From ab02f3ffc58fabb521e06e2cb3823189d118e60b Mon Sep 17 00:00:00 2001 From: Polina Date: Thu, 2 Jul 2026 19:20:03 +0000 Subject: [PATCH] =?UTF-8?q?=D0=97=D0=B0=D0=B3=D1=80=D1=83=D0=B7=D0=B8?= =?UTF-8?q?=D1=82=D1=8C=20=D1=84=D0=B0=D0=B9=D0=BB=D1=8B=20=D0=B2=20=C2=AB?= =?UTF-8?q?venv/Lib/site-packages/pip/=5Fvendor/packaking=C2=BB?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- .../pip/_vendor/packaking/__init__.py | 15 + .../pip/_vendor/packaking/_elffile.py | 108 ++++++ .../pip/_vendor/packaking/_manylinux.py | 262 +++++++++++++ .../pip/_vendor/packaking/_musllinux.py | 85 ++++ .../pip/_vendor/packaking/_parser.py | 365 ++++++++++++++++++ 5 files changed, 835 insertions(+) create mode 100644 venv/Lib/site-packages/pip/_vendor/packaking/__init__.py create mode 100644 venv/Lib/site-packages/pip/_vendor/packaking/_elffile.py create mode 100644 venv/Lib/site-packages/pip/_vendor/packaking/_manylinux.py create mode 100644 venv/Lib/site-packages/pip/_vendor/packaking/_musllinux.py create mode 100644 venv/Lib/site-packages/pip/_vendor/packaking/_parser.py diff --git a/venv/Lib/site-packages/pip/_vendor/packaking/__init__.py b/venv/Lib/site-packages/pip/_vendor/packaking/__init__.py new file mode 100644 index 0000000..21695a7 --- /dev/null +++ b/venv/Lib/site-packages/pip/_vendor/packaking/__init__.py @@ -0,0 +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. + +__title__ = "packaging" +__summary__ = "Core utilities for Python packages" +__uri__ = "https://github.com/pypa/packaging" + +__version__ = "26.0" + +__author__ = "Donald Stufft and individual contributors" +__email__ = "donald@stufft.io" + +__license__ = "BSD-2-Clause or Apache-2.0" +__copyright__ = f"2014 {__author__}" diff --git a/venv/Lib/site-packages/pip/_vendor/packaking/_elffile.py b/venv/Lib/site-packages/pip/_vendor/packaking/_elffile.py new file mode 100644 index 0000000..497b064 --- /dev/null +++ b/venv/Lib/site-packages/pip/_vendor/packaking/_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. + +ELF header: https://refspecs.linuxfoundation.org/elf/gabi4+/ch4.eheader.html +""" + +from __future__ import annotations + +import enum +import os +import struct +from typing import IO + + +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 as e: + raise ELFInvalid("unable to parse identification") from e + 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 MSB. + (2, 1): ("HHIQQQIHHH", ">IIQQQQQQ", (0, 2, 5)), # 64-bit MSB. + }[(self.capacity, self.encoding)] + except KeyError as e: + raise ELFInvalid( + f"unrecognized capacity ({self.capacity}) or encoding ({self.encoding})" + ) from e + + 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) -> str | None: + """ + 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/venv/Lib/site-packages/pip/_vendor/packaking/_manylinux.py b/venv/Lib/site-packages/pip/_vendor/packaking/_manylinux.py new file mode 100644 index 0000000..0e79e8a --- /dev/null +++ b/venv/Lib/site-packages/pip/_vendor/packaking/_manylinux.py @@ -0,0 +1,262 @@ +from __future__ import annotations + +import collections +import contextlib +import functools +import os +import re +import sys +import warnings +from typing import Generator, Iterator, NamedTuple, Sequence + +from ._elffile import EIClass, EIData, ELFFile, EMachine + +EF_ARM_ABIMASK = 0xFF000000 +EF_ARM_ABI_VER5 = 0x05000000 +EF_ARM_ABI_FLOAT_HARD = 0x00000400 + +_ALLOWED_ARCHS = { + "x86_64", + "aarch64", + "ppc64", + "ppc64le", + "s390x", + "loongarch64", + "riscv64", +} + + +# `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[ELFFile | None, 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, archs: Sequence[str]) -> bool: + if "armv7l" in archs: + return _is_linux_armhf(executable) + if "i686" in archs: + return _is_linux_i686(executable) + return any(arch in _ALLOWED_ARCHS for arch in archs) + + +# 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() -> str | None: + """ + 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 | None = 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() -> str | None: + """ + Fallback implementation of glibc_version_string using ctypes. + """ + try: + import ctypes # noqa: PLC0415 + 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() -> str | None: + """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) -> _GLibCVersion: + """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[0-9]+)\.(?P[0-9]+)", version_str) + if not m: + warnings.warn( + f"Expected glibc version with 2 components major.minor, got: {version_str}", + RuntimeWarning, + stacklevel=2, + ) + return _GLibCVersion(-1, -1) + return _GLibCVersion(int(m.group("major")), int(m.group("minor"))) + + +@functools.lru_cache +def _get_glibc_version() -> _GLibCVersion: + version_str = _glibc_version_string() + if version_str is None: + return _GLibCVersion(-1, -1) + return _parse_glibc_version(version_str) + + +# From PEP 513, PEP 600 +def _is_compatible(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: PLC0415 + 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) and hasattr(_manylinux, "manylinux1_compatible"): + return bool(_manylinux.manylinux1_compatible) + if version == _GLibCVersion(2, 12) and hasattr( + _manylinux, "manylinux2010_compatible" + ): + return bool(_manylinux.manylinux2010_compatible) + if version == _GLibCVersion(2, 17) and hasattr( + _manylinux, "manylinux2014_compatible" + ): + return bool(_manylinux.manylinux2014_compatible) + return True + + +_LEGACY_MANYLINUX_MAP: dict[_GLibCVersion, str] = { + # CentOS 7 w/ glibc 2.17 (PEP 599) + _GLibCVersion(2, 17): "manylinux2014", + # CentOS 6 w/ glibc 2.12 (PEP 571) + _GLibCVersion(2, 12): "manylinux2010", + # CentOS 5 w/ glibc 2.5 (PEP 513) + _GLibCVersion(2, 5): "manylinux1", +} + + +def platform_tags(archs: Sequence[str]) -> Iterator[str]: + """Generate manylinux tags compatible to the current platform. + + :param archs: Sequence of compatible architectures. + The first one shall be the closest to the actual architecture and 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 manylinux-compatible. + + :returns: An iterator of compatible manylinux tags. + """ + if not _have_compatible_abi(sys.executable, archs): + return + # Oldest glibc to be supported regardless of architecture is (2, 17). + too_old_glibc2 = _GLibCVersion(2, 16) + if set(archs) & {"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 arch in archs: + 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) + if _is_compatible(arch, glibc_version): + yield "manylinux_{}_{}_{}".format(*glibc_version, arch) + + # Handle the legacy manylinux1, manylinux2010, manylinux2014 tags. + if legacy_tag := _LEGACY_MANYLINUX_MAP.get(glibc_version): + yield f"{legacy_tag}_{arch}" diff --git a/venv/Lib/site-packages/pip/_vendor/packaking/_musllinux.py b/venv/Lib/site-packages/pip/_vendor/packaking/_musllinux.py new file mode 100644 index 0000000..4e8116a --- /dev/null +++ b/venv/Lib/site-packages/pip/_vendor/packaking/_musllinux.py @@ -0,0 +1,85 @@ +"""PEP 656 support. + +This module implements logic to detect if the currently running Python is +linked against musl, and what musl version is used. +""" + +from __future__ import annotations + +import functools +import re +import subprocess +import sys +from typing import Iterator, NamedTuple, Sequence + +from ._elffile import ELFFile + + +class _MuslVersion(NamedTuple): + major: int + minor: int + + +def _parse_musl_version(output: str) -> _MuslVersion | None: + 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) -> _MuslVersion | None: + """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], check=False, stderr=subprocess.PIPE, text=True) + return _parse_musl_version(proc.stderr) + + +def platform_tags(archs: Sequence[str]) -> Iterator[str]: + """Generate musllinux tags compatible to the current platform. + + :param archs: Sequence of compatible architectures. + The first one shall be the closest to the actual architecture and 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 arch in archs: + 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/venv/Lib/site-packages/pip/_vendor/packaking/_parser.py b/venv/Lib/site-packages/pip/_vendor/packaking/_parser.py new file mode 100644 index 0000000..f6c1f5c --- /dev/null +++ b/venv/Lib/site-packages/pip/_vendor/packaking/_parser.py @@ -0,0 +1,365 @@ +"""Handwritten parser of dependency specifiers. + +The docstring for each __parse_* function contains EBNF-inspired grammar representing +the implementation. +""" + +from __future__ import annotations + +import ast +from typing import List, Literal, NamedTuple, Sequence, Tuple, Union + +from ._tokenizer import DEFAULT_RULES, Tokenizer + + +class Node: + __slots__ = ("value",) + + 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.value!r})>" + + def serialize(self) -> str: + raise NotImplementedError + + +class Variable(Node): + __slots__ = () + + def serialize(self) -> str: + return str(self) + + +class Value(Node): + __slots__ = () + + def serialize(self) -> str: + return f'"{self}"' + + +class Op(Node): + __slots__ = () + + def serialize(self) -> str: + return str(self) + + +MarkerLogical = Literal["and", "or"] +MarkerVar = Union[Variable, Value] +MarkerItem = Tuple[MarkerVar, Op, MarkerVar] +MarkerAtom = Union[MarkerItem, Sequence["MarkerAtom"]] +MarkerList = List[Union["MarkerList", MarkerAtom, MarkerLogical]] + + +class ParsedRequirement(NamedTuple): + name: str + url: str + extras: list[str] + specifier: str + marker: MarkerList | None + + +# -------------------------------------------------------------------------------------- +# 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, MarkerList | None]: + """ + 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, + expected="semicolon (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, + expected=( + "comma (within version specifier), semicolon (after version specifier)" + if specifier + else "semicolon (after name with no version specifier)" + ), + ) + + return (url, specifier, marker) + + +def _parse_requirement_marker( + tokenizer: Tokenizer, *, span_start: int, expected: str +) -> MarkerList: + """ + requirement_marker = SEMICOLON marker WS? + """ + + if not tokenizer.check("SEMICOLON"): + tokenizer.raise_syntax_error( + f"Expected {expected} or end", + span_start=span_start, + span_end=None, + ) + 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_full_marker(Tokenizer(source, rules=DEFAULT_RULES)) + + +def _parse_full_marker(tokenizer: Tokenizer) -> MarkerList: + retval = _parse_marker(tokenizer) + tokenizer.expect("END", expected="end of marker expression") + return retval + + +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: # noqa: RET503 + """ + 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 in ("platform_python_implementation", "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" + )