc_minor in range(glibc_max.minor, min_minor, -1): + glibc_version = _GLibCVersion(glibc_max.major, glibc_minor) + tag = "manylinux_{}_{}".format(*glibc_version) + if _is_compatible(tag, arch, glibc_version): + yield linux.replace("linux", tag) + # Handle the legacy manylinux1, manylinux2010, manylinux2014 tags. + if glibc_version in _LEGACY_MANYLINUX_MAP: + legacy_tag = _LEGACY_MANYLINUX_MAP[glibc_version] + if _is_compatible(legacy_tag, arch, glibc_version): + yield linux.replace("linux", legacy_tag) diff --git a/third_party/python/setuptools/pkg_resources/_vendor/packaging/_musllinux.py b/third_party/python/setuptools/pkg_resources/_vendor/packaging/_musllinux.py new file mode 100644 index 0000000000..706ba600a9 --- /dev/null +++ b/third_party/python/setuptools/pkg_resources/_vendor/packaging/_musllinux.py @@ -0,0 +1,80 @@ +"""PEP 656 support. + +This module implements logic to detect if the currently running Python is +linked against musl, and what musl version is used. +""" + +import functools +import re +import subprocess +import sys +from typing import Iterator, NamedTuple, Optional + +from ._elffile import ELFFile + + +class _MuslVersion(NamedTuple): + major: int + minor: int + + +def _parse_musl_version(output: str) -> Optional[_MuslVersion]: + lines = [n for n in (n.strip() for n in output.splitlines()) if n] + if len(lines) < 2 or lines[0][:4] != "musl": + return None + m = re.match(r"Version (\d+)\.(\d+)", lines[1]) + if not m: + return None + return _MuslVersion(major=int(m.group(1)), minor=int(m.group(2))) + + +@functools.lru_cache() +def _get_musl_version(executable: str) -> Optional[_MuslVersion]: + """Detect currently-running musl runtime version. + + This is done by checking the specified executable's dynamic linking + information, and invoking the loader to parse its output for a version + string. If the loader is musl, the output would be something like:: + + musl libc (x86_64) + Version 1.2.2 + Dynamic Program Loader + """ + try: + with open(executable, "rb") as f: + ld = ELFFile(f).interpreter + except (OSError, TypeError, ValueError): + return None + if ld is None or "musl" not in ld: + return None + proc = subprocess.run([ld], stderr=subprocess.PIPE, universal_newlines=True) + return _parse_musl_version(proc.stderr) + + +def platform_tags(arch: str) -> Iterator[str]: + """Generate musllinux tags compatible to the current platform. + + :param arch: Should be the part of platform tag after the ``linux_`` + prefix, e.g. ``x86_64``. The ``linux_`` prefix is assumed as a + prerequisite for the current platform to be musllinux-compatible. + + :returns: An iterator of compatible musllinux tags. + """ + sys_musl = _get_musl_version(sys.executable) + if sys_musl is None: # Python not dynamically linked against musl. + return + for minor in range(sys_musl.minor, -1, -1): + yield f"musllinux_{sys_musl.major}_{minor}_{arch}" + + +if __name__ == "__main__": # pragma: no cover + import sysconfig + + plat = sysconfig.get_platform() + assert plat.startswith("linux-"), "not linux" + + print("plat:", plat) + print("musl:", _get_musl_version(sys.executable)) + print("tags:", end=" ") + for t in platform_tags(re.sub(r"[.-]", "_", plat.split("-", 1)[-1])): + print(t, end="\n ") diff --git a/third_party/python/setuptools/pkg_resources/_vendor/packaging/_parser.py b/third_party/python/setuptools/pkg_resources/_vendor/packaging/_parser.py new file mode 100644 index 0000000000..5a18b758fe --- /dev/null +++ b/third_party/python/setuptools/pkg_resources/_vendor/packaging/_parser.py @@ -0,0 +1,353 @@ +"""Handwritten parser of dependency specifiers. + +The docstring for each __parse_* function contains ENBF-inspired grammar representing +the implementation. +""" + +import ast +from typing import Any, List, NamedTuple, Optional, Tuple, Union + +from ._tokenizer import DEFAULT_RULES, Tokenizer + + +class Node: + def __init__(self, value: str) -> None: + self.value = value + + def __str__(self) -> str: + return self.value + + def __repr__(self) -> str: + return f"<{self.__class__.__name__}('{self}')>" + + def serialize(self) -> str: + raise NotImplementedError + + +class Variable(Node): + def serialize(self) -> str: + return str(self) + + +class Value(Node): + def serialize(self) -> str: + return f'"{self}"' + + +class Op(Node): + def serialize(self) -> str: + return str(self) + + +MarkerVar = Union[Variable, Value] +MarkerItem = Tuple[MarkerVar, Op, MarkerVar] +# MarkerAtom = Union[MarkerItem, List["MarkerAtom"]] +# MarkerList = List[Union["MarkerList", MarkerAtom, str]] +# mypy does not support recursive type definition +# https://github.com/python/mypy/issues/731 +MarkerAtom = Any +MarkerList = List[Any] + + +class ParsedRequirement(NamedTuple): + name: str + url: str + extras: List[str] + specifier: str + marker: Optional[MarkerList] + + +# -------------------------------------------------------------------------------------- +# Recursive descent parser for dependency specifier +# -------------------------------------------------------------------------------------- +def parse_requirement(source: str) -> ParsedRequirement: + return _parse_requirement(Tokenizer(source, rules=DEFAULT_RULES)) + + +def _parse_requirement(tokenizer: Tokenizer) -> ParsedRequirement: + """ + requirement = WS? IDENTIFIER WS? extras WS? requirement_details + """ + tokenizer.consume("WS") + + name_token = tokenizer.expect( + "IDENTIFIER", expected="package name at the start of dependency specifier" + ) + name = name_token.text + tokenizer.consume("WS") + + extras = _parse_extras(tokenizer) + tokenizer.consume("WS") + + url, specifier, marker = _parse_requirement_details(tokenizer) + tokenizer.expect("END", expected="end of dependency specifier") + + return ParsedRequirement(name, url, extras, specifier, marker) + + +def _parse_requirement_details( + tokenizer: Tokenizer, +) -> Tuple[str, str, Optional[MarkerList]]: + """ + requirement_details = AT URL (WS requirement_marker?)? + | specifier WS? (requirement_marker)? + """ + + specifier = "" + url = "" + marker = None + + if tokenizer.check("AT"): + tokenizer.read() + tokenizer.consume("WS") + + url_start = tokenizer.position + url = tokenizer.expect("URL", expected="URL after @").text + if tokenizer.check("END", peek=True): + return (url, specifier, marker) + + tokenizer.expect("WS", expected="whitespace after URL") + + # The input might end after whitespace. + if tokenizer.check("END", peek=True): + return (url, specifier, marker) + + marker = _parse_requirement_marker( + tokenizer, span_start=url_start, after="URL and whitespace" + ) + else: + specifier_start = tokenizer.position + specifier = _parse_specifier(tokenizer) + tokenizer.consume("WS") + + if tokenizer.check("END", peek=True): + return (url, specifier, marker) + + marker = _parse_requirement_marker( + tokenizer, + span_start=specifier_start, + after=( + "version specifier" + if specifier + else "name and no valid version specifier" + ), + ) + + return (url, specifier, marker) + + +def _parse_requirement_marker( + tokenizer: Tokenizer, *, span_start: int, after: str +) -> MarkerList: + """ + requirement_marker = SEMICOLON marker WS? + """ + + if not tokenizer.check("SEMICOLON"): + tokenizer.raise_syntax_error( + f"Expected end or semicolon (after {after})", + span_start=span_start, + ) + tokenizer.read() + + marker = _parse_marker(tokenizer) + tokenizer.consume("WS") + + return marker + + +def _parse_extras(tokenizer: Tokenizer) -> List[str]: + """ + extras = (LEFT_BRACKET wsp* extras_list? wsp* RIGHT_BRACKET)? + """ + if not tokenizer.check("LEFT_BRACKET", peek=True): + return [] + + with tokenizer.enclosing_tokens( + "LEFT_BRACKET", + "RIGHT_BRACKET", + around="extras", + ): + tokenizer.consume("WS") + extras = _parse_extras_list(tokenizer) + tokenizer.consume("WS") + + return extras + + +def _parse_extras_list(tokenizer: Tokenizer) -> List[str]: + """ + extras_list = identifier (wsp* ',' wsp* identifier)* + """ + extras: List[str] = [] + + if not tokenizer.check("IDENTIFIER"): + return extras + + extras.append(tokenizer.read().text) + + while True: + tokenizer.consume("WS") + if tokenizer.check("IDENTIFIER", peek=True): + tokenizer.raise_syntax_error("Expected comma between extra names") + elif not tokenizer.check("COMMA"): + break + + tokenizer.read() + tokenizer.consume("WS") + + extra_token = tokenizer.expect("IDENTIFIER", expected="extra name after comma") + extras.append(extra_token.text) + + return extras + + +def _parse_specifier(tokenizer: Tokenizer) -> str: + """ + specifier = LEFT_PARENTHESIS WS? version_many WS? RIGHT_PARENTHESIS + | WS? version_many WS? + """ + with tokenizer.enclosing_tokens( + "LEFT_PARENTHESIS", + "RIGHT_PARENTHESIS", + around="version specifier", + ): + tokenizer.consume("WS") + parsed_specifiers = _parse_version_many(tokenizer) + tokenizer.consume("WS") + + return parsed_specifiers + + +def _parse_version_many(tokenizer: Tokenizer) -> str: + """ + version_many = (SPECIFIER (WS? COMMA WS? SPECIFIER)*)? + """ + parsed_specifiers = "" + while tokenizer.check("SPECIFIER"): + span_start = tokenizer.position + parsed_specifiers += tokenizer.read().text + if tokenizer.check("VERSION_PREFIX_TRAIL", peek=True): + tokenizer.raise_syntax_error( + ".* suffix can only be used with `==` or `!=` operators", + span_start=span_start, + span_end=tokenizer.position + 1, + ) + if tokenizer.check("VERSION_LOCAL_LABEL_TRAIL", peek=True): + tokenizer.raise_syntax_error( + "Local version label can only be used with `==` or `!=` operators", + span_start=span_start, + span_end=tokenizer.position, + ) + tokenizer.consume("WS") + if not tokenizer.check("COMMA"): + break + parsed_specifiers += tokenizer.read().text + tokenizer.consume("WS") + + return parsed_specifiers + + +# -------------------------------------------------------------------------------------- +# Recursive descent parser for marker expression +# -------------------------------------------------------------------------------------- +def parse_marker(source: str) -> MarkerList: + return _parse_marker(Tokenizer(source, rules=DEFAULT_RULES)) + + +def _parse_marker(tokenizer: Tokenizer) -> MarkerList: + """ + marker = marker_atom (BOOLOP marker_atom)+ + """ + expression = [_parse_marker_atom(tokenizer)] + while tokenizer.check("BOOLOP"): + token = tokenizer.read() + expr_right = _parse_marker_atom(tokenizer) + expression.extend((token.text, expr_right)) + return expression + + +def _parse_marker_atom(tokenizer: Tokenizer) -> MarkerAtom: + """ + marker_atom = WS? LEFT_PARENTHESIS WS? marker WS? RIGHT_PARENTHESIS WS? + | WS? marker_item WS? + """ + + tokenizer.consume("WS") + if tokenizer.check("LEFT_PARENTHESIS", peek=True): + with tokenizer.enclosing_tokens( + "LEFT_PARENTHESIS", + "RIGHT_PARENTHESIS", + around="marker expression", + ): + tokenizer.consume("WS") + marker: MarkerAtom = _parse_marker(tokenizer) + tokenizer.consume("WS") + else: + marker = _parse_marker_item(tokenizer) + tokenizer.consume("WS") + return marker + + +def _parse_marker_item(tokenizer: Tokenizer) -> MarkerItem: + """ + marker_item = WS? marker_var WS? marker_op WS? marker_var WS? + """ + tokenizer.consume("WS") + marker_var_left = _parse_marker_var(tokenizer) + tokenizer.consume("WS") + marker_op = _parse_marker_op(tokenizer) + tokenizer.consume("WS") + marker_var_right = _parse_marker_var(tokenizer) + tokenizer.consume("WS") + return (marker_var_left, marker_op, marker_var_right) + + +def _parse_marker_var(tokenizer: Tokenizer) -> MarkerVar: + """ + marker_var = VARIABLE | QUOTED_STRING + """ + if tokenizer.check("VARIABLE"): + return process_env_var(tokenizer.read().text.replace(".", "_")) + elif tokenizer.check("QUOTED_STRING"): + return process_python_str(tokenizer.read().text) + else: + tokenizer.raise_syntax_error( + message="Expected a marker variable or quoted string" + ) + + +def process_env_var(env_var: str) -> Variable: + if ( + env_var == "platform_python_implementation" + or env_var == "python_implementation" + ): + return Variable("platform_python_implementation") + else: + return Variable(env_var) + + +def process_python_str(python_str: str) -> Value: + value = ast.literal_eval(python_str) + return Value(str(value)) + + +def _parse_marker_op(tokenizer: Tokenizer) -> Op: + """ + marker_op = IN | NOT IN | OP + """ + if tokenizer.check("IN"): + tokenizer.read() + return Op("in") + elif tokenizer.check("NOT"): + tokenizer.read() + tokenizer.expect("WS", expected="whitespace after 'not'") + tokenizer.expect("IN", expected="'in' after 'not'") + return Op("not in") + elif tokenizer.check("OP"): + return Op(tokenizer.read().text) + else: + return tokenizer.raise_syntax_error( + "Expected marker operator, one of " + "<=, <, !=, ==, >=, >, ~=, ===, in, not in" + ) diff --git a/third_party/python/setuptools/pkg_resources/_vendor/packaging/_structures.py b/third_party/python/setuptools/pkg_resources/_vendor/packaging/_structures.py index 800d5c5588..90a6465f96 100644 --- a/third_party/python/setuptools/pkg_resources/_vendor/packaging/_structures.py +++ b/third_party/python/setuptools/pkg_resources/_vendor/packaging/_structures.py @@ -1,85 +1,60 @@ # This file is dual licensed under the terms of the Apache License, Version # 2.0, and the BSD License. See the LICENSE file in the root of this repository # for complete details. -from __future__ import absolute_import, division, print_function -class InfinityType(object): - def __repr__(self): - # type: () -> str +class InfinityType: + def __repr__(self) -> str: return "Infinity" - def __hash__(self): - # type: () -> int + def __hash__(self) -> int: return hash(repr(self)) - def __lt__(self, other): - # type: (object) -> bool + def __lt__(self, other: object) -> bool: return False - def __le__(self, other): - # type: (object) -> bool + def __le__(self, other: object) -> bool: return False - def __eq__(self, other): - # type: (object) -> bool + def __eq__(self, other: object) -> bool: return isinstance(other, self.__class__) - def __ne__(self, other): - # type: (object) -> bool - return not isinstance(other, self.__class__) - - def __gt__(self, other): - # type: (object) -> bool + def __gt__(self, other: object) -> bool: return True - def __ge__(self, other): - # type: (object) -> bool + def __ge__(self, other: object) -> bool: return True - def __neg__(self): - # type: (object) -> NegativeInfinityType + def __neg__(self: object) -> "NegativeInfinityType": return NegativeInfinity Infinity = InfinityType() -class NegativeInfinityType(object): - def __repr__(self): - # type: () -> str +class NegativeInfinityType: + def __repr__(self) -> str: return "-Infinity" - def __hash__(self): - # type: () -> int + def __hash__(self) -> int: return hash(repr(self)) - def __lt__(self, other): - # type: (object) -> bool + def __lt__(self, other: object) -> bool: return True - def __le__(self, other): - # type: (object) -> bool + def __le__(self, other: object) -> bool: return True - def __eq__(self, other): - # type: (object) -> bool + def __eq__(self, other: object) -> bool: return isinstance(other, self.__class__) - def __ne__(self, other): - # type: (object) -> bool - return not isinstance(other, self.__class__) - - def __gt__(self, other): - # type: (object) -> bool + def __gt__(self, other: object) -> bool: return False - def __ge__(self, other): - # type: (object) -> bool + def __ge__(self, other: object) -> bool: return False - def __neg__(self): - # type: (object) -> InfinityType + def __neg__(self: object) -> InfinityType: return Infinity diff --git a/third_party/python/setuptools/pkg_resources/_vendor/packaging/_tokenizer.py b/third_party/python/setuptools/pkg_resources/_vendor/packaging/_tokenizer.py new file mode 100644 index 0000000000..dd0d648d49 --- /dev/null +++ b/third_party/python/setuptools/pkg_resources/_vendor/packaging/_tokenizer.py @@ -0,0 +1,192 @@ +import contextlib +import re +from dataclasses import dataclass +from typing import Dict, Iterator, NoReturn, Optional, Tuple, Union + +from .specifiers import Specifier + + +@dataclass +class Token: + name: str + text: str + position: int + + +class ParserSyntaxError(Exception): + """The provided source text could not be parsed correctly.""" + + def __init__( + self, + message: str, + *, + source: str, + span: Tuple[int, int], + ) -> None: + self.span = span + self.message = message + self.source = source + + super().__init__() + + def __str__(self) -> str: + marker = " " * self.span[0] + "~" * (self.span[1] - self.span[0]) + "^" + return "\n ".join([self.message, self.source, marker]) + + +DEFAULT_RULES: "Dict[str, Union[str, re.Pattern[str]]]" = { + "LEFT_PARENTHESIS": r"\(", + "RIGHT_PARENTHESIS": r"\)", + "LEFT_BRACKET": r"\[", + "RIGHT_BRACKET": r"\]", + "SEMICOLON": r";", + "COMMA": r",", + "QUOTED_STRING": re.compile( + r""" + ( + ('[^']*') + | + ("[^"]*") + ) + """, + re.VERBOSE, + ), + "OP": r"(===|==|~=|!=|<=|>=|<|>)", + "BOOLOP": r"\b(or|and)\b", + "IN": r"\bin\b", + "NOT": r"\bnot\b", + "VARIABLE": re.compile( + r""" + \b( + python_version + |python_full_version + |os[._]name + |sys[._]platform + |platform_(release|system) + |platform[._](version|machine|python_implementation) + |python_implementation + |implementation_(name|version) + |extra + )\b + """, + re.VERBOSE, + ), + "SPECIFIER": re.compile( + Specifier._operator_regex_str + Specifier._version_regex_str, + re.VERBOSE | re.IGNORECASE, + ), + "AT": r"\@", + "URL": r"[^ \t]+", + "IDENTIFIER": r"\b[a-zA-Z0-9][a-zA-Z0-9._-]*\b", + "VERSION_PREFIX_TRAIL": r"\.\*", + "VERSION_LOCAL_LABEL_TRAIL": r"\+[a-z0-9]+(?:[-_\.][a-z0-9]+)*", + "WS": r"[ \t]+", + "END": r"$", +} + + +class Tokenizer: + """Context-sensitive token parsing. + + Provides methods to examine the input stream to check whether the next token + matches. + """ + + def __init__( + self, + source: str, + *, + rules: "Dict[str, Union[str, re.Pattern[str]]]", + ) -> None: + self.source = source + self.rules: Dict[str, re.Pattern[str]] = { + name: re.compile(pattern) for name, pattern in rules.items() + } + self.next_token: Optional[Token] = None + self.position = 0 + + def consume(self, name: str) -> None: + """Move beyond provided token name, if at current position.""" + if self.check(name): + self.read() + + def check(self, name: str, *, peek: bool = False) -> bool: + """Check whether the next token has the provided name. + + By default, if the check succeeds, the token *must* be read before + another check. If `peek` is set to `True`, the token is not loaded and + would need to be checked again. + """ + assert ( + self.next_token is None + ), f"Cannot check for {name!r}, already have {self.next_token!r}" + assert name in self.rules, f"Unknown token name: {name!r}" + + expression = self.rules[name] + + match = expression.match(self.source, self.position) + if match is None: + return False + if not peek: + self.next_token = Token(name, match[0], self.position) + return True + + def expect(self, name: str, *, expected: str) -> Token: + """Expect a certain token name next, failing with a syntax error otherwise. + + The token is *not* read. + """ + if not self.check(name): + raise self.raise_syntax_error(f"Expected {expected}") + return self.read() + + def read(self) -> Token: + """Consume the next token and return it.""" + token = self.next_token + assert token is not None + + self.position += len(token.text) + self.next_token = None + + return token + + def raise_syntax_error( + self, + message: str, + *, + span_start: Optional[int] = None, + span_end: Optional[int] = None, + ) -> NoReturn: + """Raise ParserSyntaxError at the given position.""" + span = ( + self.position if span_start is None else span_start, + self.position if span_end is None else span_end, + ) + raise ParserSyntaxError( + message, + source=self.source, + span=span, + ) + + @contextlib.contextmanager + def enclosing_tokens( + self, open_token: str, close_token: str, *, around: str + ) -> Iterator[None]: + if self.check(open_token): + open_position = self.position + self.read() + else: + open_position = None + + yield + + if open_position is None: + return + + if not self.check(close_token): + self.raise_syntax_error( + f"Expected matching {close_token} for {open_token}, after {around}", + span_start=open_position, + ) + + self.read() diff --git a/third_party/python/setuptools/pkg_resources/_vendor/packaging/_typing.py b/third_party/python/setuptools/pkg_resources/_vendor/packaging/_typing.py deleted file mode 100644 index 77a8b9185a..0000000000 --- a/third_party/python/setuptools/pkg_resources/_vendor/packaging/_typing.py +++ /dev/null @@ -1,48 +0,0 @@ -"""For neatly implementing static typing in packaging. - -`mypy` - the static type analysis tool we use - uses the `typing` module, which -provides core functionality fundamental to mypy's functioning. - -Generally, `typing` would be imported at runtime and used in that fashion - -it acts as a no-op at runtime and does not have any run-time overhead by -design. - -As it turns out, `typing` is not vendorable - it uses separate sources for -Python 2/Python 3. Thus, this codebase can not expect it to be present. -To work around this, mypy allows the typing import to be behind a False-y -optional to prevent it from running at runtime and type-comments can be used -to remove the need for the types to be accessible directly during runtime. - -This module provides the False-y guard in a nicely named fashion so that a -curious maintainer can reach here to read this. - -In packaging, all static-typing related imports should be guarded as follows: - - from packaging._typing import TYPE_CHECKING - - if TYPE_CHECKING: - from typing import ... - -Ref: https://github.com/python/mypy/issues/3216 -""" - -__all__ = ["TYPE_CHECKING", "cast"] - -# The TYPE_CHECKING constant defined by the typing module is False at runtime -# but True while type checking. -if False: # pragma: no cover - from typing import TYPE_CHECKING -else: - TYPE_CHECKING = False - -# typing's cast syntax requires calling typing.cast at runtime, but we don't -# want to import typing at runtime. Here, we inform the type checkers that -# we're importing `typing.cast` as `cast` and re-implement typing.cast's -# runtime behavior in a block that is ignored by type checkers. -if TYPE_CHECKING: # pragma: no cover - # not executed at runtime - from typing import cast -else: - # executed at runtime - def cast(type_, value): # noqa - return value diff --git a/third_party/python/setuptools/pkg_resources/_vendor/packaging/markers.py b/third_party/python/setuptools/pkg_resources/_vendor/packaging/markers.py index fd1559c10e..8b98fca723 100644 --- a/third_party/python/setuptools/pkg_resources/_vendor/packaging/markers.py +++ b/third_party/python/setuptools/pkg_resources/_vendor/packaging/markers.py @@ -1,26 +1,24 @@ # This file is dual licensed under the terms of the Apache License, Version # 2.0, and the BSD License. See the LICENSE file in the root of this repository # for complete details. -from __future__ import absolute_import, division, print_function import operator import os import platform import sys - -from pkg_resources.extern.pyparsing import ParseException, ParseResults, stringStart, stringEnd -from pkg_resources.extern.pyparsing import ZeroOrMore, Group, Forward, QuotedString -from pkg_resources.extern.pyparsing import Literal as L # noqa - -from ._compat import string_types -from ._typing import TYPE_CHECKING -from .specifiers import Specifier, InvalidSpecifier - -if TYPE_CHECKING: # pragma: no cover - from typing import Any, Callable, Dict, List, Optional, Tuple, Union - - Operator = Callable[[str, str], bool] - +from typing import Any, Callable, Dict, List, Optional, Tuple, Union + +from ._parser import ( + MarkerAtom, + MarkerList, + Op, + Value, + Variable, + parse_marker as _parse_marker, +) +from ._tokenizer import ParserSyntaxError +from .specifiers import InvalidSpecifier, Specifier +from .utils import canonicalize_name __all__ = [ "InvalidMarker", @@ -30,6 +28,8 @@ __all__ = [ "default_environment", ] +Operator = Callable[[str, str], bool] + class InvalidMarker(ValueError): """ @@ -50,111 +50,27 @@ class UndefinedEnvironmentName(ValueError): """ -class Node(object): - def __init__(self, value): - # type: (Any) -> None - self.value = value - - def __str__(self): - # type: () -> str - return str(self.value) - - def __repr__(self): - # type: () -> str - return "<{0}({1!r})>".format(self.__class__.__name__, str(self)) - - def serialize(self): - # type: () -> str - raise NotImplementedError - - -class Variable(Node): - def serialize(self): - # type: () -> str - return str(self) - - -class Value(Node): - def serialize(self): - # type: () -> str - return '"{0}"'.format(self) - - -class Op(Node): - def serialize(self): - # type: () -> str - return str(self) - - -VARIABLE = ( - L("implementation_version") - | L("platform_python_implementation") - | L("implementation_name") - | L("python_full_version") - | L("platform_release") - | L("platform_version") - | L("platform_machine") - | L("platform_system") - | L("python_version") - | L("sys_platform") - | L("os_name") - | L("os.name") # PEP-345 - | L("sys.platform") # PEP-345 - | L("platform.version") # PEP-345 - | L("platform.machine") # PEP-345 - | L("platform.python_implementation") # PEP-345 - | L("python_implementation") # undocumented setuptools legacy - | L("extra") # PEP-508 -) -ALIASES = { - "os.name": "os_name", - "sys.platform": "sys_platform", - "platform.version": "platform_version", - "platform.machine": "platform_machine", - "platform.python_implementation": "platform_python_implementation", - "python_implementation": "platform_python_implementation", -} -VARIABLE.setParseAction(lambda s, l, t: Variable(ALIASES.get(t[0], t[0]))) - -VERSION_CMP = ( - L("===") | L("==") | L(">=") | L("<=") | L("!=") | L("~=") | L(">") | L("<") -) - -MARKER_OP = VERSION_CMP | L("not in") | L("in") -MARKER_OP.setParseAction(lambda s, l, t: Op(t[0])) - -MARKER_VALUE = QuotedString("'") | QuotedString('"') -MARKER_VALUE.setParseAction(lambda s, l, t: Value(t[0])) - -BOOLOP = L("and") | L("or") - -MARKER_VAR = VARIABLE | MARKER_VALUE - -MARKER_ITEM = Group(MARKER_VAR + MARKER_OP + MARKER_VAR) -MARKER_ITEM.setParseAction(lambda s, l, t: tuple(t[0])) - -LPAREN = L("(").suppress() -RPAREN = L(")").suppress() - -MARKER_EXPR = Forward() -MARKER_ATOM = MARKER_ITEM | Group(LPAREN + MARKER_EXPR + RPAREN) -MARKER_EXPR << MARKER_ATOM + ZeroOrMore(BOOLOP + MARKER_EXPR) - -MARKER = stringStart + MARKER_EXPR + stringEnd - - -def _coerce_parse_result(results): - # type: (Union[ParseResults, List[Any]]) -> List[Any] - if isinstance(results, ParseResults): - return [_coerce_parse_result(i) for i in results] - else: - return results +def _normalize_extra_values(results: Any) -> Any: + """ + Normalize extra values. + """ + if isinstance(results[0], tuple): + lhs, op, rhs = results[0] + if isinstance(lhs, Variable) and lhs.value == "extra": + normalized_extra = canonicalize_name(rhs.value) + rhs = Value(normalized_extra) + elif isinstance(rhs, Variable) and rhs.value == "extra": + normalized_extra = canonicalize_name(lhs.value) + lhs = Value(normalized_extra) + results[0] = lhs, op, rhs + return results -def _format_marker(marker, first=True): - # type: (Union[List[str], Tuple[Node, ...], str], Optional[bool]) -> str +def _format_marker( + marker: Union[List[str], MarkerAtom, str], first: Optional[bool] = True +) -> str: - assert isinstance(marker, (list, tuple, string_types)) + assert isinstance(marker, (list, tuple, str)) # Sometimes we have a structure like [[...]] which is a single item list # where the single item is itself it's own list. In that case we want skip @@ -179,7 +95,7 @@ def _format_marker(marker, first=True): return marker -_operators = { +_operators: Dict[str, Operator] = { "in": lambda lhs, rhs: lhs in rhs, "not in": lambda lhs, rhs: lhs not in rhs, "<": operator.lt, @@ -188,52 +104,41 @@ _operators = { "!=": operator.ne, ">=": operator.ge, ">": operator.gt, -} # type: Dict[str, Operator] +} -def _eval_op(lhs, op, rhs): - # type: (str, Op, str) -> bool +def _eval_op(lhs: str, op: Op, rhs: str) -> bool: try: spec = Specifier("".join([op.serialize(), rhs])) except InvalidSpecifier: pass else: - return spec.contains(lhs) + return spec.contains(lhs, prereleases=True) - oper = _operators.get(op.serialize()) # type: Optional[Operator] + oper: Optional[Operator] = _operators.get(op.serialize()) if oper is None: - raise UndefinedComparison( - "Undefined {0!r} on {1!r} and {2!r}.".format(op, lhs, rhs) - ) + raise UndefinedComparison(f"Undefined {op!r} on {lhs!r} and {rhs!r}.") return oper(lhs, rhs) -class Undefined(object): - pass - - -_undefined = Undefined() +def _normalize(*values: str, key: str) -> Tuple[str, ...]: + # PEP 685 – Comparison of extra names for optional distribution dependencies + # https://peps.python.org/pep-0685/ + # > When comparing extra names, tools MUST normalize the names being + # > compared using the semantics outlined in PEP 503 for names + if key == "extra": + return tuple(canonicalize_name(v) for v in values) + # other environment markers don't have such standards + return values -def _get_env(environment, name): - # type: (Dict[str, str], str) -> str - value = environment.get(name, _undefined) # type: Union[str, Undefined] - if isinstance(value, Undefined): - raise UndefinedEnvironmentName( - "{0!r} does not exist in evaluation environment.".format(name) - ) - - return value - - -def _evaluate_markers(markers, environment): - # type: (List[Any], Dict[str, str]) -> bool - groups = [[]] # type: List[List[bool]] +def _evaluate_markers(markers: MarkerList, environment: Dict[str, str]) -> bool: + groups: List[List[bool]] = [[]] for marker in markers: - assert isinstance(marker, (list, tuple, string_types)) + assert isinstance(marker, (list, tuple, str)) if isinstance(marker, list): groups[-1].append(_evaluate_markers(marker, environment)) @@ -241,12 +146,15 @@ def _evaluate_markers(markers, environment): lhs, op, rhs = marker if isinstance(lhs, Variable): - lhs_value = _get_env(environment, lhs.value) + environment_key = lhs.value + lhs_value = environment[environment_key] rhs_value = rhs.value else: lhs_value = lhs.value - rhs_value = _get_env(environment, rhs.value) + environment_key = rhs.value + rhs_value = environment[environment_key] + lhs_value, rhs_value = _normalize(lhs_value, rhs_value, key=environment_key) groups[-1].append(_eval_op(lhs_value, op, rhs_value)) else: assert marker in ["and", "or"] @@ -256,8 +164,7 @@ def _evaluate_markers(markers, environment): return any(all(item) for item in groups) -def format_full_version(info): - # type: (sys._version_info) -> str +def format_full_version(info: "sys._version_info") -> str: version = "{0.major}.{0.minor}.{0.micro}".format(info) kind = info.releaselevel if kind != "final": @@ -265,18 +172,9 @@ def format_full_version(info): return version -def default_environment(): - # type: () -> Dict[str, str] - if hasattr(sys, "implementation"): - # Ignoring the `sys.implementation` reference for type checking due to - # mypy not liking that the attribute doesn't exist in Python 2.7 when - # run with the `--py27` flag. - iver = format_full_version(sys.implementation.version) # type: ignore - implementation_name = sys.implementation.name # type: ignore - else: - iver = "0" - implementation_name = "" - +def default_environment() -> Dict[str, str]: + iver = format_full_version(sys.implementation.version) + implementation_name = sys.implementation.name return { "implementation_name": implementation_name, "implementation_version": iver, @@ -292,27 +190,48 @@ def default_environment(): } -class Marker(object): - def __init__(self, marker): - # type: (str) -> None +class Marker: + def __init__(self, marker: str) -> None: + # Note: We create a Marker object without calling this constructor in + # packaging.requirements.Requirement. If any additional logic is + # added here, make sure to mirror/adapt Requirement. try: - self._markers = _coerce_parse_result(MARKER.parseString(marker)) - except ParseException as e: - err_str = "Invalid marker: {0!r}, parse error at {1!r}".format( - marker, marker[e.loc : e.loc + 8] - ) - raise InvalidMarker(err_str) - - def __str__(self): - # type: () -> str + self._markers = _normalize_extra_values(_parse_marker(marker)) + # The attribute `_markers` can be described in terms of a recursive type: + # MarkerList = List[Union[Tuple[Node, ...], str, MarkerList]] + # + # For example, the following expression: + # python_version > "3.6" or (python_version == "3.6" and os_name == "unix") + # + # is parsed into: + # [ + # (<Variable('python_version')>, <Op('>')>, <Value('3.6')>), + # 'and', + # [ + # (<Variable('python_version')>, <Op('==')>, <Value('3.6')>), + # 'or', + # (<Variable('os_name')>, <Op('==')>, <Value('unix')>) + # ] + # ] + except ParserSyntaxError as e: + raise InvalidMarker(str(e)) from e + + def __str__(self) -> str: return _format_marker(self._markers) - def __repr__(self): - # type: () -> str - return "<Marker({0!r})>".format(str(self)) + def __repr__(self) -> str: + return f"<Marker('{self}')>" + + def __hash__(self) -> int: + return hash((self.__class__.__name__, str(self))) + + def __eq__(self, other: Any) -> bool: + if not isinstance(other, Marker): + return NotImplemented + + return str(self) == str(other) - def evaluate(self, environment=None): - # type: (Optional[Dict[str, str]]) -> bool + def evaluate(self, environment: Optional[Dict[str, str]] = None) -> bool: """Evaluate a marker. Return the boolean from evaluating the given marker against the @@ -322,7 +241,12 @@ class Marker(object): The environment is determined from the current Python process. """ current_environment = default_environment() + current_environment["extra"] = "" if environment is not None: current_environment.update(environment) + # The API used to allow setting extra to None. We need to handle this + # case for backwards compatibility. + if current_environment["extra"] is None: + current_environment["extra"] = "" return _evaluate_markers(self._markers, current_environment) diff --git a/third_party/python/setuptools/pkg_resources/_vendor/packaging/metadata.py b/third_party/python/setuptools/pkg_resources/_vendor/packaging/metadata.py new file mode 100644 index 0000000000..e76a60c395 --- /dev/null +++ b/third_party/python/setuptools/pkg_resources/_vendor/packaging/metadata.py @@ -0,0 +1,408 @@ +import email.feedparser +import email.header +import email.message +import email.parser +import email.policy +import sys +import typing +from typing import Dict, List, Optional, Tuple, Union, cast + +if sys.version_info >= (3, 8): # pragma: no cover + from typing import TypedDict +else: # pragma: no cover + if typing.TYPE_CHECKING: + from typing_extensions import TypedDict + else: + try: + from typing_extensions import TypedDict + except ImportError: + + class TypedDict: + def __init_subclass__(*_args, **_kwargs): + pass + + +# The RawMetadata class attempts to make as few assumptions about the underlying +# serialization formats as possible. The idea is that as long as a serialization +# formats offer some very basic primitives in *some* way then we can support +# serializing to and from that format. +class RawMetadata(TypedDict, total=False): + """A dictionary of raw core metadata. + + Each field in core metadata maps to a key of this dictionary (when data is + provided). The key is lower-case and underscores are used instead of dashes + compared to the equivalent core metadata field. Any core metadata field that + can be specified multiple times or can hold multiple values in a single + field have a key with a plural name. + + Core metadata fields that can be specified multiple times are stored as a + list or dict depending on which is appropriate for the field. Any fields + which hold multiple values in a single field are stored as a list. + + """ + + # Metadata 1.0 - PEP 241 + metadata_version: str + name: str + version: str + platforms: List[str] + summary: str + description: str + keywords: List[str] + home_page: str + author: str + author_email: str + license: str + + # Metadata 1.1 - PEP 314 + supported_platforms: List[str] + download_url: str + classifiers: List[str] + requires: List[str] + provides: List[str] + obsoletes: List[str] + + # Metadata 1.2 - PEP 345 + maintainer: str + maintainer_email: str + requires_dist: List[str] + provides_dist: List[str] + obsoletes_dist: List[str] + requires_python: str + requires_external: List[str] + project_urls: Dict[str, str] + + # Metadata 2.0 + # PEP 426 attempted to completely revamp the metadata format + # but got stuck without ever being able to build consensus on + # it and ultimately ended up withdrawn. + # + # However, a number of tools had started emiting METADATA with + # `2.0` Metadata-Version, so for historical reasons, this version + # was skipped. + + # Metadata 2.1 - PEP 566 + description_content_type: str + provides_extra: List[str] + + # Metadata 2.2 - PEP 643 + dynamic: List[str] + + # Metadata 2.3 - PEP 685 + # No new fields were added in PEP 685, just some edge case were + # tightened up to provide better interoptability. + + +_STRING_FIELDS = { + "author", + "author_email", + "description", + "description_content_type", + "download_url", + "home_page", + "license", + "maintainer", + "maintainer_email", + "metadata_version", + "name", + "requires_python", + "summary", + "version", +} + +_LIST_STRING_FIELDS = { + "classifiers", + "dynamic", + "obsoletes", + "obsoletes_dist", + "platforms", + "provides", + "provides_dist", + "provides_extra", + "requires", + "requires_dist", + "requires_external", + "supported_platforms", +} + + +def _parse_keywords(data: str) -> List[str]: + """Split a string of comma-separate keyboards into a list of keywords.""" + return [k.strip() for k in data.split(",")] + + +def _parse_project_urls(data: List[str]) -> Dict[str, str]: + """Parse a list of label/URL string pairings separated by a comma.""" + urls = {} + for pair in data: + # Our logic is slightly tricky here as we want to try and do + # *something* reasonable with malformed data. + # + # The main thing that we have to worry about, is data that does + # not have a ',' at all to split the label from the Value. There + # isn't a singular right answer here, and we will fail validation + # later on (if the caller is validating) so it doesn't *really* + # matter, but since the missing value has to be an empty str + # and our return value is dict[str, str], if we let the key + # be the missing value, then they'd have multiple '' values that + # overwrite each other in a accumulating dict. + # + # The other potentional issue is that it's possible to have the + # same label multiple times in the metadata, with no solid "right" + # answer with what to do in that case. As such, we'll do the only + # thing we can, which is treat the field as unparseable and add it + # to our list of unparsed fields. + parts = [p.strip() for p in pair.split(",", 1)] + parts.extend([""] * (max(0, 2 - len(parts)))) # Ensure 2 items + + # TODO: The spec doesn't say anything about if the keys should be + # considered case sensitive or not... logically they should + # be case-preserving and case-insensitive, but doing that + # would open up more cases where we might have duplicate + # entries. + label, url = parts + if label in urls: + # The label already exists in our set of urls, so this field + # is unparseable, and we can just add the whole thing to our + # unparseable data and stop processing it. + raise KeyError("duplicate labels in project urls") + urls[label] = url + + return urls + + +def _get_payload(msg: email.message.Message, source: Union[bytes, str]) -> str: + """Get the body of the message.""" + # If our source is a str, then our caller has managed encodings for us, + # and we don't need to deal with it. + if isinstance(source, str): + payload: str = msg.get_payload() + return payload + # If our source is a bytes, then we're managing the encoding and we need + # to deal with it. + else: + bpayload: bytes = msg.get_payload(decode=True) + try: + return bpayload.decode("utf8", "strict") + except UnicodeDecodeError: + raise ValueError("payload in an invalid encoding") + + +# The various parse_FORMAT functions here are intended to be as lenient as +# possible in their parsing, while still returning a correctly typed +# RawMetadata. +# +# To aid in this, we also generally want to do as little touching of the +# data as possible, except where there are possibly some historic holdovers +# that make valid data awkward to work with. +# +# While this is a lower level, intermediate format than our ``Metadata`` +# class, some light touch ups can make a massive difference in usability. + +# Map METADATA fields to RawMetadata. +_EMAIL_TO_RAW_MAPPING = { + "author": "author", + "author-email": "author_email", + "classifier": "classifiers", + "description": "description", + "description-content-type": "description_content_type", + "download-url": "download_url", + "dynamic": "dynamic", + "home-page": "home_page", + "keywords": "keywords", + "license": "license", + "maintainer": "maintainer", + "maintainer-email": "maintainer_email", + "metadata-version": "metadata_version", + "name": "name", + "obsoletes": "obsoletes", + "obsoletes-dist": "obsoletes_dist", + "platform": "platforms", + "project-url": "project_urls", + "provides": "provides", + "provides-dist": "provides_dist", + "provides-extra": "provides_extra", + "requires": "requires", + "requires-dist": "requires_dist", + "requires-external": "requires_external", + "requires-python": "requires_python", + "summary": "summary", + "supported-platform": "supported_platforms", + "version": "version", +} + + +def parse_email(data: Union[bytes, str]) -> Tuple[RawMetadata, Dict[str, List[str]]]: + """Parse a distribution's metadata. + + This function returns a two-item tuple of dicts. The first dict is of + recognized fields from the core metadata specification. Fields that can be + parsed and translated into Python's built-in types are converted + appropriately. All other fields are left as-is. Fields that are allowed to + appear multiple times are stored as lists. + + The second dict contains all other fields from the metadata. This includes + any unrecognized fields. It also includes any fields which are expected to + be parsed into a built-in type but were not formatted appropriately. Finally, + any fields that are expected to appear only once but are repeated are + included in this dict. + + """ + raw: Dict[str, Union[str, List[str], Dict[str, str]]] = {} + unparsed: Dict[str, List[str]] = {} + + if isinstance(data, str): + parsed = email.parser.Parser(policy=email.policy.compat32).parsestr(data) + else: + parsed = email.parser.BytesParser(policy=email.policy.compat32).parsebytes(data) + + # We have to wrap parsed.keys() in a set, because in the case of multiple + # values for a key (a list), the key will appear multiple times in the + # list of keys, but we're avoiding that by using get_all(). + for name in frozenset(parsed.keys()): + # Header names in RFC are case insensitive, so we'll normalize to all + # lower case to make comparisons easier. + name = name.lower() + + # We use get_all() here, even for fields that aren't multiple use, + # because otherwise someone could have e.g. two Name fields, and we + # would just silently ignore it rather than doing something about it. + headers = parsed.get_all(name) + + # The way the email module works when parsing bytes is that it + # unconditionally decodes the bytes as ascii using the surrogateescape + # handler. When you pull that data back out (such as with get_all() ), + # it looks to see if the str has any surrogate escapes, and if it does + # it wraps it in a Header object instead of returning the string. + # + # As such, we'll look for those Header objects, and fix up the encoding. + value = [] + # Flag if we have run into any issues processing the headers, thus + # signalling that the data belongs in 'unparsed'. + valid_encoding = True + for h in headers: + # It's unclear if this can return more types than just a Header or + # a str, so we'll just assert here to make sure. + assert isinstance(h, (email.header.Header, str)) + + # If it's a header object, we need to do our little dance to get + # the real data out of it. In cases where there is invalid data + # we're going to end up with mojibake, but there's no obvious, good + # way around that without reimplementing parts of the Header object + # ourselves. + # + # That should be fine since, if mojibacked happens, this key is + # going into the unparsed dict anyways. + if isinstance(h, email.header.Header): + # The Header object stores it's data as chunks, and each chunk + # can be independently encoded, so we'll need to check each + # of them. + chunks: List[Tuple[bytes, Optional[str]]] = [] + for bin, encoding in email.header.decode_header(h): + try: + bin.decode("utf8", "strict") + except UnicodeDecodeError: + # Enable mojibake. + encoding = "latin1" + valid_encoding = False + else: + encoding = "utf8" + chunks.append((bin, encoding)) + + # Turn our chunks back into a Header object, then let that + # Header object do the right thing to turn them into a + # string for us. + value.append(str(email.header.make_header(chunks))) + # This is already a string, so just add it. + else: + value.append(h) + + # We've processed all of our values to get them into a list of str, + # but we may have mojibake data, in which case this is an unparsed + # field. + if not valid_encoding: + unparsed[name] = value + continue + + raw_name = _EMAIL_TO_RAW_MAPPING.get(name) + if raw_name is None: + # This is a bit of a weird situation, we've encountered a key that + # we don't know what it means, so we don't know whether it's meant + # to be a list or not. + # + # Since we can't really tell one way or another, we'll just leave it + # as a list, even though it may be a single item list, because that's + # what makes the most sense for email headers. + unparsed[name] = value + continue + + # If this is one of our string fields, then we'll check to see if our + # value is a list of a single item. If it is then we'll assume that + # it was emitted as a single string, and unwrap the str from inside + # the list. + # + # If it's any other kind of data, then we haven't the faintest clue + # what we should parse it as, and we have to just add it to our list + # of unparsed stuff. + if raw_name in _STRING_FIELDS and len(value) == 1: + raw[raw_name] = value[0] + # If this is one of our list of string fields, then we can just assign + # the value, since email *only* has strings, and our get_all() call + # above ensures that this is a list. + elif raw_name in _LIST_STRING_FIELDS: + raw[raw_name] = value + # Special Case: Keywords + # The keywords field is implemented in the metadata spec as a str, + # but it conceptually is a list of strings, and is serialized using + # ", ".join(keywords), so we'll do some light data massaging to turn + # this into what it logically is. + elif raw_name == "keywords" and len(value) == 1: + raw[raw_name] = _parse_keywords(value[0]) + # Special Case: Project-URL + # The project urls is implemented in the metadata spec as a list of + # specially-formatted strings that represent a key and a value, which + # is fundamentally a mapping, however the email format doesn't support + # mappings in a sane way, so it was crammed into a list of strings + # instead. + # + # We will do a little light data massaging to turn this into a map as + # it logically should be. + elif raw_name == "project_urls": + try: + raw[raw_name] = _parse_project_urls(value) + except KeyError: + unparsed[name] = value + # Nothing that we've done has managed to parse this, so it'll just + # throw it in our unparseable data and move on. + else: + unparsed[name] = value + + # We need to support getting the Description from the message payload in + # addition to getting it from the the headers. This does mean, though, there + # is the possibility of it being set both ways, in which case we put both + # in 'unparsed' since we don't know which is right. + try: + payload = _get_payload(parsed, data) + except ValueError: + unparsed.setdefault("description", []).append( + parsed.get_payload(decode=isinstance(data, bytes)) + ) + else: + if payload: + # Check to see if we've already got a description, if so then both + # it, and this body move to unparseable. + if "description" in raw: + description_header = cast(str, raw.pop("description")) + unparsed.setdefault("description", []).extend( + [description_header, payload] + ) + elif "description" in unparsed: + unparsed["description"].append(payload) + else: + raw["description"] = payload + + # We need to cast our `raw` to a metadata, because a TypedDict only support + # literal key names, but we're computing our key names on purpose, but the + # way this function is implemented, our `TypedDict` can only have valid key + # names. + return cast(RawMetadata, raw), unparsed diff --git a/third_party/python/setuptools/pkg_resources/_vendor/packaging/requirements.py b/third_party/python/setuptools/pkg_resources/_vendor/packaging/requirements.py index 9495a1df1e..f34bfa85c8 100644 --- a/third_party/python/setuptools/pkg_resources/_vendor/packaging/requirements.py +++ b/third_party/python/setuptools/pkg_resources/_vendor/packaging/requirements.py @@ -1,22 +1,14 @@ # This file is dual licensed under the terms of the Apache License, Version # 2.0, and the BSD License. See the LICENSE file in the root of this repository # for complete details. -from __future__ import absolute_import, division, print_function -import string -import re +import urllib.parse +from typing import Any, List, Optional, Set -from pkg_resources.extern.pyparsing import stringStart, stringEnd, originalTextFor, ParseException -from pkg_resources.extern.pyparsing import ZeroOrMore, Word, Optional, Regex, Combine -from pkg_resources.extern.pyparsing import Literal as L # noqa -from urllib import parse as urlparse - -from ._typing import TYPE_CHECKING -from .markers import MARKER_EXPR, Marker -from .specifiers import LegacySpecifier, Specifier, SpecifierSet - -if TYPE_CHECKING: # pragma: no cover - from typing import List +from ._parser import parse_requirement as _parse_requirement +from ._tokenizer import ParserSyntaxError +from .markers import Marker, _normalize_extra_values +from .specifiers import SpecifierSet class InvalidRequirement(ValueError): @@ -25,61 +17,7 @@ class InvalidRequirement(ValueError): """ -ALPHANUM = Word(string.ascii_letters + string.digits) - -LBRACKET = L("[").suppress() -RBRACKET = L("]").suppress() -LPAREN = L("(").suppress() -RPAREN = L(")").suppress() -COMMA = L(",").suppress() -SEMICOLON = L(";").suppress() -AT = L("@").suppress() - -PUNCTUATION = Word("-_.") -IDENTIFIER_END = ALPHANUM | (ZeroOrMore(PUNCTUATION) + ALPHANUM) -IDENTIFIER = Combine(ALPHANUM + ZeroOrMore(IDENTIFIER_END)) - -NAME = IDENTIFIER("name") -EXTRA = IDENTIFIER - -URI = Regex(r"[^ ]+")("url") -URL = AT + URI - -EXTRAS_LIST = EXTRA + ZeroOrMore(COMMA + EXTRA) -EXTRAS = (LBRACKET + Optional(EXTRAS_LIST) + RBRACKET)("extras") - -VERSION_PEP440 = Regex(Specifier._regex_str, re.VERBOSE | re.IGNORECASE) -VERSION_LEGACY = Regex(LegacySpecifier._regex_str, re.VERBOSE | re.IGNORECASE) - -VERSION_ONE = VERSION_PEP440 ^ VERSION_LEGACY -VERSION_MANY = Combine( - VERSION_ONE + ZeroOrMore(COMMA + VERSION_ONE), joinString=",", adjacent=False -)("_raw_spec") -_VERSION_SPEC = Optional(((LPAREN + VERSION_MANY + RPAREN) | VERSION_MANY)) -_VERSION_SPEC.setParseAction(lambda s, l, t: t._raw_spec or "") - -VERSION_SPEC = originalTextFor(_VERSION_SPEC)("specifier") -VERSION_SPEC.setParseAction(lambda s, l, t: t[1]) - -MARKER_EXPR = originalTextFor(MARKER_EXPR())("marker") -MARKER_EXPR.setParseAction( - lambda s, l, t: Marker(s[t._original_start : t._original_end]) -) -MARKER_SEPARATOR = SEMICOLON -MARKER = MARKER_SEPARATOR + MARKER_EXPR - -VERSION_AND_MARKER = VERSION_SPEC + Optional(MARKER) -URL_AND_MARKER = URL + Optional(MARKER) - -NAMED_REQUIREMENT = NAME + Optional(EXTRAS) + (URL_AND_MARKER | VERSION_AND_MARKER) - -REQUIREMENT = stringStart + NAMED_REQUIREMENT + stringEnd -# pkg_resources.extern.pyparsing isn't thread safe during initialization, so we do it eagerly, see -# issue #104 -REQUIREMENT.parseString("x[]") - - -class Requirement(object): +class Requirement: """Parse a requirement. Parse a given requirement string into its parts, such as name, specifier, @@ -92,54 +30,66 @@ class Requirement(object): # the thing as well as the version? What about the markers? # TODO: Can we normalize the name and extra name? - def __init__(self, requirement_string): - # type: (str) -> None + def __init__(self, requirement_string: str) -> None: try: - req = REQUIREMENT.parseString(requirement_string) - except ParseException as e: - raise InvalidRequirement( - 'Parse error at "{0!r}": {1}'.format( - requirement_string[e.loc : e.loc + 8], e.msg - ) - ) - - self.name = req.name - if req.url: - parsed_url = urlparse.urlparse(req.url) + parsed = _parse_requirement(requirement_string) + except ParserSyntaxError as e: + raise InvalidRequirement(str(e)) from e + + self.name: str = parsed.name + if parsed.url: + parsed_url = urllib.parse.urlparse(parsed.url) if parsed_url.scheme == "file": - if urlparse.urlunparse(parsed_url) != req.url: + if urllib.parse.urlunparse(parsed_url) != parsed.url: raise InvalidRequirement("Invalid URL given") elif not (parsed_url.scheme and parsed_url.netloc) or ( not parsed_url.scheme and not parsed_url.netloc ): - raise InvalidRequirement("Invalid URL: {0}".format(req.url)) - self.url = req.url + raise InvalidRequirement(f"Invalid URL: {parsed.url}") + self.url: Optional[str] = parsed.url else: self.url = None - self.extras = set(req.extras.asList() if req.extras else []) - self.specifier = SpecifierSet(req.specifier) - self.marker = req.marker if req.marker else None + self.extras: Set[str] = set(parsed.extras if parsed.extras else []) + self.specifier: SpecifierSet = SpecifierSet(parsed.specifier) + self.marker: Optional[Marker] = None + if parsed.marker is not None: + self.marker = Marker.__new__(Marker) + self.marker._markers = _normalize_extra_values(parsed.marker) - def __str__(self): - # type: () -> str - parts = [self.name] # type: List[str] + def __str__(self) -> str: + parts: List[str] = [self.name] if self.extras: - parts.append("[{0}]".format(",".join(sorted(self.extras)))) + formatted_extras = ",".join(sorted(self.extras)) + parts.append(f"[{formatted_extras}]") if self.specifier: parts.append(str(self.specifier)) if self.url: - parts.append("@ {0}".format(self.url)) + parts.append(f"@ {self.url}") if self.marker: parts.append(" ") if self.marker: - parts.append("; {0}".format(self.marker)) + parts.append(f"; {self.marker}") return "".join(parts) - def __repr__(self): - # type: () -> str - return "<Requirement({0!r})>".format(str(self)) + def __repr__(self) -> str: + return f"<Requirement('{self}')>" + + def __hash__(self) -> int: + return hash((self.__class__.__name__, str(self))) + + def __eq__(self, other: Any) -> bool: + if not isinstance(other, Requirement): + return NotImplemented + + return ( + self.name == other.name + and self.extras == other.extras + and self.specifier == other.specifier + and self.url == other.url + and self.marker == other.marker + ) diff --git a/third_party/python/setuptools/pkg_resources/_vendor/packaging/specifiers.py b/third_party/python/setuptools/pkg_resources/_vendor/packaging/specifiers.py index fe09bb1dbb..ba8fe37b7f 100644 --- a/third_party/python/setuptools/pkg_resources/_vendor/packaging/specifiers.py +++ b/third_party/python/setuptools/pkg_resources/_vendor/packaging/specifiers.py @@ -1,339 +1,123 @@ # This file is dual licensed under the terms of the Apache License, Version # 2.0, and the BSD License. See the LICENSE file in the root of this repository # for complete details. -from __future__ import absolute_import, division, print_function +""" +.. testsetup:: + + from packaging.specifiers import Specifier, SpecifierSet, InvalidSpecifier + from packaging.version import Version +""" import abc -import functools import itertools import re +from typing import ( + Callable, + Iterable, + Iterator, + List, + Optional, + Set, + Tuple, + TypeVar, + Union, +) -from ._compat import string_types, with_metaclass -from ._typing import TYPE_CHECKING from .utils import canonicalize_version -from .version import Version, LegacyVersion, parse - -if TYPE_CHECKING: # pragma: no cover - from typing import ( - List, - Dict, - Union, - Iterable, - Iterator, - Optional, - Callable, - Tuple, - FrozenSet, - ) +from .version import Version + +UnparsedVersion = Union[Version, str] +UnparsedVersionVar = TypeVar("UnparsedVersionVar", bound=UnparsedVersion) +CallableOperator = Callable[[Version, str], bool] - ParsedVersion = Union[Version, LegacyVersion] - UnparsedVersion = Union[Version, LegacyVersion, str] - CallableOperator = Callable[[ParsedVersion, str], bool] + +def _coerce_version(version: UnparsedVersion) -> Version: + if not isinstance(version, Version): + version = Version(version) + return version class InvalidSpecifier(ValueError): """ - An invalid specifier was found, users should refer to PEP 440. + Raised when attempting to create a :class:`Specifier` with a specifier + string that is invalid. + + >>> Specifier("lolwat") + Traceback (most recent call last): + ... + packaging.specifiers.InvalidSpecifier: Invalid specifier: 'lolwat' """ -class BaseSpecifier(with_metaclass(abc.ABCMeta, object)): # type: ignore +class BaseSpecifier(metaclass=abc.ABCMeta): @abc.abstractmethod - def __str__(self): - # type: () -> str + def __str__(self) -> str: """ - Returns the str representation of this Specifier like object. This + Returns the str representation of this Specifier-like object. This should be representative of the Specifier itself. """ @abc.abstractmethod - def __hash__(self): - # type: () -> int + def __hash__(self) -> int: """ - Returns a hash value for this Specifier like object. + Returns a hash value for this Specifier-like object. """ @abc.abstractmethod - def __eq__(self, other): - # type: (object) -> bool + def __eq__(self, other: object) -> bool: """ - Returns a boolean representing whether or not the two Specifier like + Returns a boolean representing whether or not the two Specifier-like objects are equal. + + :param other: The other object to check against. """ + @property @abc.abstractmethod - def __ne__(self, other): - # type: (object) -> bool - """ - Returns a boolean representing whether or not the two Specifier like - objects are not equal. - """ + def prereleases(self) -> Optional[bool]: + """Whether or not pre-releases as a whole are allowed. - @abc.abstractproperty - def prereleases(self): - # type: () -> Optional[bool] - """ - Returns whether or not pre-releases as a whole are allowed by this - specifier. + This can be set to either ``True`` or ``False`` to explicitly enable or disable + prereleases or it can be set to ``None`` (the default) to use default semantics. """ @prereleases.setter - def prereleases(self, value): - # type: (bool) -> None - """ - Sets whether or not pre-releases as a whole are allowed by this - specifier. + def prereleases(self, value: bool) -> None: + """Setter for :attr:`prereleases`. + + :param value: The value to set. """ @abc.abstractmethod - def contains(self, item, prereleases=None): - # type: (str, Optional[bool]) -> bool + def contains(self, item: str, prereleases: Optional[bool] = None) -> bool: """ Determines if the given item is contained within this specifier. """ @abc.abstractmethod - def filter(self, iterable, prereleases=None): - # type: (Iterable[UnparsedVersion], Optional[bool]) -> Iterable[UnparsedVersion] + def filter( + self, iterable: Iterable[UnparsedVersionVar], prereleases: Optional[bool] = None + ) -> Iterator[UnparsedVersionVar]: """ Takes an iterable of items and filters them so that only items which are contained within this specifier are allowed in it. """ -class _IndividualSpecifier(BaseSpecifier): - - _operators = {} # type: Dict[str, str] - - def __init__(self, spec="", prereleases=None): - # type: (str, Optional[bool]) -> None - match = self._regex.search(spec) - if not match: - raise InvalidSpecifier("Invalid specifier: '{0}'".format(spec)) - - self._spec = ( - match.group("operator").strip(), - match.group("version").strip(), - ) # type: Tuple[str, str] +class Specifier(BaseSpecifier): + """This class abstracts handling of version specifiers. - # Store whether or not this Specifier should accept prereleases - self._prereleases = prereleases + .. tip:: - def __repr__(self): - # type: () -> str - pre = ( - ", prereleases={0!r}".format(self.prereleases) - if self._prereleases is not None - else "" - ) - - return "<{0}({1!r}{2})>".format(self.__class__.__name__, str(self), pre) - - def __str__(self): - # type: () -> str - return "{0}{1}".format(*self._spec) - - @property - def _canonical_spec(self): - # type: () -> Tuple[str, Union[Version, str]] - return self._spec[0], canonicalize_version(self._spec[1]) - - def __hash__(self): - # type: () -> int - return hash(self._canonical_spec) - - def __eq__(self, other): - # type: (object) -> bool - if isinstance(other, string_types): - try: - other = self.__class__(str(other)) - except InvalidSpecifier: - return NotImplemented - elif not isinstance(other, self.__class__): - return NotImplemented - - return self._canonical_spec == other._canonical_spec - - def __ne__(self, other): - # type: (object) -> bool - if isinstance(other, string_types): - try: - other = self.__class__(str(other)) - except InvalidSpecifier: - return NotImplemented - elif not isinstance(other, self.__class__): - return NotImplemented - - return self._spec != other._spec - - def _get_operator(self, op): - # type: (str) -> CallableOperator - operator_callable = getattr( - self, "_compare_{0}".format(self._operators[op]) - ) # type: CallableOperator - return operator_callable - - def _coerce_version(self, version): - # type: (UnparsedVersion) -> ParsedVersion - if not isinstance(version, (LegacyVersion, Version)): - version = parse(version) - return version - - @property - def operator(self): - # type: () -> str - return self._spec[0] - - @property - def version(self): - # type: () -> str - return self._spec[1] - - @property - def prereleases(self): - # type: () -> Optional[bool] - return self._prereleases - - @prereleases.setter - def prereleases(self, value): - # type: (bool) -> None - self._prereleases = value - - def __contains__(self, item): - # type: (str) -> bool - return self.contains(item) - - def contains(self, item, prereleases=None): - # type: (UnparsedVersion, Optional[bool]) -> bool - - # Determine if prereleases are to be allowed or not. - if prereleases is None: - prereleases = self.prereleases - - # Normalize item to a Version or LegacyVersion, this allows us to have - # a shortcut for ``"2.0" in Specifier(">=2") - normalized_item = self._coerce_version(item) - - # Determine if we should be supporting prereleases in this specifier - # or not, if we do not support prereleases than we can short circuit - # logic if this version is a prereleases. - if normalized_item.is_prerelease and not prereleases: - return False - - # Actually do the comparison to determine if this item is contained - # within this Specifier or not. - operator_callable = self._get_operator(self.operator) # type: CallableOperator - return operator_callable(normalized_item, self.version) - - def filter(self, iterable, prereleases=None): - # type: (Iterable[UnparsedVersion], Optional[bool]) -> Iterable[UnparsedVersion] - - yielded = False - found_prereleases = [] - - kw = {"prereleases": prereleases if prereleases is not None else True} - - # Attempt to iterate over all the values in the iterable and if any of - # them match, yield them. - for version in iterable: - parsed_version = self._coerce_version(version) - - if self.contains(parsed_version, **kw): - # If our version is a prerelease, and we were not set to allow - # prereleases, then we'll store it for later incase nothing - # else matches this specifier. - if parsed_version.is_prerelease and not ( - prereleases or self.prereleases - ): - found_prereleases.append(version) - # Either this is not a prerelease, or we should have been - # accepting prereleases from the beginning. - else: - yielded = True - yield version - - # Now that we've iterated over everything, determine if we've yielded - # any values, and if we have not and we have any prereleases stored up - # then we will go ahead and yield the prereleases. - if not yielded and found_prereleases: - for version in found_prereleases: - yield version - - -class LegacySpecifier(_IndividualSpecifier): - - _regex_str = r""" - (?P<operator>(==|!=|<=|>=|<|>)) - \s* - (?P<version> - [^,;\s)]* # Since this is a "legacy" specifier, and the version - # string can be just about anything, we match everything - # except for whitespace, a semi-colon for marker support, - # a closing paren since versions can be enclosed in - # them, and a comma since it's a version separator. - ) - """ - - _regex = re.compile(r"^\s*" + _regex_str + r"\s*$", re.VERBOSE | re.IGNORECASE) - - _operators = { - "==": "equal", - "!=": "not_equal", - "<=": "less_than_equal", - ">=": "greater_than_equal", - "<": "less_than", - ">": "greater_than", - } - - def _coerce_version(self, version): - # type: (Union[ParsedVersion, str]) -> LegacyVersion - if not isinstance(version, LegacyVersion): - version = LegacyVersion(str(version)) - return version - - def _compare_equal(self, prospective, spec): - # type: (LegacyVersion, str) -> bool - return prospective == self._coerce_version(spec) - - def _compare_not_equal(self, prospective, spec): - # type: (LegacyVersion, str) -> bool - return prospective != self._coerce_version(spec) - - def _compare_less_than_equal(self, prospective, spec): - # type: (LegacyVersion, str) -> bool - return prospective <= self._coerce_version(spec) - - def _compare_greater_than_equal(self, prospective, spec): - # type: (LegacyVersion, str) -> bool - return prospective >= self._coerce_version(spec) - - def _compare_less_than(self, prospective, spec): - # type: (LegacyVersion, str) -> bool - return prospective < self._coerce_version(spec) - - def _compare_greater_than(self, prospective, spec): - # type: (LegacyVersion, str) -> bool - return prospective > self._coerce_version(spec) - - -def _require_version_compare( - fn # type: (Callable[[Specifier, ParsedVersion, str], bool]) -): - # type: (...) -> Callable[[Specifier, ParsedVersion, str], bool] - @functools.wraps(fn) - def wrapped(self, prospective, spec): - # type: (Specifier, ParsedVersion, str) -> bool - if not isinstance(prospective, Version): - return False - return fn(self, prospective, spec) - - return wrapped - - -class Specifier(_IndividualSpecifier): + It is generally not required to instantiate this manually. You should instead + prefer to work with :class:`SpecifierSet` instead, which can parse + comma-separated version specifiers (which is what package metadata contains). + """ - _regex_str = r""" + _operator_regex_str = r""" (?P<operator>(~=|==|!=|<=|>=|<|>|===)) + """ + _version_regex_str = r""" (?P<version> (?: # The identity operators allow for an escape hatch that will @@ -343,8 +127,10 @@ class Specifier(_IndividualSpecifier): # but included entirely as an escape hatch. (?<====) # Only match for the identity operator \s* - [^\s]* # We just match everything, except for whitespace - # since we are only testing for strict identity. + [^\s;)]* # The arbitrary version can be just about anything, + # we match everything except for whitespace, a + # semi-colon for marker support, and a closing paren + # since versions can be enclosed in them. ) | (?: @@ -357,23 +143,23 @@ class Specifier(_IndividualSpecifier): v? (?:[0-9]+!)? # epoch [0-9]+(?:\.[0-9]+)* # release - (?: # pre release - [-_\.]? - (a|b|c|rc|alpha|beta|pre|preview) - [-_\.]? - [0-9]* - )? - (?: # post release - (?:-[0-9]+)|(?:[-_\.]?(post|rev|r)[-_\.]?[0-9]*) - )? - # You cannot use a wild card and a dev or local version - # together so group them with a | and make them optional. + # You cannot use a wild card and a pre-release, post-release, a dev or + # local version together so group them with a | and make them optional. (?: + \.\* # Wild card syntax of .* + | + (?: # pre release + [-_\.]? + (alpha|beta|preview|pre|a|b|c|rc) + [-_\.]? + [0-9]* + )? + (?: # post release + (?:-[0-9]+)|(?:[-_\.]?(post|rev|r)[-_\.]?[0-9]*) + )? (?:[-_\.]?dev[-_\.]?[0-9]*)? # dev release (?:\+[a-z0-9]+(?:[-_\.][a-z0-9]+)*)? # local - | - \.\* # Wild card syntax of .* )? ) | @@ -388,7 +174,7 @@ class Specifier(_IndividualSpecifier): [0-9]+(?:\.[0-9]+)+ # release (We have a + instead of a *) (?: # pre release [-_\.]? - (a|b|c|rc|alpha|beta|pre|preview) + (alpha|beta|preview|pre|a|b|c|rc) [-_\.]? [0-9]* )? @@ -413,7 +199,7 @@ class Specifier(_IndividualSpecifier): [0-9]+(?:\.[0-9]+)* # release (?: # pre release [-_\.]? - (a|b|c|rc|alpha|beta|pre|preview) + (alpha|beta|preview|pre|a|b|c|rc) [-_\.]? [0-9]* )? @@ -425,7 +211,10 @@ class Specifier(_IndividualSpecifier): ) """ - _regex = re.compile(r"^\s*" + _regex_str + r"\s*$", re.VERBOSE | re.IGNORECASE) + _regex = re.compile( + r"^\s*" + _operator_regex_str + _version_regex_str + r"\s*$", + re.VERBOSE | re.IGNORECASE, + ) _operators = { "~=": "compatible", @@ -438,9 +227,153 @@ class Specifier(_IndividualSpecifier): "===": "arbitrary", } - @_require_version_compare - def _compare_compatible(self, prospective, spec): - # type: (ParsedVersion, str) -> bool + def __init__(self, spec: str = "", prereleases: Optional[bool] = None) -> None: + """Initialize a Specifier instance. + + :param spec: + The string representation of a specifier which will be parsed and + normalized before use. + :param prereleases: + This tells the specifier if it should accept prerelease versions if + applicable or not. The default of ``None`` will autodetect it from the + given specifiers. + :raises InvalidSpecifier: + If the given specifier is invalid (i.e. bad syntax). + """ + match = self._regex.search(spec) + if not match: + raise InvalidSpecifier(f"Invalid specifier: '{spec}'") + + self._spec: Tuple[str, str] = ( + match.group("operator").strip(), + match.group("version").strip(), + ) + + # Store whether or not this Specifier should accept prereleases + self._prereleases = prereleases + + # https://github.com/python/mypy/pull/13475#pullrequestreview-1079784515 + @property # type: ignore[override] + def prereleases(self) -> bool: + # If there is an explicit prereleases set for this, then we'll just + # blindly use that. + if self._prereleases is not None: + return self._prereleases + + # Look at all of our specifiers and determine if they are inclusive + # operators, and if they are if they are including an explicit + # prerelease. + operator, version = self._spec + if operator in ["==", ">=", "<=", "~=", "==="]: + # The == specifier can include a trailing .*, if it does we + # want to remove before parsing. + if operator == "==" and version.endswith(".*"): + version = version[:-2] + + # Parse the version, and if it is a pre-release than this + # specifier allows pre-releases. + if Version(version).is_prerelease: + return True + + return False + + @prereleases.setter + def prereleases(self, value: bool) -> None: + self._prereleases = value + + @property + def operator(self) -> str: + """The operator of this specifier. + + >>> Specifier("==1.2.3").operator + '==' + """ + return self._spec[0] + + @property + def version(self) -> str: + """The version of this specifier. + + >>> Specifier("==1.2.3").version + '1.2.3' + """ + return self._spec[1] + + def __repr__(self) -> str: + """A representation of the Specifier that shows all internal state. + + >>> Specifier('>=1.0.0') + <Specifier('>=1.0.0')> + >>> Specifier('>=1.0.0', prereleases=False) + <Specifier('>=1.0.0', prereleases=False)> + >>> Specifier('>=1.0.0', prereleases=True) + <Specifier('>=1.0.0', prereleases=True)> + """ + pre = ( + f", prereleases={self.prereleases!r}" + if self._prereleases is not None + else "" + ) + + return f"<{self.__class__.__name__}({str(self)!r}{pre})>" + + def __str__(self) -> str: + """A string representation of the Specifier that can be round-tripped. + + >>> str(Specifier('>=1.0.0')) + '>=1.0.0' + >>> str(Specifier('>=1.0.0', prereleases=False)) + '>=1.0.0' + """ + return "{}{}".format(*self._spec) + + @property + def _canonical_spec(self) -> Tuple[str, str]: + canonical_version = canonicalize_version( + self._spec[1], + strip_trailing_zero=(self._spec[0] != "~="), + ) + return self._spec[0], canonical_version + + def __hash__(self) -> int: + return hash(self._canonical_spec) + + def __eq__(self, other: object) -> bool: + """Whether or not the two Specifier-like objects are equal. + + :param other: The other object to check against. + + The value of :attr:`prereleases` is ignored. + + >>> Specifier("==1.2.3") == Specifier("== 1.2.3.0") + True + >>> (Specifier("==1.2.3", prereleases=False) == + ... Specifier("==1.2.3", prereleases=True)) + True + >>> Specifier("==1.2.3") == "==1.2.3" + True + >>> Specifier("==1.2.3") == Specifier("==1.2.4") + False + >>> Specifier("==1.2.3") == Specifier("~=1.2.3") + False + """ + if isinstance(other, str): + try: + other = self.__class__(str(other)) + except InvalidSpecifier: + return NotImplemented + elif not isinstance(other, self.__class__): + return NotImplemented + + return self._canonical_spec == other._canonical_spec + + def _get_operator(self, op: str) -> CallableOperator: + operator_callable: CallableOperator = getattr( + self, f"_compare_{self._operators[op]}" + ) + return operator_callable + + def _compare_compatible(self, prospective: Version, spec: str) -> bool: # Compatible releases have an equivalent combination of >= and ==. That # is that ~=2.2 is equivalent to >=2.2,==2.*. This allows us to @@ -449,15 +382,9 @@ class Specifier(_IndividualSpecifier): # the other specifiers. # We want everything but the last item in the version, but we want to - # ignore post and dev releases and we want to treat the pre-release as - # it's own separate segment. + # ignore suffix segments. prefix = ".".join( - list( - itertools.takewhile( - lambda x: (not x.startswith("post") and not x.startswith("dev")), - _version_split(spec), - ) - )[:-1] + list(itertools.takewhile(_is_not_suffix, _version_split(spec)))[:-1] ) # Add the prefix notation to the end of our string @@ -467,35 +394,35 @@ class Specifier(_IndividualSpecifier): prospective, prefix ) - @_require_version_compare - def _compare_equal(self, prospective, spec): - # type: (ParsedVersion, str) -> bool + def _compare_equal(self, prospective: Version, spec: str) -> bool: # We need special logic to handle prefix matching if spec.endswith(".*"): # In the case of prefix matching we want to ignore local segment. - prospective = Version(prospective.public) + normalized_prospective = canonicalize_version( + prospective.public, strip_trailing_zero=False + ) + # Get the normalized version string ignoring the trailing .* + normalized_spec = canonicalize_version(spec[:-2], strip_trailing_zero=False) # Split the spec out by dots, and pretend that there is an implicit # dot in between a release segment and a pre-release segment. - split_spec = _version_split(spec[:-2]) # Remove the trailing .* + split_spec = _version_split(normalized_spec) # Split the prospective version out by dots, and pretend that there # is an implicit dot in between a release segment and a pre-release # segment. - split_prospective = _version_split(str(prospective)) + split_prospective = _version_split(normalized_prospective) + + # 0-pad the prospective version before shortening it to get the correct + # shortened version. + padded_prospective, _ = _pad_version(split_prospective, split_spec) # Shorten the prospective version to be the same length as the spec # so that we can determine if the specifier is a prefix of the # prospective version or not. - shortened_prospective = split_prospective[: len(split_spec)] - - # Pad out our two sides with zeros so that they both equal the same - # length. - padded_spec, padded_prospective = _pad_version( - split_spec, shortened_prospective - ) + shortened_prospective = padded_prospective[: len(split_spec)] - return padded_prospective == padded_spec + return shortened_prospective == split_spec else: # Convert our spec string into a Version spec_version = Version(spec) @@ -508,32 +435,24 @@ class Specifier(_IndividualSpecifier): return prospective == spec_version - @_require_version_compare - def _compare_not_equal(self, prospective, spec): - # type: (ParsedVersion, str) -> bool + def _compare_not_equal(self, prospective: Version, spec: str) -> bool: return not self._compare_equal(prospective, spec) - @_require_version_compare - def _compare_less_than_equal(self, prospective, spec): - # type: (ParsedVersion, str) -> bool + def _compare_less_than_equal(self, prospective: Version, spec: str) -> bool: # NB: Local version identifiers are NOT permitted in the version # specifier, so local version labels can be universally removed from # the prospective version. return Version(prospective.public) <= Version(spec) - @_require_version_compare - def _compare_greater_than_equal(self, prospective, spec): - # type: (ParsedVersion, str) -> bool + def _compare_greater_than_equal(self, prospective: Version, spec: str) -> bool: # NB: Local version identifiers are NOT permitted in the version # specifier, so local version labels can be universally removed from # the prospective version. return Version(prospective.public) >= Version(spec) - @_require_version_compare - def _compare_less_than(self, prospective, spec_str): - # type: (ParsedVersion, str) -> bool + def _compare_less_than(self, prospective: Version, spec_str: str) -> bool: # Convert our spec to a Version instance, since we'll want to work with # it as a version. @@ -558,9 +477,7 @@ class Specifier(_IndividualSpecifier): # version in the spec. return True - @_require_version_compare - def _compare_greater_than(self, prospective, spec_str): - # type: (ParsedVersion, str) -> bool + def _compare_greater_than(self, prospective: Version, spec_str: str) -> bool: # Convert our spec to a Version instance, since we'll want to work with # it as a version. @@ -591,48 +508,143 @@ class Specifier(_IndividualSpecifier): # same version in the spec. return True - def _compare_arbitrary(self, prospective, spec): - # type: (Version, str) -> bool + def _compare_arbitrary(self, prospective: Version, spec: str) -> bool: return str(prospective).lower() == str(spec).lower() - @property - def prereleases(self): - # type: () -> bool + def __contains__(self, item: Union[str, Version]) -> bool: + """Return whether or not the item is contained in this specifier. - # If there is an explicit prereleases set for this, then we'll just - # blindly use that. - if self._prereleases is not None: - return self._prereleases + :param item: The item to check for. - # Look at all of our specifiers and determine if they are inclusive - # operators, and if they are if they are including an explicit - # prerelease. - operator, version = self._spec - if operator in ["==", ">=", "<=", "~=", "==="]: - # The == specifier can include a trailing .*, if it does we - # want to remove before parsing. - if operator == "==" and version.endswith(".*"): - version = version[:-2] + This is used for the ``in`` operator and behaves the same as + :meth:`contains` with no ``prereleases`` argument passed. - # Parse the version, and if it is a pre-release than this - # specifier allows pre-releases. - if parse(version).is_prerelease: - return True + >>> "1.2.3" in Specifier(">=1.2.3") + True + >>> Version("1.2.3") in Specifier(">=1.2.3") + True + >>> "1.0.0" in Specifier(">=1.2.3") + False + >>> "1.3.0a1" in Specifier(">=1.2.3") + False + >>> "1.3.0a1" in Specifier(">=1.2.3", prereleases=True) + True + """ + return self.contains(item) - return False + def contains( + self, item: UnparsedVersion, prereleases: Optional[bool] = None + ) -> bool: + """Return whether or not the item is contained in this specifier. + + :param item: + The item to check for, which can be a version string or a + :class:`Version` instance. + :param prereleases: + Whether or not to match prereleases with this Specifier. If set to + ``None`` (the default), it uses :attr:`prereleases` to determine + whether or not prereleases are allowed. + + >>> Specifier(">=1.2.3").contains("1.2.3") + True + >>> Specifier(">=1.2.3").contains(Version("1.2.3")) + True + >>> Specifier(">=1.2.3").contains("1.0.0") + False + >>> Specifier(">=1.2.3").contains("1.3.0a1") + False + >>> Specifier(">=1.2.3", prereleases=True).contains("1.3.0a1") + True + >>> Specifier(">=1.2.3").contains("1.3.0a1", prereleases=True) + True + """ - @prereleases.setter - def prereleases(self, value): - # type: (bool) -> None - self._prereleases = value + # Determine if prereleases are to be allowed or not. + if prereleases is None: + prereleases = self.prereleases + + # Normalize item to a Version, this allows us to have a shortcut for + # "2.0" in Specifier(">=2") + normalized_item = _coerce_version(item) + + # Determine if we should be supporting prereleases in this specifier + # or not, if we do not support prereleases than we can short circuit + # logic if this version is a prereleases. + if normalized_item.is_prerelease and not prereleases: + return False + + # Actually do the comparison to determine if this item is contained + # within this Specifier or not. + operator_callable: CallableOperator = self._get_operator(self.operator) + return operator_callable(normalized_item, self.version) + + def filter( + self, iterable: Iterable[UnparsedVersionVar], prereleases: Optional[bool] = None + ) -> Iterator[UnparsedVersionVar]: + """Filter items in the given iterable, that match the specifier. + + :param iterable: + An iterable that can contain version strings and :class:`Version` instances. + The items in the iterable will be filtered according to the specifier. + :param prereleases: + Whether or not to allow prereleases in the returned iterator. If set to + ``None`` (the default), it will be intelligently decide whether to allow + prereleases or not (based on the :attr:`prereleases` attribute, and + whether the only versions matching are prereleases). + + This method is smarter than just ``filter(Specifier().contains, [...])`` + because it implements the rule from :pep:`440` that a prerelease item + SHOULD be accepted if no other versions match the given specifier. + + >>> list(Specifier(">=1.2.3").filter(["1.2", "1.3", "1.5a1"])) + ['1.3'] + >>> list(Specifier(">=1.2.3").filter(["1.2", "1.2.3", "1.3", Version("1.4")])) + ['1.2.3', '1.3', <Version('1.4')>] + >>> list(Specifier(">=1.2.3").filter(["1.2", "1.5a1"])) + ['1.5a1'] + >>> list(Specifier(">=1.2.3").filter(["1.3", "1.5a1"], prereleases=True)) + ['1.3', '1.5a1'] + >>> list(Specifier(">=1.2.3", prereleases=True).filter(["1.3", "1.5a1"])) + ['1.3', '1.5a1'] + """ + + yielded = False + found_prereleases = [] + + kw = {"prereleases": prereleases if prereleases is not None else True} + + # Attempt to iterate over all the values in the iterable and if any of + # them match, yield them. + for version in iterable: + parsed_version = _coerce_version(version) + + if self.contains(parsed_version, **kw): + # If our version is a prerelease, and we were not set to allow + # prereleases, then we'll store it for later in case nothing + # else matches this specifier. + if parsed_version.is_prerelease and not ( + prereleases or self.prereleases + ): + found_prereleases.append(version) + # Either this is not a prerelease, or we should have been + # accepting prereleases from the beginning. + else: + yielded = True + yield version + + # Now that we've iterated over everything, determine if we've yielded + # any values, and if we have not and we have any prereleases stored up + # then we will go ahead and yield the prereleases. + if not yielded and found_prereleases: + for version in found_prereleases: + yield version _prefix_regex = re.compile(r"^([0-9]+)((?:a|b|c|rc)[0-9]+)$") -def _version_split(version): - # type: (str) -> List[str] - result = [] # type: List[str] +def _version_split(version: str) -> List[str]: + result: List[str] = [] for item in version.split("."): match = _prefix_regex.search(item) if match: @@ -642,8 +654,13 @@ def _version_split(version): return result -def _pad_version(left, right): - # type: (List[str], List[str]) -> Tuple[List[str], List[str]] +def _is_not_suffix(segment: str) -> bool: + return not any( + segment.startswith(prefix) for prefix in ("dev", "a", "b", "rc", "post") + ) + + +def _pad_version(left: List[str], right: List[str]) -> Tuple[List[str], List[str]]: left_split, right_split = [], [] # Get the release segment of our versions @@ -662,21 +679,39 @@ def _pad_version(left, right): class SpecifierSet(BaseSpecifier): - def __init__(self, specifiers="", prereleases=None): - # type: (str, Optional[bool]) -> None + """This class abstracts handling of a set of version specifiers. + + It can be passed a single specifier (``>=3.0``), a comma-separated list of + specifiers (``>=3.0,!=3.1``), or no specifier at all. + """ - # Split on , to break each individual specifier into it's own item, and + def __init__( + self, specifiers: str = "", prereleases: Optional[bool] = None + ) -> None: + """Initialize a SpecifierSet instance. + + :param specifiers: + The string representation of a specifier or a comma-separated list of + specifiers which will be parsed and normalized before use. + :param prereleases: + This tells the SpecifierSet if it should accept prerelease versions if + applicable or not. The default of ``None`` will autodetect it from the + given specifiers. + + :raises InvalidSpecifier: + If the given ``specifiers`` are not parseable than this exception will be + raised. + """ + + # Split on `,` to break each individual specifier into it's own item, and # strip each item to remove leading/trailing whitespace. split_specifiers = [s.strip() for s in specifiers.split(",") if s.strip()] # Parsed each individual specifier, attempting first to make it a - # Specifier and falling back to a LegacySpecifier. - parsed = set() + # Specifier. + parsed: Set[Specifier] = set() for specifier in split_specifiers: - try: - parsed.add(Specifier(specifier)) - except InvalidSpecifier: - parsed.add(LegacySpecifier(specifier)) + parsed.add(Specifier(specifier)) # Turn our parsed specifiers into a frozen set and save them for later. self._specs = frozenset(parsed) @@ -685,27 +720,75 @@ class SpecifierSet(BaseSpecifier): # we accept prereleases or not. self._prereleases = prereleases - def __repr__(self): - # type: () -> str + @property + def prereleases(self) -> Optional[bool]: + # If we have been given an explicit prerelease modifier, then we'll + # pass that through here. + if self._prereleases is not None: + return self._prereleases + + # If we don't have any specifiers, and we don't have a forced value, + # then we'll just return None since we don't know if this should have + # pre-releases or not. + if not self._specs: + return None + + # Otherwise we'll see if any of the given specifiers accept + # prereleases, if any of them do we'll return True, otherwise False. + return any(s.prereleases for s in self._specs) + + @prereleases.setter + def prereleases(self, value: bool) -> None: + self._prereleases = value + + def __repr__(self) -> str: + """A representation of the specifier set that shows all internal state. + + Note that the ordering of the individual specifiers within the set may not + match the input string. + + >>> SpecifierSet('>=1.0.0,!=2.0.0') + <SpecifierSet('!=2.0.0,>=1.0.0')> + >>> SpecifierSet('>=1.0.0,!=2.0.0', prereleases=False) + <SpecifierSet('!=2.0.0,>=1.0.0', prereleases=False)> + >>> SpecifierSet('>=1.0.0,!=2.0.0', prereleases=True) + <SpecifierSet('!=2.0.0,>=1.0.0', prereleases=True)> + """ pre = ( - ", prereleases={0!r}".format(self.prereleases) + f", prereleases={self.prereleases!r}" if self._prereleases is not None else "" ) - return "<SpecifierSet({0!r}{1})>".format(str(self), pre) + return f"<SpecifierSet({str(self)!r}{pre})>" - def __str__(self): - # type: () -> str + def __str__(self) -> str: + """A string representation of the specifier set that can be round-tripped. + + Note that the ordering of the individual specifiers within the set may not + match the input string. + + >>> str(SpecifierSet(">=1.0.0,!=1.0.1")) + '!=1.0.1,>=1.0.0' + >>> str(SpecifierSet(">=1.0.0,!=1.0.1", prereleases=False)) + '!=1.0.1,>=1.0.0' + """ return ",".join(sorted(str(s) for s in self._specs)) - def __hash__(self): - # type: () -> int + def __hash__(self) -> int: return hash(self._specs) - def __and__(self, other): - # type: (Union[SpecifierSet, str]) -> SpecifierSet - if isinstance(other, string_types): + def __and__(self, other: Union["SpecifierSet", str]) -> "SpecifierSet": + """Return a SpecifierSet which is a combination of the two sets. + + :param other: The other object to combine with. + + >>> SpecifierSet(">=1.0.0,!=1.0.1") & '<=2.0.0,!=2.0.1' + <SpecifierSet('!=1.0.1,!=2.0.1,<=2.0.0,>=1.0.0')> + >>> SpecifierSet(">=1.0.0,!=1.0.1") & SpecifierSet('<=2.0.0,!=2.0.1') + <SpecifierSet('!=1.0.1,!=2.0.1,<=2.0.0,>=1.0.0')> + """ + if isinstance(other, str): other = SpecifierSet(other) elif not isinstance(other, SpecifierSet): return NotImplemented @@ -727,66 +810,99 @@ class SpecifierSet(BaseSpecifier): return specifier - def __eq__(self, other): - # type: (object) -> bool - if isinstance(other, (string_types, _IndividualSpecifier)): - other = SpecifierSet(str(other)) - elif not isinstance(other, SpecifierSet): - return NotImplemented + def __eq__(self, other: object) -> bool: + """Whether or not the two SpecifierSet-like objects are equal. - return self._specs == other._specs + :param other: The other object to check against. + + The value of :attr:`prereleases` is ignored. - def __ne__(self, other): - # type: (object) -> bool - if isinstance(other, (string_types, _IndividualSpecifier)): + >>> SpecifierSet(">=1.0.0,!=1.0.1") == SpecifierSet(">=1.0.0,!=1.0.1") + True + >>> (SpecifierSet(">=1.0.0,!=1.0.1", prereleases=False) == + ... SpecifierSet(">=1.0.0,!=1.0.1", prereleases=True)) + True + >>> SpecifierSet(">=1.0.0,!=1.0.1") == ">=1.0.0,!=1.0.1" + True + >>> SpecifierSet(">=1.0.0,!=1.0.1") == SpecifierSet(">=1.0.0") + False + >>> SpecifierSet(">=1.0.0,!=1.0.1") == SpecifierSet(">=1.0.0,!=1.0.2") + False + """ + if isinstance(other, (str, Specifier)): other = SpecifierSet(str(other)) elif not isinstance(other, SpecifierSet): return NotImplemented - return self._specs != other._specs + return self._specs == other._specs - def __len__(self): - # type: () -> int + def __len__(self) -> int: + """Returns the number of specifiers in this specifier set.""" return len(self._specs) - def __iter__(self): - # type: () -> Iterator[FrozenSet[_IndividualSpecifier]] - return iter(self._specs) - - @property - def prereleases(self): - # type: () -> Optional[bool] + def __iter__(self) -> Iterator[Specifier]: + """ + Returns an iterator over all the underlying :class:`Specifier` instances + in this specifier set. - # If we have been given an explicit prerelease modifier, then we'll - # pass that through here. - if self._prereleases is not None: - return self._prereleases + >>> sorted(SpecifierSet(">=1.0.0,!=1.0.1"), key=str) + [<Specifier('!=1.0.1')>, <Specifier('>=1.0.0')>] + """ + return iter(self._specs) - # If we don't have any specifiers, and we don't have a forced value, - # then we'll just return None since we don't know if this should have - # pre-releases or not. - if not self._specs: - return None + def __contains__(self, item: UnparsedVersion) -> bool: + """Return whether or not the item is contained in this specifier. - # Otherwise we'll see if any of the given specifiers accept - # prereleases, if any of them do we'll return True, otherwise False. - return any(s.prereleases for s in self._specs) + :param item: The item to check for. - @prereleases.setter - def prereleases(self, value): - # type: (bool) -> None - self._prereleases = value + This is used for the ``in`` operator and behaves the same as + :meth:`contains` with no ``prereleases`` argument passed. - def __contains__(self, item): - # type: (Union[ParsedVersion, str]) -> bool + >>> "1.2.3" in SpecifierSet(">=1.0.0,!=1.0.1") + True + >>> Version("1.2.3") in SpecifierSet(">=1.0.0,!=1.0.1") + True + >>> "1.0.1" in SpecifierSet(">=1.0.0,!=1.0.1") + False + >>> "1.3.0a1" in SpecifierSet(">=1.0.0,!=1.0.1") + False + >>> "1.3.0a1" in SpecifierSet(">=1.0.0,!=1.0.1", prereleases=True) + True + """ return self.contains(item) - def contains(self, item, prereleases=None): - # type: (Union[ParsedVersion, str], Optional[bool]) -> bool - - # Ensure that our item is a Version or LegacyVersion instance. - if not isinstance(item, (LegacyVersion, Version)): - item = parse(item) + def contains( + self, + item: UnparsedVersion, + prereleases: Optional[bool] = None, + installed: Optional[bool] = None, + ) -> bool: + """Return whether or not the item is contained in this SpecifierSet. + + :param item: + The item to check for, which can be a version string or a + :class:`Version` instance. + :param prereleases: + Whether or not to match prereleases with this SpecifierSet. If set to + ``None`` (the default), it uses :attr:`prereleases` to determine + whether or not prereleases are allowed. + + >>> SpecifierSet(">=1.0.0,!=1.0.1").contains("1.2.3") + True + >>> SpecifierSet(">=1.0.0,!=1.0.1").contains(Version("1.2.3")) + True + >>> SpecifierSet(">=1.0.0,!=1.0.1").contains("1.0.1") + False + >>> SpecifierSet(">=1.0.0,!=1.0.1").contains("1.3.0a1") + False + >>> SpecifierSet(">=1.0.0,!=1.0.1", prereleases=True).contains("1.3.0a1") + True + >>> SpecifierSet(">=1.0.0,!=1.0.1").contains("1.3.0a1", prereleases=True) + True + """ + # Ensure that our item is a Version instance. + if not isinstance(item, Version): + item = Version(item) # Determine if we're forcing a prerelease or not, if we're not forcing # one for this particular filter call, then we'll use whatever the @@ -803,6 +919,9 @@ class SpecifierSet(BaseSpecifier): if not prereleases and item.is_prerelease: return False + if installed and item.is_prerelease: + item = Version(item.base_version) + # We simply dispatch to the underlying specs here to make sure that the # given version is contained within all of them. # Note: This use of all() here means that an empty set of specifiers @@ -810,12 +929,46 @@ class SpecifierSet(BaseSpecifier): return all(s.contains(item, prereleases=prereleases) for s in self._specs) def filter( - self, - iterable, # type: Iterable[Union[ParsedVersion, str]] - prereleases=None, # type: Optional[bool] - ): - # type: (...) -> Iterable[Union[ParsedVersion, str]] - + self, iterable: Iterable[UnparsedVersionVar], prereleases: Optional[bool] = None + ) -> Iterator[UnparsedVersionVar]: + """Filter items in the given iterable, that match the specifiers in this set. + + :param iterable: + An iterable that can contain version strings and :class:`Version` instances. + The items in the iterable will be filtered according to the specifier. + :param prereleases: + Whether or not to allow prereleases in the returned iterator. If set to + ``None`` (the default), it will be intelligently decide whether to allow + prereleases or not (based on the :attr:`prereleases` attribute, and + whether the only versions matching are prereleases). + + This method is smarter than just ``filter(SpecifierSet(...).contains, [...])`` + because it implements the rule from :pep:`440` that a prerelease item + SHOULD be accepted if no other versions match the given specifier. + + >>> list(SpecifierSet(">=1.2.3").filter(["1.2", "1.3", "1.5a1"])) + ['1.3'] + >>> list(SpecifierSet(">=1.2.3").filter(["1.2", "1.3", Version("1.4")])) + ['1.3', <Version('1.4')>] + >>> list(SpecifierSet(">=1.2.3").filter(["1.2", "1.5a1"])) + [] + >>> list(SpecifierSet(">=1.2.3").filter(["1.3", "1.5a1"], prereleases=True)) + ['1.3', '1.5a1'] + >>> list(SpecifierSet(">=1.2.3", prereleases=True).filter(["1.3", "1.5a1"])) + ['1.3', '1.5a1'] + + An "empty" SpecifierSet will filter items based on the presence of prerelease + versions in the set. + + >>> list(SpecifierSet("").filter(["1.3", "1.5a1"])) + ['1.3'] + >>> list(SpecifierSet("").filter(["1.5a1"])) + ['1.5a1'] + >>> list(SpecifierSet("", prereleases=True).filter(["1.3", "1.5a1"])) + ['1.3', '1.5a1'] + >>> list(SpecifierSet("").filter(["1.3", "1.5a1"], prereleases=True)) + ['1.3', '1.5a1'] + """ # Determine if we're forcing a prerelease or not, if we're not forcing # one for this particular filter call, then we'll use whatever the # SpecifierSet thinks for whether or not we should support prereleases. @@ -828,24 +981,16 @@ class SpecifierSet(BaseSpecifier): if self._specs: for spec in self._specs: iterable = spec.filter(iterable, prereleases=bool(prereleases)) - return iterable + return iter(iterable) # If we do not have any specifiers, then we need to have a rough filter # which will filter out any pre-releases, unless there are no final - # releases, and which will filter out LegacyVersion in general. + # releases. else: - filtered = [] # type: List[Union[ParsedVersion, str]] - found_prereleases = [] # type: List[Union[ParsedVersion, str]] + filtered: List[UnparsedVersionVar] = [] + found_prereleases: List[UnparsedVersionVar] = [] for item in iterable: - # Ensure that we some kind of Version class for this item. - if not isinstance(item, (LegacyVersion, Version)): - parsed_version = parse(item) - else: - parsed_version = item - - # Filter out any item which is parsed as a LegacyVersion - if isinstance(parsed_version, LegacyVersion): - continue + parsed_version = _coerce_version(item) # Store any item which is a pre-release for later unless we've # already found a final version or we are accepting prereleases @@ -858,6 +1003,6 @@ class SpecifierSet(BaseSpecifier): # If we've found no items except for pre-releases, then we'll go # ahead and use the pre-releases if not filtered and found_prereleases and prereleases is None: - return found_prereleases + return iter(found_prereleases) - return filtered + return iter(filtered) diff --git a/third_party/python/setuptools/pkg_resources/_vendor/packaging/tags.py b/third_party/python/setuptools/pkg_resources/_vendor/packaging/tags.py index 9064910b8b..76d243414d 100644 --- a/third_party/python/setuptools/pkg_resources/_vendor/packaging/tags.py +++ b/third_party/python/setuptools/pkg_resources/_vendor/packaging/tags.py @@ -2,62 +2,45 @@ # 2.0, and the BSD License. See the LICENSE file in the root of this repository # for complete details. -from __future__ import absolute_import - -import distutils.util - -try: - from importlib.machinery import EXTENSION_SUFFIXES -except ImportError: # pragma: no cover - import imp - - EXTENSION_SUFFIXES = [x[0] for x in imp.get_suffixes()] - del imp import logging -import os import platform -import re -import struct +import subprocess import sys import sysconfig -import warnings - -from ._typing import TYPE_CHECKING, cast - -if TYPE_CHECKING: # pragma: no cover - from typing import ( - Dict, - FrozenSet, - IO, - Iterable, - Iterator, - List, - Optional, - Sequence, - Tuple, - Union, - ) - - PythonVersion = Sequence[int] - MacVersion = Tuple[int, int] - GlibcVersion = Tuple[int, int] - +from importlib.machinery import EXTENSION_SUFFIXES +from typing import ( + Dict, + FrozenSet, + Iterable, + Iterator, + List, + Optional, + Sequence, + Tuple, + Union, + cast, +) + +from . import _manylinux, _musllinux logger = logging.getLogger(__name__) -INTERPRETER_SHORT_NAMES = { +PythonVersion = Sequence[int] +MacVersion = Tuple[int, int] + +INTERPRETER_SHORT_NAMES: Dict[str, str] = { "python": "py", # Generic. "cpython": "cp", "pypy": "pp", "ironpython": "ip", "jython": "jy", -} # type: Dict[str, str] +} -_32_BIT_INTERPRETER = sys.maxsize <= 2 ** 32 +_32_BIT_INTERPRETER = sys.maxsize <= 2**32 -class Tag(object): +class Tag: """ A representation of the tag triple for a wheel. @@ -65,55 +48,53 @@ class Tag(object): is also supported. """ - __slots__ = ["_interpreter", "_abi", "_platform"] + __slots__ = ["_interpreter", "_abi", "_platform", "_hash"] - def __init__(self, interpreter, abi, platform): - # type: (str, str, str) -> None + def __init__(self, interpreter: str, abi: str, platform: str) -> None: self._interpreter = interpreter.lower() self._abi = abi.lower() self._platform = platform.lower() + # The __hash__ of every single element in a Set[Tag] will be evaluated each time + # that a set calls its `.disjoint()` method, which may be called hundreds of + # times when scanning a page of links for packages with tags matching that + # Set[Tag]. Pre-computing the value here produces significant speedups for + # downstream consumers. + self._hash = hash((self._interpreter, self._abi, self._platform)) @property - def interpreter(self): - # type: () -> str + def interpreter(self) -> str: return self._interpreter @property - def abi(self): - # type: () -> str + def abi(self) -> str: return self._abi @property - def platform(self): - # type: () -> str + def platform(self) -> str: return self._platform - def __eq__(self, other): - # type: (object) -> bool + def __eq__(self, other: object) -> bool: if not isinstance(other, Tag): return NotImplemented return ( - (self.platform == other.platform) - and (self.abi == other.abi) - and (self.interpreter == other.interpreter) + (self._hash == other._hash) # Short-circuit ASAP for perf reasons. + and (self._platform == other._platform) + and (self._abi == other._abi) + and (self._interpreter == other._interpreter) ) - def __hash__(self): - # type: () -> int - return hash((self._interpreter, self._abi, self._platform)) + def __hash__(self) -> int: + return self._hash - def __str__(self): - # type: () -> str - return "{}-{}-{}".format(self._interpreter, self._abi, self._platform) + def __str__(self) -> str: + return f"{self._interpreter}-{self._abi}-{self._platform}" - def __repr__(self): - # type: () -> str - return "<{self} @ {self_id}>".format(self=self, self_id=id(self)) + def __repr__(self) -> str: + return f"<{self} @ {id(self)}>" -def parse_tag(tag): - # type: (str) -> FrozenSet[Tag] +def parse_tag(tag: str) -> FrozenSet[Tag]: """ Parses the provided tag (e.g. `py3-none-any`) into a frozenset of Tag instances. @@ -129,25 +110,8 @@ def parse_tag(tag): return frozenset(tags) -def _warn_keyword_parameter(func_name, kwargs): - # type: (str, Dict[str, bool]) -> bool - """ - Backwards-compatibility with Python 2.7 to allow treating 'warn' as keyword-only. - """ - if not kwargs: - return False - elif len(kwargs) > 1 or "warn" not in kwargs: - kwargs.pop("warn", None) - arg = next(iter(kwargs.keys())) - raise TypeError( - "{}() got an unexpected keyword argument {!r}".format(func_name, arg) - ) - return kwargs["warn"] - - -def _get_config_var(name, warn=False): - # type: (str, bool) -> Union[int, str, None] - value = sysconfig.get_config_var(name) +def _get_config_var(name: str, warn: bool = False) -> Union[int, str, None]: + value: Union[int, str, None] = sysconfig.get_config_var(name) if value is None and warn: logger.debug( "Config variable '%s' is unset, Python ABI tag may be incorrect", name @@ -155,13 +119,11 @@ def _get_config_var(name, warn=False): return value -def _normalize_string(string): - # type: (str) -> str - return string.replace(".", "_").replace("-", "_") +def _normalize_string(string: str) -> str: + return string.replace(".", "_").replace("-", "_").replace(" ", "_") -def _abi3_applies(python_version): - # type: (PythonVersion) -> bool +def _abi3_applies(python_version: PythonVersion) -> bool: """ Determine if the Python version supports abi3. @@ -170,8 +132,7 @@ def _abi3_applies(python_version): return len(python_version) > 1 and tuple(python_version) >= (3, 2) -def _cpython_abis(py_version, warn=False): - # type: (PythonVersion, bool) -> List[str] +def _cpython_abis(py_version: PythonVersion, warn: bool = False) -> List[str]: py_version = tuple(py_version) # To allow for version comparison. abis = [] version = _version_nodot(py_version[:2]) @@ -197,7 +158,7 @@ def _cpython_abis(py_version, warn=False): elif debug: # Debug builds can also load "normal" extension modules. # We can also assume no UCS-4 or pymalloc requirement. - abis.append("cp{version}".format(version=version)) + abis.append(f"cp{version}") abis.insert( 0, "cp{version}{debug}{pymalloc}{ucs4}".format( @@ -208,12 +169,12 @@ def _cpython_abis(py_version, warn=False): def cpython_tags( - python_version=None, # type: Optional[PythonVersion] - abis=None, # type: Optional[Iterable[str]] - platforms=None, # type: Optional[Iterable[str]] - **kwargs # type: bool -): - # type: (...) -> Iterator[Tag] + python_version: Optional[PythonVersion] = None, + abis: Optional[Iterable[str]] = None, + platforms: Optional[Iterable[str]] = None, + *, + warn: bool = False, +) -> Iterator[Tag]: """ Yields the tags for a CPython interpreter. @@ -229,11 +190,10 @@ def cpython_tags( If 'abi3' or 'none' are specified in 'abis' then they will be yielded at their normal position and not at the beginning. """ - warn = _warn_keyword_parameter("cpython_tags", kwargs) if not python_version: python_version = sys.version_info[:2] - interpreter = "cp{}".format(_version_nodot(python_version[:2])) + interpreter = f"cp{_version_nodot(python_version[:2])}" if abis is None: if len(python_version) > 1: @@ -248,15 +208,13 @@ def cpython_tags( except ValueError: pass - platforms = list(platforms or _platform_tags()) + platforms = list(platforms or platform_tags()) for abi in abis: for platform_ in platforms: yield Tag(interpreter, abi, platform_) if _abi3_applies(python_version): - for tag in (Tag(interpreter, "abi3", platform_) for platform_ in platforms): - yield tag - for tag in (Tag(interpreter, "none", platform_) for platform_ in platforms): - yield tag + yield from (Tag(interpreter, "abi3", platform_) for platform_ in platforms) + yield from (Tag(interpreter, "none", platform_) for platform_ in platforms) if _abi3_applies(python_version): for minor_version in range(python_version[1] - 1, 1, -1): @@ -267,20 +225,54 @@ def cpython_tags( yield Tag(interpreter, "abi3", platform_) -def _generic_abi(): - # type: () -> Iterator[str] - abi = sysconfig.get_config_var("SOABI") - if abi: - yield _normalize_string(abi) +def _generic_abi() -> List[str]: + """ + Return the ABI tag based on EXT_SUFFIX. + """ + # The following are examples of `EXT_SUFFIX`. + # We want to keep the parts which are related to the ABI and remove the + # parts which are related to the platform: + # - linux: '.cpython-310-x86_64-linux-gnu.so' => cp310 + # - mac: '.cpython-310-darwin.so' => cp310 + # - win: '.cp310-win_amd64.pyd' => cp310 + # - win: '.pyd' => cp37 (uses _cpython_abis()) + # - pypy: '.pypy38-pp73-x86_64-linux-gnu.so' => pypy38_pp73 + # - graalpy: '.graalpy-38-native-x86_64-darwin.dylib' + # => graalpy_38_native + + ext_suffix = _get_config_var("EXT_SUFFIX", warn=True) + if not isinstance(ext_suffix, str) or ext_suffix[0] != ".": + raise SystemError("invalid sysconfig.get_config_var('EXT_SUFFIX')") + parts = ext_suffix.split(".") + if len(parts) < 3: + # CPython3.7 and earlier uses ".pyd" on Windows. + return _cpython_abis(sys.version_info[:2]) + soabi = parts[1] + if soabi.startswith("cpython"): + # non-windows + abi = "cp" + soabi.split("-")[1] + elif soabi.startswith("cp"): + # windows + abi = soabi.split("-")[0] + elif soabi.startswith("pypy"): + abi = "-".join(soabi.split("-")[:2]) + elif soabi.startswith("graalpy"): + abi = "-".join(soabi.split("-")[:3]) + elif soabi: + # pyston, ironpython, others? + abi = soabi + else: + return [] + return [_normalize_string(abi)] def generic_tags( - interpreter=None, # type: Optional[str] - abis=None, # type: Optional[Iterable[str]] - platforms=None, # type: Optional[Iterable[str]] - **kwargs # type: bool -): - # type: (...) -> Iterator[Tag] + interpreter: Optional[str] = None, + abis: Optional[Iterable[str]] = None, + platforms: Optional[Iterable[str]] = None, + *, + warn: bool = False, +) -> Iterator[Tag]: """ Yields the tags for a generic interpreter. @@ -289,15 +281,15 @@ def generic_tags( The "none" ABI will be added if it was not explicitly provided. """ - warn = _warn_keyword_parameter("generic_tags", kwargs) if not interpreter: interp_name = interpreter_name() interp_version = interpreter_version(warn=warn) interpreter = "".join([interp_name, interp_version]) if abis is None: abis = _generic_abi() - platforms = list(platforms or _platform_tags()) - abis = list(abis) + else: + abis = list(abis) + platforms = list(platforms or platform_tags()) if "none" not in abis: abis.append("none") for abi in abis: @@ -305,8 +297,7 @@ def generic_tags( yield Tag(interpreter, abi, platform_) -def _py_interpreter_range(py_version): - # type: (PythonVersion) -> Iterator[str] +def _py_interpreter_range(py_version: PythonVersion) -> Iterator[str]: """ Yields Python versions in descending order. @@ -314,19 +305,18 @@ def _py_interpreter_range(py_version): all previous versions of that major version. """ if len(py_version) > 1: - yield "py{version}".format(version=_version_nodot(py_version[:2])) - yield "py{major}".format(major=py_version[0]) + yield f"py{_version_nodot(py_version[:2])}" + yield f"py{py_version[0]}" if len(py_version) > 1: for minor in range(py_version[1] - 1, -1, -1): - yield "py{version}".format(version=_version_nodot((py_version[0], minor))) + yield f"py{_version_nodot((py_version[0], minor))}" def compatible_tags( - python_version=None, # type: Optional[PythonVersion] - interpreter=None, # type: Optional[str] - platforms=None, # type: Optional[Iterable[str]] -): - # type: (...) -> Iterator[Tag] + python_version: Optional[PythonVersion] = None, + interpreter: Optional[str] = None, + platforms: Optional[Iterable[str]] = None, +) -> Iterator[Tag]: """ Yields the sequence of tags that are compatible with a specific version of Python. @@ -337,7 +327,7 @@ def compatible_tags( """ if not python_version: python_version = sys.version_info[:2] - platforms = list(platforms or _platform_tags()) + platforms = list(platforms or platform_tags()) for version in _py_interpreter_range(python_version): for platform_ in platforms: yield Tag(version, "none", platform_) @@ -347,8 +337,7 @@ def compatible_tags( yield Tag(version, "none", "any") -def _mac_arch(arch, is_32bit=_32_BIT_INTERPRETER): - # type: (str, bool) -> str +def _mac_arch(arch: str, is_32bit: bool = _32_BIT_INTERPRETER) -> str: if not is_32bit: return arch @@ -358,8 +347,7 @@ def _mac_arch(arch, is_32bit=_32_BIT_INTERPRETER): return "i386" -def _mac_binary_formats(version, cpu_arch): - # type: (MacVersion, str) -> List[str] +def _mac_binary_formats(version: MacVersion, cpu_arch: str) -> List[str]: formats = [cpu_arch] if cpu_arch == "x86_64": if version < (10, 4): @@ -382,12 +370,18 @@ def _mac_binary_formats(version, cpu_arch): return [] formats.extend(["fat32", "fat"]) - formats.append("universal") + if cpu_arch in {"arm64", "x86_64"}: + formats.append("universal2") + + if cpu_arch in {"x86_64", "i386", "ppc64", "ppc", "intel"}: + formats.append("universal") + return formats -def mac_platforms(version=None, arch=None): - # type: (Optional[MacVersion], Optional[str]) -> Iterator[str] +def mac_platforms( + version: Optional[MacVersion] = None, arch: Optional[str] = None +) -> Iterator[str]: """ Yields the platform tags for a macOS system. @@ -396,292 +390,101 @@ def mac_platforms(version=None, arch=None): generate platform tags for. Both parameters default to the appropriate value for the current system. """ - version_str, _, cpu_arch = platform.mac_ver() # type: ignore + version_str, _, cpu_arch = platform.mac_ver() if version is None: version = cast("MacVersion", tuple(map(int, version_str.split(".")[:2]))) + if version == (10, 16): + # When built against an older macOS SDK, Python will report macOS 10.16 + # instead of the real version. + version_str = subprocess.run( + [ + sys.executable, + "-sS", + "-c", + "import platform; print(platform.mac_ver()[0])", + ], + check=True, + env={"SYSTEM_VERSION_COMPAT": "0"}, + stdout=subprocess.PIPE, + universal_newlines=True, + ).stdout + version = cast("MacVersion", tuple(map(int, version_str.split(".")[:2]))) else: version = version if arch is None: arch = _mac_arch(cpu_arch) else: arch = arch - for minor_version in range(version[1], -1, -1): - compat_version = version[0], minor_version - binary_formats = _mac_binary_formats(compat_version, arch) - for binary_format in binary_formats: - yield "macosx_{major}_{minor}_{binary_format}".format( - major=compat_version[0], - minor=compat_version[1], - binary_format=binary_format, - ) - - -# From PEP 513. -def _is_manylinux_compatible(name, glibc_version): - # type: (str, GlibcVersion) -> bool - # Check for presence of _manylinux module. - try: - import _manylinux # noqa - return bool(getattr(_manylinux, name + "_compatible")) - except (ImportError, AttributeError): - # Fall through to heuristic check below. - pass - - return _have_compatible_glibc(*glibc_version) - - -def _glibc_version_string(): - # type: () -> Optional[str] - # Returns glibc version string, or None if not using glibc. - return _glibc_version_string_confstr() or _glibc_version_string_ctypes() - - -def _glibc_version_string_confstr(): - # type: () -> Optional[str] - """ - Primary implementation of glibc_version_string using os.confstr. - """ - # os.confstr is quite a bit faster than ctypes.DLL. It's also less likely - # to be broken or missing. This strategy is used in the standard library - # platform module. - # https://github.com/python/cpython/blob/fcf1d003bf4f0100c9d0921ff3d70e1127ca1b71/Lib/platform.py#L175-L183 - try: - # os.confstr("CS_GNU_LIBC_VERSION") returns a string like "glibc 2.17". - version_string = os.confstr( # type: ignore[attr-defined] # noqa: F821 - "CS_GNU_LIBC_VERSION" - ) - assert version_string is not None - _, version = version_string.split() # type: Tuple[str, str] - except (AssertionError, AttributeError, OSError, ValueError): - # os.confstr() or CS_GNU_LIBC_VERSION not available (or a bad value)... - return None - return version + if (10, 0) <= version and version < (11, 0): + # Prior to Mac OS 11, each yearly release of Mac OS bumped the + # "minor" version number. The major version was always 10. + for minor_version in range(version[1], -1, -1): + compat_version = 10, minor_version + binary_formats = _mac_binary_formats(compat_version, arch) + for binary_format in binary_formats: + yield "macosx_{major}_{minor}_{binary_format}".format( + major=10, minor=minor_version, binary_format=binary_format + ) + if version >= (11, 0): + # Starting with Mac OS 11, each yearly release bumps the major version + # number. The minor versions are now the midyear updates. + for major_version in range(version[0], 10, -1): + compat_version = major_version, 0 + binary_formats = _mac_binary_formats(compat_version, arch) + for binary_format in binary_formats: + yield "macosx_{major}_{minor}_{binary_format}".format( + major=major_version, minor=0, binary_format=binary_format + ) -def _glibc_version_string_ctypes(): - # type: () -> Optional[str] - """ - Fallback implementation of glibc_version_string using ctypes. - """ - try: - import ctypes - except ImportError: - return None - - # ctypes.CDLL(None) internally calls dlopen(NULL), and as the dlopen - # manpage says, "If filename is NULL, then the returned handle is for the - # main program". This way we can let the linker do the work to figure out - # which libc our process is actually using. - # - # Note: typeshed is wrong here so we are ignoring this line. - process_namespace = ctypes.CDLL(None) # type: ignore - try: - gnu_get_libc_version = process_namespace.gnu_get_libc_version - except AttributeError: - # Symbol doesn't exist -> therefore, we are not linked to - # glibc. - return None - - # Call gnu_get_libc_version, which returns a string like "2.5" - gnu_get_libc_version.restype = ctypes.c_char_p - version_str = gnu_get_libc_version() # type: str - # py2 / py3 compatibility: - if not isinstance(version_str, str): - version_str = version_str.decode("ascii") - - return version_str - - -# Separated out from have_compatible_glibc for easier unit testing. -def _check_glibc_version(version_str, required_major, minimum_minor): - # type: (str, int, int) -> bool - # Parse string and check against requested version. - # - # We use a regexp instead of str.split because we want to discard any - # random junk that might come after the minor version -- this might happen - # in patched/forked versions of glibc (e.g. Linaro's version of glibc - # uses version strings like "2.20-2014.11"). See gh-3588. - m = re.match(r"(?P<major>[0-9]+)\.(?P<minor>[0-9]+)", version_str) - if not m: - warnings.warn( - "Expected glibc version with 2 components major.minor," - " got: %s" % version_str, - RuntimeWarning, - ) - return False - return ( - int(m.group("major")) == required_major - and int(m.group("minor")) >= minimum_minor - ) + if version >= (11, 0): + # Mac OS 11 on x86_64 is compatible with binaries from previous releases. + # Arm64 support was introduced in 11.0, so no Arm binaries from previous + # releases exist. + # + # However, the "universal2" binary format can have a + # macOS version earlier than 11.0 when the x86_64 part of the binary supports + # that version of macOS. + if arch == "x86_64": + for minor_version in range(16, 3, -1): + compat_version = 10, minor_version + binary_formats = _mac_binary_formats(compat_version, arch) + for binary_format in binary_formats: + yield "macosx_{major}_{minor}_{binary_format}".format( + major=compat_version[0], + minor=compat_version[1], + binary_format=binary_format, + ) + else: + for minor_version in range(16, 3, -1): + compat_version = 10, minor_version + binary_format = "universal2" + yield "macosx_{major}_{minor}_{binary_format}".format( + major=compat_version[0], + minor=compat_version[1], + binary_format=binary_format, + ) -def _have_compatible_glibc(required_major, minimum_minor): - # type: (int, int) -> bool - version_str = _glibc_version_string() - if version_str is None: - return False - return _check_glibc_version(version_str, required_major, minimum_minor) - - -# Python does not provide platform information at sufficient granularity to -# identify the architecture of the running executable in some cases, so we -# determine it dynamically by reading the information from the running -# process. This only applies on Linux, which uses the ELF format. -class _ELFFileHeader(object): - # https://en.wikipedia.org/wiki/Executable_and_Linkable_Format#File_header - class _InvalidELFFileHeader(ValueError): - """ - An invalid ELF file header was found. - """ - - ELF_MAGIC_NUMBER = 0x7F454C46 - ELFCLASS32 = 1 - ELFCLASS64 = 2 - ELFDATA2LSB = 1 - ELFDATA2MSB = 2 - EM_386 = 3 - EM_S390 = 22 - EM_ARM = 40 - EM_X86_64 = 62 - EF_ARM_ABIMASK = 0xFF000000 - EF_ARM_ABI_VER5 = 0x05000000 - EF_ARM_ABI_FLOAT_HARD = 0x00000400 - - def __init__(self, file): - # type: (IO[bytes]) -> None - def unpack(fmt): - # type: (str) -> int - try: - (result,) = struct.unpack( - fmt, file.read(struct.calcsize(fmt)) - ) # type: (int, ) - except struct.error: - raise _ELFFileHeader._InvalidELFFileHeader() - return result - - self.e_ident_magic = unpack(">I") - if self.e_ident_magic != self.ELF_MAGIC_NUMBER: - raise _ELFFileHeader._InvalidELFFileHeader() - self.e_ident_class = unpack("B") - if self.e_ident_class not in {self.ELFCLASS32, self.ELFCLASS64}: - raise _ELFFileHeader._InvalidELFFileHeader() - self.e_ident_data = unpack("B") - if self.e_ident_data not in {self.ELFDATA2LSB, self.ELFDATA2MSB}: - raise _ELFFileHeader._InvalidELFFileHeader() - self.e_ident_version = unpack("B") - self.e_ident_osabi = unpack("B") - self.e_ident_abiversion = unpack("B") - self.e_ident_pad = file.read(7) - format_h = "<H" if self.e_ident_data == self.ELFDATA2LSB else ">H" - format_i = "<I" if self.e_ident_data == self.ELFDATA2LSB else ">I" - format_q = "<Q" if self.e_ident_data == self.ELFDATA2LSB else ">Q" - format_p = format_i if self.e_ident_class == self.ELFCLASS32 else format_q - self.e_type = unpack(format_h) - self.e_machine = unpack(format_h) - self.e_version = unpack(format_i) - self.e_entry = unpack(format_p) - self.e_phoff = unpack(format_p) - self.e_shoff = unpack(format_p) - self.e_flags = unpack(format_i) - self.e_ehsize = unpack(format_h) - self.e_phentsize = unpack(format_h) - self.e_phnum = unpack(format_h) - self.e_shentsize = unpack(format_h) - self.e_shnum = unpack(format_h) - self.e_shstrndx = unpack(format_h) - - -def _get_elf_header(): - # type: () -> Optional[_ELFFileHeader] - try: - with open(sys.executable, "rb") as f: - elf_header = _ELFFileHeader(f) - except (IOError, OSError, TypeError, _ELFFileHeader._InvalidELFFileHeader): - return None - return elf_header - - -def _is_linux_armhf(): - # type: () -> bool - # hard-float ABI can be detected from the ELF header of the running - # process - # https://static.docs.arm.com/ihi0044/g/aaelf32.pdf - elf_header = _get_elf_header() - if elf_header is None: - return False - result = elf_header.e_ident_class == elf_header.ELFCLASS32 - result &= elf_header.e_ident_data == elf_header.ELFDATA2LSB - result &= elf_header.e_machine == elf_header.EM_ARM - result &= ( - elf_header.e_flags & elf_header.EF_ARM_ABIMASK - ) == elf_header.EF_ARM_ABI_VER5 - result &= ( - elf_header.e_flags & elf_header.EF_ARM_ABI_FLOAT_HARD - ) == elf_header.EF_ARM_ABI_FLOAT_HARD - return result - - -def _is_linux_i686(): - # type: () -> bool - elf_header = _get_elf_header() - if elf_header is None: - return False - result = elf_header.e_ident_class == elf_header.ELFCLASS32 - result &= elf_header.e_ident_data == elf_header.ELFDATA2LSB - result &= elf_header.e_machine == elf_header.EM_386 - return result - - -def _have_compatible_manylinux_abi(arch): - # type: (str) -> bool - if arch == "armv7l": - return _is_linux_armhf() - if arch == "i686": - return _is_linux_i686() - return True - - -def _linux_platforms(is_32bit=_32_BIT_INTERPRETER): - # type: (bool) -> Iterator[str] - linux = _normalize_string(distutils.util.get_platform()) +def _linux_platforms(is_32bit: bool = _32_BIT_INTERPRETER) -> Iterator[str]: + linux = _normalize_string(sysconfig.get_platform()) if is_32bit: if linux == "linux_x86_64": linux = "linux_i686" elif linux == "linux_aarch64": linux = "linux_armv7l" - manylinux_support = [] _, arch = linux.split("_", 1) - if _have_compatible_manylinux_abi(arch): - if arch in {"x86_64", "i686", "aarch64", "armv7l", "ppc64", "ppc64le", "s390x"}: - manylinux_support.append( - ("manylinux2014", (2, 17)) - ) # CentOS 7 w/ glibc 2.17 (PEP 599) - if arch in {"x86_64", "i686"}: - manylinux_support.append( - ("manylinux2010", (2, 12)) - ) # CentOS 6 w/ glibc 2.12 (PEP 571) - manylinux_support.append( - ("manylinux1", (2, 5)) - ) # CentOS 5 w/ glibc 2.5 (PEP 513) - manylinux_support_iter = iter(manylinux_support) - for name, glibc_version in manylinux_support_iter: - if _is_manylinux_compatible(name, glibc_version): - yield linux.replace("linux", name) - break - # Support for a later manylinux implies support for an earlier version. - for name, _ in manylinux_support_iter: - yield linux.replace("linux", name) + yield from _manylinux.platform_tags(linux, arch) + yield from _musllinux.platform_tags(arch) yield linux -def _generic_platforms(): - # type: () -> Iterator[str] - yield _normalize_string(distutils.util.get_platform()) +def _generic_platforms() -> Iterator[str]: + yield _normalize_string(sysconfig.get_platform()) -def _platform_tags(): - # type: () -> Iterator[str] +def platform_tags() -> Iterator[str]: """ Provides the platform tags for this installation. """ @@ -693,25 +496,21 @@ def _platform_tags(): return _generic_platforms() -def interpreter_name(): - # type: () -> str +def interpreter_name() -> str: """ Returns the name of the running interpreter. + + Some implementations have a reserved, two-letter abbreviation which will + be returned when appropriate. """ - try: - name = sys.implementation.name # type: ignore - except AttributeError: # pragma: no cover - # Python 2.7 compatibility. - name = platform.python_implementation().lower() + name = sys.implementation.name return INTERPRETER_SHORT_NAMES.get(name) or name -def interpreter_version(**kwargs): - # type: (bool) -> str +def interpreter_version(*, warn: bool = False) -> str: """ Returns the version of the running interpreter. """ - warn = _warn_keyword_parameter("interpreter_version", kwargs) version = _get_config_var("py_version_nodot", warn=warn) if version: version = str(version) @@ -720,32 +519,28 @@ def interpreter_version(**kwargs): return version -def _version_nodot(version): - # type: (PythonVersion) -> str - if any(v >= 10 for v in version): - sep = "_" - else: - sep = "" - return sep.join(map(str, version)) +def _version_nodot(version: PythonVersion) -> str: + return "".join(map(str, version)) -def sys_tags(**kwargs): - # type: (bool) -> Iterator[Tag] +def sys_tags(*, warn: bool = False) -> Iterator[Tag]: """ Returns the sequence of tag triples for the running interpreter. The order of the sequence corresponds to priority order for the interpreter, from most to least important. """ - warn = _warn_keyword_parameter("sys_tags", kwargs) interp_name = interpreter_name() if interp_name == "cp": - for tag in cpython_tags(warn=warn): - yield tag + yield from cpython_tags(warn=warn) else: - for tag in generic_tags(): - yield tag + yield from generic_tags() - for tag in compatible_tags(): - yield tag + if interp_name == "pp": + interp = "pp3" + elif interp_name == "cp": + interp = "cp" + interpreter_version(warn=warn) + else: + interp = None + yield from compatible_tags(interpreter=interp) diff --git a/third_party/python/setuptools/pkg_resources/_vendor/packaging/utils.py b/third_party/python/setuptools/pkg_resources/_vendor/packaging/utils.py index 19579c1a0f..33c613b749 100644 --- a/third_party/python/setuptools/pkg_resources/_vendor/packaging/utils.py +++ b/third_party/python/setuptools/pkg_resources/_vendor/packaging/utils.py @@ -1,65 +1,141 @@ # This file is dual licensed under the terms of the Apache License, Version # 2.0, and the BSD License. See the LICENSE file in the root of this repository # for complete details. -from __future__ import absolute_import, division, print_function import re +from typing import FrozenSet, NewType, Tuple, Union, cast -from ._typing import TYPE_CHECKING, cast +from .tags import Tag, parse_tag from .version import InvalidVersion, Version -if TYPE_CHECKING: # pragma: no cover - from typing import NewType, Union +BuildTag = Union[Tuple[()], Tuple[int, str]] +NormalizedName = NewType("NormalizedName", str) + + +class InvalidWheelFilename(ValueError): + """ + An invalid wheel filename was found, users should refer to PEP 427. + """ + + +class InvalidSdistFilename(ValueError): + """ + An invalid sdist filename was found, users should refer to the packaging user guide. + """ - NormalizedName = NewType("NormalizedName", str) _canonicalize_regex = re.compile(r"[-_.]+") +# PEP 427: The build number must start with a digit. +_build_tag_regex = re.compile(r"(\d+)(.*)") -def canonicalize_name(name): - # type: (str) -> NormalizedName +def canonicalize_name(name: str) -> NormalizedName: # This is taken from PEP 503. value = _canonicalize_regex.sub("-", name).lower() - return cast("NormalizedName", value) + return cast(NormalizedName, value) -def canonicalize_version(_version): - # type: (str) -> Union[Version, str] +def canonicalize_version( + version: Union[Version, str], *, strip_trailing_zero: bool = True +) -> str: """ This is very similar to Version.__str__, but has one subtle difference with the way it handles the release segment. """ - - try: - version = Version(_version) - except InvalidVersion: - # Legacy versions cannot be normalized - return _version + if isinstance(version, str): + try: + parsed = Version(version) + except InvalidVersion: + # Legacy versions cannot be normalized + return version + else: + parsed = version parts = [] # Epoch - if version.epoch != 0: - parts.append("{0}!".format(version.epoch)) + if parsed.epoch != 0: + parts.append(f"{parsed.epoch}!") # Release segment - # NB: This strips trailing '.0's to normalize - parts.append(re.sub(r"(\.0)+$", "", ".".join(str(x) for x in version.release))) + release_segment = ".".join(str(x) for x in parsed.release) + if strip_trailing_zero: + # NB: This strips trailing '.0's to normalize + release_segment = re.sub(r"(\.0)+$", "", release_segment) + parts.append(release_segment) # Pre-release - if version.pre is not None: - parts.append("".join(str(x) for x in version.pre)) + if parsed.pre is not None: + parts.append("".join(str(x) for x in parsed.pre)) # Post-release - if version.post is not None: - parts.append(".post{0}".format(version.post)) + if parsed.post is not None: + parts.append(f".post{parsed.post}") # Development release - if version.dev is not None: - parts.append(".dev{0}".format(version.dev)) + if parsed.dev is not None: + parts.append(f".dev{parsed.dev}") # Local version segment - if version.local is not None: - parts.append("+{0}".format(version.local)) + if parsed.local is not None: + parts.append(f"+{parsed.local}") return "".join(parts) + + +def parse_wheel_filename( + filename: str, +) -> Tuple[NormalizedName, Version, BuildTag, FrozenSet[Tag]]: + if not filename.endswith(".whl"): + raise InvalidWheelFilename( + f"Invalid wheel filename (extension must be '.whl'): {filename}" + ) + + filename = filename[:-4] + dashes = filename.count("-") + if dashes not in (4, 5): + raise InvalidWheelFilename( + f"Invalid wheel filename (wrong number of parts): {filename}" + ) + + parts = filename.split("-", dashes - 2) + name_part = parts[0] + # See PEP 427 for the rules on escaping the project name + if "__" in name_part or re.match(r"^[\w\d._]*$", name_part, re.UNICODE) is None: + raise InvalidWheelFilename(f"Invalid project name: {filename}") + name = canonicalize_name(name_part) + version = Version(parts[1]) + if dashes == 5: + build_part = parts[2] + build_match = _build_tag_regex.match(build_part) + if build_match is None: + raise InvalidWheelFilename( + f"Invalid build number: {build_part} in '{filename}'" + ) + build = cast(BuildTag, (int(build_match.group(1)), build_match.group(2))) + else: + build = () + tags = parse_tag(parts[-1]) + return (name, version, build, tags) + + +def parse_sdist_filename(filename: str) -> Tuple[NormalizedName, Version]: + if filename.endswith(".tar.gz"): + file_stem = filename[: -len(".tar.gz")] + elif filename.endswith(".zip"): + file_stem = filename[: -len(".zip")] + else: + raise InvalidSdistFilename( + f"Invalid sdist filename (extension must be '.tar.gz' or '.zip'):" + f" {filename}" + ) + + # We are requiring a PEP 440 version, which cannot contain dashes, + # so we split on the last dash. + name_part, sep, version_part = file_stem.rpartition("-") + if not sep: + raise InvalidSdistFilename(f"Invalid sdist filename: {filename}") + + name = canonicalize_name(name_part) + version = Version(version_part) + return (name, version) diff --git a/third_party/python/setuptools/pkg_resources/_vendor/packaging/version.py b/third_party/python/setuptools/pkg_resources/_vendor/packaging/version.py index 00371e86a8..b30e8cbf84 100644 --- a/third_party/python/setuptools/pkg_resources/_vendor/packaging/version.py +++ b/third_party/python/setuptools/pkg_resources/_vendor/packaging/version.py @@ -1,238 +1,116 @@ # This file is dual licensed under the terms of the Apache License, Version # 2.0, and the BSD License. See the LICENSE file in the root of this repository # for complete details. -from __future__ import absolute_import, division, print_function +""" +.. testsetup:: + + from packaging.version import parse, Version +""" import collections import itertools import re - -from ._structures import Infinity, NegativeInfinity -from ._typing import TYPE_CHECKING - -if TYPE_CHECKING: # pragma: no cover - from typing import Callable, Iterator, List, Optional, SupportsInt, Tuple, Union - - from ._structures import InfinityType, NegativeInfinityType - - InfiniteTypes = Union[InfinityType, NegativeInfinityType] - PrePostDevType = Union[InfiniteTypes, Tuple[str, int]] - SubLocalType = Union[InfiniteTypes, int, str] - LocalType = Union[ - NegativeInfinityType, - Tuple[ - Union[ - SubLocalType, - Tuple[SubLocalType, str], - Tuple[NegativeInfinityType, SubLocalType], - ], - ..., +from typing import Any, Callable, Optional, SupportsInt, Tuple, Union + +from ._structures import Infinity, InfinityType, NegativeInfinity, NegativeInfinityType + +__all__ = ["VERSION_PATTERN", "parse", "Version", "InvalidVersion"] + +InfiniteTypes = Union[InfinityType, NegativeInfinityType] +PrePostDevType = Union[InfiniteTypes, Tuple[str, int]] +SubLocalType = Union[InfiniteTypes, int, str] +LocalType = Union[ + NegativeInfinityType, + Tuple[ + Union[ + SubLocalType, + Tuple[SubLocalType, str], + Tuple[NegativeInfinityType, SubLocalType], ], - ] - CmpKey = Tuple[ - int, Tuple[int, ...], PrePostDevType, PrePostDevType, PrePostDevType, LocalType - ] - LegacyCmpKey = Tuple[int, Tuple[str, ...]] - VersionComparisonMethod = Callable[ - [Union[CmpKey, LegacyCmpKey], Union[CmpKey, LegacyCmpKey]], bool - ] - -__all__ = ["parse", "Version", "LegacyVersion", "InvalidVersion", "VERSION_PATTERN"] - + ..., + ], +] +CmpKey = Tuple[ + int, Tuple[int, ...], PrePostDevType, PrePostDevType, PrePostDevType, LocalType +] +VersionComparisonMethod = Callable[[CmpKey, CmpKey], bool] _Version = collections.namedtuple( "_Version", ["epoch", "release", "dev", "pre", "post", "local"] ) -def parse(version): - # type: (str) -> Union[LegacyVersion, Version] - """ - Parse the given version string and return either a :class:`Version` object - or a :class:`LegacyVersion` object depending on if the given version is - a valid PEP 440 version or a legacy version. - """ - try: - return Version(version) - except InvalidVersion: - return LegacyVersion(version) +def parse(version: str) -> "Version": + """Parse the given version string. + >>> parse('1.0.dev1') + <Version('1.0.dev1')> -class InvalidVersion(ValueError): + :param version: The version string to parse. + :raises InvalidVersion: When the version string is not a valid version. """ - An invalid version was found, users should refer to PEP 440. - """ - - -class _BaseVersion(object): - _key = None # type: Union[CmpKey, LegacyCmpKey] - - def __hash__(self): - # type: () -> int - return hash(self._key) + return Version(version) - def __lt__(self, other): - # type: (_BaseVersion) -> bool - return self._compare(other, lambda s, o: s < o) - def __le__(self, other): - # type: (_BaseVersion) -> bool - return self._compare(other, lambda s, o: s <= o) +class InvalidVersion(ValueError): + """Raised when a version string is not a valid version. - def __eq__(self, other): - # type: (object) -> bool - return self._compare(other, lambda s, o: s == o) + >>> Version("invalid") + Traceback (most recent call last): + ... + packaging.version.InvalidVersion: Invalid version: 'invalid' + """ - def __ge__(self, other): - # type: (_BaseVersion) -> bool - return self._compare(other, lambda s, o: s >= o) - def __gt__(self, other): - # type: (_BaseVersion) -> bool - return self._compare(other, lambda s, o: s > o) +class _BaseVersion: + _key: Tuple[Any, ...] - def __ne__(self, other): - # type: (object) -> bool - return self._compare(other, lambda s, o: s != o) + def __hash__(self) -> int: + return hash(self._key) - def _compare(self, other, method): - # type: (object, VersionComparisonMethod) -> Union[bool, NotImplemented] + # Please keep the duplicated `isinstance` check + # in the six comparisons hereunder + # unless you find a way to avoid adding overhead function calls. + def __lt__(self, other: "_BaseVersion") -> bool: if not isinstance(other, _BaseVersion): return NotImplemented - return method(self._key, other._key) - - -class LegacyVersion(_BaseVersion): - def __init__(self, version): - # type: (str) -> None - self._version = str(version) - self._key = _legacy_cmpkey(self._version) - - def __str__(self): - # type: () -> str - return self._version - - def __repr__(self): - # type: () -> str - return "<LegacyVersion({0})>".format(repr(str(self))) - - @property - def public(self): - # type: () -> str - return self._version - - @property - def base_version(self): - # type: () -> str - return self._version - - @property - def epoch(self): - # type: () -> int - return -1 - - @property - def release(self): - # type: () -> None - return None - - @property - def pre(self): - # type: () -> None - return None - - @property - def post(self): - # type: () -> None - return None - - @property - def dev(self): - # type: () -> None - return None + return self._key < other._key - @property - def local(self): - # type: () -> None - return None - - @property - def is_prerelease(self): - # type: () -> bool - return False - - @property - def is_postrelease(self): - # type: () -> bool - return False - - @property - def is_devrelease(self): - # type: () -> bool - return False - - -_legacy_version_component_re = re.compile(r"(\d+ | [a-z]+ | \.| -)", re.VERBOSE) - -_legacy_version_replacement_map = { - "pre": "c", - "preview": "c", - "-": "final-", - "rc": "c", - "dev": "@", -} - - -def _parse_version_parts(s): - # type: (str) -> Iterator[str] - for part in _legacy_version_component_re.split(s): - part = _legacy_version_replacement_map.get(part, part) - - if not part or part == ".": - continue + def __le__(self, other: "_BaseVersion") -> bool: + if not isinstance(other, _BaseVersion): + return NotImplemented - if part[:1] in "0123456789": - # pad for numeric comparison - yield part.zfill(8) - else: - yield "*" + part + return self._key <= other._key - # ensure that alpha/beta/candidate are before final - yield "*final" + def __eq__(self, other: object) -> bool: + if not isinstance(other, _BaseVersion): + return NotImplemented + return self._key == other._key -def _legacy_cmpkey(version): - # type: (str) -> LegacyCmpKey + def __ge__(self, other: "_BaseVersion") -> bool: + if not isinstance(other, _BaseVersion): + return NotImplemented - # We hardcode an epoch of -1 here. A PEP 440 version can only have a epoch - # greater than or equal to 0. This will effectively put the LegacyVersion, - # which uses the defacto standard originally implemented by setuptools, - # as before all PEP 440 versions. - epoch = -1 + return self._key >= other._key - # This scheme is taken from pkg_resources.parse_version setuptools prior to - # it's adoption of the packaging library. - parts = [] # type: List[str] - for part in _parse_version_parts(version.lower()): - if part.startswith("*"): - # remove "-" before a prerelease tag - if part < "*final": - while parts and parts[-1] == "*final-": - parts.pop() + def __gt__(self, other: "_BaseVersion") -> bool: + if not isinstance(other, _BaseVersion): + return NotImplemented - # remove trailing zeros from each series of numeric parts - while parts and parts[-1] == "00000000": - parts.pop() + return self._key > other._key - parts.append(part) + def __ne__(self, other: object) -> bool: + if not isinstance(other, _BaseVersion): + return NotImplemented - return epoch, tuple(parts) + return self._key != other._key # Deliberately not anchored to the start and end of the string, to make it # easier for 3rd party code to reuse -VERSION_PATTERN = r""" +_VERSION_PATTERN = r""" v? (?: (?:(?P<epoch>[0-9]+)!)? # epoch @@ -263,18 +141,61 @@ VERSION_PATTERN = r""" (?:\+(?P<local>[a-z0-9]+(?:[-_\.][a-z0-9]+)*))? # local version """ +VERSION_PATTERN = _VERSION_PATTERN +""" +A string containing the regular expression used to match a valid version. + +The pattern is not anchored at either end, and is intended for embedding in larger +expressions (for example, matching a version number as part of a file name). The +regular expression should be compiled with the ``re.VERBOSE`` and ``re.IGNORECASE`` +flags set. + +:meta hide-value: +""" + class Version(_BaseVersion): + """This class abstracts handling of a project's versions. + + A :class:`Version` instance is comparison aware and can be compared and + sorted using the standard Python interfaces. + + >>> v1 = Version("1.0a5") + >>> v2 = Version("1.0") + >>> v1 + <Version('1.0a5')> + >>> v2 + <Version('1.0')> + >>> v1 < v2 + True + >>> v1 == v2 + False + >>> v1 > v2 + False + >>> v1 >= v2 + False + >>> v1 <= v2 + True + """ _regex = re.compile(r"^\s*" + VERSION_PATTERN + r"\s*$", re.VERBOSE | re.IGNORECASE) + _key: CmpKey - def __init__(self, version): - # type: (str) -> None + def __init__(self, version: str) -> None: + """Initialize a Version object. + + :param version: + The string representation of a version which will be parsed and normalized + before use. + :raises InvalidVersion: + If the ``version`` does not conform to PEP 440 in any way then this + exception will be raised. + """ # Validate the version and parse it into pieces match = self._regex.search(version) if not match: - raise InvalidVersion("Invalid version: '{0}'".format(version)) + raise InvalidVersion(f"Invalid version: '{version}'") # Store the parsed out pieces of the version self._version = _Version( @@ -298,17 +219,25 @@ class Version(_BaseVersion): self._version.local, ) - def __repr__(self): - # type: () -> str - return "<Version({0})>".format(repr(str(self))) + def __repr__(self) -> str: + """A representation of the Version that shows all internal state. + + >>> Version('1.0.0') + <Version('1.0.0')> + """ + return f"<Version('{self}')>" - def __str__(self): - # type: () -> str + def __str__(self) -> str: + """A string representation of the version that can be rounded-tripped. + + >>> str(Version("1.0a5")) + '1.0a5' + """ parts = [] # Epoch if self.epoch != 0: - parts.append("{0}!".format(self.epoch)) + parts.append(f"{self.epoch}!") # Release segment parts.append(".".join(str(x) for x in self.release)) @@ -319,67 +248,131 @@ class Version(_BaseVersion): # Post-release if self.post is not None: - parts.append(".post{0}".format(self.post)) + parts.append(f".post{self.post}") # Development release if self.dev is not None: - parts.append(".dev{0}".format(self.dev)) + parts.append(f".dev{self.dev}") # Local version segment if self.local is not None: - parts.append("+{0}".format(self.local)) + parts.append(f"+{self.local}") return "".join(parts) @property - def epoch(self): - # type: () -> int - _epoch = self._version.epoch # type: int + def epoch(self) -> int: + """The epoch of the version. + + >>> Version("2.0.0").epoch + 0 + >>> Version("1!2.0.0").epoch + 1 + """ + _epoch: int = self._version.epoch return _epoch @property - def release(self): - # type: () -> Tuple[int, ...] - _release = self._version.release # type: Tuple[int, ...] + def release(self) -> Tuple[int, ...]: + """The components of the "release" segment of the version. + + >>> Version("1.2.3").release + (1, 2, 3) + >>> Version("2.0.0").release + (2, 0, 0) + >>> Version("1!2.0.0.post0").release + (2, 0, 0) + + Includes trailing zeroes but not the epoch or any pre-release / development / + post-release suffixes. + """ + _release: Tuple[int, ...] = self._version.release return _release @property - def pre(self): - # type: () -> Optional[Tuple[str, int]] - _pre = self._version.pre # type: Optional[Tuple[str, int]] + def pre(self) -> Optional[Tuple[str, int]]: + """The pre-release segment of the version. + + >>> print(Version("1.2.3").pre) + None + >>> Version("1.2.3a1").pre + ('a', 1) + >>> Version("1.2.3b1").pre + ('b', 1) + >>> Version("1.2.3rc1").pre + ('rc', 1) + """ + _pre: Optional[Tuple[str, int]] = self._version.pre return _pre @property - def post(self): - # type: () -> Optional[Tuple[str, int]] + def post(self) -> Optional[int]: + """The post-release number of the version. + + >>> print(Version("1.2.3").post) + None + >>> Version("1.2.3.post1").post + 1 + """ return self._version.post[1] if self._version.post else None @property - def dev(self): - # type: () -> Optional[Tuple[str, int]] + def dev(self) -> Optional[int]: + """The development number of the version. + + >>> print(Version("1.2.3").dev) + None + >>> Version("1.2.3.dev1").dev + 1 + """ return self._version.dev[1] if self._version.dev else None @property - def local(self): - # type: () -> Optional[str] + def local(self) -> Optional[str]: + """The local version segment of the version. + + >>> print(Version("1.2.3").local) + None + >>> Version("1.2.3+abc").local + 'abc' + """ if self._version.local: return ".".join(str(x) for x in self._version.local) else: return None @property - def public(self): - # type: () -> str + def public(self) -> str: + """The public portion of the version. + + >>> Version("1.2.3").public + '1.2.3' + >>> Version("1.2.3+abc").public + '1.2.3' + >>> Version("1.2.3+abc.dev1").public + '1.2.3' + """ return str(self).split("+", 1)[0] @property - def base_version(self): - # type: () -> str + def base_version(self) -> str: + """The "base version" of the version. + + >>> Version("1.2.3").base_version + '1.2.3' + >>> Version("1.2.3+abc").base_version + '1.2.3' + >>> Version("1!1.2.3+abc.dev1").base_version + '1!1.2.3' + + The "base version" is the public version of the project without any pre or post + release markers. + """ parts = [] # Epoch if self.epoch != 0: - parts.append("{0}!".format(self.epoch)) + parts.append(f"{self.epoch}!") # Release segment parts.append(".".join(str(x) for x in self.release)) @@ -387,41 +380,79 @@ class Version(_BaseVersion): return "".join(parts) @property - def is_prerelease(self): - # type: () -> bool + def is_prerelease(self) -> bool: + """Whether this version is a pre-release. + + >>> Version("1.2.3").is_prerelease + False + >>> Version("1.2.3a1").is_prerelease + True + >>> Version("1.2.3b1").is_prerelease + True + >>> Version("1.2.3rc1").is_prerelease + True + >>> Version("1.2.3dev1").is_prerelease + True + """ return self.dev is not None or self.pre is not None @property - def is_postrelease(self): - # type: () -> bool + def is_postrelease(self) -> bool: + """Whether this version is a post-release. + + >>> Version("1.2.3").is_postrelease + False + >>> Version("1.2.3.post1").is_postrelease + True + """ return self.post is not None @property - def is_devrelease(self): - # type: () -> bool + def is_devrelease(self) -> bool: + """Whether this version is a development release. + + >>> Version("1.2.3").is_devrelease + False + >>> Version("1.2.3.dev1").is_devrelease + True + """ return self.dev is not None @property - def major(self): - # type: () -> int + def major(self) -> int: + """The first item of :attr:`release` or ``0`` if unavailable. + + >>> Version("1.2.3").major + 1 + """ return self.release[0] if len(self.release) >= 1 else 0 @property - def minor(self): - # type: () -> int + def minor(self) -> int: + """The second item of :attr:`release` or ``0`` if unavailable. + + >>> Version("1.2.3").minor + 2 + >>> Version("1").minor + 0 + """ return self.release[1] if len(self.release) >= 2 else 0 @property - def micro(self): - # type: () -> int + def micro(self) -> int: + """The third item of :attr:`release` or ``0`` if unavailable. + + >>> Version("1.2.3").micro + 3 + >>> Version("1").micro + 0 + """ return self.release[2] if len(self.release) >= 3 else 0 def _parse_letter_version( - letter, # type: str - number, # type: Union[str, bytes, SupportsInt] -): - # type: (...) -> Optional[Tuple[str, int]] + letter: str, number: Union[str, bytes, SupportsInt] +) -> Optional[Tuple[str, int]]: if letter: # We consider there to be an implicit 0 in a pre-release if there is @@ -458,8 +489,7 @@ def _parse_letter_version( _local_version_separators = re.compile(r"[\._-]") -def _parse_local_version(local): - # type: (str) -> Optional[LocalType] +def _parse_local_version(local: str) -> Optional[LocalType]: """ Takes a string like abc.1.twelve and turns it into ("abc", 1, "twelve"). """ @@ -472,14 +502,13 @@ def _parse_local_version(local): def _cmpkey( - epoch, # type: int - release, # type: Tuple[int, ...] - pre, # type: Optional[Tuple[str, int]] - post, # type: Optional[Tuple[str, int]] - dev, # type: Optional[Tuple[str, int]] - local, # type: Optional[Tuple[SubLocalType]] -): - # type: (...) -> CmpKey + epoch: int, + release: Tuple[int, ...], + pre: Optional[Tuple[str, int]], + post: Optional[Tuple[str, int]], + dev: Optional[Tuple[str, int]], + local: Optional[Tuple[SubLocalType]], +) -> CmpKey: # When we compare a release version, we want to compare it with all of the # trailing zeros removed. So we'll use a reverse the list, drop all the now @@ -495,7 +524,7 @@ def _cmpkey( # if there is not a pre or a post segment. If we have one of those then # the normal sorting rules will handle this case correctly. if pre is None and post is None and dev is not None: - _pre = NegativeInfinity # type: PrePostDevType + _pre: PrePostDevType = NegativeInfinity # Versions without a pre-release (except as noted above) should sort after # those with one. elif pre is None: @@ -505,21 +534,21 @@ def _cmpkey( # Versions without a post segment should sort before those with one. if post is None: - _post = NegativeInfinity # type: PrePostDevType + _post: PrePostDevType = NegativeInfinity else: _post = post # Versions without a development segment should sort after those with one. if dev is None: - _dev = Infinity # type: PrePostDevType + _dev: PrePostDevType = Infinity else: _dev = dev if local is None: # Versions without a local segment should sort before those with one. - _local = NegativeInfinity # type: LocalType + _local: LocalType = NegativeInfinity else: # Versions with a local segment need that segment parsed to implement # the sorting rules in PEP440. |