"""For modules related to installing packages.
"""

import errno
import json
import operator
import os
import shutil
import site
from optparse import SUPPRESS_HELP, Values
from typing import List, Optional

from pip._vendor.packaging.utils import canonicalize_name
from pip._vendor.rich import print_json

from pip._internal.cache import WheelCache
from pip._internal.cli import cmdoptions
from pip._internal.cli.cmdoptions import make_target_python
from pip._internal.cli.req_command import (
    RequirementCommand,
    with_cleanup,
)
from pip._internal.cli.status_codes import ERROR, SUCCESS
from pip._internal.exceptions import CommandError, InstallationError
from pip._internal.locations import get_scheme
from pip._internal.metadata import get_environment
from pip._internal.models.installation_report import InstallationReport
from pip._internal.operations.build.build_tracker import get_build_tracker
from pip._internal.operations.check import ConflictDetails, check_install_conflicts
from pip._internal.req import install_given_reqs
from pip._internal.req.req_install import (
    InstallRequirement,
    check_legacy_setup_py_options,
)
from pip._internal.utils.compat import WINDOWS
from pip._internal.utils.filesystem import test_writable_dir
from pip._internal.utils.logging import getLogger
from pip._internal.utils.misc import (
    check_externally_managed,
    ensure_dir,
    get_pip_version,
    protect_pip_from_modification_on_windows,
    warn_if_run_as_root,
    write_output,
)
from pip._internal.utils.temp_dir import TempDirectory
from pip._internal.utils.virtualenv import (
    running_under_virtualenv,
    virtualenv_no_global,
)
from pip._internal.wheel_builder import build, should_build_for_install_command

logger = getLogger(__name__)


class InstallCommand(RequirementCommand):
    """
    Install packages from:

    - PyPI (and other indexes) using requirement specifiers.
    - VCS project urls.
    - Local project directories.
    - Local or remote source archives.

    pip also supports installing from "requirements files", which provide
    an easy way to specify a whole environment to be installed.
    """

    usage = """
      %prog [options] <requirement specifier> [package-index-options] ...
      %prog [options] -r <requirements file> [package-index-options] ...
      %prog [options] [-e] <vcs project url> ...
      %prog [options] [-e] <local project path> ...
      %prog [options] <archive url/path> ..."""

    def add_options(self) -> None:
        self.cmd_opts.add_option(cmdoptions.requirements())
        self.cmd_opts.add_option(cmdoptions.constraints())
        self.cmd_opts.add_option(cmdoptions.no_deps())
        self.cmd_opts.add_option(cmdoptions.pre())

        self.cmd_opts.add_option(cmdoptions.editable())
        self.cmd_opts.add_option(
            "--dry-run",
            action="store_true",
            dest="dry_run",
            default=False,
            help=(
                "Don't actually install anything, just print what would be. "
                "Can be used in combination with --ignore-installed "
                "to 'resolve' the requirements."
            ),
        )
        self.cmd_opts.add_option(
            "-t",
            "--target",
            dest="target_dir",
            metavar="dir",
            default=None,
            help=(
                "Install packages into <dir>. "
                "By default this will not replace existing files/folders in "
                "<dir>. Use --upgrade to replace existing packages in <dir> "
                "with new versions."
            ),
        )
        cmdoptions.add_target_python_options(self.cmd_opts)

        self.cmd_opts.add_option(
            "--user",
            dest="use_user_site",
            action="store_true",
            help=(
                "Install to the Python user install directory for your "
                "platform. Typically ~/.local/, or %APPDATA%\\Python on "
                "Windows. (See the Python documentation for site.USER_BASE "
                "for full details.)"
            ),
        )
        self.cmd_opts.add_option(
            "--no-user",
            dest="use_user_site",
            action="store_false",
            help=SUPPRESS_HELP,
        )
        self.cmd_opts.add_option(
            "--root",
            dest="root_path",
            metavar="dir",
            default=None,
            help="Install everything relative to this alternate root directory.",
        )
        self.cmd_opts.add_option(
            "--prefix",
            dest="prefix_path",
            metavar="dir",
            default=None,
            help=(
                "Installation prefix where lib, bin and other top-level "
                "folders are placed. Note that the resulting installation may "
                "contain scripts and other resources which reference the "
                "Python interpreter of pip, and not that of ``--prefix``. "
                "See also the ``--python`` option if the intention is to "
                "install packages into another (possibly pip-free) "
                "environment."
            ),
        )

        self.cmd_opts.add_option(cmdoptions.src())

        self.cmd_opts.add_option(
            "-U",
            "--upgrade",
            dest="upgrade",
            action="store_true",
            help=(
                "Upgrade all specified packages to the newest available "
                "version. The handling of dependencies depends on the "
                "upgrade-strategy used."
            ),
        )

        self.cmd_opts.add_option(
            "--upgrade-strategy",
            dest="upgrade_strategy",
            default="only-if-needed",
            choices=["only-if-needed", "eager"],
            help=(
                "Determines how dependency upgrading should be handled "
                "[default: %default]. "
                '"eager" - dependencies are upgraded regardless of '
                "whether the currently installed version satisfies the "
                "requirements of the upgraded package(s). "
                '"only-if-needed" -  are upgraded only when they do not '
                "satisfy the requirements of the upgraded package(s)."
            ),
        )

        self.cmd_opts.add_option(
            "--force-reinstall",
            dest="force_reinstall",
            action="store_true",
            help="Reinstall all packages even if they are already up-to-date.",
        )

        self.cmd_opts.add_option(
            "-I",
            "--ignore-installed",
            dest="ignore_installed",
            action="store_true",
            help=(
                "Ignore the installed packages, overwriting them. "
                "This can break your system if the existing package "
                "is of a different version or was installed "
                "with a different package manager!"
            ),
        )

        self.cmd_opts.add_option(cmdoptions.ignore_requires_python())
        self.cmd_opts.add_option(cmdoptions.no_build_isolation())
        self.cmd_opts.add_option(cmdoptions.use_pep517())
        self.cmd_opts.add_option(cmdoptions.no_use_pep517())
        self.cmd_opts.add_option(cmdoptions.check_build_deps())
        self.cmd_opts.add_option(cmdoptions.override_externally_managed())

        self.cmd_opts.add_option(cmdoptions.config_settings())
        self.cmd_opts.add_option(cmdoptions.global_options())

        self.cmd_opts.add_option(
            "--compile",
            action="store_true",
            dest="compile",
            default=True,
            help="Compile Python source files to bytecode",
        )

        self.cmd_opts.add_option(
            "--no-compile",
            action="store_false",
            dest="compile",
            help="Do not compile Python source files to bytecode",
        )

        self.cmd_opts.add_option(
            "--no-warn-script-location",
            action="store_false",
            dest="warn_script_location",
            default=True,
            help="Do not warn when installing scripts outside PATH",
        )
        self.cmd_opts.add_option(
            "--no-warn-conflicts",
            action="store_false",
            dest="warn_about_conflicts",
            default=True,
            help="Do not warn about broken dependencies",
        )
        self.cmd_opts.add_option(cmdoptions.no_binary())
        self.cmd_opts.add_option(cmdoptions.only_binary())
        self.cmd_opts.add_option(cmdoptions.prefer_binary())
        self.cmd_opts.add_option(cmdoptions.require_hashes())
        self.cmd_opts.add_option(cmdoptions.progress_bar())
        self.cmd_opts.add_option(cmdoptions.root_user_action())

        index_opts = cmdoptions.make_option_group(
            cmdoptions.index_group,
            self.parser,
        )

        self.parser.insert_option_group(0, index_opts)
        self.parser.insert_option_group(0, self.cmd_opts)

        self.cmd_opts.add_option(
            "--report",
            dest="json_report_file",
            metavar="file",
            default=None,
            help=(
                "Generate a JSON file describing what pip did to install "
                "the provided requirements. "
                "Can be used in combination with --dry-run and --ignore-installed "
                "to 'resolve' the requirements. "
                "When - is used as file name it writes to stdout. "
                "When writing to stdout, please combine with the --quiet option "
                "to avoid mixing pip logging output with JSON output."
            ),
        )

    @with_cleanup
    def run(self, options: Values, args: List[str]) -> int:
        if options.use_user_site and options.target_dir is not None:
            raise CommandError("Can not combine '--user' and '--target'")

        # Check whether the environment we're installing into is externally
        # managed, as specified in PEP 668. Specifying --root, --target, or
        # --prefix disables the check, since there's no reliable way to locate
        # the EXTERNALLY-MANAGED file for those cases. An exception is also
        # made specifically for "--dry-run --report" for convenience.
        installing_into_current_environment = (
            not (options.dry_run and options.json_report_file)
            and options.root_path is None
            and options.target_dir is None
            and options.prefix_path is None
        )
        if (
            installing_into_current_environment
            and not options.override_externally_managed
        ):
            check_externally_managed()

        upgrade_strategy = "to-satisfy-only"
        if options.upgrade:
            upgrade_strategy = options.upgrade_strategy

        cmdoptions.check_dist_restriction(options, check_target=True)

        logger.verbose("Using %s", get_pip_version())
        options.use_user_site = decide_user_install(
            options.use_user_site,
            prefix_path=options.prefix_path,
            target_dir=options.target_dir,
            root_path=options.root_path,
            isolated_mode=options.isolated_mode,
        )

        target_temp_dir: Optional[TempDirectory] = None
        target_temp_dir_path: Optional[str] = None
        if options.target_dir:
            options.ignore_installed = True
            options.target_dir = os.path.abspath(options.target_dir)
            if (
                # fmt: off
                os.path.exists(options.target_dir) and
                not os.path.isdir(options.target_dir)
                # fmt: on
            ):
                raise CommandError(
                    "Target path exists but is not a directory, will not continue."
                )

            # Create a target directory for using with the target option
            target_temp_dir = TempDirectory(kind="target")
            target_temp_dir_path = target_temp_dir.path
            self.enter_context(target_temp_dir)

        global_options = options.global_options or []

        session = self.get_default_session(options)

        target_python = make_target_python(options)
        finder = self._build_package_finder(
            options=options,
            session=session,
            target_python=target_python,
            ignore_requires_python=options.ignore_requires_python,
        )
        build_tracker = self.enter_context(get_build_tracker())

        directory = TempDirectory(
            delete=not options.no_clean,
            kind="install",
            globally_managed=True,
        )

        try:
            reqs = self.get_requirements(args, options, finder, session)
            check_legacy_setup_py_options(options, reqs)

            wheel_cache = WheelCache(options.cache_dir)

            # Only when installing is it permitted to use PEP 660.
            # In other circumstances (pip wheel, pip download) we generate
            # regular (i.e. non editable) metadata and wheels.
            for req in reqs:
                req.permit_editable_wheels = True

            preparer = self.make_requirement_preparer(
                temp_build_dir=directory,
                options=options,
                build_tracker=build_tracker,
                session=session,
                finder=finder,
                use_user_site=options.use_user_site,
                verbosity=self.verbosity,
            )
            resolver = self.make_resolver(
                preparer=preparer,
                finder=finder,
                options=options,
                wheel_cache=wheel_cache,
                use_user_site=options.use_user_site,
                ignore_installed=options.ignore_installed,
                ignore_requires_python=options.ignore_requires_python,
                force_reinstall=options.force_reinstall,
                upgrade_strategy=upgrade_strategy,
                use_pep517=options.use_pep517,
                py_version_info=options.python_version,
            )

            self.trace_basic_info(finder)

            requirement_set = resolver.resolve(
                reqs, check_supported_wheels=not options.target_dir
            )

            if options.json_report_file:
                report = InstallationReport(requirement_set.requirements_to_install)
                if options.json_report_file == "-":
                    print_json(data=report.to_dict())
                else:
                    with open(options.json_report_file, "w", encoding="utf-8") as f:
                        json.dump(report.to_dict(), f, indent=2, ensure_ascii=False)

            if options.dry_run:
                would_install_items = sorted(
                    (r.metadata["name"], r.metadata["version"])
                    for r in requirement_set.requirements_to_install
                )
                if would_install_items:
                    write_output(
                        "Would install %s",
                        " ".join("-".join(item) for item in would_install_items),
                    )
                return SUCCESS

            try:
                pip_req = requirement_set.get_requirement("pip")
            except KeyError:
                modifying_pip = False
            else:
                # If we're not replacing an already installed pip,
                # we're not modifying it.
                modifying_pip = pip_req.satisfied_by is None
                if modifying_pip:
                    # Eagerly import this module to avoid crashes. Otherwise, this
                    # module would be imported *after* pip was replaced, resulting in
                    # crashes if the new self_outdated_check module was incompatible
                    # with the rest of pip that's already imported.
                    import pip._internal.self_outdated_check  # noqa: F401
            protect_pip_from_modification_on_windows(modifying_pip=modifying_pip)

            reqs_to_build = [
                r
                for r in requirement_set.requirements.values()
                if should_build_for_install_command(r)
            ]

            _, build_failures = build(
                reqs_to_build,
                wheel_cache=wheel_cache,
                verify=True,
                build_options=[],
                global_options=global_options,
            )

            if build_failures:
                raise InstallationError(
                    "ERROR: Failed to build installable wheels for some "
                    "pyproject.toml based projects ({})".format(
                        ", ".join(r.name for r in build_failures)  # type: ignore
                    )
                )

            to_install = resolver.get_installation_order(requirement_set)

            # Check for conflicts in the package set we're installing.
            conflicts: Optional[ConflictDetails] = None
            should_warn_about_conflicts = (
                not options.ignore_dependencies and options.warn_about_conflicts
            )
            if should_warn_about_conflicts:
                conflicts = self._determine_conflicts(to_install)

            # Don't warn about script install locations if
            # --target or --prefix has been specified
            warn_script_location = options.warn_script_location
            if options.target_dir or options.prefix_path:
                warn_script_location = False

            installed = install_given_reqs(
                to_install,
                global_options,
                root=options.root_path,
                home=target_temp_dir_path,
                prefix=options.prefix_path,
                warn_script_location=warn_script_location,
                use_user_site=options.use_user_site,
                pycompile=options.compile,
            )

            lib_locations = get_lib_location_guesses(
                user=options.use_user_site,
                home=target_temp_dir_path,
                root=options.root_path,
                prefix=options.prefix_path,
                isolated=options.isolated_mode,
            )
            env = get_environment(lib_locations)

            # Display a summary of installed packages, with extra care to
            # display a package name as it was requested by the user.
            installed.sort(key=operator.attrgetter("name"))
            summary = []
            installed_versions = {}
            for distribution in env.iter_all_distributions():
                installed_versions[distribution.canonical_name] = distribution.version
            for package in installed:
                display_name = package.name
                version = installed_versions.get(canonicalize_name(display_name), None)
                if version:
                    text = f"{display_name}-{version}"
                else:
                    text = display_name
                summary.append(text)

            if conflicts is not None:
                self._warn_about_conflicts(
                    conflicts,
                    resolver_variant=self.determine_resolver_variant(options),
                )

            installed_desc = " ".join(summary)
            if installed_desc:
                write_output(
                    "Successfully installed %s",
                    installed_desc,
                )
        except OSError as error:
            show_traceback = self.verbosity >= 1

            message = create_os_error_message(
                error,
                show_traceback,
                options.use_user_site,
            )
            logger.error(message, exc_info=show_traceback)

            return ERROR

        if options.target_dir:
            assert target_temp_dir
            self._handle_target_dir(
                options.target_dir, target_temp_dir, options.upgrade
            )
        if options.root_user_action == "warn":
            warn_if_run_as_root()
        return SUCCESS

    def _handle_target_dir(
        self, target_dir: str, target_temp_dir: TempDirectory, upgrade: bool
    ) -> None:
        ensure_dir(target_dir)

        # Checking both purelib and platlib directories for installed
        # packages to be moved to target directory
        lib_dir_list = []

        # Checking both purelib and platlib directories for installed
        # packages to be moved to target directory
        scheme = get_scheme("", home=target_temp_dir.path)
        purelib_dir = scheme.purelib
        platlib_dir = scheme.platlib
        data_dir = scheme.data

        if os.path.exists(purelib_dir):
            lib_dir_list.append(purelib_dir)
        if os.path.exists(platlib_dir) and platlib_dir != purelib_dir:
            lib_dir_list.append(platlib_dir)
        if os.path.exists(data_dir):
            lib_dir_list.append(data_dir)

        for lib_dir in lib_dir_list:
            for item in os.listdir(lib_dir):
                if lib_dir == data_dir:
                    ddir = os.path.join(data_dir, item)
                    if any(s.startswith(ddir) for s in lib_dir_list[:-1]):
                        continue
                target_item_dir = os.path.join(target_dir, item)
                if os.path.exists(target_item_dir):
                    if not upgrade:
                        logger.warning(
                            "Target directory %s already exists. Specify "
                            "--upgrade to force replacement.",
                            target_item_dir,
                        )
                        continue
                    if os.path.islink(target_item_dir):
                        logger.warning(
                            "Target directory %s already exists and is "
                            "a link. pip will not automatically replace "
                            "links, please remove if replacement is "
                            "desired.",
                            target_item_dir,
                        )
                        continue
                    if os.path.isdir(target_item_dir):
                        shutil.rmtree(target_item_dir)
                    else:
                        os.remove(target_item_dir)

                shutil.move(os.path.join(lib_dir, item), target_item_dir)

    def _determine_conflicts(
        self, to_install: List[InstallRequirement]
    ) -> Optional[ConflictDetails]:
        try:
            return check_install_conflicts(to_install)
        except Exception:
            logger.exception(
                "Error while checking for conflicts. Please file an issue on "
                "pip's issue tracker: https://github.com/pypa/pip/issues/new"
            )
            return None

    def _warn_about_conflicts(
        self, conflict_details: ConflictDetails, resolver_variant: str
    ) -> None:
        package_set, (missing, conflicting) = conflict_details
        if not missing and not conflicting:
            return

        parts: List[str] = []
        if resolver_variant == "legacy":
            parts.append(
                "pip's legacy dependency resolver does not consider dependency "
                "conflicts when selecting packages. This behaviour is the "
                "source of the following dependency conflicts."
            )
        else:
            assert resolver_variant == "resolvelib"
            parts.append(
                "pip's dependency resolver does not currently take into account "
                "all the packages that are installed. This behaviour is the "
                "source of the following dependency conflicts."
            )

        # NOTE: There is some duplication here, with commands/check.py
        for project_name in missing:
            version = package_set[project_name][0]
            for dependency in missing[project_name]:
                message = (
                    f"{project_name} {version} requires {dependency[1]}, "
                    "which is not installed."
                )
                parts.append(message)

        for project_name in conflicting:
            version = package_set[project_name][0]
            for dep_name, dep_version, req in conflicting[project_name]:
                message = (
                    "{name} {version} requires {requirement}, but {you} have "
                    "{dep_name} {dep_version} which is incompatible."
                ).format(
                    name=project_name,
                    version=version,
                    requirement=req,
                    dep_name=dep_name,
                    dep_version=dep_version,
                    you=("you" if resolver_variant == "resolvelib" else "you'll"),
                )
                parts.append(message)

        logger.critical("\n".join(parts))


def get_lib_location_guesses(
    user: bool = False,
    home: Optional[str] = None,
    root: Optional[str] = None,
    isolated: bool = False,
    prefix: Optional[str] = None,
) -> List[str]:
    scheme = get_scheme(
        "",
        user=user,
        home=home,
        root=root,
        isolated=isolated,
        prefix=prefix,
    )
    return [scheme.purelib, scheme.platlib]


def site_packages_writable(root: Optional[str], isolated: bool) -> bool:
    return all(
        test_writable_dir(d)
        for d in set(get_lib_location_guesses(root=root, isolated=isolated))
    )


def decide_user_install(
    use_user_site: Optional[bool],
    prefix_path: Optional[str] = None,
    target_dir: Optional[str] = None,
    root_path: Optional[str] = None,
    isolated_mode: bool = False,
) -> bool:
    """Determine whether to do a user install based on the input options.

    If use_user_site is False, no additional checks are done.
    If use_user_site is True, it is checked for compatibility with other
    options.
    If use_user_site is None, the default behaviour depends on the environment,
    which is provided by the other arguments.
    """
    # In some cases (config from tox), use_user_site can be set to an integer
    # rather than a bool, which 'use_user_site is False' wouldn't catch.
    if (use_user_site is not None) and (not use_user_site):
        logger.debug("Non-user install by explicit request")
        return False

    if use_user_site:
        if prefix_path:
            raise CommandError(
                "Can not combine '--user' and '--prefix' as they imply "
                "different installation locations"
            )
        if virtualenv_no_global():
            raise InstallationError(
                "Can not perform a '--user' install. User site-packages "
                "are not visible in this virtualenv."
            )
        logger.debug("User install by explicit request")
        return True

    # If we are here, user installs have not been explicitly requested/avoided
    assert use_user_site is None

    # user install incompatible with --prefix/--target
    if prefix_path or target_dir:
        logger.debug("Non-user install due to --prefix or --target option")
        return False

    # If user installs are not enabled, choose a non-user install
    if not site.ENABLE_USER_SITE:
        logger.debug("Non-user install because user site-packages disabled")
        return False

    # If we have permission for a non-user install, do that,
    # otherwise do a user install.
    if site_packages_writable(root=root_path, isolated=isolated_mode):
        logger.debug("Non-user install because site-packages writeable")
        return False

    logger.info(
        "Defaulting to user installation because normal site-packages "
        "is not writeable"
    )
    return True


def create_os_error_message(
    error: OSError, show_traceback: bool, using_user_site: bool
) -> str:
    """Format an error message for an OSError

    It may occur anytime during the execution of the install command.
    """
    parts = []

    # Mention the error if we are not going to show a traceback
    parts.append("Could not install packages due to an OSError")
    if not show_traceback:
        parts.append(": ")
        parts.append(str(error))
    else:
        parts.append(".")

    # Spilt the error indication from a helper message (if any)
    parts[-1] += "\n"

    # Suggest useful actions to the user:
    #  (1) using user site-packages or (2) verifying the permissions
    if error.errno == errno.EACCES:
        user_option_part = "Consider using the `--user` option"
        permissions_part = "Check the permissions"

        if not running_under_virtualenv() and not using_user_site:
            parts.extend(
                [
                    user_option_part,
                    " or ",
                    permissions_part.lower(),
                ]
            )
        else:
            parts.append(permissions_part)
        parts.append(".\n")

    # Suggest the user to enable Long Paths if path length is
    # more than 260
    if (
        WINDOWS
        and error.errno == errno.ENOENT
        and error.filename
        and len(error.filename) > 260
    ):
        parts.append(
            "HINT: This error might have occurred since "
            "this system does not have Windows Long Path "
            "support enabled. You can find information on "
            "how to enable this at "
            "https://pip.pypa.io/warnings/enable-long-paths\n"
        )

    return "".join(parts).strip() + "\n"

"""
    pygments.util
    ~~~~~~~~~~~~~

    Utility functions.

    :copyright: Copyright 2006-2024 by the Pygments team, see AUTHORS.
    :license: BSD, see LICENSE for details.
"""

import re
from io import TextIOWrapper


split_path_re = re.compile(r'[/\\ ]')
doctype_lookup_re = re.compile(r'''
    <!DOCTYPE\s+(
     [a-zA-Z_][a-zA-Z0-9]*
     (?: \s+      # optional in HTML5
     [a-zA-Z_][a-zA-Z0-9]*\s+
     "[^"]*")?
     )
     [^>]*>
''', re.DOTALL | re.MULTILINE | re.VERBOSE)
tag_re = re.compile(r'<(.+?)(\s.*?)?>.*?</.+?>',
                    re.IGNORECASE | re.DOTALL | re.MULTILINE)
xml_decl_re = re.compile(r'\s*<\?xml[^>]*\?>', re.I)


class ClassNotFound(ValueError):
    """Raised if one of the lookup functions didn't find a matching class."""


class OptionError(Exception):
    """
    This exception will be raised by all option processing functions if
    the type or value of the argument is not correct.
    """

def get_choice_opt(options, optname, allowed, default=None, normcase=False):
    """
    If the key `optname` from the dictionary is not in the sequence
    `allowed`, raise an error, otherwise return it.
    """
    string = options.get(optname, default)
    if normcase:
        string = string.lower()
    if string not in allowed:
        raise OptionError('Value for option {} must be one of {}'.format(optname, ', '.join(map(str, allowed))))
    return string


def get_bool_opt(options, optname, default=None):
    """
    Intuitively, this is `options.get(optname, default)`, but restricted to
    Boolean value. The Booleans can be represented as string, in order to accept
    Boolean value from the command line arguments. If the key `optname` is
    present in the dictionary `options` and is not associated with a Boolean,
    raise an `OptionError`. If it is absent, `default` is returned instead.

    The valid string values for ``True`` are ``1``, ``yes``, ``true`` and
    ``on``, the ones for ``False`` are ``0``, ``no``, ``false`` and ``off``
    (matched case-insensitively).
    """
    string = options.get(optname, default)
    if isinstance(string, bool):
        return string
    elif isinstance(string, int):
        return bool(string)
    elif not isinstance(string, str):
        raise OptionError(f'Invalid type {string!r} for option {optname}; use '
                          '1/0, yes/no, true/false, on/off')
    elif string.lower() in ('1', 'yes', 'true', 'on'):
        return True
    elif string.lower() in ('0', 'no', 'false', 'off'):
        return False
    else:
        raise OptionError(f'Invalid value {string!r} for option {optname}; use '
                          '1/0, yes/no, true/false, on/off')


def get_int_opt(options, optname, default=None):
    """As :func:`get_bool_opt`, but interpret the value as an integer."""
    string = options.get(optname, default)
    try:
        return int(string)
    except TypeError:
        raise OptionError(f'Invalid type {string!r} for option {optname}; you '
                          'must give an integer value')
    except ValueError:
        raise OptionError(f'Invalid value {string!r} for option {optname}; you '
                          'must give an integer value')

def get_list_opt(options, optname, default=None):
    """
    If the key `optname` from the dictionary `options` is a string,
    split it at whitespace and return it. If it is already a list
    or a tuple, it is returned as a list.
    """
    val = options.get(optname, default)
    if isinstance(val, str):
        return val.split()
    elif isinstance(val, (list, tuple)):
        return list(val)
    else:
        raise OptionError(f'Invalid type {val!r} for option {optname}; you '
                          'must give a list value')


def docstring_headline(obj):
    if not obj.__doc__:
        return ''
    res = []
    for line in obj.__doc__.strip().splitlines():
        if line.strip():
            res.append(" " + line.strip())
        else:
            break
    return ''.join(res).lstrip()


def make_analysator(f):
    """Return a static text analyser function that returns float values."""
    def text_analyse(text):
        try:
            rv = f(text)
        except Exception:
            return 0.0
        if not rv:
            return 0.0
        try:
            return min(1.0, max(0.0, float(rv)))
        except (ValueError, TypeError):
            return 0.0
    text_analyse.__doc__ = f.__doc__
    return staticmethod(text_analyse)


def shebang_matches(text, regex):
    r"""Check if the given regular expression matches the last part of the
    shebang if one exists.

        >>> from pygments.util import shebang_matches
        >>> shebang_matches('#!/usr/bin/env python', r'python(2\.\d)?')
        True
        >>> shebang_matches('#!/usr/bin/python2.4', r'python(2\.\d)?')
        True
        >>> shebang_matches('#!/usr/bin/python-ruby', r'python(2\.\d)?')
        False
        >>> shebang_matches('#!/usr/bin/python/ruby', r'python(2\.\d)?')
        False
        >>> shebang_matches('#!/usr/bin/startsomethingwith python',
        ...                 r'python(2\.\d)?')
        True

    It also checks for common windows executable file extensions::

        >>> shebang_matches('#!C:\\Python2.4\\Python.exe', r'python(2\.\d)?')
        True

    Parameters (``'-f'`` or ``'--foo'`` are ignored so ``'perl'`` does
    the same as ``'perl -e'``)

    Note that this method automatically searches the whole string (eg:
    the regular expression is wrapped in ``'^$'``)
    """
    index = text.find('\n')
    if index >= 0:
        first_line = text[:index].lower()
    else:
        first_line = text.lower()
    if first_line.startswith('#!'):
        try:
            found = [x for x in split_path_re.split(first_line[2:].strip())
                     if x and not x.startswith('-')][-1]
        except IndexError:
            return False
        regex = re.compile(rf'^{regex}(\.(exe|cmd|bat|bin))?$', re.IGNORECASE)
        if regex.search(found) is not None:
            return True
    return False


def doctype_matches(text, regex):
    """Check if the doctype matches a regular expression (if present).

    Note that this method only checks the first part of a DOCTYPE.
    eg: 'html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"'
    """
    m = doctype_lookup_re.search(text)
    if m is None:
        return False
    doctype = m.group(1)
    return re.compile(regex, re.I).match(doctype.strip()) is not None


def html_doctype_matches(text):
    """Check if the file looks like it has a html doctype."""
    return doctype_matches(text, r'html')


_looks_like_xml_cache = {}


def looks_like_xml(text):
    """Check if a doctype exists or if we have some tags."""
    if xml_decl_re.match(text):
        return True
    key = hash(text)
    try:
        return _looks_like_xml_cache[key]
    except KeyError:
        m = doctype_lookup_re.search(text)
        if m is not None:
            return True
        rv = tag_re.search(text[:1000]) is not None
        _looks_like_xml_cache[key] = rv
        return rv


def surrogatepair(c):
    """Given a unicode character code with length greater than 16 bits,
    return the two 16 bit surrogate pair.
    """
    # From example D28 of:
    # http://www.unicode.org/book/ch03.pdf
    return (0xd7c0 + (c >> 10), (0xdc00 + (c & 0x3ff)))


def format_lines(var_name, seq, raw=False, indent_level=0):
    """Formats a sequence of strings for output."""
    lines = []
    base_indent = ' ' * indent_level * 4
    inner_indent = ' ' * (indent_level + 1) * 4
    lines.append(base_indent + var_name + ' = (')
    if raw:
        # These should be preformatted reprs of, say, tuples.
        for i in seq:
            lines.append(inner_indent + i + ',')
    else:
        for i in seq:
            # Force use of single quotes
            r = repr(i + '"')
            lines.append(inner_indent + r[:-2] + r[-1] + ',')
    lines.append(base_indent + ')')
    return '\n'.join(lines)


def duplicates_removed(it, already_seen=()):
    """
    Returns a list with duplicates removed from the iterable `it`.

    Order is preserved.
    """
    lst = []
    seen = set()
    for i in it:
        if i in seen or i in already_seen:
            continue
        lst.append(i)
        seen.add(i)
    return lst


class Future:
    """Generic class to defer some work.

    Handled specially in RegexLexerMeta, to support regex string construction at
    first use.
    """
    def get(self):
        raise NotImplementedError


def guess_decode(text):
    """Decode *text* with guessed encoding.

    First try UTF-8; this should fail for non-UTF-8 encodings.
    Then try the preferred locale encoding.
    Fall back to latin-1, which always works.
    """
    try:
        text = text.decode('utf-8')
        return text, 'utf-8'
    except UnicodeDecodeError:
        try:
            import locale
            prefencoding = locale.getpreferredencoding()
            text = text.decode()
            return text, prefencoding
        except (UnicodeDecodeError, LookupError):
            text = text.decode('latin1')
            return text, 'latin1'


def guess_decode_from_terminal(text, term):
    """Decode *text* coming from terminal *term*.

    First try the terminal encoding, if given.
    Then try UTF-8.  Then try the preferred locale encoding.
    Fall back to latin-1, which always works.
    """
    if getattr(term, 'encoding', None):
        try:
            text = text.decode(term.encoding)
        except UnicodeDecodeError:
            pass
        else:
            return text, term.encoding
    return guess_decode(text)


def terminal_encoding(term):
    """Return our best guess of encoding for the given *term*."""
    if getattr(term, 'encoding', None):
        return term.encoding
    import locale
    return locale.getpreferredencoding()


class UnclosingTextIOWrapper(TextIOWrapper):
    # Don't close underlying buffer on destruction.
    def close(self):
        self.flush()

"""
For types associated with installation schemes.

For a general overview of available schemes and their context, see
https://docs.python.org/3/install/index.html#alternate-installation.
"""

from dataclasses import dataclass

SCHEME_KEYS = ["platlib", "purelib", "headers", "scripts", "data"]


@dataclass(frozen=True)
class Scheme:
    """A Scheme holds paths which are used as the base directories for
    artifacts associated with a Python package.
    """

    __slots__ = SCHEME_KEYS

    platlib: str
    purelib: str
    headers: str
    scripts: str
    data: str

"""Network Authentication Helpers

Contains interface (MultiDomainBasicAuth) and associated glue code for
providing credentials in the context of network requests.
"""

import logging
import os
import shutil
import subprocess
import sysconfig
import typing
import urllib.parse
from abc import ABC, abstractmethod
from functools import lru_cache
from os.path import commonprefix
from pathlib import Path
from typing import Any, Dict, List, NamedTuple, Optional, Tuple

from pip._vendor.requests.auth import AuthBase, HTTPBasicAuth
from pip._vendor.requests.models import Request, Response
from pip._vendor.requests.utils import get_netrc_auth

from pip._internal.utils.logging import getLogger
from pip._internal.utils.misc import (
    ask,
    ask_input,
    ask_password,
    remove_auth_from_url,
    split_auth_netloc_from_url,
)
from pip._internal.vcs.versioncontrol import AuthInfo

logger = getLogger(__name__)

KEYRING_DISABLED = False


class Credentials(NamedTuple):
    url: str
    username: str
    password: str


class KeyRingBaseProvider(ABC):
    """Keyring base provider interface"""

    has_keyring: bool

    @abstractmethod
    def get_auth_info(
        self, url: str, username: Optional[str]
    ) -> Optional[AuthInfo]: ...

    @abstractmethod
    def save_auth_info(self, url: str, username: str, password: str) -> None: ...


class KeyRingNullProvider(KeyRingBaseProvider):
    """Keyring null provider"""

    has_keyring = False

    def get_auth_info(self, url: str, username: Optional[str]) -> Optional[AuthInfo]:
        return None

    def save_auth_info(self, url: str, username: str, password: str) -> None:
        return None


class KeyRingPythonProvider(KeyRingBaseProvider):
    """Keyring interface which uses locally imported `keyring`"""

    has_keyring = True

    def __init__(self) -> None:
        import keyring

        self.keyring = keyring

    def get_auth_info(self, url: str, username: Optional[str]) -> Optional[AuthInfo]:
        # Support keyring's get_credential interface which supports getting
        # credentials without a username. This is only available for
        # keyring>=15.2.0.
        if hasattr(self.keyring, "get_credential"):
            logger.debug("Getting credentials from keyring for %s", url)
            cred = self.keyring.get_credential(url, username)
            if cred is not None:
                return cred.username, cred.password
            return None

        if username is not None:
            logger.debug("Getting password from keyring for %s", url)
            password = self.keyring.get_password(url, username)
            if password:
                return username, password
        return None

    def save_auth_info(self, url: str, username: str, password: str) -> None:
        self.keyring.set_password(url, username, password)


class KeyRingCliProvider(KeyRingBaseProvider):
    """Provider which uses `keyring` cli

    Instead of calling the keyring package installed alongside pip
    we call keyring on the command line which will enable pip to
    use which ever installation of keyring is available first in
    PATH.
    """

    has_keyring = True

    def __init__(self, cmd: str) -> None:
        self.keyring = cmd

    def get_auth_info(self, url: str, username: Optional[str]) -> Optional[AuthInfo]:
        # This is the default implementation of keyring.get_credential
        # https://github.com/jaraco/keyring/blob/97689324abcf01bd1793d49063e7ca01e03d7d07/keyring/backend.py#L134-L139
        if username is not None:
            password = self._get_password(url, username)
            if password is not None:
                return username, password
        return None

    def save_auth_info(self, url: str, username: str, password: str) -> None:
        return self._set_password(url, username, password)

    def _get_password(self, service_name: str, username: str) -> Optional[str]:
        """Mirror the implementation of keyring.get_password using cli"""
        if self.keyring is None:
            return None

        cmd = [self.keyring, "get", service_name, username]
        env = os.environ.copy()
        env["PYTHONIOENCODING"] = "utf-8"
        res = subprocess.run(
            cmd,
            stdin=subprocess.DEVNULL,
            stdout=subprocess.PIPE,
            env=env,
        )
        if res.returncode:
            return None
        return res.stdout.decode("utf-8").strip(os.linesep)

    def _set_password(self, service_name: str, username: str, password: str) -> None:
        """Mirror the implementation of keyring.set_password using cli"""
        if self.keyring is None:
            return None
        env = os.environ.copy()
        env["PYTHONIOENCODING"] = "utf-8"
        subprocess.run(
            [self.keyring, "set", service_name, username],
            input=f"{password}{os.linesep}".encode(),
            env=env,
            check=True,
        )
        return None


@lru_cache(maxsize=None)
def get_keyring_provider(provider: str) -> KeyRingBaseProvider:
    logger.verbose("Keyring provider requested: %s", provider)

    # keyring has previously failed and been disabled
    if KEYRING_DISABLED:
        provider = "disabled"
    if provider in ["import", "auto"]:
        try:
            impl = KeyRingPythonProvider()
            logger.verbose("Keyring provider set: import")
            return impl
        except ImportError:
            pass
        except Exception as exc:
            # In the event of an unexpected exception
            # we should warn the user
            msg = "Installed copy of keyring fails with exception %s"
            if provider == "auto":
                msg = msg + ", trying to find a keyring executable as a fallback"
            logger.warning(msg, exc, exc_info=logger.isEnabledFor(logging.DEBUG))
    if provider in ["subprocess", "auto"]:
        cli = shutil.which("keyring")
        if cli and cli.startswith(sysconfig.get_path("scripts")):
            # all code within this function is stolen from shutil.which implementation
            @typing.no_type_check
            def PATH_as_shutil_which_determines_it() -> str:
                path = os.environ.get("PATH", None)
                if path is None:
                    try:
                        path = os.confstr("CS_PATH")
                    except (AttributeError, ValueError):
                        # os.confstr() or CS_PATH is not available
                        path = os.defpath
                # bpo-35755: Don't use os.defpath if the PATH environment variable is
                # set to an empty string

                return path

            scripts = Path(sysconfig.get_path("scripts"))

            paths = []
            for path in PATH_as_shutil_which_determines_it().split(os.pathsep):
                p = Path(path)
                try:
                    if not p.samefile(scripts):
                        paths.append(path)
                except FileNotFoundError:
                    pass

            path = os.pathsep.join(paths)

            cli = shutil.which("keyring", path=path)

        if cli:
            logger.verbose("Keyring provider set: subprocess with executable %s", cli)
            return KeyRingCliProvider(cli)

    logger.verbose("Keyring provider set: disabled")
    return KeyRingNullProvider()


class MultiDomainBasicAuth(AuthBase):
    def __init__(
        self,
        prompting: bool = True,
        index_urls: Optional[List[str]] = None,
        keyring_provider: str = "auto",
    ) -> None:
        self.prompting = prompting
        self.index_urls = index_urls
        self.keyring_provider = keyring_provider  # type: ignore[assignment]
        self.passwords: Dict[str, AuthInfo] = {}
        # When the user is prompted to enter credentials and keyring is
        # available, we will offer to save them. If the user accepts,
        # this value is set to the credentials they entered. After the
        # request authenticates, the caller should call
        # ``save_credentials`` to save these.
        self._credentials_to_save: Optional[Credentials] = None

    @property
    def keyring_provider(self) -> KeyRingBaseProvider:
        return get_keyring_provider(self._keyring_provider)

    @keyring_provider.setter
    def keyring_provider(self, provider: str) -> None:
        # The free function get_keyring_provider has been decorated with
        # functools.cache. If an exception occurs in get_keyring_auth that
        # cache will be cleared and keyring disabled, take that into account
        # if you want to remove this indirection.
        self._keyring_provider = provider

    @property
    def use_keyring(self) -> bool:
        # We won't use keyring when --no-input is passed unless
        # a specific provider is requested because it might require
        # user interaction
        return self.prompting or self._keyring_provider not in ["auto", "disabled"]

    def _get_keyring_auth(
        self,
        url: Optional[str],
        username: Optional[str],
    ) -> Optional[AuthInfo]:
        """Return the tuple auth for a given url from keyring."""
        # Do nothing if no url was provided
        if not url:
            return None

        try:
            return self.keyring_provider.get_auth_info(url, username)
        except Exception as exc:
            # Log the full exception (with stacktrace) at debug, so it'll only
            # show up when running in verbose mode.
            logger.debug("Keyring is skipped due to an exception", exc_info=True)
            # Always log a shortened version of the exception.
            logger.warning(
                "Keyring is skipped due to an exception: %s",
                str(exc),
            )
            global KEYRING_DISABLED
            KEYRING_DISABLED = True
            get_keyring_provider.cache_clear()
            return None

    def _get_index_url(self, url: str) -> Optional[str]:
        """Return the original index URL matching the requested URL.

        Cached or dynamically generated credentials may work against
        the original index URL rather than just the netloc.

        The provided url should have had its username and password
        removed already. If the original index url had credentials then
        they will be included in the return value.

        Returns None if no matching index was found, or if --no-index
        was specified by the user.
        """
        if not url or not self.index_urls:
            return None

        url = remove_auth_from_url(url).rstrip("/") + "/"
        parsed_url = urllib.parse.urlsplit(url)

        candidates = []

        for index in self.index_urls:
            index = index.rstrip("/") + "/"
            parsed_index = urllib.parse.urlsplit(remove_auth_from_url(index))
            if parsed_url == parsed_index:
                return index

            if parsed_url.netloc != parsed_index.netloc:
                continue

            candidate = urllib.parse.urlsplit(index)
            candidates.append(candidate)

        if not candidates:
            return None

        candidates.sort(
            reverse=True,
            key=lambda candidate: commonprefix(
                [
                    parsed_url.path,
                    candidate.path,
                ]
            ).rfind("/"),
        )

        return urllib.parse.urlunsplit(candidates[0])

    def _get_new_credentials(
        self,
        original_url: str,
        *,
        allow_netrc: bool = True,
        allow_keyring: bool = False,
    ) -> AuthInfo:
        """Find and return credentials for the specified URL."""
        # Split the credentials and netloc from the url.
        url, netloc, url_user_password = split_auth_netloc_from_url(
            original_url,
        )

        # Start with the credentials embedded in the url
        username, password = url_user_password
        if username is not None and password is not None:
            logger.debug("Found credentials in url for %s", netloc)
            return url_user_password

        # Find a matching index url for this request
        index_url = self._get_index_url(url)
        if index_url:
            # Split the credentials from the url.
            index_info = split_auth_netloc_from_url(index_url)
            if index_info:
                index_url, _, index_url_user_password = index_info
                logger.debug("Found index url %s", index_url)

        # If an index URL was found, try its embedded credentials
        if index_url and index_url_user_password[0] is not None:
            username, password = index_url_user_password
            if username is not None and password is not None:
                logger.debug("Found credentials in index url for %s", netloc)
                return index_url_user_password

        # Get creds from netrc if we still don't have them
        if allow_netrc:
            netrc_auth = get_netrc_auth(original_url)
            if netrc_auth:
                logger.debug("Found credentials in netrc for %s", netloc)
                return netrc_auth

        # If we don't have a password and keyring is available, use it.
        if allow_keyring:
            # The index url is more specific than the netloc, so try it first
            # fmt: off
            kr_auth = (
                self._get_keyring_auth(index_url, username) or
                self._get_keyring_auth(netloc, username)
            )
            # fmt: on
            if kr_auth:
                logger.debug("Found credentials in keyring for %s", netloc)
                return kr_auth

        return username, password

    def _get_url_and_credentials(
        self, original_url: str
    ) -> Tuple[str, Optional[str], Optional[str]]:
        """Return the credentials to use for the provided URL.

        If allowed, netrc and keyring may be used to obtain the
        correct credentials.

        Returns (url_without_credentials, username, password). Note
        that even if the original URL contains credentials, this
        function may return a different username and password.
        """
        url, netloc, _ = split_auth_netloc_from_url(original_url)

        # Try to get credentials from original url
        username, password = self._get_new_credentials(original_url)

        # If credentials not found, use any stored credentials for this netloc.
        # Do this if either the username or the password is missing.
        # This accounts for the situation in which the user has specified
        # the username in the index url, but the password comes from keyring.
        if (username is None or password is None) and netloc in self.passwords:
            un, pw = self.passwords[netloc]
            # It is possible that the cached credentials are for a different username,
            # in which case the cache should be ignored.
            if username is None or username == un:
                username, password = un, pw

        if username is not None or password is not None:
            # Convert the username and password if they're None, so that
            # this netloc will show up as "cached" in the conditional above.
            # Further, HTTPBasicAuth doesn't accept None, so it makes sense to
            # cache the value that is going to be used.
            username = username or ""
            password = password or ""

            # Store any acquired credentials.
            self.passwords[netloc] = (username, password)

        assert (
            # Credentials were found
            (username is not None and password is not None)
            # Credentials were not found
            or (username is None and password is None)
        ), f"Could not load credentials from url: {original_url}"

        return url, username, password

    def __call__(self, req: Request) -> Request:
        # Get credentials for this request
        url, username, password = self._get_url_and_credentials(req.url)

        # Set the url of the request to the url without any credentials
        req.url = url

        if username is not None and password is not None:
            # Send the basic auth with this request
            req = HTTPBasicAuth(username, password)(req)

        # Attach a hook to handle 401 responses
        req.register_hook("response", self.handle_401)

        return req

    # Factored out to allow for easy patching in tests
    def _prompt_for_password(
        self, netloc: str
    ) -> Tuple[Optional[str], Optional[str], bool]:
        username = ask_input(f"User for {netloc}: ") if self.prompting else None
        if not username:
            return None, None, False
        if self.use_keyring:
            auth = self._get_keyring_auth(netloc, username)
            if auth and auth[0] is not None and auth[1] is not None:
                return auth[0], auth[1], False
        password = ask_password("Password: ")
        return username, password, True

    # Factored out to allow for easy patching in tests
    def _should_save_password_to_keyring(self) -> bool:
        if (
            not self.prompting
            or not self.use_keyring
            or not self.keyring_provider.has_keyring
        ):
            return False
        return ask("Save credentials to keyring [y/N]: ", ["y", "n"]) == "y"

    def handle_401(self, resp: Response, **kwargs: Any) -> Response:
        # We only care about 401 responses, anything else we want to just
        #   pass through the actual response
        if resp.status_code != 401:
            return resp

        username, password = None, None

        # Query the keyring for credentials:
        if self.use_keyring:
            username, password = self._get_new_credentials(
                resp.url,
                allow_netrc=False,
                allow_keyring=True,
            )

        # We are not able to prompt the user so simply return the response
        if not self.prompting and not username and not password:
            return resp

        parsed = urllib.parse.urlparse(resp.url)

        # Prompt the user for a new username and password
        save = False
        if not username and not password:
            username, password, save = self._prompt_for_password(parsed.netloc)

        # Store the new username and password to use for future requests
        self._credentials_to_save = None
        if username is not None and password is not None:
            self.passwords[parsed.netloc] = (username, password)

            # Prompt to save the password to keyring
            if save and self._should_save_password_to_keyring():
                self._credentials_to_save = Credentials(
                    url=parsed.netloc,
                    username=username,
                    password=password,
                )

        # Consume content and release the original connection to allow our new
        #   request to reuse the same one.
        # The result of the assignment isn't used, it's just needed to consume
        # the content.
        _ = resp.content
        resp.raw.release_conn()

        # Add our new username and password to the request
        req = HTTPBasicAuth(username or "", password or "")(resp.request)
        req.register_hook("response", self.warn_on_401)

        # On successful request, save the credentials that were used to
        # keyring. (Note that if the user responded "no" above, this member
        # is not set and nothing will be saved.)
        if self._credentials_to_save:
            req.register_hook("response", self.save_credentials)

        # Send our new request
        new_resp = resp.connection.send(req, **kwargs)
        new_resp.history.append(resp)

        return new_resp

    def warn_on_401(self, resp: Response, **kwargs: Any) -> None:
        """Response callback to warn about incorrect credentials."""
        if resp.status_code == 401:
            logger.warning(
                "401 Error, Credentials not correct for %s",
                resp.request.url,
            )

    def save_credentials(self, resp: Response, **kwargs: Any) -> None:
        """Response callback to save credentials on success."""
        assert (
            self.keyring_provider.has_keyring
        ), "should never reach here without keyring"

        creds = self._credentials_to_save
        self._credentials_to_save = None
        if creds and resp.status_code < 400:
            try:
                logger.info("Saving credentials to keyring")
                self.keyring_provider.save_auth_info(
                    creds.url, creds.username, creds.password
                )
            except Exception:
                logger.exception("Failed to save credentials")

# -*- coding: utf-8 -*-
#
# Copyright (C) 2012-2023 The Python Software Foundation.
# See LICENSE.txt and CONTRIBUTORS.txt.
#
"""PEP 376 implementation."""

from __future__ import unicode_literals

import base64
import codecs
import contextlib
import hashlib
import logging
import os
import posixpath
import sys
import zipimport

from . import DistlibException, resources
from .compat import StringIO
from .version import get_scheme, UnsupportedVersionError
from .metadata import (Metadata, METADATA_FILENAME, WHEEL_METADATA_FILENAME, LEGACY_METADATA_FILENAME)
from .util import (parse_requirement, cached_property, parse_name_and_version, read_exports, write_exports, CSVReader,
                   CSVWriter)

__all__ = [
    'Distribution', 'BaseInstalledDistribution', 'InstalledDistribution', 'EggInfoDistribution', 'DistributionPath'
]

logger = logging.getLogger(__name__)

EXPORTS_FILENAME = 'pydist-exports.json'
COMMANDS_FILENAME = 'pydist-commands.json'

DIST_FILES = ('INSTALLER', METADATA_FILENAME, 'RECORD', 'REQUESTED', 'RESOURCES', EXPORTS_FILENAME, 'SHARED')

DISTINFO_EXT = '.dist-info'


class _Cache(object):
    """
    A simple cache mapping names and .dist-info paths to distributions
    """

    def __init__(self):
        """
        Initialise an instance. There is normally one for each DistributionPath.
        """
        self.name = {}
        self.path = {}
        self.generated = False

    def clear(self):
        """
        Clear the cache, setting it to its initial state.
        """
        self.name.clear()
        self.path.clear()
        self.generated = False

    def add(self, dist):
        """
        Add a distribution to the cache.
        :param dist: The distribution to add.
        """
        if dist.path not in self.path:
            self.path[dist.path] = dist
            self.name.setdefault(dist.key, []).append(dist)


class DistributionPath(object):
    """
    Represents a set of distributions installed on a path (typically sys.path).
    """

    def __init__(self, path=None, include_egg=False):
        """
        Create an instance from a path, optionally including legacy (distutils/
        setuptools/distribute) distributions.
        :param path: The path to use, as a list of directories. If not specified,
                     sys.path is used.
        :param include_egg: If True, this instance will look for and return legacy
                            distributions as well as those based on PEP 376.
        """
        if path is None:
            path = sys.path
        self.path = path
        self._include_dist = True
        self._include_egg = include_egg

        self._cache = _Cache()
        self._cache_egg = _Cache()
        self._cache_enabled = True
        self._scheme = get_scheme('default')

    def _get_cache_enabled(self):
        return self._cache_enabled

    def _set_cache_enabled(self, value):
        self._cache_enabled = value

    cache_enabled = property(_get_cache_enabled, _set_cache_enabled)

    def clear_cache(self):
        """
        Clears the internal cache.
        """
        self._cache.clear()
        self._cache_egg.clear()

    def _yield_distributions(self):
        """
        Yield .dist-info and/or .egg(-info) distributions.
        """
        # We need to check if we've seen some resources already, because on
        # some Linux systems (e.g. some Debian/Ubuntu variants) there are
        # symlinks which alias other files in the environment.
        seen = set()
        for path in self.path:
            finder = resources.finder_for_path(path)
            if finder is None:
                continue
            r = finder.find('')
            if not r or not r.is_container:
                continue
            rset = sorted(r.resources)
            for entry in rset:
                r = finder.find(entry)
                if not r or r.path in seen:
                    continue
                try:
                    if self._include_dist and entry.endswith(DISTINFO_EXT):
                        possible_filenames = [METADATA_FILENAME, WHEEL_METADATA_FILENAME, LEGACY_METADATA_FILENAME]
                        for metadata_filename in possible_filenames:
                            metadata_path = posixpath.join(entry, metadata_filename)
                            pydist = finder.find(metadata_path)
                            if pydist:
                                break
                        else:
                            continue

                        with contextlib.closing(pydist.as_stream()) as stream:
                            metadata = Metadata(fileobj=stream, scheme='legacy')
                        logger.debug('Found %s', r.path)
                        seen.add(r.path)
                        yield new_dist_class(r.path, metadata=metadata, env=self)
                    elif self._include_egg and entry.endswith(('.egg-info', '.egg')):
                        logger.debug('Found %s', r.path)
                        seen.add(r.path)
                        yield old_dist_class(r.path, self)
                except Exception as e:
                    msg = 'Unable to read distribution at %s, perhaps due to bad metadata: %s'
                    logger.warning(msg, r.path, e)
                    import warnings
                    warnings.warn(msg % (r.path, e), stacklevel=2)

    def _generate_cache(self):
        """
        Scan the path for distributions and populate the cache with
        those that are found.
        """
        gen_dist = not self._cache.generated
        gen_egg = self._include_egg and not self._cache_egg.generated
        if gen_dist or gen_egg:
            for dist in self._yield_distributions():
                if isinstance(dist, InstalledDistribution):
                    self._cache.add(dist)
                else:
                    self._cache_egg.add(dist)

            if gen_dist:
                self._cache.generated = True
            if gen_egg:
                self._cache_egg.generated = True

    @classmethod
    def distinfo_dirname(cls, name, version):
        """
        The *name* and *version* parameters are converted into their
        filename-escaped form, i.e. any ``'-'`` characters are replaced
        with ``'_'`` other than the one in ``'dist-info'`` and the one
        separating the name from the version number.

        :parameter name: is converted to a standard distribution name by replacing
                         any runs of non- alphanumeric characters with a single
                         ``'-'``.
        :type name: string
        :parameter version: is converted to a standard version string. Spaces
                            become dots, and all other non-alphanumeric characters
                            (except dots) become dashes, with runs of multiple
                            dashes condensed to a single dash.
        :type version: string
        :returns: directory name
        :rtype: string"""
        name = name.replace('-', '_')
        return '-'.join([name, version]) + DISTINFO_EXT

    def get_distributions(self):
        """
        Provides an iterator that looks for distributions and returns
        :class:`InstalledDistribution` or
        :class:`EggInfoDistribution` instances for each one of them.

        :rtype: iterator of :class:`InstalledDistribution` and
                :class:`EggInfoDistribution` instances
        """
        if not self._cache_enabled:
            for dist in self._yield_distributions():
                yield dist
        else:
            self._generate_cache()

            for dist in self._cache.path.values():
                yield dist

            if self._include_egg:
                for dist in self._cache_egg.path.values():
                    yield dist

    def get_distribution(self, name):
        """
        Looks for a named distribution on the path.

        This function only returns the first result found, as no more than one
        value is expected. If nothing is found, ``None`` is returned.

        :rtype: :class:`InstalledDistribution`, :class:`EggInfoDistribution`
                or ``None``
        """
        result = None
        name = name.lower()
        if not self._cache_enabled:
            for dist in self._yield_distributions():
                if dist.key == name:
                    result = dist
                    break
        else:
            self._generate_cache()

            if name in self._cache.name:
                result = self._cache.name[name][0]
            elif self._include_egg and name in self._cache_egg.name:
                result = self._cache_egg.name[name][0]
        return result

    def provides_distribution(self, name, version=None):
        """
        Iterates over all distributions to find which distributions provide *name*.
        If a *version* is provided, it will be used to filter the results.

        This function only returns the first result found, since no more than
        one values are expected. If the directory is not found, returns ``None``.

        :parameter version: a version specifier that indicates the version
                            required, conforming to the format in ``PEP-345``

        :type name: string
        :type version: string
        """
        matcher = None
        if version is not None:
            try:
                matcher = self._scheme.matcher('%s (%s)' % (name, version))
            except ValueError:
                raise DistlibException('invalid name or version: %r, %r' % (name, version))

        for dist in self.get_distributions():
            # We hit a problem on Travis where enum34 was installed and doesn't
            # have a provides attribute ...
            if not hasattr(dist, 'provides'):
                logger.debug('No "provides": %s', dist)
            else:
                provided = dist.provides

                for p in provided:
                    p_name, p_ver = parse_name_and_version(p)
                    if matcher is None:
                        if p_name == name:
                            yield dist
                            break
                    else:
                        if p_name == name and matcher.match(p_ver):
                            yield dist
                            break

    def get_file_path(self, name, relative_path):
        """
        Return the path to a resource file.
        """
        dist = self.get_distribution(name)
        if dist is None:
            raise LookupError('no distribution named %r found' % name)
        return dist.get_resource_path(relative_path)

    def get_exported_entries(self, category, name=None):
        """
        Return all of the exported entries in a particular category.

        :param category: The category to search for entries.
        :param name: If specified, only entries with that name are returned.
        """
        for dist in self.get_distributions():
            r = dist.exports
            if category in r:
                d = r[category]
                if name is not None:
                    if name in d:
                        yield d[name]
                else:
                    for v in d.values():
                        yield v


class Distribution(object):
    """
    A base class for distributions, whether installed or from indexes.
    Either way, it must have some metadata, so that's all that's needed
    for construction.
    """

    build_time_dependency = False
    """
    Set to True if it's known to be only a build-time dependency (i.e.
    not needed after installation).
    """

    requested = False
    """A boolean that indicates whether the ``REQUESTED`` metadata file is
    present (in other words, whether the package was installed by user
    request or it was installed as a dependency)."""

    def __init__(self, metadata):
        """
        Initialise an instance.
        :param metadata: The instance of :class:`Metadata` describing this
        distribution.
        """
        self.metadata = metadata
        self.name = metadata.name
        self.key = self.name.lower()  # for case-insensitive comparisons
        self.version = metadata.version
        self.locator = None
        self.digest = None
        self.extras = None  # additional features requested
        self.context = None  # environment marker overrides
        self.download_urls = set()
        self.digests = {}

    @property
    def source_url(self):
        """
        The source archive download URL for this distribution.
        """
        return self.metadata.source_url

    download_url = source_url  # Backward compatibility

    @property
    def name_and_version(self):
        """
        A utility property which displays the name and version in parentheses.
        """
        return '%s (%s)' % (self.name, self.version)

    @property
    def provides(self):
        """
        A set of distribution names and versions provided by this distribution.
        :return: A set of "name (version)" strings.
        """
        plist = self.metadata.provides
        s = '%s (%s)' % (self.name, self.version)
        if s not in plist:
            plist.append(s)
        return plist

    def _get_requirements(self, req_attr):
        md = self.metadata
        reqts = getattr(md, req_attr)
        logger.debug('%s: got requirements %r from metadata: %r', self.name, req_attr, reqts)
        return set(md.get_requirements(reqts, extras=self.extras, env=self.context))

    @property
    def run_requires(self):
        return self._get_requirements('run_requires')

    @property
    def meta_requires(self):
        return self._get_requirements('meta_requires')

    @property
    def build_requires(self):
        return self._get_requirements('build_requires')

    @property
    def test_requires(self):
        return self._get_requirements('test_requires')

    @property
    def dev_requires(self):
        return self._get_requirements('dev_requires')

    def matches_requirement(self, req):
        """
        Say if this instance matches (fulfills) a requirement.
        :param req: The requirement to match.
        :rtype req: str
        :return: True if it matches, else False.
        """
        # Requirement may contain extras - parse to lose those
        # from what's passed to the matcher
        r = parse_requirement(req)
        scheme = get_scheme(self.metadata.scheme)
        try:
            matcher = scheme.matcher(r.requirement)
        except UnsupportedVersionError:
            # XXX compat-mode if cannot read the version
            logger.warning('could not read version %r - using name only', req)
            name = req.split()[0]
            matcher = scheme.matcher(name)

        name = matcher.key  # case-insensitive

        result = False
        for p in self.provides:
            p_name, p_ver = parse_name_and_version(p)
            if p_name != name:
                continue
            try:
                result = matcher.match(p_ver)
                break
            except UnsupportedVersionError:
                pass
        return result

    def __repr__(self):
        """
        Return a textual representation of this instance,
        """
        if self.source_url:
            suffix = ' [%s]' % self.source_url
        else:
            suffix = ''
        return '<Distribution %s (%s)%s>' % (self.name, self.version, suffix)

    def __eq__(self, other):
        """
        See if this distribution is the same as another.
        :param other: The distribution to compare with. To be equal to one
                      another. distributions must have the same type, name,
                      version and source_url.
        :return: True if it is the same, else False.
        """
        if type(other) is not type(self):
            result = False
        else:
            result = (self.name == other.name and self.version == other.version and self.source_url == other.source_url)
        return result

    def __hash__(self):
        """
        Compute hash in a way which matches the equality test.
        """
        return hash(self.name) + hash(self.version) + hash(self.source_url)


class BaseInstalledDistribution(Distribution):
    """
    This is the base class for installed distributions (whether PEP 376 or
    legacy).
    """

    hasher = None

    def __init__(self, metadata, path, env=None):
        """
        Initialise an instance.
        :param metadata: An instance of :class:`Metadata` which describes the
                         distribution. This will normally have been initialised
                         from a metadata file in the ``path``.
        :param path:     The path of the ``.dist-info`` or ``.egg-info``
                         directory for the distribution.
        :param env:      This is normally the :class:`DistributionPath`
                         instance where this distribution was found.
        """
        super(BaseInstalledDistribution, self).__init__(metadata)
        self.path = path
        self.dist_path = env

    def get_hash(self, data, hasher=None):
        """
        Get the hash of some data, using a particular hash algorithm, if
        specified.

        :param data: The data to be hashed.
        :type data: bytes
        :param hasher: The name of a hash implementation, supported by hashlib,
                       or ``None``. Examples of valid values are ``'sha1'``,
                       ``'sha224'``, ``'sha384'``, '``sha256'``, ``'md5'`` and
                       ``'sha512'``. If no hasher is specified, the ``hasher``
                       attribute of the :class:`InstalledDistribution` instance
                       is used. If the hasher is determined to be ``None``, MD5
                       is used as the hashing algorithm.
        :returns: The hash of the data. If a hasher was explicitly specified,
                  the returned hash will be prefixed with the specified hasher
                  followed by '='.
        :rtype: str
        """
        if hasher is None:
            hasher = self.hasher
        if hasher is None:
            hasher = hashlib.md5
            prefix = ''
        else:
            hasher = getattr(hashlib, hasher)
            prefix = '%s=' % self.hasher
        digest = hasher(data).digest()
        digest = base64.urlsafe_b64encode(digest).rstrip(b'=').decode('ascii')
        return '%s%s' % (prefix, digest)


class InstalledDistribution(BaseInstalledDistribution):
    """
    Created with the *path* of the ``.dist-info`` directory provided to the
    constructor. It reads the metadata contained in ``pydist.json`` when it is
    instantiated., or uses a passed in Metadata instance (useful for when
    dry-run mode is being used).
    """

    hasher = 'sha256'

    def __init__(self, path, metadata=None, env=None):
        self.modules = []
        self.finder = finder = resources.finder_for_path(path)
        if finder is None:
            raise ValueError('finder unavailable for %s' % path)
        if env and env._cache_enabled and path in env._cache.path:
            metadata = env._cache.path[path].metadata
        elif metadata is None:
            r = finder.find(METADATA_FILENAME)
            # Temporary - for Wheel 0.23 support
            if r is None:
                r = finder.find(WHEEL_METADATA_FILENAME)
            # Temporary - for legacy support
            if r is None:
                r = finder.find(LEGACY_METADATA_FILENAME)
            if r is None:
                raise ValueError('no %s found in %s' % (METADATA_FILENAME, path))
            with contextlib.closing(r.as_stream()) as stream:
                metadata = Metadata(fileobj=stream, scheme='legacy')

        super(InstalledDistribution, self).__init__(metadata, path, env)

        if env and env._cache_enabled:
            env._cache.add(self)

        r = finder.find('REQUESTED')
        self.requested = r is not None
        p = os.path.join(path, 'top_level.txt')
        if os.path.exists(p):
            with open(p, 'rb') as f:
                data = f.read().decode('utf-8')
            self.modules = data.splitlines()

    def __repr__(self):
        return '<InstalledDistribution %r %s at %r>' % (self.name, self.version, self.path)

    def __str__(self):
        return "%s %s" % (self.name, self.version)

    def _get_records(self):
        """
        Get the list of installed files for the distribution
        :return: A list of tuples of path, hash and size. Note that hash and
                 size might be ``None`` for some entries. The path is exactly
                 as stored in the file (which is as in PEP 376).
        """
        results = []
        r = self.get_distinfo_resource('RECORD')
        with contextlib.closing(r.as_stream()) as stream:
            with CSVReader(stream=stream) as record_reader:
                # Base location is parent dir of .dist-info dir
                # base_location = os.path.dirname(self.path)
                # base_location = os.path.abspath(base_location)
                for row in record_reader:
                    missing = [None for i in range(len(row), 3)]
                    path, checksum, size = row + missing
                    # if not os.path.isabs(path):
                    #     path = path.replace('/', os.sep)
                    #     path = os.path.join(base_location, path)
                    results.append((path, checksum, size))
        return results

    @cached_property
    def exports(self):
        """
        Return the information exported by this distribution.
        :return: A dictionary of exports, mapping an export category to a dict
                 of :class:`ExportEntry` instances describing the individual
                 export entries, and keyed by name.
        """
        result = {}
        r = self.get_distinfo_resource(EXPORTS_FILENAME)
        if r:
            result = self.read_exports()
        return result

    def read_exports(self):
        """
        Read exports data from a file in .ini format.

        :return: A dictionary of exports, mapping an export category to a list
                 of :class:`ExportEntry` instances describing the individual
                 export entries.
        """
        result = {}
        r = self.get_distinfo_resource(EXPORTS_FILENAME)
        if r:
            with contextlib.closing(r.as_stream()) as stream:
                result = read_exports(stream)
        return result

    def write_exports(self, exports):
        """
        Write a dictionary of exports to a file in .ini format.
        :param exports: A dictionary of exports, mapping an export category to
                        a list of :class:`ExportEntry` instances describing the
                        individual export entries.
        """
        rf = self.get_distinfo_file(EXPORTS_FILENAME)
        with open(rf, 'w') as f:
            write_exports(exports, f)

    def get_resource_path(self, relative_path):
        """
        NOTE: This API may change in the future.

        Return the absolute path to a resource file with the given relative
        path.

        :param relative_path: The path, relative to .dist-info, of the resource
                              of interest.
        :return: The absolute path where the resource is to be found.
        """
        r = self.get_distinfo_resource('RESOURCES')
        with contextlib.closing(r.as_stream()) as stream:
            with CSVReader(stream=stream) as resources_reader:
                for relative, destination in resources_reader:
                    if relative == relative_path:
                        return destination
        raise KeyError('no resource file with relative path %r '
                       'is installed' % relative_path)

    def list_installed_files(self):
        """
        Iterates over the ``RECORD`` entries and returns a tuple
        ``(path, hash, size)`` for each line.

        :returns: iterator of (path, hash, size)
        """
        for result in self._get_records():
            yield result

    def write_installed_files(self, paths, prefix, dry_run=False):
        """
        Writes the ``RECORD`` file, using the ``paths`` iterable passed in. Any
        existing ``RECORD`` file is silently overwritten.

        prefix is used to determine when to write absolute paths.
        """
        prefix = os.path.join(prefix, '')
        base = os.path.dirname(self.path)
        base_under_prefix = base.startswith(prefix)
        base = os.path.join(base, '')
        record_path = self.get_distinfo_file('RECORD')
        logger.info('creating %s', record_path)
        if dry_run:
            return None
        with CSVWriter(record_path) as writer:
            for path in paths:
                if os.path.isdir(path) or path.endswith(('.pyc', '.pyo')):
                    # do not put size and hash, as in PEP-376
                    hash_value = size = ''
                else:
                    size = '%d' % os.path.getsize(path)
                    with open(path, 'rb') as fp:
                        hash_value = self.get_hash(fp.read())
                if path.startswith(base) or (base_under_prefix and path.startswith(prefix)):
                    path = os.path.relpath(path, base)
                writer.writerow((path, hash_value, size))

            # add the RECORD file itself
            if record_path.startswith(base):
                record_path = os.path.relpath(record_path, base)
            writer.writerow((record_path, '', ''))
        return record_path

    def check_installed_files(self):
        """
        Checks that the hashes and sizes of the files in ``RECORD`` are
        matched by the files themselves. Returns a (possibly empty) list of
        mismatches. Each entry in the mismatch list will be a tuple consisting
        of the path, 'exists', 'size' or 'hash' according to what didn't match
        (existence is checked first, then size, then hash), the expected
        value and the actual value.
        """
        mismatches = []
        base = os.path.dirname(self.path)
        record_path = self.get_distinfo_file('RECORD')
        for path, hash_value, size in self.list_installed_files():
            if not os.path.isabs(path):
                path = os.path.join(base, path)
            if path == record_path:
                continue
            if not os.path.exists(path):
                mismatches.append((path, 'exists', True, False))
            elif os.path.isfile(path):
                actual_size = str(os.path.getsize(path))
                if size and actual_size != size:
                    mismatches.append((path, 'size', size, actual_size))
                elif hash_value:
                    if '=' in hash_value:
                        hasher = hash_value.split('=', 1)[0]
                    else:
                        hasher = None

                    with open(path, 'rb') as f:
                        actual_hash = self.get_hash(f.read(), hasher)
                        if actual_hash != hash_value:
                            mismatches.append((path, 'hash', hash_value, actual_hash))
        return mismatches

    @cached_property
    def shared_locations(self):
        """
        A dictionary of shared locations whose keys are in the set 'prefix',
        'purelib', 'platlib', 'scripts', 'headers', 'data' and 'namespace'.
        The corresponding value is the absolute path of that category for
        this distribution, and takes into account any paths selected by the
        user at installation time (e.g. via command-line arguments). In the
        case of the 'namespace' key, this would be a list of absolute paths
        for the roots of namespace packages in this distribution.

        The first time this property is accessed, the relevant information is
        read from the SHARED file in the .dist-info directory.
        """
        result = {}
        shared_path = os.path.join(self.path, 'SHARED')
        if os.path.isfile(shared_path):
            with codecs.open(shared_path, 'r', encoding='utf-8') as f:
                lines = f.read().splitlines()
            for line in lines:
                key, value = line.split('=', 1)
                if key == 'namespace':
                    result.setdefault(key, []).append(value)
                else:
                    result[key] = value
        return result

    def write_shared_locations(self, paths, dry_run=False):
        """
        Write shared location information to the SHARED file in .dist-info.
        :param paths: A dictionary as described in the documentation for
        :meth:`shared_locations`.
        :param dry_run: If True, the action is logged but no file is actually
                        written.
        :return: The path of the file written to.
        """
        shared_path = os.path.join(self.path, 'SHARED')
        logger.info('creating %s', shared_path)
        if dry_run:
            return None
        lines = []
        for key in ('prefix', 'lib', 'headers', 'scripts', 'data'):
            path = paths[key]
            if os.path.isdir(paths[key]):
                lines.append('%s=%s' % (key, path))
        for ns in paths.get('namespace', ()):
            lines.append('namespace=%s' % ns)

        with codecs.open(shared_path, 'w', encoding='utf-8') as f:
            f.write('\n'.join(lines))
        return shared_path

    def get_distinfo_resource(self, path):
        if path not in DIST_FILES:
            raise DistlibException('invalid path for a dist-info file: '
                                   '%r at %r' % (path, self.path))
        finder = resources.finder_for_path(self.path)
        if finder is None:
            raise DistlibException('Unable to get a finder for %s' % self.path)
        return finder.find(path)

    def get_distinfo_file(self, path):
        """
        Returns a path located under the ``.dist-info`` directory. Returns a
        string representing the path.

        :parameter path: a ``'/'``-separated path relative to the
                         ``.dist-info`` directory or an absolute path;
                         If *path* is an absolute path and doesn't start
                         with the ``.dist-info`` directory path,
                         a :class:`DistlibException` is raised
        :type path: str
        :rtype: str
        """
        # Check if it is an absolute path  # XXX use relpath, add tests
        if path.find(os.sep) >= 0:
            # it's an absolute path?
            distinfo_dirname, path = path.split(os.sep)[-2:]
            if distinfo_dirname != self.path.split(os.sep)[-1]:
                raise DistlibException('dist-info file %r does not belong to the %r %s '
                                       'distribution' % (path, self.name, self.version))

        # The file must be relative
        if path not in DIST_FILES:
            raise DistlibException('invalid path for a dist-info file: '
                                   '%r at %r' % (path, self.path))

        return os.path.join(self.path, path)

    def list_distinfo_files(self):
        """
        Iterates over the ``RECORD`` entries and returns paths for each line if
        the path is pointing to a file located in the ``.dist-info`` directory
        or one of its subdirectories.

        :returns: iterator of paths
        """
        base = os.path.dirname(self.path)
        for path, checksum, size in self._get_records():
            # XXX add separator or use real relpath algo
            if not os.path.isabs(path):
                path = os.path.join(base, path)
            if path.startswith(self.path):
                yield path

    def __eq__(self, other):
        return (isinstance(other, InstalledDistribution) and self.path == other.path)

    # See http://docs.python.org/reference/datamodel#object.__hash__
    __hash__ = object.__hash__


class EggInfoDistribution(BaseInstalledDistribution):
    """Created with the *path* of the ``.egg-info`` directory or file provided
    to the constructor. It reads the metadata contained in the file itself, or
    if the given path happens to be a directory, the metadata is read from the
    file ``PKG-INFO`` under that directory."""

    requested = True  # as we have no way of knowing, assume it was
    shared_locations = {}

    def __init__(self, path, env=None):

        def set_name_and_version(s, n, v):
            s.name = n
            s.key = n.lower()  # for case-insensitive comparisons
            s.version = v

        self.path = path
        self.dist_path = env
        if env and env._cache_enabled and path in env._cache_egg.path:
            metadata = env._cache_egg.path[path].metadata
            set_name_and_version(self, metadata.name, metadata.version)
        else:
            metadata = self._get_metadata(path)

            # Need to be set before caching
            set_name_and_version(self, metadata.name, metadata.version)

            if env and env._cache_enabled:
                env._cache_egg.add(self)
        super(EggInfoDistribution, self).__init__(metadata, path, env)

    def _get_metadata(self, path):
        requires = None

        def parse_requires_data(data):
            """Create a list of dependencies from a requires.txt file.

            *data*: the contents of a setuptools-produced requires.txt file.
            """
            reqs = []
            lines = data.splitlines()
            for line in lines:
                line = line.strip()
                # sectioned files have bare newlines (separating sections)
                if not line:  # pragma: no cover
                    continue
                if line.startswith('['):  # pragma: no cover
                    logger.warning('Unexpected line: quitting requirement scan: %r', line)
                    break
                r = parse_requirement(line)
                if not r:  # pragma: no cover
                    logger.warning('Not recognised as a requirement: %r', line)
                    continue
                if r.extras:  # pragma: no cover
                    logger.warning('extra requirements in requires.txt are '
                                   'not supported')
                if not r.constraints:
                    reqs.append(r.name)
                else:
                    cons = ', '.join('%s%s' % c for c in r.constraints)
                    reqs.append('%s (%s)' % (r.name, cons))
            return reqs

        def parse_requires_path(req_path):
            """Create a list of dependencies from a requires.txt file.

            *req_path*: the path to a setuptools-produced requires.txt file.
            """

            reqs = []
            try:
                with codecs.open(req_path, 'r', 'utf-8') as fp:
                    reqs = parse_requires_data(fp.read())
            except IOError:
                pass
            return reqs

        tl_path = tl_data = None
        if path.endswith('.egg'):
            if os.path.isdir(path):
                p = os.path.join(path, 'EGG-INFO')
                meta_path = os.path.join(p, 'PKG-INFO')
                metadata = Metadata(path=meta_path, scheme='legacy')
                req_path = os.path.join(p, 'requires.txt')
                tl_path = os.path.join(p, 'top_level.txt')
                requires = parse_requires_path(req_path)
            else:
                # FIXME handle the case where zipfile is not available
                zipf = zipimport.zipimporter(path)
                fileobj = StringIO(zipf.get_data('EGG-INFO/PKG-INFO').decode('utf8'))
                metadata = Metadata(fileobj=fileobj, scheme='legacy')
                try:
                    data = zipf.get_data('EGG-INFO/requires.txt')
                    tl_data = zipf.get_data('EGG-INFO/top_level.txt').decode('utf-8')
                    requires = parse_requires_data(data.decode('utf-8'))
                except IOError:
                    requires = None
        elif path.endswith('.egg-info'):
            if os.path.isdir(path):
                req_path = os.path.join(path, 'requires.txt')
                requires = parse_requires_path(req_path)
                path = os.path.join(path, 'PKG-INFO')
                tl_path = os.path.join(path, 'top_level.txt')
            metadata = Metadata(path=path, scheme='legacy')
        else:
            raise DistlibException('path must end with .egg-info or .egg, '
                                   'got %r' % path)

        if requires:
            metadata.add_requirements(requires)
        # look for top-level modules in top_level.txt, if present
        if tl_data is None:
            if tl_path is not None and os.path.exists(tl_path):
                with open(tl_path, 'rb') as f:
                    tl_data = f.read().decode('utf-8')
        if not tl_data:
            tl_data = []
        else:
            tl_data = tl_data.splitlines()
        self.modules = tl_data
        return metadata

    def __repr__(self):
        return '<EggInfoDistribution %r %s at %r>' % (self.name, self.version, self.path)

    def __str__(self):
        return "%s %s" % (self.name, self.version)

    def check_installed_files(self):
        """
        Checks that the hashes and sizes of the files in ``RECORD`` are
        matched by the files themselves. Returns a (possibly empty) list of
        mismatches. Each entry in the mismatch list will be a tuple consisting
        of the path, 'exists', 'size' or 'hash' according to what didn't match
        (existence is checked first, then size, then hash), the expected
        value and the actual value.
        """
        mismatches = []
        record_path = os.path.join(self.path, 'installed-files.txt')
        if os.path.exists(record_path):
            for path, _, _ in self.list_installed_files():
                if path == record_path:
                    continue
                if not os.path.exists(path):
                    mismatches.append((path, 'exists', True, False))
        return mismatches

    def list_installed_files(self):
        """
        Iterates over the ``installed-files.txt`` entries and returns a tuple
        ``(path, hash, size)`` for each line.

        :returns: a list of (path, hash, size)
        """

        def _md5(path):
            f = open(path, 'rb')
            try:
                content = f.read()
            finally:
                f.close()
            return hashlib.md5(content).hexdigest()

        def _size(path):
            return os.stat(path).st_size

        record_path = os.path.join(self.path, 'installed-files.txt')
        result = []
        if os.path.exists(record_path):
            with codecs.open(record_path, 'r', encoding='utf-8') as f:
                for line in f:
                    line = line.strip()
                    p = os.path.normpath(os.path.join(self.path, line))
                    # "./" is present as a marker between installed files
                    # and installation metadata files
                    if not os.path.exists(p):
                        logger.warning('Non-existent file: %s', p)
                        if p.endswith(('.pyc', '.pyo')):
                            continue
                        # otherwise fall through and fail
                    if not os.path.isdir(p):
                        result.append((p, _md5(p), _size(p)))
            result.append((record_path, None, None))
        return result

    def list_distinfo_files(self, absolute=False):
        """
        Iterates over the ``installed-files.txt`` entries and returns paths for
        each line if the path is pointing to a file located in the
        ``.egg-info`` directory or one of its subdirectories.

        :parameter absolute: If *absolute* is ``True``, each returned path is
                          transformed into a local absolute path. Otherwise the
                          raw value from ``installed-files.txt`` is returned.
        :type absolute: boolean
        :returns: iterator of paths
        """
        record_path = os.path.join(self.path, 'installed-files.txt')
        if os.path.exists(record_path):
            skip = True
            with codecs.open(record_path, 'r', encoding='utf-8') as f:
                for line in f:
                    line = line.strip()
                    if line == './':
                        skip = False
                        continue
                    if not skip:
                        p = os.path.normpath(os.path.join(self.path, line))
                        if p.startswith(self.path):
                            if absolute:
                                yield p
                            else:
                                yield line

    def __eq__(self, other):
        return (isinstance(other, EggInfoDistribution) and self.path == other.path)

    # See http://docs.python.org/reference/datamodel#object.__hash__
    __hash__ = object.__hash__


new_dist_class = InstalledDistribution
old_dist_class = EggInfoDistribution


class DependencyGraph(object):
    """
    Represents a dependency graph between distributions.

    The dependency relationships are stored in an ``adjacency_list`` that maps
    distributions to a list of ``(other, label)`` tuples where  ``other``
    is a distribution and the edge is labeled with ``label`` (i.e. the version
    specifier, if such was provided). Also, for more efficient traversal, for
    every distribution ``x``, a list of predecessors is kept in
    ``reverse_list[x]``. An edge from distribution ``a`` to
    distribution ``b`` means that ``a`` depends on ``b``. If any missing
    dependencies are found, they are stored in ``missing``, which is a
    dictionary that maps distributions to a list of requirements that were not
    provided by any other distributions.
    """

    def __init__(self):
        self.adjacency_list = {}
        self.reverse_list = {}
        self.missing = {}

    def add_distribution(self, distribution):
        """Add the *distribution* to the graph.

        :type distribution: :class:`distutils2.database.InstalledDistribution`
                            or :class:`distutils2.database.EggInfoDistribution`
        """
        self.adjacency_list[distribution] = []
        self.reverse_list[distribution] = []
        # self.missing[distribution] = []

    def add_edge(self, x, y, label=None):
        """Add an edge from distribution *x* to distribution *y* with the given
        *label*.

        :type x: :class:`distutils2.database.InstalledDistribution` or
                 :class:`distutils2.database.EggInfoDistribution`
        :type y: :class:`distutils2.database.InstalledDistribution` or
                 :class:`distutils2.database.EggInfoDistribution`
        :type label: ``str`` or ``None``
        """
        self.adjacency_list[x].append((y, label))
        # multiple edges are allowed, so be careful
        if x not in self.reverse_list[y]:
            self.reverse_list[y].append(x)

    def add_missing(self, distribution, requirement):
        """
        Add a missing *requirement* for the given *distribution*.

        :type distribution: :class:`distutils2.database.InstalledDistribution`
                            or :class:`distutils2.database.EggInfoDistribution`
        :type requirement: ``str``
        """
        logger.debug('%s missing %r', distribution, requirement)
        self.missing.setdefault(distribution, []).append(requirement)

    def _repr_dist(self, dist):
        return '%s %s' % (dist.name, dist.version)

    def repr_node(self, dist, level=1):
        """Prints only a subgraph"""
        output = [self._repr_dist(dist)]
        for other, label in self.adjacency_list[dist]:
            dist = self._repr_dist(other)
            if label is not None:
                dist = '%s [%s]' % (dist, label)
            output.append('    ' * level + str(dist))
            suboutput = self.repr_node(other, level + 1)
            subs = suboutput.split('\n')
            output.extend(subs[1:])
        return '\n'.join(output)

    def to_dot(self, f, skip_disconnected=True):
        """Writes a DOT output for the graph to the provided file *f*.

        If *skip_disconnected* is set to ``True``, then all distributions
        that are not dependent on any other distribution are skipped.

        :type f: has to support ``file``-like operations
        :type skip_disconnected: ``bool``
        """
        disconnected = []

        f.write("digraph dependencies {\n")
        for dist, adjs in self.adjacency_list.items():
            if len(adjs) == 0 and not skip_disconnected:
                disconnected.append(dist)
            for other, label in adjs:
                if label is not None:
                    f.write('"%s" -> "%s" [label="%s"]\n' % (dist.name, other.name, label))
                else:
                    f.write('"%s" -> "%s"\n' % (dist.name, other.name))
        if not skip_disconnected and len(disconnected) > 0:
            f.write('subgraph disconnected {\n')
            f.write('label = "Disconnected"\n')
            f.write('bgcolor = red\n')

            for dist in disconnected:
                f.write('"%s"' % dist.name)
                f.write('\n')
            f.write('}\n')
        f.write('}\n')

    def topological_sort(self):
        """
        Perform a topological sort of the graph.
        :return: A tuple, the first element of which is a topologically sorted
                 list of distributions, and the second element of which is a
                 list of distributions that cannot be sorted because they have
                 circular dependencies and so form a cycle.
        """
        result = []
        # Make a shallow copy of the adjacency list
        alist = {}
        for k, v in self.adjacency_list.items():
            alist[k] = v[:]
        while True:
            # See what we can remove in this run
            to_remove = []
            for k, v in list(alist.items())[:]:
                if not v:
                    to_remove.append(k)
                    del alist[k]
            if not to_remove:
                # What's left in alist (if anything) is a cycle.
                break
            # Remove from the adjacency list of others
            for k, v in alist.items():
                alist[k] = [(d, r) for d, r in v if d not in to_remove]
            logger.debug('Moving to result: %s', ['%s (%s)' % (d.name, d.version) for d in to_remove])
            result.extend(to_remove)
        return result, list(alist.keys())

    def __repr__(self):
        """Representation of the graph"""
        output = []
        for dist, adjs in self.adjacency_list.items():
            output.append(self.repr_node(dist))
        return '\n'.join(output)


def make_graph(dists, scheme='default'):
    """Makes a dependency graph from the given distributions.

    :parameter dists: a list of distributions
    :type dists: list of :class:`distutils2.database.InstalledDistribution` and
                 :class:`distutils2.database.EggInfoDistribution` instances
    :rtype: a :class:`DependencyGraph` instance
    """
    scheme = get_scheme(scheme)
    graph = DependencyGraph()
    provided = {}  # maps names to lists of (version, dist) tuples

    # first, build the graph and find out what's provided
    for dist in dists:
        graph.add_distribution(dist)

        for p in dist.provides:
            name, version = parse_name_and_version(p)
            logger.debug('Add to provided: %s, %s, %s', name, version, dist)
            provided.setdefault(name, []).append((version, dist))

    # now make the edges
    for dist in dists:
        requires = (dist.run_requires | dist.meta_requires | dist.build_requires | dist.dev_requires)
        for req in requires:
            try:
                matcher = scheme.matcher(req)
            except UnsupportedVersionError:
                # XXX compat-mode if cannot read the version
                logger.warning('could not read version %r - using name only', req)
                name = req.split()[0]
                matcher = scheme.matcher(name)

            name = matcher.key  # case-insensitive

            matched = False
            if name in provided:
                for version, provider in provided[name]:
                    try:
                        match = matcher.match(version)
                    except UnsupportedVersionError:
                        match = False

                    if match:
                        graph.add_edge(dist, provider, req)
                        matched = True
                        break
            if not matched:
                graph.add_missing(dist, req)
    return graph


def get_dependent_dists(dists, dist):
    """Recursively generate a list of distributions from *dists* that are
    dependent on *dist*.

    :param dists: a list of distributions
    :param dist: a distribution, member of *dists* for which we are interested
    """
    if dist not in dists:
        raise DistlibException('given distribution %r is not a member '
                               'of the list' % dist.name)
    graph = make_graph(dists)

    dep = [dist]  # dependent distributions
    todo = graph.reverse_list[dist]  # list of nodes we should inspect

    while todo:
        d = todo.pop()
        dep.append(d)
        for succ in graph.reverse_list[d]:
            if succ not in dep:
                todo.append(succ)

    dep.pop(0)  # remove dist from dep, was there to prevent infinite loops
    return dep


def get_required_dists(dists, dist):
    """Recursively generate a list of distributions from *dists* that are
    required by *dist*.

    :param dists: a list of distributions
    :param dist: a distribution, member of *dists* for which we are interested
                 in finding the dependencies.
    """
    if dist not in dists:
        raise DistlibException('given distribution %r is not a member '
                               'of the list' % dist.name)
    graph = make_graph(dists)

    req = set()  # required distributions
    todo = graph.adjacency_list[dist]  # list of nodes we should inspect
    seen = set(t[0] for t in todo)  # already added to todo

    while todo:
        d = todo.pop()[0]
        req.add(d)
        pred_list = graph.adjacency_list[d]
        for pred in pred_list:
            d = pred[0]
            if d not in req and d not in seen:
                seen.add(d)
                todo.append(pred)
    return req


def make_dist(name, version, **kwargs):
    """
    A convenience method for making a dist given just a name and version.
    """
    summary = kwargs.pop('summary', 'Placeholder for summary')
    md = Metadata(**kwargs)
    md.name = name
    md.version = version
    md.summary = summary or 'Placeholder for summary'
    return Distribution(md)