| 1 |
from __future__ import annotations |
| 2 |
|
| 3 |
import os |
| 4 |
import re |
| 5 |
import abc |
| 6 |
import sys |
| 7 |
import json |
| 8 |
import email |
| 9 |
import types |
| 10 |
import inspect |
| 11 |
import pathlib |
| 12 |
import zipfile |
| 13 |
import operator |
| 14 |
import textwrap |
| 15 |
import warnings |
| 16 |
import functools |
| 17 |
import itertools |
| 18 |
import posixpath |
| 19 |
import collections |
| 20 |
|
| 21 |
from . import _meta |
| 22 |
from ._collections import FreezableDefaultDict, Pair |
| 23 |
from ._functools import method_cache, pass_none |
| 24 |
from ._itertools import always_iterable, unique_everseen |
| 25 |
from ._meta import PackageMetadata, SimplePath |
| 26 |
|
| 27 |
from contextlib import suppress |
| 28 |
from importlib import import_module |
| 29 |
from importlib.abc import MetaPathFinder |
| 30 |
from itertools import starmap |
| 31 |
from typing import Any, Iterable, List, Mapping, Match, Optional, Set, cast |
| 32 |
|
| 33 |
__all__ = [ |
| 34 |
'Distribution', |
| 35 |
'DistributionFinder', |
| 36 |
'PackageMetadata', |
| 37 |
'PackageNotFoundError', |
| 38 |
'distribution', |
| 39 |
'distributions', |
| 40 |
'entry_points', |
| 41 |
'files', |
| 42 |
'metadata', |
| 43 |
'packages_distributions', |
| 44 |
'requires', |
| 45 |
'version', |
| 46 |
] |
| 47 |
|
| 48 |
|
| 49 |
class PackageNotFoundError(ModuleNotFoundError): |
| 50 |
"""The package was not found.""" |
| 51 |
|
| 52 |
def __str__(self) -> str: |
| 53 |
return f"No package metadata was found for {self.name}" |
| 54 |
|
| 55 |
@property |
| 56 |
def name(self) -> str: # type: ignore[override] |
| 57 |
(name,) = self.args |
| 58 |
return name |
| 59 |
|
| 60 |
|
| 61 |
class Sectioned: |
| 62 |
""" |
| 63 |
A simple entry point config parser for performance |
| 64 |
|
| 65 |
>>> for item in Sectioned.read(Sectioned._sample): |
| 66 |
... print(item) |
| 67 |
Pair(name='sec1', value='# comments ignored') |
| 68 |
Pair(name='sec1', value='a = 1') |
| 69 |
Pair(name='sec1', value='b = 2') |
| 70 |
Pair(name='sec2', value='a = 2') |
| 71 |
|
| 72 |
>>> res = Sectioned.section_pairs(Sectioned._sample) |
| 73 |
>>> item = next(res) |
| 74 |
>>> item.name |
| 75 |
'sec1' |
| 76 |
>>> item.value |
| 77 |
Pair(name='a', value='1') |
| 78 |
>>> item = next(res) |
| 79 |
>>> item.value |
| 80 |
Pair(name='b', value='2') |
| 81 |
>>> item = next(res) |
| 82 |
>>> item.name |
| 83 |
'sec2' |
| 84 |
>>> item.value |
| 85 |
Pair(name='a', value='2') |
| 86 |
>>> list(res) |
| 87 |
[] |
| 88 |
""" |
| 89 |
|
| 90 |
_sample = textwrap.dedent( |
| 91 |
""" |
| 92 |
[sec1] |
| 93 |
# comments ignored |
| 94 |
a = 1 |
| 95 |
b = 2 |
| 96 |
|
| 97 |
[sec2] |
| 98 |
a = 2 |
| 99 |
""" |
| 100 |
).lstrip() |
| 101 |
|
| 102 |
@classmethod |
| 103 |
def section_pairs(cls, text): |
| 104 |
return ( |
| 105 |
section._replace(value=Pair.parse(section.value)) |
| 106 |
for section in cls.read(text, filter_=cls.valid) |
| 107 |
if section.name is not None |
| 108 |
) |
| 109 |
|
| 110 |
@staticmethod |
| 111 |
def read(text, filter_=None): |
| 112 |
lines = filter(filter_, map(str.strip, text.splitlines())) |
| 113 |
name = None |
| 114 |
for value in lines: |
| 115 |
section_match = value.startswith('[') and value.endswith(']') |
| 116 |
if section_match: |
| 117 |
name = value.strip('[]') |
| 118 |
continue |
| 119 |
yield Pair(name, value) |
| 120 |
|
| 121 |
@staticmethod |
| 122 |
def valid(line: str): |
| 123 |
return line and not line.startswith('#') |
| 124 |
|
| 125 |
|
| 126 |
class EntryPoint: |
| 127 |
"""An entry point as defined by Python packaging conventions. |
| 128 |
|
| 129 |
See `the packaging docs on entry points |
| 130 |
<https://packaging.python.org/specifications/entry-points/>`_ |
| 131 |
for more information. |
| 132 |
|
| 133 |
>>> ep = EntryPoint( |
| 134 |
... name=None, group=None, value='package.module:attr [extra1, extra2]') |
| 135 |
>>> ep.module |
| 136 |
'package.module' |
| 137 |
>>> ep.attr |
| 138 |
'attr' |
| 139 |
>>> ep.extras |
| 140 |
['extra1', 'extra2'] |
| 141 |
""" |
| 142 |
|
| 143 |
pattern = re.compile( |
| 144 |
r'(?P<module>[\w.]+)\s*' |
| 145 |
r'(:\s*(?P<attr>[\w.]+)\s*)?' |
| 146 |
r'((?P<extras>\[.*\])\s*)?$' |
| 147 |
) |
| 148 |
""" |
| 149 |
A regular expression describing the syntax for an entry point, |
| 150 |
which might look like: |
| 151 |
|
| 152 |
- module |
| 153 |
- package.module |
| 154 |
- package.module:attribute |
| 155 |
- package.module:object.attribute |
| 156 |
- package.module:attr [extra1, extra2] |
| 157 |
|
| 158 |
Other combinations are possible as well. |
| 159 |
|
| 160 |
The expression is lenient about whitespace around the ':', |
| 161 |
following the attr, and following any extras. |
| 162 |
""" |
| 163 |
|
| 164 |
name: str |
| 165 |
value: str |
| 166 |
group: str |
| 167 |
|
| 168 |
dist: Optional[Distribution] = None |
| 169 |
|
| 170 |
def __init__(self, name: str, value: str, group: str) -> None: |
| 171 |
vars(self).update(name=name, value=value, group=group) |
| 172 |
|
| 173 |
def load(self) -> Any: |
| 174 |
"""Load the entry point from its definition. If only a module |
| 175 |
is indicated by the value, return that module. Otherwise, |
| 176 |
return the named object. |
| 177 |
""" |
| 178 |
match = cast(Match, self.pattern.match(self.value)) |
| 179 |
module = import_module(match.group('module')) |
| 180 |
attrs = filter(None, (match.group('attr') or '').split('.')) |
| 181 |
return functools.reduce(getattr, attrs, module) |
| 182 |
|
| 183 |
@property |
| 184 |
def module(self) -> str: |
| 185 |
match = self.pattern.match(self.value) |
| 186 |
assert match is not None |
| 187 |
return match.group('module') |
| 188 |
|
| 189 |
@property |
| 190 |
def attr(self) -> str: |
| 191 |
match = self.pattern.match(self.value) |
| 192 |
assert match is not None |
| 193 |
return match.group('attr') |
| 194 |
|
| 195 |
@property |
| 196 |
def extras(self) -> List[str]: |
| 197 |
match = self.pattern.match(self.value) |
| 198 |
assert match is not None |
| 199 |
return re.findall(r'\w+', match.group('extras') or '') |
| 200 |
|
| 201 |
def _for(self, dist): |
| 202 |
vars(self).update(dist=dist) |
| 203 |
return self |
| 204 |
|
| 205 |
def matches(self, **params): |
| 206 |
""" |
| 207 |
EntryPoint matches the given parameters. |
| 208 |
|
| 209 |
>>> ep = EntryPoint(group='foo', name='bar', value='bing:bong [extra1, extra2]') |
| 210 |
>>> ep.matches(group='foo') |
| 211 |
True |
| 212 |
>>> ep.matches(name='bar', value='bing:bong [extra1, extra2]') |
| 213 |
True |
| 214 |
>>> ep.matches(group='foo', name='other') |
| 215 |
False |
| 216 |
>>> ep.matches() |
| 217 |
True |
| 218 |
>>> ep.matches(extras=['extra1', 'extra2']) |
| 219 |
True |
| 220 |
>>> ep.matches(module='bing') |
| 221 |
True |
| 222 |
>>> ep.matches(attr='bong') |
| 223 |
True |
| 224 |
""" |
| 225 |
attrs = (getattr(self, param) for param in params) |
| 226 |
return all(map(operator.eq, params.values(), attrs)) |
| 227 |
|
| 228 |
def _key(self): |
| 229 |
return self.name, self.value, self.group |
| 230 |
|
| 231 |
def __lt__(self, other): |
| 232 |
return self._key() < other._key() |
| 233 |
|
| 234 |
def __eq__(self, other): |
| 235 |
return self._key() == other._key() |
| 236 |
|
| 237 |
def __setattr__(self, name, value): |
| 238 |
raise AttributeError("EntryPoint objects are immutable.") |
| 239 |
|
| 240 |
def __repr__(self): |
| 241 |
return ( |
| 242 |
f'EntryPoint(name={self.name!r}, value={self.value!r}, ' |
| 243 |
f'group={self.group!r})' |
| 244 |
) |
| 245 |
|
| 246 |
def __hash__(self) -> int: |
| 247 |
return hash(self._key()) |
| 248 |
|
| 249 |
|
| 250 |
class EntryPoints(tuple): |
| 251 |
""" |
| 252 |
An immutable collection of selectable EntryPoint objects. |
| 253 |
""" |
| 254 |
|
| 255 |
__slots__ = () |
| 256 |
|
| 257 |
def __getitem__(self, name: str) -> EntryPoint: # type: ignore[override] |
| 258 |
""" |
| 259 |
Get the EntryPoint in self matching name. |
| 260 |
""" |
| 261 |
try: |
| 262 |
return next(iter(self.select(name=name))) |
| 263 |
except StopIteration: |
| 264 |
raise KeyError(name) |
| 265 |
|
| 266 |
def __repr__(self): |
| 267 |
""" |
| 268 |
Repr with classname and tuple constructor to |
| 269 |
signal that we deviate from regular tuple behavior. |
| 270 |
""" |
| 271 |
return '%s(%r)' % (self.__class__.__name__, tuple(self)) |
| 272 |
|
| 273 |
def select(self, **params) -> EntryPoints: |
| 274 |
""" |
| 275 |
Select entry points from self that match the |
| 276 |
given parameters (typically group and/or name). |
| 277 |
""" |
| 278 |
return EntryPoints(ep for ep in self if ep.matches(**params)) |
| 279 |
|
| 280 |
@property |
| 281 |
def names(self) -> Set[str]: |
| 282 |
""" |
| 283 |
Return the set of all names of all entry points. |
| 284 |
""" |
| 285 |
return {ep.name for ep in self} |
| 286 |
|
| 287 |
@property |
| 288 |
def groups(self) -> Set[str]: |
| 289 |
""" |
| 290 |
Return the set of all groups of all entry points. |
| 291 |
""" |
| 292 |
return {ep.group for ep in self} |
| 293 |
|
| 294 |
@classmethod |
| 295 |
def _from_text_for(cls, text, dist): |
| 296 |
return cls(ep._for(dist) for ep in cls._from_text(text)) |
| 297 |
|
| 298 |
@staticmethod |
| 299 |
def _from_text(text): |
| 300 |
return ( |
| 301 |
EntryPoint(name=item.value.name, value=item.value.value, group=item.name) |
| 302 |
for item in Sectioned.section_pairs(text or '') |
| 303 |
) |
| 304 |
|
| 305 |
|
| 306 |
class PackagePath(pathlib.PurePosixPath): |
| 307 |
"""A reference to a path in a package""" |
| 308 |
|
| 309 |
hash: Optional[FileHash] |
| 310 |
size: int |
| 311 |
dist: Distribution |
| 312 |
|
| 313 |
def read_text(self, encoding: str = 'utf-8') -> str: # type: ignore[override] |
| 314 |
return self.locate().read_text(encoding=encoding) |
| 315 |
|
| 316 |
def read_binary(self) -> bytes: |
| 317 |
return self.locate().read_bytes() |
| 318 |
|
| 319 |
def locate(self) -> SimplePath: |
| 320 |
"""Return a path-like object for this path""" |
| 321 |
return self.dist.locate_file(self) |
| 322 |
|
| 323 |
|
| 324 |
class FileHash: |
| 325 |
def __init__(self, spec: str) -> None: |
| 326 |
self.mode, _, self.value = spec.partition('=') |
| 327 |
|
| 328 |
def __repr__(self) -> str: |
| 329 |
return f'<FileHash mode: {self.mode} value: {self.value}>' |
| 330 |
|
| 331 |
|
| 332 |
class DeprecatedNonAbstract: |
| 333 |
# Required until Python 3.14 |
| 334 |
def __new__(cls, *args, **kwargs): |
| 335 |
all_names = { |
| 336 |
name for subclass in inspect.getmro(cls) for name in vars(subclass) |
| 337 |
} |
| 338 |
abstract = { |
| 339 |
name |
| 340 |
for name in all_names |
| 341 |
if getattr(getattr(cls, name), '__isabstractmethod__', False) |
| 342 |
} |
| 343 |
if abstract: |
| 344 |
warnings.warn( |
| 345 |
f"Unimplemented abstract methods {abstract}", |
| 346 |
DeprecationWarning, |
| 347 |
stacklevel=2, |
| 348 |
) |
| 349 |
return super().__new__(cls) |
| 350 |
|
| 351 |
|
| 352 |
class Distribution(DeprecatedNonAbstract): |
| 353 |
""" |
| 354 |
An abstract Python distribution package. |
| 355 |
|
| 356 |
Custom providers may derive from this class and define |
| 357 |
the abstract methods to provide a concrete implementation |
| 358 |
for their environment. Some providers may opt to override |
| 359 |
the default implementation of some properties to bypass |
| 360 |
the file-reading mechanism. |
| 361 |
""" |
| 362 |
|
| 363 |
@abc.abstractmethod |
| 364 |
def read_text(self, filename) -> Optional[str]: |
| 365 |
"""Attempt to load metadata file given by the name. |
| 366 |
|
| 367 |
Python distribution metadata is organized by blobs of text |
| 368 |
typically represented as "files" in the metadata directory |
| 369 |
(e.g. package-1.0.dist-info). These files include things |
| 370 |
like: |
| 371 |
|
| 372 |
- METADATA: The distribution metadata including fields |
| 373 |
like Name and Version and Description. |
| 374 |
- entry_points.txt: A series of entry points as defined in |
| 375 |
`the entry points spec <https://packaging.python.org/en/latest/specifications/entry-points/#file-format>`_. |
| 376 |
- RECORD: A record of files according to |
| 377 |
`this recording spec <https://packaging.python.org/en/latest/specifications/recording-installed-packages/#the-record-file>`_. |
| 378 |
|
| 379 |
A package may provide any set of files, including those |
| 380 |
not listed here or none at all. |
| 381 |
|
| 382 |
:param filename: The name of the file in the distribution info. |
| 383 |
:return: The text if found, otherwise None. |
| 384 |
""" |
| 385 |
|
| 386 |
@abc.abstractmethod |
| 387 |
def locate_file(self, path: str | os.PathLike[str]) -> SimplePath: |
| 388 |
""" |
| 389 |
Given a path to a file in this distribution, return a SimplePath |
| 390 |
to it. |
| 391 |
""" |
| 392 |
|
| 393 |
@classmethod |
| 394 |
def from_name(cls, name: str) -> Distribution: |
| 395 |
"""Return the Distribution for the given package name. |
| 396 |
|
| 397 |
:param name: The name of the distribution package to search for. |
| 398 |
:return: The Distribution instance (or subclass thereof) for the named |
| 399 |
package, if found. |
| 400 |
:raises PackageNotFoundError: When the named package's distribution |
| 401 |
metadata cannot be found. |
| 402 |
:raises ValueError: When an invalid value is supplied for name. |
| 403 |
""" |
| 404 |
if not name: |
| 405 |
raise ValueError("A distribution name is required.") |
| 406 |
try: |
| 407 |
return next(iter(cls.discover(name=name))) |
| 408 |
except StopIteration: |
| 409 |
raise PackageNotFoundError(name) |
| 410 |
|
| 411 |
@classmethod |
| 412 |
def discover( |
| 413 |
cls, *, context: Optional[DistributionFinder.Context] = None, **kwargs |
| 414 |
) -> Iterable[Distribution]: |
| 415 |
"""Return an iterable of Distribution objects for all packages. |
| 416 |
|
| 417 |
Pass a ``context`` or pass keyword arguments for constructing |
| 418 |
a context. |
| 419 |
|
| 420 |
:context: A ``DistributionFinder.Context`` object. |
| 421 |
:return: Iterable of Distribution objects for packages matching |
| 422 |
the context. |
| 423 |
""" |
| 424 |
if context and kwargs: |
| 425 |
raise ValueError("cannot accept context and kwargs") |
| 426 |
context = context or DistributionFinder.Context(**kwargs) |
| 427 |
return itertools.chain.from_iterable( |
| 428 |
resolver(context) for resolver in cls._discover_resolvers() |
| 429 |
) |
| 430 |
|
| 431 |
@staticmethod |
| 432 |
def at(path: str | os.PathLike[str]) -> Distribution: |
| 433 |
"""Return a Distribution for the indicated metadata path. |
| 434 |
|
| 435 |
:param path: a string or path-like object |
| 436 |
:return: a concrete Distribution instance for the path |
| 437 |
""" |
| 438 |
return PathDistribution(pathlib.Path(path)) |
| 439 |
|
| 440 |
@staticmethod |
| 441 |
def _discover_resolvers(): |
| 442 |
"""Search the meta_path for resolvers (MetadataPathFinders).""" |
| 443 |
declared = ( |
| 444 |
getattr(finder, 'find_distributions', None) for finder in sys.meta_path |
| 445 |
) |
| 446 |
return filter(None, declared) |
| 447 |
|
| 448 |
@property |
| 449 |
def metadata(self) -> _meta.PackageMetadata: |
| 450 |
"""Return the parsed metadata for this Distribution. |
| 451 |
|
| 452 |
The returned object will have keys that name the various bits of |
| 453 |
metadata per the |
| 454 |
`Core metadata specifications <https://packaging.python.org/en/latest/specifications/core-metadata/#core-metadata>`_. |
| 455 |
|
| 456 |
Custom providers may provide the METADATA file or override this |
| 457 |
property. |
| 458 |
""" |
| 459 |
# deferred for performance (python/cpython#109829) |
| 460 |
from . import _adapters |
| 461 |
|
| 462 |
opt_text = ( |
| 463 |
self.read_text('METADATA') |
| 464 |
or self.read_text('PKG-INFO') |
| 465 |
# This last clause is here to support old egg-info files. Its |
| 466 |
# effect is to just end up using the PathDistribution's self._path |
| 467 |
# (which points to the egg-info file) attribute unchanged. |
| 468 |
or self.read_text('') |
| 469 |
) |
| 470 |
text = cast(str, opt_text) |
| 471 |
return _adapters.Message(email.message_from_string(text)) |
| 472 |
|
| 473 |
@property |
| 474 |
def name(self) -> str: |
| 475 |
"""Return the 'Name' metadata for the distribution package.""" |
| 476 |
return self.metadata['Name'] |
| 477 |
|
| 478 |
@property |
| 479 |
def _normalized_name(self): |
| 480 |
"""Return a normalized version of the name.""" |
| 481 |
return Prepared.normalize(self.name) |
| 482 |
|
| 483 |
@property |
| 484 |
def version(self) -> str: |
| 485 |
"""Return the 'Version' metadata for the distribution package.""" |
| 486 |
return self.metadata['Version'] |
| 487 |
|
| 488 |
@property |
| 489 |
def entry_points(self) -> EntryPoints: |
| 490 |
""" |
| 491 |
Return EntryPoints for this distribution. |
| 492 |
|
| 493 |
Custom providers may provide the ``entry_points.txt`` file |
| 494 |
or override this property. |
| 495 |
""" |
| 496 |
return EntryPoints._from_text_for(self.read_text('entry_points.txt'), self) |
| 497 |
|
| 498 |
@property |
| 499 |
def files(self) -> Optional[List[PackagePath]]: |
| 500 |
"""Files in this distribution. |
| 501 |
|
| 502 |
:return: List of PackagePath for this distribution or None |
| 503 |
|
| 504 |
Result is `None` if the metadata file that enumerates files |
| 505 |
(i.e. RECORD for dist-info, or installed-files.txt or |
| 506 |
SOURCES.txt for egg-info) is missing. |
| 507 |
Result may be empty if the metadata exists but is empty. |
| 508 |
|
| 509 |
Custom providers are recommended to provide a "RECORD" file (in |
| 510 |
``read_text``) or override this property to allow for callers to be |
| 511 |
able to resolve filenames provided by the package. |
| 512 |
""" |
| 513 |
|
| 514 |
def make_file(name, hash=None, size_str=None): |
| 515 |
result = PackagePath(name) |
| 516 |
result.hash = FileHash(hash) if hash else None |
| 517 |
result.size = int(size_str) if size_str else None |
| 518 |
result.dist = self |
| 519 |
return result |
| 520 |
|
| 521 |
@pass_none |
| 522 |
def make_files(lines): |
| 523 |
# Delay csv import, since Distribution.files is not as widely used |
| 524 |
# as other parts of importlib.metadata |
| 525 |
import csv |
| 526 |
|
| 527 |
return starmap(make_file, csv.reader(lines)) |
| 528 |
|
| 529 |
@pass_none |
| 530 |
def skip_missing_files(package_paths): |
| 531 |
return list(filter(lambda path: path.locate().exists(), package_paths)) |
| 532 |
|
| 533 |
return skip_missing_files( |
| 534 |
make_files( |
| 535 |
self._read_files_distinfo() |
| 536 |
or self._read_files_egginfo_installed() |
| 537 |
or self._read_files_egginfo_sources() |
| 538 |
) |
| 539 |
) |
| 540 |
|
| 541 |
def _read_files_distinfo(self): |
| 542 |
""" |
| 543 |
Read the lines of RECORD. |
| 544 |
""" |
| 545 |
text = self.read_text('RECORD') |
| 546 |
return text and text.splitlines() |
| 547 |
|
| 548 |
def _read_files_egginfo_installed(self): |
| 549 |
""" |
| 550 |
Read installed-files.txt and return lines in a similar |
| 551 |
CSV-parsable format as RECORD: each file must be placed |
| 552 |
relative to the site-packages directory and must also be |
| 553 |
quoted (since file names can contain literal commas). |
| 554 |
|
| 555 |
This file is written when the package is installed by pip, |
| 556 |
but it might not be written for other installation methods. |
| 557 |
Assume the file is accurate if it exists. |
| 558 |
""" |
| 559 |
text = self.read_text('installed-files.txt') |
| 560 |
# Prepend the .egg-info/ subdir to the lines in this file. |
| 561 |
# But this subdir is only available from PathDistribution's |
| 562 |
# self._path. |
| 563 |
subdir = getattr(self, '_path', None) |
| 564 |
if not text or not subdir: |
| 565 |
return |
| 566 |
|
| 567 |
paths = ( |
| 568 |
(subdir / name) |
| 569 |
.resolve() |
| 570 |
.relative_to(self.locate_file('').resolve(), walk_up=True) |
| 571 |
.as_posix() |
| 572 |
for name in text.splitlines() |
| 573 |
) |
| 574 |
return map('"{}"'.format, paths) |
| 575 |
|
| 576 |
def _read_files_egginfo_sources(self): |
| 577 |
""" |
| 578 |
Read SOURCES.txt and return lines in a similar CSV-parsable |
| 579 |
format as RECORD: each file name must be quoted (since it |
| 580 |
might contain literal commas). |
| 581 |
|
| 582 |
Note that SOURCES.txt is not a reliable source for what |
| 583 |
files are installed by a package. This file is generated |
| 584 |
for a source archive, and the files that are present |
| 585 |
there (e.g. setup.py) may not correctly reflect the files |
| 586 |
that are present after the package has been installed. |
| 587 |
""" |
| 588 |
text = self.read_text('SOURCES.txt') |
| 589 |
return text and map('"{}"'.format, text.splitlines()) |
| 590 |
|
| 591 |
@property |
| 592 |
def requires(self) -> Optional[List[str]]: |
| 593 |
"""Generated requirements specified for this Distribution""" |
| 594 |
reqs = self._read_dist_info_reqs() or self._read_egg_info_reqs() |
| 595 |
return reqs and list(reqs) |
| 596 |
|
| 597 |
def _read_dist_info_reqs(self): |
| 598 |
return self.metadata.get_all('Requires-Dist') |
| 599 |
|
| 600 |
def _read_egg_info_reqs(self): |
| 601 |
source = self.read_text('requires.txt') |
| 602 |
return pass_none(self._deps_from_requires_text)(source) |
| 603 |
|
| 604 |
@classmethod |
| 605 |
def _deps_from_requires_text(cls, source): |
| 606 |
return cls._convert_egg_info_reqs_to_simple_reqs(Sectioned.read(source)) |
| 607 |
|
| 608 |
@staticmethod |
| 609 |
def _convert_egg_info_reqs_to_simple_reqs(sections): |
| 610 |
""" |
| 611 |
Historically, setuptools would solicit and store 'extra' |
| 612 |
requirements, including those with environment markers, |
| 613 |
in separate sections. More modern tools expect each |
| 614 |
dependency to be defined separately, with any relevant |
| 615 |
extras and environment markers attached directly to that |
| 616 |
requirement. This method converts the former to the |
| 617 |
latter. See _test_deps_from_requires_text for an example. |
| 618 |
""" |
| 619 |
|
| 620 |
def make_condition(name): |
| 621 |
return name and f'extra == "{name}"' |
| 622 |
|
| 623 |
def quoted_marker(section): |
| 624 |
section = section or '' |
| 625 |
extra, sep, markers = section.partition(':') |
| 626 |
if extra and markers: |
| 627 |
markers = f'({markers})' |
| 628 |
conditions = list(filter(None, [markers, make_condition(extra)])) |
| 629 |
return '; ' + ' and '.join(conditions) if conditions else '' |
| 630 |
|
| 631 |
def url_req_space(req): |
| 632 |
""" |
| 633 |
PEP 508 requires a space between the url_spec and the quoted_marker. |
| 634 |
Ref python/importlib_metadata#357. |
| 635 |
""" |
| 636 |
# '@' is uniquely indicative of a url_req. |
| 637 |
return ' ' * ('@' in req) |
| 638 |
|
| 639 |
for section in sections: |
| 640 |
space = url_req_space(section.value) |
| 641 |
yield section.value + space + quoted_marker(section.name) |
| 642 |
|
| 643 |
@property |
| 644 |
def origin(self): |
| 645 |
return self._load_json('direct_url.json') |
| 646 |
|
| 647 |
def _load_json(self, filename): |
| 648 |
return pass_none(json.loads)( |
| 649 |
self.read_text(filename), |
| 650 |
object_hook=lambda data: types.SimpleNamespace(**data), |
| 651 |
) |
| 652 |
|
| 653 |
|
| 654 |
class DistributionFinder(MetaPathFinder): |
| 655 |
""" |
| 656 |
A MetaPathFinder capable of discovering installed distributions. |
| 657 |
|
| 658 |
Custom providers should implement this interface in order to |
| 659 |
supply metadata. |
| 660 |
""" |
| 661 |
|
| 662 |
class Context: |
| 663 |
""" |
| 664 |
Keyword arguments presented by the caller to |
| 665 |
``distributions()`` or ``Distribution.discover()`` |
| 666 |
to narrow the scope of a search for distributions |
| 667 |
in all DistributionFinders. |
| 668 |
|
| 669 |
Each DistributionFinder may expect any parameters |
| 670 |
and should attempt to honor the canonical |
| 671 |
parameters defined below when appropriate. |
| 672 |
|
| 673 |
This mechanism gives a custom provider a means to |
| 674 |
solicit additional details from the caller beyond |
| 675 |
"name" and "path" when searching distributions. |
| 676 |
For example, imagine a provider that exposes suites |
| 677 |
of packages in either a "public" or "private" ``realm``. |
| 678 |
A caller may wish to query only for distributions in |
| 679 |
a particular realm and could call |
| 680 |
``distributions(realm="private")`` to signal to the |
| 681 |
custom provider to only include distributions from that |
| 682 |
realm. |
| 683 |
""" |
| 684 |
|
| 685 |
name = None |
| 686 |
""" |
| 687 |
Specific name for which a distribution finder should match. |
| 688 |
A name of ``None`` matches all distributions. |
| 689 |
""" |
| 690 |
|
| 691 |
def __init__(self, **kwargs): |
| 692 |
vars(self).update(kwargs) |
| 693 |
|
| 694 |
@property |
| 695 |
def path(self) -> List[str]: |
| 696 |
""" |
| 697 |
The sequence of directory path that a distribution finder |
| 698 |
should search. |
| 699 |
|
| 700 |
Typically refers to Python installed package paths such as |
| 701 |
"site-packages" directories and defaults to ``sys.path``. |
| 702 |
""" |
| 703 |
return vars(self).get('path', sys.path) |
| 704 |
|
| 705 |
@abc.abstractmethod |
| 706 |
def find_distributions(self, context=Context()) -> Iterable[Distribution]: |
| 707 |
""" |
| 708 |
Find distributions. |
| 709 |
|
| 710 |
Return an iterable of all Distribution instances capable of |
| 711 |
loading the metadata for packages matching the ``context``, |
| 712 |
a DistributionFinder.Context instance. |
| 713 |
""" |
| 714 |
|
| 715 |
|
| 716 |
class FastPath: |
| 717 |
""" |
| 718 |
Micro-optimized class for searching a root for children. |
| 719 |
|
| 720 |
Root is a path on the file system that may contain metadata |
| 721 |
directories either as natural directories or within a zip file. |
| 722 |
|
| 723 |
>>> FastPath('').children() |
| 724 |
['...'] |
| 725 |
|
| 726 |
FastPath objects are cached and recycled for any given root. |
| 727 |
|
| 728 |
>>> FastPath('foobar') is FastPath('foobar') |
| 729 |
True |
| 730 |
""" |
| 731 |
|
| 732 |
@functools.lru_cache() # type: ignore |
| 733 |
def __new__(cls, root): |
| 734 |
return super().__new__(cls) |
| 735 |
|
| 736 |
def __init__(self, root): |
| 737 |
self.root = root |
| 738 |
|
| 739 |
def joinpath(self, child): |
| 740 |
return pathlib.Path(self.root, child) |
| 741 |
|
| 742 |
def children(self): |
| 743 |
with suppress(Exception): |
| 744 |
return os.listdir(self.root or '.') |
| 745 |
with suppress(Exception): |
| 746 |
return self.zip_children() |
| 747 |
return [] |
| 748 |
|
| 749 |
def zip_children(self): |
| 750 |
zip_path = zipfile.Path(self.root) |
| 751 |
names = zip_path.root.namelist() |
| 752 |
self.joinpath = zip_path.joinpath |
| 753 |
|
| 754 |
return dict.fromkeys(child.split(posixpath.sep, 1)[0] for child in names) |
| 755 |
|
| 756 |
def search(self, name): |
| 757 |
return self.lookup(self.mtime).search(name) |
| 758 |
|
| 759 |
@property |
| 760 |
def mtime(self): |
| 761 |
with suppress(OSError): |
| 762 |
return os.stat(self.root).st_mtime |
| 763 |
self.lookup.cache_clear() |
| 764 |
|
| 765 |
@method_cache |
| 766 |
def lookup(self, mtime): |
| 767 |
return Lookup(self) |
| 768 |
|
| 769 |
|
| 770 |
class Lookup: |
| 771 |
""" |
| 772 |
A micro-optimized class for searching a (fast) path for metadata. |
| 773 |
""" |
| 774 |
|
| 775 |
def __init__(self, path: FastPath): |
| 776 |
""" |
| 777 |
Calculate all of the children representing metadata. |
| 778 |
|
| 779 |
From the children in the path, calculate early all of the |
| 780 |
children that appear to represent metadata (infos) or legacy |
| 781 |
metadata (eggs). |
| 782 |
""" |
| 783 |
|
| 784 |
base = os.path.basename(path.root).lower() |
| 785 |
base_is_egg = base.endswith(".egg") |
| 786 |
self.infos = FreezableDefaultDict(list) |
| 787 |
self.eggs = FreezableDefaultDict(list) |
| 788 |
|
| 789 |
for child in path.children(): |
| 790 |
low = child.lower() |
| 791 |
if low.endswith((".dist-info", ".egg-info")): |
| 792 |
# rpartition is faster than splitext and suitable for this purpose. |
| 793 |
name = low.rpartition(".")[0].partition("-")[0] |
| 794 |
normalized = Prepared.normalize(name) |
| 795 |
self.infos[normalized].append(path.joinpath(child)) |
| 796 |
elif base_is_egg and low == "egg-info": |
| 797 |
name = base.rpartition(".")[0].partition("-")[0] |
| 798 |
legacy_normalized = Prepared.legacy_normalize(name) |
| 799 |
self.eggs[legacy_normalized].append(path.joinpath(child)) |
| 800 |
|
| 801 |
self.infos.freeze() |
| 802 |
self.eggs.freeze() |
| 803 |
|
| 804 |
def search(self, prepared: Prepared): |
| 805 |
""" |
| 806 |
Yield all infos and eggs matching the Prepared query. |
| 807 |
""" |
| 808 |
infos = ( |
| 809 |
self.infos[prepared.normalized] |
| 810 |
if prepared |
| 811 |
else itertools.chain.from_iterable(self.infos.values()) |
| 812 |
) |
| 813 |
eggs = ( |
| 814 |
self.eggs[prepared.legacy_normalized] |
| 815 |
if prepared |
| 816 |
else itertools.chain.from_iterable(self.eggs.values()) |
| 817 |
) |
| 818 |
return itertools.chain(infos, eggs) |
| 819 |
|
| 820 |
|
| 821 |
class Prepared: |
| 822 |
""" |
| 823 |
A prepared search query for metadata on a possibly-named package. |
| 824 |
|
| 825 |
Pre-calculates the normalization to prevent repeated operations. |
| 826 |
|
| 827 |
>>> none = Prepared(None) |
| 828 |
>>> none.normalized |
| 829 |
>>> none.legacy_normalized |
| 830 |
>>> bool(none) |
| 831 |
False |
| 832 |
>>> sample = Prepared('Sample__Pkg-name.foo') |
| 833 |
>>> sample.normalized |
| 834 |
'sample_pkg_name_foo' |
| 835 |
>>> sample.legacy_normalized |
| 836 |
'sample__pkg_name.foo' |
| 837 |
>>> bool(sample) |
| 838 |
True |
| 839 |
""" |
| 840 |
|
| 841 |
normalized = None |
| 842 |
legacy_normalized = None |
| 843 |
|
| 844 |
def __init__(self, name: Optional[str]): |
| 845 |
self.name = name |
| 846 |
if name is None: |
| 847 |
return |
| 848 |
self.normalized = self.normalize(name) |
| 849 |
self.legacy_normalized = self.legacy_normalize(name) |
| 850 |
|
| 851 |
@staticmethod |
| 852 |
def normalize(name): |
| 853 |
""" |
| 854 |
PEP 503 normalization plus dashes as underscores. |
| 855 |
""" |
| 856 |
return re.sub(r"[-_.]+", "-", name).lower().replace('-', '_') |
| 857 |
|
| 858 |
@staticmethod |
| 859 |
def legacy_normalize(name): |
| 860 |
""" |
| 861 |
Normalize the package name as found in the convention in |
| 862 |
older packaging tools versions and specs. |
| 863 |
""" |
| 864 |
return name.lower().replace('-', '_') |
| 865 |
|
| 866 |
def __bool__(self): |
| 867 |
return bool(self.name) |
| 868 |
|
| 869 |
|
| 870 |
class MetadataPathFinder(DistributionFinder): |
| 871 |
@classmethod |
| 872 |
def find_distributions( |
| 873 |
cls, context=DistributionFinder.Context() |
| 874 |
) -> Iterable[PathDistribution]: |
| 875 |
""" |
| 876 |
Find distributions. |
| 877 |
|
| 878 |
Return an iterable of all Distribution instances capable of |
| 879 |
loading the metadata for packages matching ``context.name`` |
| 880 |
(or all names if ``None`` indicated) along the paths in the list |
| 881 |
of directories ``context.path``. |
| 882 |
""" |
| 883 |
found = cls._search_paths(context.name, context.path) |
| 884 |
return map(PathDistribution, found) |
| 885 |
|
| 886 |
@classmethod |
| 887 |
def _search_paths(cls, name, paths): |
| 888 |
"""Find metadata directories in paths heuristically.""" |
| 889 |
prepared = Prepared(name) |
| 890 |
return itertools.chain.from_iterable( |
| 891 |
path.search(prepared) for path in map(FastPath, paths) |
| 892 |
) |
| 893 |
|
| 894 |
@classmethod |
| 895 |
def invalidate_caches(cls) -> None: |
| 896 |
FastPath.__new__.cache_clear() |
| 897 |
|
| 898 |
|
| 899 |
class PathDistribution(Distribution): |
| 900 |
def __init__(self, path: SimplePath) -> None: |
| 901 |
"""Construct a distribution. |
| 902 |
|
| 903 |
:param path: SimplePath indicating the metadata directory. |
| 904 |
""" |
| 905 |
self._path = path |
| 906 |
|
| 907 |
def read_text(self, filename: str | os.PathLike[str]) -> Optional[str]: |
| 908 |
with suppress( |
| 909 |
FileNotFoundError, |
| 910 |
IsADirectoryError, |
| 911 |
KeyError, |
| 912 |
NotADirectoryError, |
| 913 |
PermissionError, |
| 914 |
): |
| 915 |
return self._path.joinpath(filename).read_text(encoding='utf-8') |
| 916 |
|
| 917 |
return None |
| 918 |
|
| 919 |
read_text.__doc__ = Distribution.read_text.__doc__ |
| 920 |
|
| 921 |
def locate_file(self, path: str | os.PathLike[str]) -> SimplePath: |
| 922 |
return self._path.parent / path |
| 923 |
|
| 924 |
@property |
| 925 |
def _normalized_name(self): |
| 926 |
""" |
| 927 |
Performance optimization: where possible, resolve the |
| 928 |
normalized name from the file system path. |
| 929 |
""" |
| 930 |
stem = os.path.basename(str(self._path)) |
| 931 |
return ( |
| 932 |
pass_none(Prepared.normalize)(self._name_from_stem(stem)) |
| 933 |
or super()._normalized_name |
| 934 |
) |
| 935 |
|
| 936 |
@staticmethod |
| 937 |
def _name_from_stem(stem): |
| 938 |
""" |
| 939 |
>>> PathDistribution._name_from_stem('foo-3.0.egg-info') |
| 940 |
'foo' |
| 941 |
>>> PathDistribution._name_from_stem('CherryPy-3.0.dist-info') |
| 942 |
'CherryPy' |
| 943 |
>>> PathDistribution._name_from_stem('face.egg-info') |
| 944 |
'face' |
| 945 |
>>> PathDistribution._name_from_stem('foo.bar') |
| 946 |
""" |
| 947 |
filename, ext = os.path.splitext(stem) |
| 948 |
if ext not in ('.dist-info', '.egg-info'): |
| 949 |
return |
| 950 |
name, sep, rest = filename.partition('-') |
| 951 |
return name |
| 952 |
|
| 953 |
|
| 954 |
def distribution(distribution_name: str) -> Distribution: |
| 955 |
"""Get the ``Distribution`` instance for the named package. |
| 956 |
|
| 957 |
:param distribution_name: The name of the distribution package as a string. |
| 958 |
:return: A ``Distribution`` instance (or subclass thereof). |
| 959 |
""" |
| 960 |
return Distribution.from_name(distribution_name) |
| 961 |
|
| 962 |
|
| 963 |
def distributions(**kwargs) -> Iterable[Distribution]: |
| 964 |
"""Get all ``Distribution`` instances in the current environment. |
| 965 |
|
| 966 |
:return: An iterable of ``Distribution`` instances. |
| 967 |
""" |
| 968 |
return Distribution.discover(**kwargs) |
| 969 |
|
| 970 |
|
| 971 |
def metadata(distribution_name: str) -> _meta.PackageMetadata: |
| 972 |
"""Get the metadata for the named package. |
| 973 |
|
| 974 |
:param distribution_name: The name of the distribution package to query. |
| 975 |
:return: A PackageMetadata containing the parsed metadata. |
| 976 |
""" |
| 977 |
return Distribution.from_name(distribution_name).metadata |
| 978 |
|
| 979 |
|
| 980 |
def version(distribution_name: str) -> str: |
| 981 |
"""Get the version string for the named package. |
| 982 |
|
| 983 |
:param distribution_name: The name of the distribution package to query. |
| 984 |
:return: The version string for the package as defined in the package's |
| 985 |
"Version" metadata key. |
| 986 |
""" |
| 987 |
return distribution(distribution_name).version |
| 988 |
|
| 989 |
|
| 990 |
_unique = functools.partial( |
| 991 |
unique_everseen, |
| 992 |
key=operator.attrgetter('_normalized_name'), |
| 993 |
) |
| 994 |
""" |
| 995 |
Wrapper for ``distributions`` to return unique distributions by name. |
| 996 |
""" |
| 997 |
|
| 998 |
|
| 999 |
def entry_points(**params) -> EntryPoints: |
| 1000 |
"""Return EntryPoint objects for all installed packages. |
| 1001 |
|
| 1002 |
Pass selection parameters (group or name) to filter the |
| 1003 |
result to entry points matching those properties (see |
| 1004 |
EntryPoints.select()). |
| 1005 |
|
| 1006 |
:return: EntryPoints for all installed packages. |
| 1007 |
""" |
| 1008 |
eps = itertools.chain.from_iterable( |
| 1009 |
dist.entry_points for dist in _unique(distributions()) |
| 1010 |
) |
| 1011 |
return EntryPoints(eps).select(**params) |
| 1012 |
|
| 1013 |
|
| 1014 |
def files(distribution_name: str) -> Optional[List[PackagePath]]: |
| 1015 |
"""Return a list of files for the named package. |
| 1016 |
|
| 1017 |
:param distribution_name: The name of the distribution package to query. |
| 1018 |
:return: List of files composing the distribution. |
| 1019 |
""" |
| 1020 |
return distribution(distribution_name).files |
| 1021 |
|
| 1022 |
|
| 1023 |
def requires(distribution_name: str) -> Optional[List[str]]: |
| 1024 |
""" |
| 1025 |
Return a list of requirements for the named package. |
| 1026 |
|
| 1027 |
:return: An iterable of requirements, suitable for |
| 1028 |
packaging.requirement.Requirement. |
| 1029 |
""" |
| 1030 |
return distribution(distribution_name).requires |
| 1031 |
|
| 1032 |
|
| 1033 |
def packages_distributions() -> Mapping[str, List[str]]: |
| 1034 |
""" |
| 1035 |
Return a mapping of top-level packages to their |
| 1036 |
distributions. |
| 1037 |
|
| 1038 |
>>> import collections.abc |
| 1039 |
>>> pkgs = packages_distributions() |
| 1040 |
>>> all(isinstance(dist, collections.abc.Sequence) for dist in pkgs.values()) |
| 1041 |
True |
| 1042 |
""" |
| 1043 |
pkg_to_dist = collections.defaultdict(list) |
| 1044 |
for dist in distributions(): |
| 1045 |
for pkg in _top_level_declared(dist) or _top_level_inferred(dist): |
| 1046 |
pkg_to_dist[pkg].append(dist.metadata['Name']) |
| 1047 |
return dict(pkg_to_dist) |
| 1048 |
|
| 1049 |
|
| 1050 |
def _top_level_declared(dist): |
| 1051 |
return (dist.read_text('top_level.txt') or '').split() |
| 1052 |
|
| 1053 |
|
| 1054 |
def _topmost(name: PackagePath) -> Optional[str]: |
| 1055 |
""" |
| 1056 |
Return the top-most parent as long as there is a parent. |
| 1057 |
""" |
| 1058 |
top, *rest = name.parts |
| 1059 |
return top if rest else None |
| 1060 |
|
| 1061 |
|
| 1062 |
def _get_toplevel_name(name: PackagePath) -> str: |
| 1063 |
""" |
| 1064 |
Infer a possibly importable module name from a name presumed on |
| 1065 |
sys.path. |
| 1066 |
|
| 1067 |
>>> _get_toplevel_name(PackagePath('foo.py')) |
| 1068 |
'foo' |
| 1069 |
>>> _get_toplevel_name(PackagePath('foo')) |
| 1070 |
'foo' |
| 1071 |
>>> _get_toplevel_name(PackagePath('foo.pyc')) |
| 1072 |
'foo' |
| 1073 |
>>> _get_toplevel_name(PackagePath('foo/__init__.py')) |
| 1074 |
'foo' |
| 1075 |
>>> _get_toplevel_name(PackagePath('foo.pth')) |
| 1076 |
'foo.pth' |
| 1077 |
>>> _get_toplevel_name(PackagePath('foo.dist-info')) |
| 1078 |
'foo.dist-info' |
| 1079 |
""" |
| 1080 |
return _topmost(name) or ( |
| 1081 |
# python/typeshed#10328 |
| 1082 |
inspect.getmodulename(name) # type: ignore |
| 1083 |
or str(name) |
| 1084 |
) |
| 1085 |
|
| 1086 |
|
| 1087 |
def _top_level_inferred(dist): |
| 1088 |
opt_names = set(map(_get_toplevel_name, always_iterable(dist.files))) |
| 1089 |
|
| 1090 |
def importable_name(name): |
| 1091 |
return '.' not in name |
| 1092 |
|
| 1093 |
return filter(importable_name, opt_names) |