• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
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 InvalidName(ValueError):
16    """
17    An invalid distribution name; users should refer to the packaging user guide.
18    """
19
20
21class InvalidWheelFilename(ValueError):
22    """
23    An invalid wheel filename was found, users should refer to PEP 427.
24    """
25
26
27class InvalidSdistFilename(ValueError):
28    """
29    An invalid sdist filename was found, users should refer to the packaging user guide.
30    """
31
32
33# Core metadata spec for `Name`
34_validate_regex = re.compile(
35    r"^([A-Z0-9]|[A-Z0-9][A-Z0-9._-]*[A-Z0-9])$", re.IGNORECASE
36)
37_canonicalize_regex = re.compile(r"[-_.]+")
38_normalized_regex = re.compile(r"^([a-z0-9]|[a-z0-9]([a-z0-9-](?!--))*[a-z0-9])$")
39# PEP 427: The build number must start with a digit.
40_build_tag_regex = re.compile(r"(\d+)(.*)")
41
42
43def canonicalize_name(name: str, *, validate: bool = False) -> NormalizedName:
44    if validate and not _validate_regex.match(name):
45        raise InvalidName(f"name is invalid: {name!r}")
46    # This is taken from PEP 503.
47    value = _canonicalize_regex.sub("-", name).lower()
48    return cast(NormalizedName, value)
49
50
51def is_normalized_name(name: str) -> bool:
52    return _normalized_regex.match(name) is not None
53
54
55def canonicalize_version(
56    version: Union[Version, str], *, strip_trailing_zero: bool = True
57) -> str:
58    """
59    This is very similar to Version.__str__, but has one subtle difference
60    with the way it handles the release segment.
61    """
62    if isinstance(version, str):
63        try:
64            parsed = Version(version)
65        except InvalidVersion:
66            # Legacy versions cannot be normalized
67            return version
68    else:
69        parsed = version
70
71    parts = []
72
73    # Epoch
74    if parsed.epoch != 0:
75        parts.append(f"{parsed.epoch}!")
76
77    # Release segment
78    release_segment = ".".join(str(x) for x in parsed.release)
79    if strip_trailing_zero:
80        # NB: This strips trailing '.0's to normalize
81        release_segment = re.sub(r"(\.0)+$", "", release_segment)
82    parts.append(release_segment)
83
84    # Pre-release
85    if parsed.pre is not None:
86        parts.append("".join(str(x) for x in parsed.pre))
87
88    # Post-release
89    if parsed.post is not None:
90        parts.append(f".post{parsed.post}")
91
92    # Development release
93    if parsed.dev is not None:
94        parts.append(f".dev{parsed.dev}")
95
96    # Local version segment
97    if parsed.local is not None:
98        parts.append(f"+{parsed.local}")
99
100    return "".join(parts)
101
102
103def parse_wheel_filename(
104    filename: str,
105) -> Tuple[NormalizedName, Version, BuildTag, FrozenSet[Tag]]:
106    if not filename.endswith(".whl"):
107        raise InvalidWheelFilename(
108            f"Invalid wheel filename (extension must be '.whl'): {filename}"
109        )
110
111    filename = filename[:-4]
112    dashes = filename.count("-")
113    if dashes not in (4, 5):
114        raise InvalidWheelFilename(
115            f"Invalid wheel filename (wrong number of parts): {filename}"
116        )
117
118    parts = filename.split("-", dashes - 2)
119    name_part = parts[0]
120    # See PEP 427 for the rules on escaping the project name.
121    if "__" in name_part or re.match(r"^[\w\d._]*$", name_part, re.UNICODE) is None:
122        raise InvalidWheelFilename(f"Invalid project name: {filename}")
123    name = canonicalize_name(name_part)
124
125    try:
126        version = Version(parts[1])
127    except InvalidVersion as e:
128        raise InvalidWheelFilename(
129            f"Invalid wheel filename (invalid version): {filename}"
130        ) from e
131
132    if dashes == 5:
133        build_part = parts[2]
134        build_match = _build_tag_regex.match(build_part)
135        if build_match is None:
136            raise InvalidWheelFilename(
137                f"Invalid build number: {build_part} in '{filename}'"
138            )
139        build = cast(BuildTag, (int(build_match.group(1)), build_match.group(2)))
140    else:
141        build = ()
142    tags = parse_tag(parts[-1])
143    return (name, version, build, tags)
144
145
146def parse_sdist_filename(filename: str) -> Tuple[NormalizedName, Version]:
147    if filename.endswith(".tar.gz"):
148        file_stem = filename[: -len(".tar.gz")]
149    elif filename.endswith(".zip"):
150        file_stem = filename[: -len(".zip")]
151    else:
152        raise InvalidSdistFilename(
153            f"Invalid sdist filename (extension must be '.tar.gz' or '.zip'):"
154            f" {filename}"
155        )
156
157    # We are requiring a PEP 440 version, which cannot contain dashes,
158    # so we split on the last dash.
159    name_part, sep, version_part = file_stem.rpartition("-")
160    if not sep:
161        raise InvalidSdistFilename(f"Invalid sdist filename: {filename}")
162
163    name = canonicalize_name(name_part)
164
165    try:
166        version = Version(version_part)
167    except InvalidVersion as e:
168        raise InvalidSdistFilename(
169            f"Invalid sdist filename (invalid version): {filename}"
170        ) from e
171
172    return (name, version)
173