1# This file is dual licensed under the terms of the Apache License, Version 2# 2.0, and the BSD License. See the LICENSE file in the root of this repository 3# for complete details. 4 5import re 6from typing import FrozenSet, NewType, Tuple, Union, cast 7 8from .tags import Tag, parse_tag 9from .version import InvalidVersion, Version 10 11BuildTag = Union[Tuple[()], Tuple[int, str]] 12NormalizedName = NewType("NormalizedName", str) 13 14 15class InvalidWheelFilename(ValueError): 16 """ 17 An invalid wheel filename was found, users should refer to PEP 427. 18 """ 19 20 21class InvalidSdistFilename(ValueError): 22 """ 23 An invalid sdist filename was found, users should refer to the packaging user guide. 24 """ 25 26 27_canonicalize_regex = re.compile(r"[-_.]+") 28# PEP 427: The build number must start with a digit. 29_build_tag_regex = re.compile(r"(\d+)(.*)") 30 31 32def canonicalize_name(name: str) -> NormalizedName: 33 # This is taken from PEP 503. 34 value = _canonicalize_regex.sub("-", name).lower() 35 return cast(NormalizedName, value) 36 37 38def canonicalize_version(version: Union[Version, str]) -> str: 39 """ 40 This is very similar to Version.__str__, but has one subtle difference 41 with the way it handles the release segment. 42 """ 43 if isinstance(version, str): 44 try: 45 parsed = Version(version) 46 except InvalidVersion: 47 # Legacy versions cannot be normalized 48 return version 49 else: 50 parsed = version 51 52 parts = [] 53 54 # Epoch 55 if parsed.epoch != 0: 56 parts.append(f"{parsed.epoch}!") 57 58 # Release segment 59 # NB: This strips trailing '.0's to normalize 60 parts.append(re.sub(r"(\.0)+$", "", ".".join(str(x) for x in parsed.release))) 61 62 # Pre-release 63 if parsed.pre is not None: 64 parts.append("".join(str(x) for x in parsed.pre)) 65 66 # Post-release 67 if parsed.post is not None: 68 parts.append(f".post{parsed.post}") 69 70 # Development release 71 if parsed.dev is not None: 72 parts.append(f".dev{parsed.dev}") 73 74 # Local version segment 75 if parsed.local is not None: 76 parts.append(f"+{parsed.local}") 77 78 return "".join(parts) 79 80 81def parse_wheel_filename( 82 filename: str, 83) -> Tuple[NormalizedName, Version, BuildTag, FrozenSet[Tag]]: 84 if not filename.endswith(".whl"): 85 raise InvalidWheelFilename( 86 f"Invalid wheel filename (extension must be '.whl'): {filename}" 87 ) 88 89 filename = filename[:-4] 90 dashes = filename.count("-") 91 if dashes not in (4, 5): 92 raise InvalidWheelFilename( 93 f"Invalid wheel filename (wrong number of parts): {filename}" 94 ) 95 96 parts = filename.split("-", dashes - 2) 97 name_part = parts[0] 98 # See PEP 427 for the rules on escaping the project name 99 if "__" in name_part or re.match(r"^[\w\d._]*$", name_part, re.UNICODE) is None: 100 raise InvalidWheelFilename(f"Invalid project name: {filename}") 101 name = canonicalize_name(name_part) 102 version = Version(parts[1]) 103 if dashes == 5: 104 build_part = parts[2] 105 build_match = _build_tag_regex.match(build_part) 106 if build_match is None: 107 raise InvalidWheelFilename( 108 f"Invalid build number: {build_part} in '{filename}'" 109 ) 110 build = cast(BuildTag, (int(build_match.group(1)), build_match.group(2))) 111 else: 112 build = () 113 tags = parse_tag(parts[-1]) 114 return (name, version, build, tags) 115 116 117def parse_sdist_filename(filename: str) -> Tuple[NormalizedName, Version]: 118 if filename.endswith(".tar.gz"): 119 file_stem = filename[: -len(".tar.gz")] 120 elif filename.endswith(".zip"): 121 file_stem = filename[: -len(".zip")] 122 else: 123 raise InvalidSdistFilename( 124 f"Invalid sdist filename (extension must be '.tar.gz' or '.zip'):" 125 f" {filename}" 126 ) 127 128 # We are requiring a PEP 440 version, which cannot contain dashes, 129 # so we split on the last dash. 130 name_part, sep, version_part = file_stem.rpartition("-") 131 if not sep: 132 raise InvalidSdistFilename(f"Invalid sdist filename: {filename}") 133 134 name = canonicalize_name(name_part) 135 version = Version(version_part) 136 return (name, version) 137