|
| 1 | +# |
| 2 | +# Copyright (c) nexB Inc. and others. All rights reserved. |
| 3 | +# ScanCode is a trademark of nexB Inc. |
| 4 | +# SPDX-License-Identifier: Apache-2.0 |
| 5 | +# See http://www.apache.org/licenses/LICENSE-2.0 for the license text. |
| 6 | +# See https://github.com/nexB/scancode-toolkit for support or download. |
| 7 | +# See https://aboutcode.org for more information about nexB OSS projects. |
| 8 | +# |
| 9 | + |
| 10 | +import collections |
| 11 | +import operator |
| 12 | +import os |
| 13 | +from typing import List |
| 14 | + |
| 15 | +import packaging.markers |
| 16 | +import packaging.requirements |
| 17 | +import packaging.specifiers |
| 18 | +import packaging.utils |
| 19 | +import packaging.version |
| 20 | +import requests |
| 21 | +from packageurl import PackageURL |
| 22 | +from packaging.requirements import Requirement |
| 23 | +from resolvelib import AbstractProvider |
| 24 | +from resolvelib import Resolver |
| 25 | +from resolvelib.reporters import BaseReporter |
| 26 | + |
| 27 | +from _packagedcode.pypi import PypiWheelHandler |
| 28 | +from python_inspector.utils_pypi import CACHE_THIRDPARTY_DIR |
| 29 | +from python_inspector.utils_pypi import PYPI_PUBLIC_REPO |
| 30 | +from python_inspector.utils_pypi import PYPI_SIMPLE_URL |
| 31 | +from python_inspector.utils_pypi import Environment |
| 32 | +from python_inspector.utils_pypi import PypiSimpleRepository |
| 33 | +from python_inspector.utils_pypi import download_wheel |
| 34 | + |
| 35 | +Candidate = collections.namedtuple("Candidate", "name version extras") |
| 36 | + |
| 37 | + |
| 38 | +def get_response(url): |
| 39 | + """ |
| 40 | + Return a response for the given url. |
| 41 | + """ |
| 42 | + resp = requests.get(url) |
| 43 | + if resp.status_code == 200: |
| 44 | + return resp.json() |
| 45 | + return None |
| 46 | + |
| 47 | + |
| 48 | +def is_valid_version(parsed_version, requirements, identifier, bad_versions): |
| 49 | + """ |
| 50 | + Return True if the parsed_version is valid for the given identifier. |
| 51 | + """ |
| 52 | + if ( |
| 53 | + any(parsed_version not in r.specifier for r in requirements[identifier]) |
| 54 | + or parsed_version in bad_versions |
| 55 | + ): |
| 56 | + return False |
| 57 | + return True |
| 58 | + |
| 59 | + |
| 60 | +class PythonInputProvider(AbstractProvider): |
| 61 | + def __init__(self, environment=None, repos=[]): |
| 62 | + self.environment = environment |
| 63 | + self.repos = repos |
| 64 | + self.versions_by_package = {} |
| 65 | + self.dependencies_by_purl = {} |
| 66 | + |
| 67 | + def identify(self, requirement_or_candidate): |
| 68 | + name = packaging.utils.canonicalize_name(requirement_or_candidate.name) |
| 69 | + if requirement_or_candidate.extras: |
| 70 | + extras_str = ",".join(sorted(requirement_or_candidate.extras)) |
| 71 | + return "{}[{}]".format(name, extras_str) |
| 72 | + return name |
| 73 | + |
| 74 | + def get_preference( |
| 75 | + self, |
| 76 | + identifier, |
| 77 | + resolutions, |
| 78 | + candidates, |
| 79 | + information, |
| 80 | + backtrack_causes, |
| 81 | + ): |
| 82 | + transitive = all(p is not None for _, p in information[identifier]) |
| 83 | + return (transitive, identifier) |
| 84 | + |
| 85 | + def get_versions_for_package(self, name, repo=None): |
| 86 | + """ |
| 87 | + Return a list of versions for a package. |
| 88 | + """ |
| 89 | + versions = [] |
| 90 | + if repo and self.environment: |
| 91 | + for version, package in repo._get_package_versions_map(name).items(): |
| 92 | + wheels = package.get_supported_wheels(environment=self.environment) |
| 93 | + if list(wheels): |
| 94 | + versions.append(version) |
| 95 | + else: |
| 96 | + if name not in self.versions_by_package: |
| 97 | + api_url = f"https://pypi.org/pypi/{name}/json" |
| 98 | + resp = get_response(api_url) |
| 99 | + if not resp: |
| 100 | + self.versions_by_package[name] = [] |
| 101 | + releases = resp.get("releases") or {} |
| 102 | + self.versions_by_package[name] = releases.keys() or [] |
| 103 | + versions = self.versions_by_package[name] |
| 104 | + return versions |
| 105 | + |
| 106 | + def get_requirements_for_package(self, purl, candidate): |
| 107 | + """ |
| 108 | + Yield requirements for a package. |
| 109 | + """ |
| 110 | + if self.repos and self.environment: |
| 111 | + wheels = download_wheel( |
| 112 | + name=candidate.name, |
| 113 | + version=str(candidate.version), |
| 114 | + environment=self.environment, |
| 115 | + repos=self.repos, |
| 116 | + ) |
| 117 | + for wheel in wheels: |
| 118 | + deps = list(PypiWheelHandler.parse(os.path.join(CACHE_THIRDPARTY_DIR, wheel))) |
| 119 | + assert len(deps) == 1 |
| 120 | + deps = deps[0].dependencies |
| 121 | + for dep in deps: |
| 122 | + if dep.scope == "install": |
| 123 | + yield packaging.requirements.Requirement(str(dep.extracted_requirement)) |
| 124 | + else: |
| 125 | + if str(purl) not in self.dependencies_by_purl: |
| 126 | + api_url = f"https://pypi.org/pypi/{purl.name}/{purl.version}/json" |
| 127 | + resp = get_response(api_url) |
| 128 | + if not resp: |
| 129 | + self.dependencies_by_purl[str(purl)] = [] |
| 130 | + info = resp.get("info") or {} |
| 131 | + requires_dist = info.get("requires_dist") or [] |
| 132 | + self.dependencies_by_purl[str(purl)] = requires_dist |
| 133 | + for dependency in self.dependencies_by_purl[str(purl)]: |
| 134 | + yield packaging.requirements.Requirement(dependency) |
| 135 | + |
| 136 | + def get_candidates(self, all_versions, requirements, identifier, bad_versions, name, extras): |
| 137 | + """ |
| 138 | + Generate candidates for the given identifier. |
| 139 | + """ |
| 140 | + for version in all_versions: |
| 141 | + parsed_version = packaging.version.parse(version) |
| 142 | + if not is_valid_version(parsed_version, requirements, identifier, bad_versions): |
| 143 | + continue |
| 144 | + yield Candidate(name=name, version=parsed_version, extras=extras) |
| 145 | + |
| 146 | + def _iter_matches(self, identifier, requirements, incompatibilities): |
| 147 | + """ |
| 148 | + Return a list of candidates for the given identifier. |
| 149 | + """ |
| 150 | + name, _, _ = identifier.partition("[") |
| 151 | + bad_versions = {c.version for c in incompatibilities[identifier]} |
| 152 | + extras = {e for r in requirements[identifier] for e in r.extras} |
| 153 | + if not self.repos: |
| 154 | + all_versions = self.get_versions_for_package(name) |
| 155 | + yield from self.get_candidates( |
| 156 | + all_versions, requirements, identifier, bad_versions, name, extras |
| 157 | + ) |
| 158 | + else: |
| 159 | + for repo in self.repos: |
| 160 | + all_versions = self.get_versions_for_package(name, repo) |
| 161 | + yield from self.get_candidates( |
| 162 | + all_versions, requirements, identifier, bad_versions, name, extras |
| 163 | + ) |
| 164 | + |
| 165 | + def find_matches(self, identifier, requirements, incompatibilities): |
| 166 | + candidates = sorted( |
| 167 | + self._iter_matches(identifier, requirements, incompatibilities), |
| 168 | + key=operator.attrgetter("version"), |
| 169 | + reverse=True, |
| 170 | + ) |
| 171 | + return candidates |
| 172 | + |
| 173 | + def is_satisfied_by(self, requirement, candidate): |
| 174 | + return candidate.version in requirement.specifier |
| 175 | + |
| 176 | + def _iter_dependencies(self, candidate): |
| 177 | + """ |
| 178 | + Yield dependencies for the given candidate. |
| 179 | + """ |
| 180 | + name = packaging.utils.canonicalize_name(candidate.name) |
| 181 | + if candidate.extras: |
| 182 | + r = f"{name}=={candidate.version}" |
| 183 | + yield packaging.requirements.Requirement(r) |
| 184 | + purl = PackageURL( |
| 185 | + type="pypi", |
| 186 | + name=name, |
| 187 | + version=str(candidate.version), |
| 188 | + ) |
| 189 | + for r in self.get_requirements_for_package(purl, candidate): |
| 190 | + if r.marker is None: |
| 191 | + yield r |
| 192 | + else: |
| 193 | + if r.marker.evaluate({"extra": ""}): |
| 194 | + yield r |
| 195 | + |
| 196 | + def get_dependencies(self, candidate): |
| 197 | + return list(self._iter_dependencies(candidate)) |
| 198 | + |
| 199 | + |
| 200 | +def get_all_srcs(mapping, graph): |
| 201 | + """ |
| 202 | + Return a list of all sources in the graph. |
| 203 | + """ |
| 204 | + for name in mapping.keys(): |
| 205 | + if list(graph.iter_parents(name)) == [None]: |
| 206 | + yield name |
| 207 | + |
| 208 | + |
| 209 | +def dfs(mapping, graph, src): |
| 210 | + """ |
| 211 | + Return a recursive mapping of dependencies. |
| 212 | + """ |
| 213 | + children = list(graph.iter_children(src)) |
| 214 | + src_purl = PackageURL( |
| 215 | + type="pypi", |
| 216 | + name=src, |
| 217 | + version=str(mapping[src].version), |
| 218 | + ) |
| 219 | + if not children: |
| 220 | + return dict(package=str(src_purl), dependencies=[]) |
| 221 | + |
| 222 | + return dict( |
| 223 | + package=str(src_purl), |
| 224 | + dependencies=sorted([dfs(mapping, graph, c) for c in children], key=lambda d: d["package"]), |
| 225 | + ) |
| 226 | + |
| 227 | + |
| 228 | +def format_resolution(result): |
| 229 | + """ |
| 230 | + Return a formatted resolution. |
| 231 | + """ |
| 232 | + mapping = result.mapping |
| 233 | + graph = result.graph |
| 234 | + as_list = [ |
| 235 | + str( |
| 236 | + PackageURL( |
| 237 | + type="pypi", |
| 238 | + name=name, |
| 239 | + version=str(candidate.version), |
| 240 | + ) |
| 241 | + ) |
| 242 | + for name, candidate in mapping.items() |
| 243 | + ] |
| 244 | + |
| 245 | + as_parent_children = [] |
| 246 | + parents = mapping.keys() |
| 247 | + for parent in parents: |
| 248 | + parent_purl = PackageURL( |
| 249 | + type="pypi", |
| 250 | + name=parent, |
| 251 | + version=str(mapping[parent].version), |
| 252 | + ) |
| 253 | + dependencies = [] |
| 254 | + for dependency in graph.iter_children(parent): |
| 255 | + dep_purl = PackageURL( |
| 256 | + type="pypi", |
| 257 | + name=dependency, |
| 258 | + version=str(mapping[dependency].version), |
| 259 | + ) |
| 260 | + dependencies.append(str(dep_purl)) |
| 261 | + dependencies.sort() |
| 262 | + parent_children = dict(package=str(parent_purl), dependencies=dependencies) |
| 263 | + as_parent_children.append(parent_children) |
| 264 | + |
| 265 | + srcs = list(get_all_srcs(mapping=mapping, graph=graph)) |
| 266 | + dependencies = [] |
| 267 | + for src in srcs: |
| 268 | + dependencies.append(dfs(mapping=mapping, graph=graph, src=src)) |
| 269 | + |
| 270 | + as_list.sort() |
| 271 | + as_parent_children.sort(key=lambda d: d["package"]) |
| 272 | + dependencies.sort(key=lambda d: d["package"]) |
| 273 | + as_tree = dict(dependencies=dependencies) |
| 274 | + return as_list, as_parent_children, as_tree |
| 275 | + |
| 276 | + |
| 277 | +def pypi_simple_repo_in_repos(repos: PypiSimpleRepository): |
| 278 | + """ |
| 279 | + Return True if simple pypi index_url is present in any of the repos |
| 280 | + """ |
| 281 | + for repo in repos: |
| 282 | + if repo.index_url == PYPI_SIMPLE_URL: |
| 283 | + return True |
| 284 | + return False |
| 285 | + |
| 286 | + |
| 287 | +def resolution( |
| 288 | + requirements: List[Requirement], |
| 289 | + environment: Environment = None, |
| 290 | + repos: List[PypiSimpleRepository] = [], |
| 291 | + return_as_parent_children: bool = True, |
| 292 | + return_as_tree: bool = False, |
| 293 | + return_as_list: bool = False, |
| 294 | +): |
| 295 | + """ |
| 296 | + Return a resolution for the given requirements. |
| 297 | + """ |
| 298 | + if repos and not pypi_simple_repo_in_repos(repos): |
| 299 | + repos.append(PYPI_PUBLIC_REPO) |
| 300 | + resolver = Resolver(PythonInputProvider(environment, repos), BaseReporter()) |
| 301 | + as_list, as_parent_children, as_tree = format_resolution(resolver.resolve(requirements)) |
| 302 | + if return_as_parent_children: |
| 303 | + return as_parent_children |
| 304 | + if return_as_tree: |
| 305 | + return as_tree |
| 306 | + if return_as_list: |
| 307 | + return as_list |
0 commit comments