"""
    pygments.formatters.rtf
    ~~~~~~~~~~~~~~~~~~~~~~~

    A formatter that generates RTF files.

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

from collections import OrderedDict
from pip._vendor.pygments.formatter import Formatter
from pip._vendor.pygments.style import _ansimap
from pip._vendor.pygments.util import get_bool_opt, get_int_opt, get_list_opt, surrogatepair


__all__ = ['RtfFormatter']


class RtfFormatter(Formatter):
    """
    Format tokens as RTF markup. This formatter automatically outputs full RTF
    documents with color information and other useful stuff. Perfect for Copy and
    Paste into Microsoft(R) Word(R) documents.

    Please note that ``encoding`` and ``outencoding`` options are ignored.
    The RTF format is ASCII natively, but handles unicode characters correctly
    thanks to escape sequences.

    .. versionadded:: 0.6

    Additional options accepted:

    `style`
        The style to use, can be a string or a Style subclass (default:
        ``'default'``).

    `fontface`
        The used font family, for example ``Bitstream Vera Sans``. Defaults to
        some generic font which is supposed to have fixed width.

    `fontsize`
        Size of the font used. Size is specified in half points. The
        default is 24 half-points, giving a size 12 font.

        .. versionadded:: 2.0

    `linenos`
        Turn on line numbering (default: ``False``).

        .. versionadded:: 2.18

    `lineno_fontsize`
        Font size for line numbers. Size is specified in half points
        (default: `fontsize`). 

        .. versionadded:: 2.18

    `lineno_padding`
        Number of spaces between the (inline) line numbers and the
        source code (default: ``2``).

        .. versionadded:: 2.18

    `linenostart`
        The line number for the first line (default: ``1``).

        .. versionadded:: 2.18

    `linenostep`
        If set to a number n > 1, only every nth line number is printed.

        .. versionadded:: 2.18

    `lineno_color`
        Color for line numbers specified as a hex triplet, e.g. ``'5e5e5e'``. 
        Defaults to the style's line number color if it is a hex triplet, 
        otherwise ansi bright black.

        .. versionadded:: 2.18

    `hl_lines`
        Specify a list of lines to be highlighted, as line numbers separated by
        spaces, e.g. ``'3 7 8'``. The line numbers are relative to the input 
        (i.e. the first line is line 1) unless `hl_linenostart` is set.

        .. versionadded:: 2.18

    `hl_color`
        Color for highlighting the lines specified in `hl_lines`, specified as 
        a hex triplet (default: style's `highlight_color`).

        .. versionadded:: 2.18

    `hl_linenostart`
        If set to ``True`` line numbers in `hl_lines` are specified
        relative to `linenostart` (default ``False``).

        .. versionadded:: 2.18
    """
    name = 'RTF'
    aliases = ['rtf']
    filenames = ['*.rtf']

    def __init__(self, **options):
        r"""
        Additional options accepted:

        ``fontface``
            Name of the font used. Could for example be ``'Courier New'``
            to further specify the default which is ``'\fmodern'``. The RTF
            specification claims that ``\fmodern`` are "Fixed-pitch serif
            and sans serif fonts". Hope every RTF implementation thinks
            the same about modern...

        """
        Formatter.__init__(self, **options)
        self.fontface = options.get('fontface') or ''
        self.fontsize = get_int_opt(options, 'fontsize', 0)
        self.linenos = get_bool_opt(options, 'linenos', False)
        self.lineno_fontsize = get_int_opt(options, 'lineno_fontsize',
                                           self.fontsize)
        self.lineno_padding = get_int_opt(options, 'lineno_padding', 2)
        self.linenostart = abs(get_int_opt(options, 'linenostart', 1))
        self.linenostep = abs(get_int_opt(options, 'linenostep', 1))
        self.hl_linenostart = get_bool_opt(options, 'hl_linenostart', False)

        self.hl_color = options.get('hl_color', '')
        if not self.hl_color:
            self.hl_color = self.style.highlight_color

        self.hl_lines = []
        for lineno in get_list_opt(options, 'hl_lines', []):
            try:
                lineno = int(lineno)
                if self.hl_linenostart:
                    lineno = lineno - self.linenostart + 1
                self.hl_lines.append(lineno)
            except ValueError:
                pass

        self.lineno_color = options.get('lineno_color', '')
        if not self.lineno_color:
            if  self.style.line_number_color == 'inherit':
                # style color is the css value 'inherit'
                # default to ansi bright-black
                self.lineno_color = _ansimap['ansibrightblack']
            else:
                # style color is assumed to be a hex triplet as other
                # colors in pygments/style.py
                self.lineno_color = self.style.line_number_color

        self.color_mapping = self._create_color_mapping()

    def _escape(self, text):
        return text.replace('\\', '\\\\') \
                   .replace('{', '\\{') \
                   .replace('}', '\\}')

    def _escape_text(self, text):
        # empty strings, should give a small performance improvement
        if not text:
            return ''

        # escape text
        text = self._escape(text)

        buf = []
        for c in text:
            cn = ord(c)
            if cn < (2**7):
                # ASCII character
                buf.append(str(c))
            elif (2**7) <= cn < (2**16):
                # single unicode escape sequence
                buf.append('{\\u%d}' % cn)
            elif (2**16) <= cn:
                # RTF limits unicode to 16 bits.
                # Force surrogate pairs
                buf.append('{\\u%d}{\\u%d}' % surrogatepair(cn))

        return ''.join(buf).replace('\n', '\\par')

    @staticmethod
    def hex_to_rtf_color(hex_color):
        if hex_color[0] == "#":
            hex_color = hex_color[1:]

        return '\\red%d\\green%d\\blue%d;' % (
                        int(hex_color[0:2], 16),
                        int(hex_color[2:4], 16),
                        int(hex_color[4:6], 16)
                    )

    def _split_tokens_on_newlines(self, tokensource):
        """
        Split tokens containing newline characters into multiple token
        each representing a line of the input file. Needed for numbering
        lines of e.g. multiline comments.
        """
        for ttype, value in tokensource:
            if value == '\n':
                yield (ttype, value)
            elif "\n" in value:
                lines = value.split("\n")
                for line in lines[:-1]:
                    yield (ttype, line+"\n")
                if lines[-1]:
                    yield (ttype, lines[-1])
            else:
                yield (ttype, value)

    def _create_color_mapping(self):
        """
        Create a mapping of style hex colors to index/offset in
        the RTF color table.
        """
        color_mapping = OrderedDict()
        offset = 1

        if self.linenos:
            color_mapping[self.lineno_color] = offset
            offset += 1

        if self.hl_lines:
            color_mapping[self.hl_color] = offset
            offset += 1

        for _, style in self.style:
            for color in style['color'], style['bgcolor'], style['border']:
                if color and color not in color_mapping:
                    color_mapping[color] = offset
                    offset += 1

        return color_mapping

    @property
    def _lineno_template(self):
        if self.lineno_fontsize != self.fontsize:
            return '{{\\fs{} \\cf{} %s{}}}'.format(self.lineno_fontsize,
                          self.color_mapping[self.lineno_color],
                          " " * self.lineno_padding)

        return '{{\\cf{} %s{}}}'.format(self.color_mapping[self.lineno_color],
                      " " * self.lineno_padding)

    @property
    def _hl_open_str(self):
        return rf'{{\highlight{self.color_mapping[self.hl_color]} '

    @property
    def _rtf_header(self):
        lines = []
        # rtf 1.8 header
        lines.append('{\\rtf1\\ansi\\uc0\\deff0'
                     '{\\fonttbl{\\f0\\fmodern\\fprq1\\fcharset0%s;}}'
                     % (self.fontface and ' '
                        + self._escape(self.fontface) or ''))

        # color table
        lines.append('{\\colortbl;')
        for color, _ in self.color_mapping.items():
            lines.append(self.hex_to_rtf_color(color))
        lines.append('}')

        # font and fontsize
        lines.append('\\f0\\sa0')
        if self.fontsize:
            lines.append('\\fs%d' % self.fontsize)

        # ensure Libre Office Writer imports and renders consecutive
        # space characters the same width, needed for line numbering.
        # https://bugs.documentfoundation.org/show_bug.cgi?id=144050
        lines.append('\\dntblnsbdb')

        return lines

    def format_unencoded(self, tokensource, outfile):
        for line in self._rtf_header:
            outfile.write(line + "\n")

        tokensource = self._split_tokens_on_newlines(tokensource)

        # first pass of tokens to count lines, needed for line numbering
        if self.linenos:
            line_count = 0
            tokens = [] # for copying the token source generator
            for ttype, value in tokensource:
                tokens.append((ttype, value))
                if value.endswith("\n"):
                    line_count += 1

            # width of line number strings (for padding with spaces)
            linenos_width = len(str(line_count+self.linenostart-1))

            tokensource = tokens

        # highlight stream
        lineno = 1
        start_new_line = True
        for ttype, value in tokensource:
            if start_new_line and lineno in self.hl_lines:
                outfile.write(self._hl_open_str)

            if start_new_line and self.linenos:
                if (lineno-self.linenostart+1)%self.linenostep == 0:
                    current_lineno = lineno + self.linenostart - 1
                    lineno_str = str(current_lineno).rjust(linenos_width)
                else:
                    lineno_str = "".rjust(linenos_width)
                outfile.write(self._lineno_template % lineno_str)

            while not self.style.styles_token(ttype) and ttype.parent:
                ttype = ttype.parent
            style = self.style.style_for_token(ttype)
            buf = []
            if style['bgcolor']:
                buf.append('\\cb%d' % self.color_mapping[style['bgcolor']])
            if style['color']:
                buf.append('\\cf%d' % self.color_mapping[style['color']])
            if style['bold']:
                buf.append('\\b')
            if style['italic']:
                buf.append('\\i')
            if style['underline']:
                buf.append('\\ul')
            if style['border']:
                buf.append('\\chbrdr\\chcfpat%d' %
                           self.color_mapping[style['border']])
            start = ''.join(buf)
            if start:
                outfile.write(f'{{{start} ')
            outfile.write(self._escape_text(value))
            if start:
                outfile.write('}')
            start_new_line = False

            # complete line of input
            if value.endswith("\n"):
                # close line highlighting
                if lineno in self.hl_lines:
                    outfile.write('}')
                # newline in RTF file after closing }
                outfile.write("\n")

                start_new_line = True
                lineno += 1

        outfile.write('}\n')

"""Prepares a distribution for installation
"""

# The following comment should be removed at some point in the future.
# mypy: strict-optional=False

import mimetypes
import os
import shutil
from dataclasses import dataclass
from pathlib import Path
from typing import Dict, Iterable, List, Optional

from pip._vendor.packaging.utils import canonicalize_name

from pip._internal.distributions import make_distribution_for_install_requirement
from pip._internal.distributions.installed import InstalledDistribution
from pip._internal.exceptions import (
    DirectoryUrlHashUnsupported,
    HashMismatch,
    HashUnpinned,
    InstallationError,
    MetadataInconsistent,
    NetworkConnectionError,
    VcsHashUnsupported,
)
from pip._internal.index.package_finder import PackageFinder
from pip._internal.metadata import BaseDistribution, get_metadata_distribution
from pip._internal.models.direct_url import ArchiveInfo
from pip._internal.models.link import Link
from pip._internal.models.wheel import Wheel
from pip._internal.network.download import BatchDownloader, Downloader
from pip._internal.network.lazy_wheel import (
    HTTPRangeRequestUnsupported,
    dist_from_wheel_url,
)
from pip._internal.network.session import PipSession
from pip._internal.operations.build.build_tracker import BuildTracker
from pip._internal.req.req_install import InstallRequirement
from pip._internal.utils._log import getLogger
from pip._internal.utils.direct_url_helpers import (
    direct_url_for_editable,
    direct_url_from_link,
)
from pip._internal.utils.hashes import Hashes, MissingHashes
from pip._internal.utils.logging import indent_log
from pip._internal.utils.misc import (
    display_path,
    hash_file,
    hide_url,
    redact_auth_from_requirement,
)
from pip._internal.utils.temp_dir import TempDirectory
from pip._internal.utils.unpacking import unpack_file
from pip._internal.vcs import vcs

logger = getLogger(__name__)


def _get_prepared_distribution(
    req: InstallRequirement,
    build_tracker: BuildTracker,
    finder: PackageFinder,
    build_isolation: bool,
    check_build_deps: bool,
) -> BaseDistribution:
    """Prepare a distribution for installation."""
    abstract_dist = make_distribution_for_install_requirement(req)
    tracker_id = abstract_dist.build_tracker_id
    if tracker_id is not None:
        with build_tracker.track(req, tracker_id):
            abstract_dist.prepare_distribution_metadata(
                finder, build_isolation, check_build_deps
            )
    return abstract_dist.get_metadata_distribution()


def unpack_vcs_link(link: Link, location: str, verbosity: int) -> None:
    vcs_backend = vcs.get_backend_for_scheme(link.scheme)
    assert vcs_backend is not None
    vcs_backend.unpack(location, url=hide_url(link.url), verbosity=verbosity)


@dataclass
class File:
    path: str
    content_type: Optional[str] = None

    def __post_init__(self) -> None:
        if self.content_type is None:
            self.content_type = mimetypes.guess_type(self.path)[0]


def get_http_url(
    link: Link,
    download: Downloader,
    download_dir: Optional[str] = None,
    hashes: Optional[Hashes] = None,
) -> File:
    temp_dir = TempDirectory(kind="unpack", globally_managed=True)
    # If a download dir is specified, is the file already downloaded there?
    already_downloaded_path = None
    if download_dir:
        already_downloaded_path = _check_download_dir(link, download_dir, hashes)

    if already_downloaded_path:
        from_path = already_downloaded_path
        content_type = None
    else:
        # let's download to a tmp dir
        from_path, content_type = download(link, temp_dir.path)
        if hashes:
            hashes.check_against_path(from_path)

    return File(from_path, content_type)


def get_file_url(
    link: Link, download_dir: Optional[str] = None, hashes: Optional[Hashes] = None
) -> File:
    """Get file and optionally check its hash."""
    # If a download dir is specified, is the file already there and valid?
    already_downloaded_path = None
    if download_dir:
        already_downloaded_path = _check_download_dir(link, download_dir, hashes)

    if already_downloaded_path:
        from_path = already_downloaded_path
    else:
        from_path = link.file_path

    # If --require-hashes is off, `hashes` is either empty, the
    # link's embedded hash, or MissingHashes; it is required to
    # match. If --require-hashes is on, we are satisfied by any
    # hash in `hashes` matching: a URL-based or an option-based
    # one; no internet-sourced hash will be in `hashes`.
    if hashes:
        hashes.check_against_path(from_path)
    return File(from_path, None)


def unpack_url(
    link: Link,
    location: str,
    download: Downloader,
    verbosity: int,
    download_dir: Optional[str] = None,
    hashes: Optional[Hashes] = None,
) -> Optional[File]:
    """Unpack link into location, downloading if required.

    :param hashes: A Hashes object, one of whose embedded hashes must match,
        or HashMismatch will be raised. If the Hashes is empty, no matches are
        required, and unhashable types of requirements (like VCS ones, which
        would ordinarily raise HashUnsupported) are allowed.
    """
    # non-editable vcs urls
    if link.is_vcs:
        unpack_vcs_link(link, location, verbosity=verbosity)
        return None

    assert not link.is_existing_dir()

    # file urls
    if link.is_file:
        file = get_file_url(link, download_dir, hashes=hashes)

    # http urls
    else:
        file = get_http_url(
            link,
            download,
            download_dir,
            hashes=hashes,
        )

    # unpack the archive to the build dir location. even when only downloading
    # archives, they have to be unpacked to parse dependencies, except wheels
    if not link.is_wheel:
        unpack_file(file.path, location, file.content_type)

    return file


def _check_download_dir(
    link: Link,
    download_dir: str,
    hashes: Optional[Hashes],
    warn_on_hash_mismatch: bool = True,
) -> Optional[str]:
    """Check download_dir for previously downloaded file with correct hash
    If a correct file is found return its path else None
    """
    download_path = os.path.join(download_dir, link.filename)

    if not os.path.exists(download_path):
        return None

    # If already downloaded, does its hash match?
    logger.info("File was already downloaded %s", download_path)
    if hashes:
        try:
            hashes.check_against_path(download_path)
        except HashMismatch:
            if warn_on_hash_mismatch:
                logger.warning(
                    "Previously-downloaded file %s has bad hash. Re-downloading.",
                    download_path,
                )
            os.unlink(download_path)
            return None
    return download_path


class RequirementPreparer:
    """Prepares a Requirement"""

    def __init__(
        self,
        build_dir: str,
        download_dir: Optional[str],
        src_dir: str,
        build_isolation: bool,
        check_build_deps: bool,
        build_tracker: BuildTracker,
        session: PipSession,
        progress_bar: str,
        finder: PackageFinder,
        require_hashes: bool,
        use_user_site: bool,
        lazy_wheel: bool,
        verbosity: int,
        legacy_resolver: bool,
    ) -> None:
        super().__init__()

        self.src_dir = src_dir
        self.build_dir = build_dir
        self.build_tracker = build_tracker
        self._session = session
        self._download = Downloader(session, progress_bar)
        self._batch_download = BatchDownloader(session, progress_bar)
        self.finder = finder

        # Where still-packed archives should be written to. If None, they are
        # not saved, and are deleted immediately after unpacking.
        self.download_dir = download_dir

        # Is build isolation allowed?
        self.build_isolation = build_isolation

        # Should check build dependencies?
        self.check_build_deps = check_build_deps

        # Should hash-checking be required?
        self.require_hashes = require_hashes

        # Should install in user site-packages?
        self.use_user_site = use_user_site

        # Should wheels be downloaded lazily?
        self.use_lazy_wheel = lazy_wheel

        # How verbose should underlying tooling be?
        self.verbosity = verbosity

        # Are we using the legacy resolver?
        self.legacy_resolver = legacy_resolver

        # Memoized downloaded files, as mapping of url: path.
        self._downloaded: Dict[str, str] = {}

        # Previous "header" printed for a link-based InstallRequirement
        self._previous_requirement_header = ("", "")

    def _log_preparing_link(self, req: InstallRequirement) -> None:
        """Provide context for the requirement being prepared."""
        if req.link.is_file and not req.is_wheel_from_cache:
            message = "Processing %s"
            information = str(display_path(req.link.file_path))
        else:
            message = "Collecting %s"
            information = redact_auth_from_requirement(req.req) if req.req else str(req)

        # If we used req.req, inject requirement source if available (this
        # would already be included if we used req directly)
        if req.req and req.comes_from:
            if isinstance(req.comes_from, str):
                comes_from: Optional[str] = req.comes_from
            else:
                comes_from = req.comes_from.from_path()
            if comes_from:
                information += f" (from {comes_from})"

        if (message, information) != self._previous_requirement_header:
            self._previous_requirement_header = (message, information)
            logger.info(message, information)

        if req.is_wheel_from_cache:
            with indent_log():
                logger.info("Using cached %s", req.link.filename)

    def _ensure_link_req_src_dir(
        self, req: InstallRequirement, parallel_builds: bool
    ) -> None:
        """Ensure source_dir of a linked InstallRequirement."""
        # Since source_dir is only set for editable requirements.
        if req.link.is_wheel:
            # We don't need to unpack wheels, so no need for a source
            # directory.
            return
        assert req.source_dir is None
        if req.link.is_existing_dir():
            # build local directories in-tree
            req.source_dir = req.link.file_path
            return

        # We always delete unpacked sdists after pip runs.
        req.ensure_has_source_dir(
            self.build_dir,
            autodelete=True,
            parallel_builds=parallel_builds,
        )
        req.ensure_pristine_source_checkout()

    def _get_linked_req_hashes(self, req: InstallRequirement) -> Hashes:
        # By the time this is called, the requirement's link should have
        # been checked so we can tell what kind of requirements req is
        # and raise some more informative errors than otherwise.
        # (For example, we can raise VcsHashUnsupported for a VCS URL
        # rather than HashMissing.)
        if not self.require_hashes:
            return req.hashes(trust_internet=True)

        # We could check these first 2 conditions inside unpack_url
        # and save repetition of conditions, but then we would
        # report less-useful error messages for unhashable
        # requirements, complaining that there's no hash provided.
        if req.link.is_vcs:
            raise VcsHashUnsupported()
        if req.link.is_existing_dir():
            raise DirectoryUrlHashUnsupported()

        # Unpinned packages are asking for trouble when a new version
        # is uploaded.  This isn't a security check, but it saves users
        # a surprising hash mismatch in the future.
        # file:/// URLs aren't pinnable, so don't complain about them
        # not being pinned.
        if not req.is_direct and not req.is_pinned:
            raise HashUnpinned()

        # If known-good hashes are missing for this requirement,
        # shim it with a facade object that will provoke hash
        # computation and then raise a HashMissing exception
        # showing the user what the hash should be.
        return req.hashes(trust_internet=False) or MissingHashes()

    def _fetch_metadata_only(
        self,
        req: InstallRequirement,
    ) -> Optional[BaseDistribution]:
        if self.legacy_resolver:
            logger.debug(
                "Metadata-only fetching is not used in the legacy resolver",
            )
            return None
        if self.require_hashes:
            logger.debug(
                "Metadata-only fetching is not used as hash checking is required",
            )
            return None
        # Try PEP 658 metadata first, then fall back to lazy wheel if unavailable.
        return self._fetch_metadata_using_link_data_attr(
            req
        ) or self._fetch_metadata_using_lazy_wheel(req.link)

    def _fetch_metadata_using_link_data_attr(
        self,
        req: InstallRequirement,
    ) -> Optional[BaseDistribution]:
        """Fetch metadata from the data-dist-info-metadata attribute, if possible."""
        # (1) Get the link to the metadata file, if provided by the backend.
        metadata_link = req.link.metadata_link()
        if metadata_link is None:
            return None
        assert req.req is not None
        logger.verbose(
            "Obtaining dependency information for %s from %s",
            req.req,
            metadata_link,
        )
        # (2) Download the contents of the METADATA file, separate from the dist itself.
        metadata_file = get_http_url(
            metadata_link,
            self._download,
            hashes=metadata_link.as_hashes(),
        )
        with open(metadata_file.path, "rb") as f:
            metadata_contents = f.read()
        # (3) Generate a dist just from those file contents.
        metadata_dist = get_metadata_distribution(
            metadata_contents,
            req.link.filename,
            req.req.name,
        )
        # (4) Ensure the Name: field from the METADATA file matches the name from the
        #     install requirement.
        #
        #     NB: raw_name will fall back to the name from the install requirement if
        #     the Name: field is not present, but it's noted in the raw_name docstring
        #     that that should NEVER happen anyway.
        if canonicalize_name(metadata_dist.raw_name) != canonicalize_name(req.req.name):
            raise MetadataInconsistent(
                req, "Name", req.req.name, metadata_dist.raw_name
            )
        return metadata_dist

    def _fetch_metadata_using_lazy_wheel(
        self,
        link: Link,
    ) -> Optional[BaseDistribution]:
        """Fetch metadata using lazy wheel, if possible."""
        # --use-feature=fast-deps must be provided.
        if not self.use_lazy_wheel:
            return None
        if link.is_file or not link.is_wheel:
            logger.debug(
                "Lazy wheel is not used as %r does not point to a remote wheel",
                link,
            )
            return None

        wheel = Wheel(link.filename)
        name = canonicalize_name(wheel.name)
        logger.info(
            "Obtaining dependency information from %s %s",
            name,
            wheel.version,
        )
        url = link.url.split("#", 1)[0]
        try:
            return dist_from_wheel_url(name, url, self._session)
        except HTTPRangeRequestUnsupported:
            logger.debug("%s does not support range requests", url)
            return None

    def _complete_partial_requirements(
        self,
        partially_downloaded_reqs: Iterable[InstallRequirement],
        parallel_builds: bool = False,
    ) -> None:
        """Download any requirements which were only fetched by metadata."""
        # Download to a temporary directory. These will be copied over as
        # needed for downstream 'download', 'wheel', and 'install' commands.
        temp_dir = TempDirectory(kind="unpack", globally_managed=True).path

        # Map each link to the requirement that owns it. This allows us to set
        # `req.local_file_path` on the appropriate requirement after passing
        # all the links at once into BatchDownloader.
        links_to_fully_download: Dict[Link, InstallRequirement] = {}
        for req in partially_downloaded_reqs:
            assert req.link
            links_to_fully_download[req.link] = req

        batch_download = self._batch_download(
            links_to_fully_download.keys(),
            temp_dir,
        )
        for link, (filepath, _) in batch_download:
            logger.debug("Downloading link %s to %s", link, filepath)
            req = links_to_fully_download[link]
            # Record the downloaded file path so wheel reqs can extract a Distribution
            # in .get_dist().
            req.local_file_path = filepath
            # Record that the file is downloaded so we don't do it again in
            # _prepare_linked_requirement().
            self._downloaded[req.link.url] = filepath

            # If this is an sdist, we need to unpack it after downloading, but the
            # .source_dir won't be set up until we are in _prepare_linked_requirement().
            # Add the downloaded archive to the install requirement to unpack after
            # preparing the source dir.
            if not req.is_wheel:
                req.needs_unpacked_archive(Path(filepath))

        # This step is necessary to ensure all lazy wheels are processed
        # successfully by the 'download', 'wheel', and 'install' commands.
        for req in partially_downloaded_reqs:
            self._prepare_linked_requirement(req, parallel_builds)

    def prepare_linked_requirement(
        self, req: InstallRequirement, parallel_builds: bool = False
    ) -> BaseDistribution:
        """Prepare a requirement to be obtained from req.link."""
        assert req.link
        self._log_preparing_link(req)
        with indent_log():
            # Check if the relevant file is already available
            # in the download directory
            file_path = None
            if self.download_dir is not None and req.link.is_wheel:
                hashes = self._get_linked_req_hashes(req)
                file_path = _check_download_dir(
                    req.link,
                    self.download_dir,
                    hashes,
                    # When a locally built wheel has been found in cache, we don't warn
                    # about re-downloading when the already downloaded wheel hash does
                    # not match. This is because the hash must be checked against the
                    # original link, not the cached link. It that case the already
                    # downloaded file will be removed and re-fetched from cache (which
                    # implies a hash check against the cache entry's origin.json).
                    warn_on_hash_mismatch=not req.is_wheel_from_cache,
                )

            if file_path is not None:
                # The file is already available, so mark it as downloaded
                self._downloaded[req.link.url] = file_path
            else:
                # The file is not available, attempt to fetch only metadata
                metadata_dist = self._fetch_metadata_only(req)
                if metadata_dist is not None:
                    req.needs_more_preparation = True
                    return metadata_dist

            # None of the optimizations worked, fully prepare the requirement
            return self._prepare_linked_requirement(req, parallel_builds)

    def prepare_linked_requirements_more(
        self, reqs: Iterable[InstallRequirement], parallel_builds: bool = False
    ) -> None:
        """Prepare linked requirements more, if needed."""
        reqs = [req for req in reqs if req.needs_more_preparation]
        for req in reqs:
            # Determine if any of these requirements were already downloaded.
            if self.download_dir is not None and req.link.is_wheel:
                hashes = self._get_linked_req_hashes(req)
                file_path = _check_download_dir(req.link, self.download_dir, hashes)
                if file_path is not None:
                    self._downloaded[req.link.url] = file_path
                    req.needs_more_preparation = False

        # Prepare requirements we found were already downloaded for some
        # reason. The other downloads will be completed separately.
        partially_downloaded_reqs: List[InstallRequirement] = []
        for req in reqs:
            if req.needs_more_preparation:
                partially_downloaded_reqs.append(req)
            else:
                self._prepare_linked_requirement(req, parallel_builds)

        # TODO: separate this part out from RequirementPreparer when the v1
        # resolver can be removed!
        self._complete_partial_requirements(
            partially_downloaded_reqs,
            parallel_builds=parallel_builds,
        )

    def _prepare_linked_requirement(
        self, req: InstallRequirement, parallel_builds: bool
    ) -> BaseDistribution:
        assert req.link
        link = req.link

        hashes = self._get_linked_req_hashes(req)

        if hashes and req.is_wheel_from_cache:
            assert req.download_info is not None
            assert link.is_wheel
            assert link.is_file
            # We need to verify hashes, and we have found the requirement in the cache
            # of locally built wheels.
            if (
                isinstance(req.download_info.info, ArchiveInfo)
                and req.download_info.info.hashes
                and hashes.has_one_of(req.download_info.info.hashes)
            ):
                # At this point we know the requirement was built from a hashable source
                # artifact, and we verified that the cache entry's hash of the original
                # artifact matches one of the hashes we expect. We don't verify hashes
                # against the cached wheel, because the wheel is not the original.
                hashes = None
            else:
                logger.warning(
                    "The hashes of the source archive found in cache entry "
                    "don't match, ignoring cached built wheel "
                    "and re-downloading source."
                )
                req.link = req.cached_wheel_source_link
                link = req.link

        self._ensure_link_req_src_dir(req, parallel_builds)

        if link.is_existing_dir():
            local_file = None
        elif link.url not in self._downloaded:
            try:
                local_file = unpack_url(
                    link,
                    req.source_dir,
                    self._download,
                    self.verbosity,
                    self.download_dir,
                    hashes,
                )
            except NetworkConnectionError as exc:
                raise InstallationError(
                    f"Could not install requirement {req} because of HTTP "
                    f"error {exc} for URL {link}"
                )
        else:
            file_path = self._downloaded[link.url]
            if hashes:
                hashes.check_against_path(file_path)
            local_file = File(file_path, content_type=None)

        # If download_info is set, we got it from the wheel cache.
        if req.download_info is None:
            # Editables don't go through this function (see
            # prepare_editable_requirement).
            assert not req.editable
            req.download_info = direct_url_from_link(link, req.source_dir)
            # Make sure we have a hash in download_info. If we got it as part of the
            # URL, it will have been verified and we can rely on it. Otherwise we
            # compute it from the downloaded file.
            # FIXME: https://github.com/pypa/pip/issues/11943
            if (
                isinstance(req.download_info.info, ArchiveInfo)
                and not req.download_info.info.hashes
                and local_file
            ):
                hash = hash_file(local_file.path)[0].hexdigest()
                # We populate info.hash for backward compatibility.
                # This will automatically populate info.hashes.
                req.download_info.info.hash = f"sha256={hash}"

        # For use in later processing,
        # preserve the file path on the requirement.
        if local_file:
            req.local_file_path = local_file.path

        dist = _get_prepared_distribution(
            req,
            self.build_tracker,
            self.finder,
            self.build_isolation,
            self.check_build_deps,
        )
        return dist

    def save_linked_requirement(self, req: InstallRequirement) -> None:
        assert self.download_dir is not None
        assert req.link is not None
        link = req.link
        if link.is_vcs or (link.is_existing_dir() and req.editable):
            # Make a .zip of the source_dir we already created.
            req.archive(self.download_dir)
            return

        if link.is_existing_dir():
            logger.debug(
                "Not copying link to destination directory "
                "since it is a directory: %s",
                link,
            )
            return
        if req.local_file_path is None:
            # No distribution was downloaded for this requirement.
            return

        download_location = os.path.join(self.download_dir, link.filename)
        if not os.path.exists(download_location):
            shutil.copy(req.local_file_path, download_location)
            download_path = display_path(download_location)
            logger.info("Saved %s", download_path)

    def prepare_editable_requirement(
        self,
        req: InstallRequirement,
    ) -> BaseDistribution:
        """Prepare an editable requirement."""
        assert req.editable, "cannot prepare a non-editable req as editable"

        logger.info("Obtaining %s", req)

        with indent_log():
            if self.require_hashes:
                raise InstallationError(
                    f"The editable requirement {req} cannot be installed when "
                    "requiring hashes, because there is no single file to "
                    "hash."
                )
            req.ensure_has_source_dir(self.src_dir)
            req.update_editable()
            assert req.source_dir
            req.download_info = direct_url_for_editable(req.unpacked_source_directory)

            dist = _get_prepared_distribution(
                req,
                self.build_tracker,
                self.finder,
                self.build_isolation,
                self.check_build_deps,
            )

            req.check_if_exists(self.use_user_site)

        return dist

    def prepare_installed_requirement(
        self,
        req: InstallRequirement,
        skip_reason: str,
    ) -> BaseDistribution:
        """Prepare an already-installed requirement."""
        assert req.satisfied_by, "req should have been satisfied but isn't"
        assert skip_reason is not None, (
            "did not get skip reason skipped but req.satisfied_by "
            f"is set to {req.satisfied_by}"
        )
        logger.info(
            "Requirement %s: %s (%s)", skip_reason, req, req.satisfied_by.version
        )
        with indent_log():
            if self.require_hashes:
                logger.debug(
                    "Since it is already installed, we are trusting this "
                    "package without checking its hash. To ensure a "
                    "completely repeatable environment, install into an "
                    "empty virtualenv."
                )
            return InstalledDistribution(req).get_metadata_distribution()

"""Lazy ZIP over HTTP"""

__all__ = ["HTTPRangeRequestUnsupported", "dist_from_wheel_url"]

from bisect import bisect_left, bisect_right
from contextlib import contextmanager
from tempfile import NamedTemporaryFile
from typing import Any, Dict, Generator, List, Optional, Tuple
from zipfile import BadZipFile, ZipFile

from pip._vendor.packaging.utils import canonicalize_name
from pip._vendor.requests.models import CONTENT_CHUNK_SIZE, Response

from pip._internal.metadata import BaseDistribution, MemoryWheel, get_wheel_distribution
from pip._internal.network.session import PipSession
from pip._internal.network.utils import HEADERS, raise_for_status, response_chunks


class HTTPRangeRequestUnsupported(Exception):
    pass


def dist_from_wheel_url(name: str, url: str, session: PipSession) -> BaseDistribution:
    """Return a distribution object from the given wheel URL.

    This uses HTTP range requests to only fetch the portion of the wheel
    containing metadata, just enough for the object to be constructed.
    If such requests are not supported, HTTPRangeRequestUnsupported
    is raised.
    """
    with LazyZipOverHTTP(url, session) as zf:
        # For read-only ZIP files, ZipFile only needs methods read,
        # seek, seekable and tell, not the whole IO protocol.
        wheel = MemoryWheel(zf.name, zf)  # type: ignore
        # After context manager exit, wheel.name
        # is an invalid file by intention.
        return get_wheel_distribution(wheel, canonicalize_name(name))


class LazyZipOverHTTP:
    """File-like object mapped to a ZIP file over HTTP.

    This uses HTTP range requests to lazily fetch the file's content,
    which is supposed to be fed to ZipFile.  If such requests are not
    supported by the server, raise HTTPRangeRequestUnsupported
    during initialization.
    """

    def __init__(
        self, url: str, session: PipSession, chunk_size: int = CONTENT_CHUNK_SIZE
    ) -> None:
        head = session.head(url, headers=HEADERS)
        raise_for_status(head)
        assert head.status_code == 200
        self._session, self._url, self._chunk_size = session, url, chunk_size
        self._length = int(head.headers["Content-Length"])
        self._file = NamedTemporaryFile()
        self.truncate(self._length)
        self._left: List[int] = []
        self._right: List[int] = []
        if "bytes" not in head.headers.get("Accept-Ranges", "none"):
            raise HTTPRangeRequestUnsupported("range request is not supported")
        self._check_zip()

    @property
    def mode(self) -> str:
        """Opening mode, which is always rb."""
        return "rb"

    @property
    def name(self) -> str:
        """Path to the underlying file."""
        return self._file.name

    def seekable(self) -> bool:
        """Return whether random access is supported, which is True."""
        return True

    def close(self) -> None:
        """Close the file."""
        self._file.close()

    @property
    def closed(self) -> bool:
        """Whether the file is closed."""
        return self._file.closed

    def read(self, size: int = -1) -> bytes:
        """Read up to size bytes from the object and return them.

        As a convenience, if size is unspecified or -1,
        all bytes until EOF are returned.  Fewer than
        size bytes may be returned if EOF is reached.
        """
        download_size = max(size, self._chunk_size)
        start, length = self.tell(), self._length
        stop = length if size < 0 else min(start + download_size, length)
        start = max(0, stop - download_size)
        self._download(start, stop - 1)
        return self._file.read(size)

    def readable(self) -> bool:
        """Return whether the file is readable, which is True."""
        return True

    def seek(self, offset: int, whence: int = 0) -> int:
        """Change stream position and return the new absolute position.

        Seek to offset relative position indicated by whence:
        * 0: Start of stream (the default).  pos should be >= 0;
        * 1: Current position - pos may be negative;
        * 2: End of stream - pos usually negative.
        """
        return self._file.seek(offset, whence)

    def tell(self) -> int:
        """Return the current position."""
        return self._file.tell()

    def truncate(self, size: Optional[int] = None) -> int:
        """Resize the stream to the given size in bytes.

        If size is unspecified resize to the current position.
        The current stream position isn't changed.

        Return the new file size.
        """
        return self._file.truncate(size)

    def writable(self) -> bool:
        """Return False."""
        return False

    def __enter__(self) -> "LazyZipOverHTTP":
        self._file.__enter__()
        return self

    def __exit__(self, *exc: Any) -> None:
        self._file.__exit__(*exc)

    @contextmanager
    def _stay(self) -> Generator[None, None, None]:
        """Return a context manager keeping the position.

        At the end of the block, seek back to original position.
        """
        pos = self.tell()
        try:
            yield
        finally:
            self.seek(pos)

    def _check_zip(self) -> None:
        """Check and download until the file is a valid ZIP."""
        end = self._length - 1
        for start in reversed(range(0, end, self._chunk_size)):
            self._download(start, end)
            with self._stay():
                try:
                    # For read-only ZIP files, ZipFile only needs
                    # methods read, seek, seekable and tell.
                    ZipFile(self)
                except BadZipFile:
                    pass
                else:
                    break

    def _stream_response(
        self, start: int, end: int, base_headers: Dict[str, str] = HEADERS
    ) -> Response:
        """Return HTTP response to a range request from start to end."""
        headers = base_headers.copy()
        headers["Range"] = f"bytes={start}-{end}"
        # TODO: Get range requests to be correctly cached
        headers["Cache-Control"] = "no-cache"
        return self._session.get(self._url, headers=headers, stream=True)

    def _merge(
        self, start: int, end: int, left: int, right: int
    ) -> Generator[Tuple[int, int], None, None]:
        """Return a generator of intervals to be fetched.

        Args:
            start (int): Start of needed interval
            end (int): End of needed interval
            left (int): Index of first overlapping downloaded data
            right (int): Index after last overlapping downloaded data
        """
        lslice, rslice = self._left[left:right], self._right[left:right]
        i = start = min([start] + lslice[:1])
        end = max([end] + rslice[-1:])
        for j, k in zip(lslice, rslice):
            if j > i:
                yield i, j - 1
            i = k + 1
        if i <= end:
            yield i, end
        self._left[left:right], self._right[left:right] = [start], [end]

    def _download(self, start: int, end: int) -> None:
        """Download bytes from start to end inclusively."""
        with self._stay():
            left = bisect_left(self._right, start)
            right = bisect_right(self._left, end)
            for start, end in self._merge(start, end, left, right):
                response = self._stream_response(start, end)
                response.raise_for_status()
                self.seek(start)
                for chunk in response_chunks(response, self._chunk_size):
                    self._file.write(chunk)