diff --git a/src/packaging/_pyproject.py b/src/packaging/_pyproject.py new file mode 100644 index 00000000..cfc9962f --- /dev/null +++ b/src/packaging/_pyproject.py @@ -0,0 +1,445 @@ +# 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. + +from __future__ import annotations + +import dataclasses +import pathlib +import re +import typing + +import packaging.requirements + +from .errors import ErrorCollector + +if typing.TYPE_CHECKING: + from collections.abc import Generator, Iterable, Sequence + + from .project_table import ContactTable, Dynamic, ProjectTable + from .requirements import Requirement + + +__all__ = [ + "License", + "Readme", +] + + +def __dir__() -> list[str]: + return __all__ + + +@dataclasses.dataclass(frozen=True) +class License: + """ + This represents a classic license, which contains text, and optionally a + file path. Modern licenses are just SPDX identifiers, which are strings. + """ + + text: str + file: pathlib.Path | None + + +@dataclasses.dataclass(frozen=True) +class Readme: + """ + This represents a readme, which contains text and a content type, and + optionally a file path. + """ + + text: str + file: pathlib.Path | None + content_type: str + + +T = typing.TypeVar("T") + + +@dataclasses.dataclass +class PyProjectReader(ErrorCollector): + """Class for reading pyproject.toml fields with error collection. + + Unrelated errors are collected and raised at once if the `collect_errors` + parameter is set to `True`. Some methods will return None if an error was + raised. Most of them expect a non-None value as input to enforce the caller + to handle missing vs. error correctly. The exact design is based on usage, + as this is an internal class. + """ + + def ensure_str(self, value: str, key: str) -> str | None: + """Ensure that a value is a string.""" + if isinstance(value, str): + return value + + msg = "Field {key} has an invalid type, expecting a string" + self.config_error(msg, key=key, got_type=type(value)) + return None + + def ensure_list(self, val: list[T], key: str) -> list[T] | None: + """Ensure that a value is a list of strings.""" + if not isinstance(val, list): + msg = "Field {key} has an invalid type, expecting a list of strings" + self.config_error(msg, key=key, got_type=type(val)) + return None + for item in val: + if not isinstance(item, str): + msg = "Field {key} contains item with invalid type, expecting a string" + self.config_error(msg, key=key, got_type=type(item)) + return None + + return val + + def ensure_dict(self, val: dict[str, str], key: str) -> dict[str, str] | None: + """Ensure that a value is a dictionary of strings.""" + if not isinstance(val, dict): + msg = "Field {key} has an invalid type, expecting a table of strings" + self.config_error(msg, key=key, got_type=type(val)) + return None + for subkey, item in val.items(): + if not isinstance(item, str): + msg = "Field {key} has an invalid type, expecting a string" + self.config_error(msg, key=f"{key}.{subkey}", got_type=type(item)) + return None + return val + + def ensure_people( + self, val: Sequence[ContactTable], key: str + ) -> list[tuple[str, str | None]]: + """Ensure that a value is a list of tables with optional "name" and "email" keys.""" + if not isinstance(val, list): + msg = ( + "Field {key} has an invalid type, expecting a list of " + 'tables containing the "name" and/or "email" keys' + ) + self.config_error(msg, key=key, got_type=type(val)) + return [] + for each in val: + if not isinstance(each, dict): + msg = ( + "Field {key} has an invalid type, expecting a list of " + 'tables containing the "name" and/or "email" keys' + " (got list with {type_name})" + ) + self.config_error(msg, key=key, type_name=type(each).__name__) + return [] + for value in each.values(): + if not isinstance(value, str): + msg = ( + "Field {key} has an invalid type, expecting a list of " + 'tables containing the "name" and/or "email" keys' + " (got list with dict with {type_name})" + ) + self.config_error(msg, key=key, type_name=type(value).__name__) + return [] + extra_keys = set(each) - {"name", "email"} + if extra_keys: + msg = ( + "Field {key} has an invalid type, expecting a list of " + 'tables containing the "name" and/or "email" keys' + " (got list with dict with extra keys {extra_keys})" + ) + self.config_error( + msg, + key=key, + extra_keys=", ".join(sorted(f'"{k}"' for k in extra_keys)), + ) + return [] + return [(entry.get("name", "Unknown"), entry.get("email")) for entry in val] + + def get_license( + self, project: ProjectTable, project_dir: pathlib.Path + ) -> License | str | None: + """Get the license field from the project table. Handles PEP 639 style license too. + + None is returned if the license field is not present or if an error occurred. + """ + val = project.get("license") + if val is None: + return None + if isinstance(val, str): + return val + + if isinstance(val, dict): + _license = self.ensure_dict(val, "project.license") # type: ignore[arg-type] + if _license is None: + return None + else: + msg = "Field {key} has an invalid type, expecting a string or table of strings" + self.config_error(msg, key="project.license", got_type=type(val)) + return None + + for field in _license: + if field not in ("file", "text"): + msg = "Unexpected field {key}" + self.config_error(msg, key=f"project.license.{field}") + return None + + file: pathlib.Path | None = None + filename = _license.get("file") + text = _license.get("text") + + if (filename and text) or (not filename and not text): + msg = ( + 'Invalid {key} contents, expecting a string or one key "file" or "text"' + ) + self.config_error(msg, key="project.license", got=_license) + return None + + if filename: + file = project_dir.joinpath(filename) + if not file.is_file(): + msg = f"License file not found ({filename!r})" + self.config_error(msg, key="project.license.file") + return None + text = file.read_text(encoding="utf-8") + + assert text is not None + return License(text, file) + + def get_license_files( + self, project: ProjectTable, project_dir: pathlib.Path + ) -> list[pathlib.Path] | None: + """Get the license-files list of files from the project table. + + Returns None if an error occurred (including invalid globs, etc) or if + not present. + """ + license_files = project.get("license-files") + if license_files is None: + return None + if self.ensure_list(license_files, "project.license-files") is None: + return None + + return list(self._get_files_from_globs(project_dir, license_files)) + + def get_readme( + self, project: ProjectTable, project_dir: pathlib.Path + ) -> Readme | None: + """Get the text of the readme from the project table. + + Returns None if an error occurred or if the readme field is not present. + """ + if "readme" not in project: + return None + + filename: str | None = None + file: pathlib.Path | None = None + text: str | None = None + content_type: str | None = None + + readme = project["readme"] + if isinstance(readme, str): + # readme is a file + text = None + filename = readme + if filename.endswith(".md"): + content_type = "text/markdown" + elif filename.endswith(".rst"): + content_type = "text/x-rst" + else: + msg = "Could not infer content type for readme file {filename!r}" + self.config_error(msg, key="project.readme", filename=filename) + return None + elif isinstance(readme, dict): + # readme is a dict containing either 'file' or 'text', and content-type + for field in readme: + if field not in ("content-type", "file", "text"): + msg = "Unexpected field {key}" + self.config_error(msg, key=f"project.readme.{field}") + return None + + content_type_raw = readme.get("content-type") + if content_type_raw is not None: + content_type = self.ensure_str( + content_type_raw, "project.readme.content-type" + ) + if content_type is None: + return None + filename_raw = readme.get("file") + if filename_raw is not None: + filename = self.ensure_str(filename_raw, "project.readme.file") + if filename is None: + return None + + text_raw = readme.get("text") + if text_raw is not None: + text = self.ensure_str(text_raw, "project.readme.text") + if text is None: + return None + + if (filename and text) or (not filename and not text): + msg = 'Invalid {key} contents, expecting either "file" or "text"' + self.config_error(msg, key="project.readme", got=readme) + return None + if not content_type: + msg = "Field {key} missing" + self.config_error(msg, key="project.readme.content-type") + return None + else: + msg = "Field {key} has an invalid type, expecting either a string or table of strings" + self.config_error(msg, key="project.readme", got_type=type(readme)) + return None + + if filename: + file = project_dir.joinpath(filename) + if not file.is_file(): + msg = "Readme file not found ({filename!r})" + self.config_error(msg, key="project.readme.file", filename=filename) + return None + text = file.read_text(encoding="utf-8") + + assert text is not None + return Readme(text, file, content_type) + + def get_dependencies(self, project: ProjectTable) -> list[Requirement]: + """Get the dependencies from the project table.""" + + requirement_strings: list[str] | None = None + requirement_strings_raw = project.get("dependencies") + if requirement_strings_raw is not None: + requirement_strings = self.ensure_list( + requirement_strings_raw, "project.dependencies" + ) + if requirement_strings is None: + return [] + + requirements: list[Requirement] = [] + for req in requirement_strings: + try: + requirements.append(packaging.requirements.Requirement(req)) + except packaging.requirements.InvalidRequirement as e: + msg = "Field {key} contains an invalid PEP 508 requirement string {req!r} ({error!r})" + self.config_error(msg, key="project.dependencies", req=req, error=e) + return [] + return requirements + + def get_optional_dependencies( + self, + project: ProjectTable, + ) -> dict[str, list[Requirement]]: + """Get the optional dependencies from the project table.""" + + val = project.get("optional-dependencies") + if not val: + return {} + + requirements_dict: dict[str, list[Requirement]] = {} + if not isinstance(val, dict): + msg = "Field {key} has an invalid type, expecting a table of PEP 508 requirement strings" + self.config_error( + msg, key="project.optional-dependencies", got_type=type(val) + ) + return {} + for extra, requirements in val.copy().items(): + assert isinstance(extra, str) + if not isinstance(requirements, list): + msg = "Field {key} has an invalid type, expecting a table of PEP 508 requirement strings" + self.config_error( + msg, + key=f"project.optional-dependencies.{extra}", + got_type=type(requirements), + ) + return {} + requirements_dict[extra] = [] + for req in requirements: + if not isinstance(req, str): + msg = "Field {key} has an invalid type, expecting a PEP 508 requirement string" + self.config_error( + msg, + key=f"project.optional-dependencies.{extra}", + got_type=type(req), + ) + return {} + try: + requirements_dict[extra].append( + packaging.requirements.Requirement(req) + ) + except packaging.requirements.InvalidRequirement as e: + msg = ( + "Field {key} contains " + "an invalid PEP 508 requirement string {req!r} ({error!r})" + ) + self.config_error( + msg, + key=f"project.optional-dependencies.{extra}", + req=req, + error=e, + ) + return {} + return dict(requirements_dict) + + def get_entrypoints(self, project: ProjectTable) -> dict[str, dict[str, str]]: + """Get the entrypoints from the project table.""" + + val = project.get("entry-points", None) + if val is None: + return {} + if not isinstance(val, dict): + msg = "Field {key} has an invalid type, expecting a table of entrypoint sections" + self.config_error(msg, key="project.entry-points", got_type=type(val)) + return {} + for section, entrypoints in val.items(): + assert isinstance(section, str) + if not re.match(r"^\w+(\.\w+)*$", section): + msg = ( + "Field {key} has an invalid value, expecting a name " + "containing only alphanumeric, underscore, or dot characters" + ) + self.config_error(msg, key="project.entry-points", got=section) + return {} + if not isinstance(entrypoints, dict): + msg = ( + "Field {key} has an invalid type, expecting a table of entrypoints" + ) + self.config_error( + msg, + key=f"project.entry-points.{section}", + got_type=type(entrypoints), + ) + return {} + for name, entrypoint in entrypoints.items(): + assert isinstance(name, str) + if not isinstance(entrypoint, str): + msg = "Field {key} has an invalid type, expecting a string" + self.config_error( + msg, + key=f"project.entry-points.{section}.{name}", + got_type=type(entrypoint), + ) + return {} + return val + + def get_dynamic(self, project: ProjectTable) -> list[Dynamic]: + """Get the dynamic fields from the project table. + + Returns an empty list if the field is not present or if an error occurred. + """ + dynamic = project.get("dynamic", []) + + self.ensure_list(dynamic, "project.dynamic") + + if "name" in dynamic: + msg = "Unsupported field 'name' in {key}" + self.config_error(msg, key="project.dynamic") + return [] + + return dynamic + + def _get_files_from_globs( + self, project_dir: pathlib.Path, globs: Iterable[str] + ) -> Generator[pathlib.Path, None, None]: + """Given a list of globs, get files that match.""" + + for glob in globs: + if glob.startswith(("..", "/")): + msg = "{glob!r} is an invalid {key} glob: the pattern must match files within the project directory" + self.config_error(msg, key="project.license-files", glob=glob) + break + files = [f for f in project_dir.glob(glob) if f.is_file()] + if not files: + msg = "Every pattern in {key} must match at least one file: {glob!r} did not match any" + self.config_error(msg, key="project.license-files", glob=glob) + break + for f in files: + yield f.relative_to(project_dir) diff --git a/src/packaging/errors.py b/src/packaging/errors.py new file mode 100644 index 00000000..17cd6e15 --- /dev/null +++ b/src/packaging/errors.py @@ -0,0 +1,88 @@ +from __future__ import annotations + +import contextlib +import dataclasses +import sys +from collections.abc import Generator +from typing import Any, NoReturn + +__all__ = ["ExceptionGroup", "ConfigurationError", "ConfigurationWarning"] + + +if sys.version_info >= (3, 11): # pragma: no cover + from builtins import ExceptionGroup +else: # pragma: no cover + + class ExceptionGroup(Exception): + """A minimal implementation of :external:exc:`ExceptionGroup` from Python 3.11. + + If :external:exc:`ExceptionGroup` is already defined by Python itself, + that version is used instead. + """ + + message: str + exceptions: list[Exception] + + def __init__(self, message: str, exceptions: list[Exception]) -> None: + self.message = message + self.exceptions = exceptions + + def __repr__(self) -> str: + return f"{self.__class__.__name__}({self.message!r}, {self.exceptions!r})" + + +class ConfigurationError(Exception): + """Error in the backend metadata. Has an optional key attribute, which will be non-None + if the error is related to a single key in the pyproject.toml file.""" + + def __init__(self, msg: str, *, key: str | None = None): + super().__init__(msg) + self._key = key + + @property + def key(self) -> str | None: # pragma: no cover + return self._key + + +class ConfigurationWarning(UserWarning): + """Warnings about backend metadata.""" + + +@dataclasses.dataclass +class ErrorCollector: + """ + Collect errors and raise them as a group at the end (if collect_errors is True), + otherwise raise them immediately. + """ + + errors: list[Exception] = dataclasses.field(default_factory=list) + + def config_error( + self, + msg: str, + *, + key: str | None = None, + got: Any = None, + got_type: type[Any] | None = None, + **kwargs: Any, + ) -> None: + """Raise a configuration error, or add it to the error list.""" + msg = msg.format(key=f'"{key}"', **kwargs) + if got is not None: + msg = f"{msg} (got {got!r})" + if got_type is not None: + msg = f"{msg} (got {got_type.__name__})" + + self.errors.append(ConfigurationError(msg, key=key)) + + def finalize(self, msg: str) -> NoReturn: + """Raise a group exception if there are any errors.""" + raise ExceptionGroup(msg, self.errors) + + @contextlib.contextmanager + def collect(self) -> Generator[None, None, None]: + """Support nesting; add any grouped errors to the error list.""" + try: + yield + except ExceptionGroup as error: + self.errors.extend(error.exceptions) diff --git a/src/packaging/metadata.py b/src/packaging/metadata.py index 09912ee5..4520a4c8 100644 --- a/src/packaging/metadata.py +++ b/src/packaging/metadata.py @@ -1,6 +1,5 @@ from __future__ import annotations -import builtins import email.feedparser import email.header import email.message @@ -19,33 +18,12 @@ from . import licenses, requirements, specifiers, utils from . import version as version_module +from .errors import ExceptionGroup from .licenses import NormalizedLicenseExpression T = typing.TypeVar("T") -if "ExceptionGroup" in builtins.__dict__: # pragma: no cover - ExceptionGroup = ExceptionGroup -else: # pragma: no cover - - class ExceptionGroup(Exception): - """A minimal implementation of :external:exc:`ExceptionGroup` from Python 3.11. - - If :external:exc:`ExceptionGroup` is already defined by Python itself, - that version is used instead. - """ - - message: str - exceptions: list[Exception] - - def __init__(self, message: str, exceptions: list[Exception]) -> None: - self.message = message - self.exceptions = exceptions - - def __repr__(self) -> str: - return f"{self.__class__.__name__}({self.message!r}, {self.exceptions!r})" - - class InvalidMetadata(ValueError): """A metadata field contains invalid data.""" @@ -170,6 +148,7 @@ class RawMetadata(TypedDict, total=False): _DICT_FIELDS = { "project_urls", } +ALL_FIELDS = _STRING_FIELDS | _LIST_FIELDS | _DICT_FIELDS def _parse_keywords(data: str) -> list[str]: diff --git a/src/packaging/project.py b/src/packaging/project.py new file mode 100644 index 00000000..68d943f6 --- /dev/null +++ b/src/packaging/project.py @@ -0,0 +1,495 @@ +# 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. + +from __future__ import annotations + +import copy +import dataclasses +import email.message +import email.policy +import email.utils +import os +import os.path +import pathlib +import sys +import typing +import warnings + +from . import markers, specifiers, utils +from . import metadata as packaging_metadata +from . import version as packaging_version +from ._pyproject import License, PyProjectReader, Readme +from .errors import ConfigurationError, ConfigurationWarning, ErrorCollector + +if typing.TYPE_CHECKING: + from collections.abc import Mapping + from typing import Any + + from .requirements import Requirement + + if sys.version_info < (3, 11): + from typing_extensions import Self + else: + from typing import Self + + from .project_table import Dynamic, PyProjectTable + +__all__ = [ + "ConfigurationError", + "License", + "Readme", + "StandardMetadata", + "extras_build_system", + "extras_project", + "extras_top_level", +] + +KNOWN_TOPLEVEL_FIELDS = {"build-system", "project", "tool", "dependency-groups"} +KNOWN_BUILD_SYSTEM_FIELDS = {"backend-path", "build-backend", "requires"} +KNOWN_PROJECT_FIELDS = { + "authors", + "classifiers", + "dependencies", + "description", + "dynamic", + "entry-points", + "gui-scripts", + "keywords", + "license", + "license-files", + "maintainers", + "name", + "optional-dependencies", + "readme", + "requires-python", + "scripts", + "urls", + "version", +} +PRE_SPDX_METADATA_VERSIONS = {"2.1", "2.2", "2.3"} + + +def extras_top_level(pyproject_table: Mapping[str, Any]) -> set[str]: + """ + Return any extra keys in the top-level of the pyproject table. + """ + return set(pyproject_table) - KNOWN_TOPLEVEL_FIELDS + + +def extras_build_system(pyproject_table: Mapping[str, Any]) -> set[str]: + """ + Return any extra keys in the build-system table. + """ + return set(pyproject_table.get("build-system", [])) - KNOWN_BUILD_SYSTEM_FIELDS + + +def extras_project(pyproject_table: Mapping[str, Any]) -> set[str]: + """ + Return any extra keys in the project table. + """ + return set(pyproject_table.get("project", [])) - KNOWN_PROJECT_FIELDS + + +@dataclasses.dataclass +class StandardMetadata: + """ + This class represents the standard metadata fields for a project. It can be + used to read metadata from a pyproject.toml table, validate it, and write it + to an RFC822 message or JSON. + """ + + name: str + version: packaging_version.Version | None = None + description: str | None = None + license: License | str | None = None + license_files: list[pathlib.Path] | None = None + readme: Readme | None = None + requires_python: specifiers.SpecifierSet | None = None + dependencies: list[Requirement] = dataclasses.field(default_factory=list) + optional_dependencies: dict[str, list[Requirement]] = dataclasses.field( + default_factory=dict + ) + entrypoints: dict[str, dict[str, str]] = dataclasses.field(default_factory=dict) + authors: list[tuple[str, str | None]] = dataclasses.field(default_factory=list) + maintainers: list[tuple[str, str | None]] = dataclasses.field(default_factory=list) + urls: dict[str, str] = dataclasses.field(default_factory=dict) + classifiers: list[str] = dataclasses.field(default_factory=list) + keywords: list[str] = dataclasses.field(default_factory=list) + scripts: dict[str, str] = dataclasses.field(default_factory=dict) + gui_scripts: dict[str, str] = dataclasses.field(default_factory=dict) + dynamic: list[Dynamic] = dataclasses.field(default_factory=list) + """ + This field is used to track dynamic fields. You can't set a field not in this list. + """ + + def __post_init__(self) -> None: + self.validate() + + @property + def canonical_name(self) -> str: + """ + Return the canonical name of the project. + """ + return utils.canonicalize_name(self.name) + + @classmethod + def from_pyproject( + cls, + data: Mapping[str, Any], + project_dir: str | os.PathLike[str] = os.path.curdir, + ) -> Self: + """ + Read metadata from a pyproject.toml table. This is the main method for + creating an instance of this class. It also supports two additional + fields: ``allow_extra_keys`` to control what happens when extra keys are + present in the pyproject table, and ``all_errors``, to raise all errors + in an ExceptionGroup instead of raising the first one. + """ + pyproject = PyProjectReader() + + pyproject_table: PyProjectTable = data # type: ignore[assignment] + if "project" not in pyproject_table: + msg = "Section {key} missing in pyproject.toml" + pyproject.config_error(msg, key="project") + pyproject.finalize("Failed to parse pyproject.toml") + msg = "Unreachable code" # pragma: no cover + raise AssertionError(msg) # pragma: no cover + + project = pyproject_table["project"] + project_dir = pathlib.Path(project_dir) + + extra_keys = extras_project(data) + if extra_keys: + extra_keys_str = ", ".join(sorted(f"{k!r}" for k in extra_keys)) + msg = "Extra keys present in {key}: {extra_keys}" + pyproject.config_error( + msg, + key="project", + extra_keys=extra_keys_str, + ) + + dynamic = pyproject.get_dynamic(project) + + for field in dynamic: + if field in data["project"]: + msg = 'Field {key} declared as dynamic in "project.dynamic" but is defined' + pyproject.config_error(msg, key=f"project.{field}") + + raw_name = project.get("name") + name = "UNKNOWN" + if raw_name is None: + msg = "Field {key} missing" + pyproject.config_error(msg, key="project.name") + else: + tmp_name = pyproject.ensure_str(raw_name, "project.name") + if tmp_name is not None: + name = tmp_name + + version: packaging_version.Version | None = packaging_version.Version("0.0.0") + raw_version = project.get("version") + if raw_version is not None: + version_string = pyproject.ensure_str(raw_version, "project.version") + if version_string is not None: + try: + version = ( + packaging_version.Version(version_string) + if version_string + else None + ) + except packaging_version.InvalidVersion: + msg = "Invalid {key} value, expecting a valid PEP 440 version" + pyproject.config_error( + msg, key="project.version", got=version_string + ) + elif "version" not in dynamic: + msg = ( + "Field {key} missing and 'version' not specified in \"project.dynamic\"" + ) + pyproject.config_error(msg, key="project.version") + + # Description fills Summary, which cannot be multiline + # However, throwing an error isn't backward compatible, + # so leave it up to the users for now. + project_description_raw = project.get("description") + description = ( + pyproject.ensure_str(project_description_raw, "project.description") + if project_description_raw is not None + else None + ) + + requires_python_raw = project.get("requires-python") + requires_python = None + if requires_python_raw is not None: + requires_python_string = pyproject.ensure_str( + requires_python_raw, "project.requires-python" + ) + if requires_python_string is not None: + try: + requires_python = specifiers.SpecifierSet(requires_python_string) + except specifiers.InvalidSpecifier: + msg = "Invalid {key} value, expecting a valid specifier set" + pyproject.config_error( + msg, key="project.requires-python", got=requires_python_string + ) + + self = None + with pyproject.collect(): + self = cls( + name=name, + version=version, + description=description, + license=pyproject.get_license(project, project_dir), + license_files=pyproject.get_license_files(project, project_dir), + readme=pyproject.get_readme(project, project_dir), + requires_python=requires_python, + dependencies=pyproject.get_dependencies(project), + optional_dependencies=pyproject.get_optional_dependencies(project), + entrypoints=pyproject.get_entrypoints(project), + authors=pyproject.ensure_people( + project.get("authors", []), "project.authors" + ), + maintainers=pyproject.ensure_people( + project.get("maintainers", []), "project.maintainers" + ), + urls=pyproject.ensure_dict(project.get("urls", {}), "project.urls") + or {}, + classifiers=pyproject.ensure_list( + project.get("classifiers", []), "project.classifiers" + ) + or [], + keywords=pyproject.ensure_list( + project.get("keywords", []), "project.keywords" + ) + or [], + scripts=pyproject.ensure_dict( + project.get("scripts", {}), "project.scripts" + ) + or {}, + gui_scripts=pyproject.ensure_dict( + project.get("gui-scripts", {}), "project.gui-scripts" + ) + or {}, + dynamic=dynamic, + ) + + pyproject.finalize("Failed to parse pyproject.toml") + assert self is not None + return self + + def validate_metdata(self, metadata_version: str) -> None: + errors = ErrorCollector() + + if not self.version: + msg = "Missing {key} field" + errors.config_error(msg, key="project.version") + + if metadata_version not in PRE_SPDX_METADATA_VERSIONS: + if isinstance(self.license, License): + warnings.warn( + 'Set "project.license" to an SPDX license expression for metadata >= 2.4', + ConfigurationWarning, + stacklevel=2, + ) + elif any(c.startswith("License ::") for c in self.classifiers): + warnings.warn( + "'License ::' classifiers are deprecated for metadata >= 2.4" + ', use a SPDX license expression for "project.license" instead', + ConfigurationWarning, + stacklevel=2, + ) + + if ( + isinstance(self.license, str) + and metadata_version in PRE_SPDX_METADATA_VERSIONS + ): + msg = "Setting {key} to an SPDX license expression is supported only when emitting metadata version >= 2.4" + errors.config_error(msg, key="project.license") + + if ( + self.license_files is not None + and metadata_version in PRE_SPDX_METADATA_VERSIONS + ): + msg = "{key} is supported only when emitting metadata version >= 2.4" + errors.config_error(msg, key="project.license-files") + + errors.finalize("Metadata validation failed") + + def validate(self) -> None: + """ + Validate metadata for consistency and correctness. Will also produce + warnings if ``warn`` is given. Respects ``all_errors``. This is called + when loading a pyproject.toml, and when making metadata. Checks: + + - ``metadata_version`` is a known version or None + - ``name`` is a valid project name + - ``license_files`` can't be used with classic ``license`` + - License classifiers can't be used with SPDX license + - ``description`` is a single line (warning) + - ``license`` is not an SPDX license expression if metadata_version >= 2.4 (warning) + - License classifiers deprecated for metadata_version >= 2.4 (warning) + - ``license`` is an SPDX license expression if metadata_version >= 2.4 + - ``license_files`` is supported only for metadata_version >= 2.4 + - ``project_url`` can't contain keys over 32 characters + """ + errors = ErrorCollector() + + try: + utils.canonicalize_name(self.name, validate=True) + except utils.InvalidName: + msg = ( + "Invalid project name {name!r}. A valid name consists only of ASCII letters and " + "numbers, period, underscore and hyphen. It must start and end with a letter or number" + ) + errors.config_error(msg, key="project.name", name=self.name) + + if self.license_files is not None and isinstance(self.license, License): + msg = '{key} must not be used when "project.license" is not a SPDX license expression' + errors.config_error(msg, key="project.license-files") + + if isinstance(self.license, str) and any( + c.startswith("License ::") for c in self.classifiers + ): + msg = "Setting {key} to an SPDX license expression is not compatible with 'License ::' classifiers" + errors.config_error(msg, key="project.license") + + if self.description and "\n" in self.description: + msg = ( + 'The one-line summary "project.description" should not contain more ' + "than one line. Readers might merge or truncate newlines." + ) + errors.config_error(msg, key="project.description") + + for name in self.urls: + if len(name) > 32: + msg = "{key} names cannot be more than 32 characters long" + errors.config_error(msg, key="project.urls", got=name) + + errors.finalize("[project] table validation failed") + + def metadata( + self, metadata_version: str, dynamic_metadata: list[str] + ) -> packaging_metadata.Metadata: + """ + Return an Message with the metadata. + """ + self.validate_metdata(metadata_version) + + assert self.version is not None + message = packaging_metadata.RawMetadata( + metadata_version=metadata_version, name=self.name, version=str(self.version) + ) + + # skip 'Platform' + # skip 'Supported-Platform' + if self.description: + message["summary"] = self.description + if self.keywords: + message["keywords"] = self.keywords + # skip 'Home-page' + # skip 'Download-URL' + if authors := _name_list(self.authors): + message["author"] = authors + + if authors_email := _email_list(self.authors): + message["author_email"] = authors_email + + if maintainers := _name_list(self.maintainers): + message["maintainer"] = maintainers + + if maintainers_email := _email_list(self.maintainers): + message["maintainer_email"] = maintainers_email + + if isinstance(self.license, License): + message["license"] = self.license.text + elif isinstance(self.license, str): + message["license_expression"] = self.license + + if self.license_files is not None: + license_files = [ + os.fspath(license_file.as_posix()) + for license_file in sorted(set(self.license_files)) + ] + message["license_files"] = license_files + elif ( + metadata_version not in PRE_SPDX_METADATA_VERSIONS + and isinstance(self.license, License) + and self.license.file + ): + message["license_files"] = [os.fspath(self.license.file.as_posix())] + + if self.classifiers: + message["classifiers"] = self.classifiers + # skip 'Provides-Dist' + # skip 'Obsoletes-Dist' + # skip 'Requires-External' + if self.urls: + message["project_urls"] = self.urls + if self.requires_python: + message["requires_python"] = str(self.requires_python) + if self.dependencies: + message["requires_dist"] = [str(d) for d in self.dependencies] + for extra, requirements in self.optional_dependencies.items(): + norm_extra = extra.replace(".", "-").replace("_", "-").lower() + message.get("provides_extra", []).append(norm_extra) + message.get("requires_dist", []).extend( + str(_build_extra_req(norm_extra, requirement)) + for requirement in requirements + ) + if self.readme: + if self.readme.content_type: + message["description_content_type"] = self.readme.content_type + message["description"] = self.readme.text + # Core Metadata 2.2 + if metadata_version != "2.1": + for field in dynamic_metadata: + if field.lower() in {"name", "version", "dynamic"}: + msg = f"Field cannot be set as dynamic metadata: {field}" + raise ConfigurationError(msg) + if field.lower() not in packaging_metadata.ALL_FIELDS: + msg = f"Field is not known: {field}" + raise ConfigurationError(msg) + message["dynamic"] = dynamic_metadata + + return packaging_metadata.Metadata.from_raw(message) + + +def _name_list(people: list[tuple[str, str | None]]) -> str | None: + """ + Build a comma-separated list of names. + """ + return ", ".join(name for name, email_ in people if not email_) or None + + +def _email_list(people: list[tuple[str, str | None]]) -> str | None: + """ + Build a comma-separated list of emails. + """ + return ( + ", ".join( + email.utils.formataddr((name, _email)) for name, _email in people if _email + ) + or None + ) + + +def _build_extra_req( + extra: str, + requirement: Requirement, +) -> Requirement: + """ + Build a new requirement with an extra marker. + """ + requirement = copy.copy(requirement) + if requirement.marker: + if "or" in requirement.marker._markers: + requirement.marker = markers.Marker( + f"({requirement.marker}) and extra == {extra!r}" + ) + else: + requirement.marker = markers.Marker( + f"{requirement.marker} and extra == {extra!r}" + ) + else: + requirement.marker = markers.Marker(f"extra == {extra!r}") + return requirement diff --git a/src/packaging/project_table.py b/src/packaging/project_table.py new file mode 100644 index 00000000..1a614578 --- /dev/null +++ b/src/packaging/project_table.py @@ -0,0 +1,142 @@ +# 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. + +from __future__ import annotations + +import sys +import typing +from typing import Any, Dict, List, Literal, TypedDict, Union + +if typing.TYPE_CHECKING: + if sys.version_info < (3, 11): + from typing_extensions import Required + else: + from typing import Required + + +__all__ = [ + "BuildSystemTable", + "ContactTable", + "Dynamic", + "IncludeGroupTable", + "LicenseTable", + "ProjectTable", + "PyProjectTable", + "ReadmeTable", +] + + +def __dir__() -> list[str]: + return __all__ + + +class ContactTable(TypedDict, total=False): + name: str + email: str + + +class LicenseTable(TypedDict, total=False): + text: str + file: str + + +ReadmeTable = TypedDict( + "ReadmeTable", {"file": str, "text": str, "content-type": str}, total=False +) + +Dynamic = Literal[ + "authors", + "classifiers", + "dependencies", + "description", + "dynamic", + "entry-points", + "gui-scripts", + "keywords", + "license", + "maintainers", + "optional-dependencies", + "readme", + "requires-python", + "scripts", + "urls", + "version", +] + +ProjectTable = TypedDict( + "ProjectTable", + { + "name": Required[str], + "version": str, + "description": str, + "license": Union[LicenseTable, str], + "license-files": List[str], + "readme": Union[str, ReadmeTable], + "requires-python": str, + "dependencies": List[str], + "optional-dependencies": Dict[str, List[str]], + "entry-points": Dict[str, Dict[str, str]], + "authors": List[ContactTable], + "maintainers": List[ContactTable], + "urls": Dict[str, str], + "classifiers": List[str], + "keywords": List[str], + "scripts": Dict[str, str], + "gui-scripts": Dict[str, str], + "dynamic": List[Dynamic], + }, + total=False, +) + +BuildSystemTable = TypedDict( + "BuildSystemTable", + { + "build-backend": str, + "requires": List[str], + "backend-path": List[str], + }, + total=False, +) + +# total=False here because this could be +# extended in the future +IncludeGroupTable = TypedDict( + "IncludeGroupTable", + {"include-group": str}, + total=False, +) + +PyProjectTable = TypedDict( + "PyProjectTable", + { + "build-system": BuildSystemTable, + "project": ProjectTable, + "tool": Dict[str, Any], + "dependency-groups": Dict[str, List[Union[str, IncludeGroupTable]]], + }, + total=False, +) + +# Tests for type checking +if typing.TYPE_CHECKING: + PyProjectTable( + { + "build-system": BuildSystemTable( + {"build-backend": "one", "requires": ["two"]} + ), + "project": ProjectTable( + { + "name": "one", + "version": "0.1.0", + } + ), + "tool": {"thing": object()}, + "dependency-groups": { + "one": [ + "one", + IncludeGroupTable({"include-group": "two"}), + ] + }, + } + )