from __future__ import annotations

import functools
import re
import sys

from contextlib import suppress
from pathlib import Path
from typing import TYPE_CHECKING
from urllib.parse import unquote
from urllib.parse import urlsplit
from urllib.request import url2pathname

from poetry.core.constraints.version import Version
from poetry.core.constraints.version import VersionRange
from poetry.core.constraints.version import parse_marker_version_constraint
from poetry.core.version.markers import SingleMarker
from poetry.core.version.markers import SingleMarkerLike
from poetry.core.version.markers import dnf


if TYPE_CHECKING:
    from poetry.core.constraints.generic import BaseConstraint
    from poetry.core.constraints.version import VersionConstraint
    from poetry.core.version.markers import BaseMarker

    # Even though we've `from __future__ import annotations`, mypy doesn't seem to like
    # this as `dict[str, ...]`
    ConvertedMarkers = dict[str, list[list[tuple[str, str]]]]


BZ2_EXTENSIONS = (".tar.bz2", ".tbz")
XZ_EXTENSIONS = (".tar.xz", ".txz", ".tlz", ".tar.lz", ".tar.lzma")
ZIP_EXTENSIONS = (".zip", ".whl")
TAR_EXTENSIONS = (".tar.gz", ".tgz", ".tar")
ARCHIVE_EXTENSIONS = ZIP_EXTENSIONS + BZ2_EXTENSIONS + TAR_EXTENSIONS + XZ_EXTENSIONS
SUPPORTED_EXTENSIONS: tuple[str, ...] = ZIP_EXTENSIONS + TAR_EXTENSIONS

with suppress(ImportError):
    import bz2  # noqa: F401

    SUPPORTED_EXTENSIONS += BZ2_EXTENSIONS

with suppress(ImportError):
    # Only for Python 3.3+
    import lzma  # noqa: F401

    SUPPORTED_EXTENSIONS += XZ_EXTENSIONS


def path_to_url(path: str | Path) -> str:
    """
    Convert a path to a file: URL.  The path will be made absolute unless otherwise
    specified and have quoted path parts.
    """
    return Path(path).absolute().as_uri()


def url_to_path(url: str) -> Path:
    """
    Convert an RFC8089 file URI to path.

    The logic used here is borrowed from pip
    https://github.com/pypa/pip/blob/4d1932fcdd1974c820ea60b3286984ebb0c3beaa/src/pip/_internal/utils/urls.py#L31
    """
    if not url.startswith("file:"):
        raise ValueError(f"{url} is not a valid file URI")

    _, netloc, path, _, _ = urlsplit(url)

    if not netloc or netloc == "localhost":
        # According to RFC 8089, same as empty authority.
        netloc = ""
    elif netloc not in {".", ".."} and sys.platform == "win32":
        # If we have a UNC path, prepend UNC share notation.
        netloc = "\\\\" + netloc
    else:
        raise ValueError(
            f"non-local file URIs are not supported on this platform: {url}"
        )

    return Path(url2pathname(netloc + unquote(path)))


def is_url(name: str) -> bool:
    if ":" not in name:
        return False
    scheme = name.split(":", 1)[0].lower()

    return scheme in [
        "http",
        "https",
        "file",
        "ftp",
        "ssh",
        "git",
        "hg",
        "bzr",
        "sftp",
        "svn",
        "ssh",
    ]


def strip_extras(path: str) -> tuple[Path, str | None]:
    m = re.match(r"^(.+)(\[[^\]]+\])$", path)
    extras = None
    if m:
        path_no_extras = m.group(1)
        extras = m.group(2)
    else:
        path_no_extras = path

    return Path(path_no_extras), extras


@functools.cache
def cached_is_dir(path: Path) -> bool:
    """A cached version of `Path.is_dir`."""
    return path.is_dir()


@functools.cache
def is_python_project(path: Path) -> bool:
    """Return true if the directory is a Python project"""
    if not cached_is_dir(path):
        return False

    setup_py = path / "setup.py"
    setup_cfg = path / "setup.cfg"
    setuptools_project = setup_py.exists() or setup_cfg.exists()

    pyproject = (path / "pyproject.toml").exists()

    return pyproject or setuptools_project


def is_archive_file(name: str | Path) -> bool:
    """Return True if `name` is a considered as an archive file."""
    ext = splitext(name)[1].lower()
    return ext in ARCHIVE_EXTENSIONS


def splitext(path: str | Path) -> tuple[str, str]:
    """Like pathlib.Path.stem and suffix, but take off .tar too"""
    if isinstance(path, str):
        path = Path(path)
    base, ext = path.stem, path.suffix
    if base.lower().endswith(".tar"):
        ext = f"{base[-4:]}{ext}"
        base = base[:-4]
    return base, ext


def convert_markers(marker: BaseMarker) -> ConvertedMarkers:
    from poetry.core.version.markers import MarkerUnion
    from poetry.core.version.markers import MultiMarker
    from poetry.core.version.markers import SingleMarker

    requirements: ConvertedMarkers = {}
    marker = dnf(marker)
    conjunctions = marker.markers if isinstance(marker, MarkerUnion) else [marker]
    group_count = len(conjunctions)

    def add_constraint(
        marker_name: str, constraint: tuple[str, str], group_index: int
    ) -> None:
        # python_full_version is equivalent to python_version
        # for Poetry so we merge them
        if marker_name == "python_full_version":
            marker_name = "python_version"
        if marker_name not in requirements:
            requirements[marker_name] = [[] for _ in range(group_count)]
        requirements[marker_name][group_index].append(constraint)

    for i, sub_marker in enumerate(conjunctions):
        if isinstance(sub_marker, MultiMarker):
            for m in sub_marker.markers:
                assert isinstance(m, SingleMarkerLike)
                if isinstance(m, SingleMarker):
                    add_constraint(m.name, (m.operator, m.value), i)
                else:
                    add_constraint(m.name, ("", str(m.constraint)), i)
        elif isinstance(sub_marker, SingleMarkerLike):
            if isinstance(sub_marker, SingleMarker):
                add_constraint(
                    sub_marker.name, (sub_marker.operator, sub_marker.value), i
                )
            else:
                add_constraint(sub_marker.name, ("", str(sub_marker.constraint)), i)

    for group_name in requirements:
        # remove duplicates
        seen = []
        for r in requirements[group_name]:
            if r not in seen:
                seen.append(r)
        requirements[group_name] = seen

    return requirements


def contains_group_without_marker(markers: ConvertedMarkers, marker_name: str) -> bool:
    return marker_name not in markers or [] in markers[marker_name]


def create_nested_marker(
    name: str,
    constraint: BaseConstraint | VersionConstraint,
) -> str:
    from poetry.core.constraints.generic import Constraint
    from poetry.core.constraints.generic import MultiConstraint
    from poetry.core.constraints.generic import UnionConstraint
    from poetry.core.constraints.version import VersionUnion

    if constraint.is_any():
        return ""

    if isinstance(constraint, (MultiConstraint, UnionConstraint)):
        multi_parts = []
        for c in constraint.constraints:
            multi = isinstance(c, (MultiConstraint, UnionConstraint))
            multi_parts.append((multi, create_nested_marker(name, c)))

        glue = " and "
        if isinstance(constraint, UnionConstraint):
            parts = [f"({part[1]})" if part[0] else part[1] for part in multi_parts]
            glue = " or "
        else:
            parts = [part[1] for part in multi_parts]

        marker = glue.join(parts)
    elif isinstance(constraint, Constraint):
        marker = f'{name} {constraint.operator} "{constraint.value}"'
    elif isinstance(constraint, VersionUnion):
        parts = [create_nested_marker(name, c) for c in constraint.ranges]
        glue = " or "
        parts = [f"({part})" for part in parts]
        marker = glue.join(parts)
    elif isinstance(constraint, Version):
        if name == "python_version" and constraint.precision >= 3:
            name = "python_full_version"

        marker = f'{name} == "{constraint.text}"'
    else:
        assert isinstance(constraint, VersionRange), (
            f"Unexpected constraint of type {type(constraint)}"
        )
        min_name = max_name = name

        parts = []

        # `python_version` is a special case: to keep the constructed marker equivalent
        # to the constraint we need to be careful with the precision.
        #
        # PEP 440 tells us that when we come to make the comparison the release
        # segment will be zero padded: eg "<= 3.10" is equivalent to "<= 3.10.0".
        #
        # But "python_version <= 3.10" is _not_ equivalent to "python_version <= 3.10.0"
        # - see normalize_python_version_markers.
        #
        # A similar issue arises for a constraint like "> 3.6".
        if constraint.min is not None:
            op = ">=" if constraint.include_min else ">"
            version = constraint.min
            if min_name == "python_version" and version.precision >= 3:
                min_name = "python_full_version"

            if (
                min_name == "python_version"
                and not constraint.include_min
                and version.precision < 3
            ):
                padding = ".0" * (3 - version.precision)
                part = f'python_full_version > "{version}{padding}"'
            else:
                part = f'{min_name} {op} "{version.stable}"'

            parts.append(part)

        if constraint.max is not None:
            op = "<=" if constraint.include_max else "<"
            version = constraint.max
            if max_name == "python_version" and version.precision >= 3:
                max_name = "python_full_version"

            if (
                max_name == "python_version"
                and constraint.include_max
                and version.precision < 3
            ):
                padding = ".0" * (3 - version.precision)
                part = f'python_full_version <= "{version}{padding}"'
            else:
                part = f'{max_name} {op} "{version.stable}"'

            parts.append(part)

        marker = " and ".join(parts)

    return marker


def get_python_constraint_from_marker(
    marker: BaseMarker,
) -> VersionConstraint:
    from poetry.core.constraints.version import EmptyConstraint
    from poetry.core.constraints.version import VersionRange

    python_marker = marker.only("python_version", "python_full_version")
    if python_marker.is_any():
        return VersionRange()

    if python_marker.is_empty():
        return EmptyConstraint()

    markers = convert_markers(marker)
    if contains_group_without_marker(markers, "python_version"):
        # groups are in disjunctive normal form (DNF),
        # an empty group means that python_version does not appear in this group,
        # which means that python_version is arbitrary for this group
        return VersionRange()

    python_version_markers = markers["python_version"]
    normalized = normalize_python_version_markers(python_version_markers)
    constraint = parse_marker_version_constraint(normalized)
    return constraint


def normalize_python_version_markers(  # NOSONAR
    disjunction: list[list[tuple[str, str]]],
) -> str:
    ors = []
    for or_ in disjunction:
        ands = []
        for op, version in or_:
            # Expand python version
            if op == "==" and "*" not in version and version.count(".") < 2:
                version = "~" + version
                op = ""

            elif op == "!=" and "*" not in version and version.count(".") < 2:
                version += ".*"

            elif op in ("<=", ">"):
                # Make adjustments on encountering versions with less than full
                # precision.
                #
                # Per PEP-508:
                # python_version <-> '.'.join(platform.python_version_tuple()[:2])
                #
                # So for two digits of precision we make the following adjustments:
                # - `python_version > "x.y"` requires version >= x.(y+1).anything
                # - `python_version <= "x.y"` requires version < x.(y+1).anything
                #
                # Treatment when we see a single digit of precision is less clear: is
                # that even a legitimate marker?
                #
                # Experiment suggests that pip behaviour is essentially to make a
                # lexicographical comparison, for example `python_version > "3"` is
                # satisfied by version 3.anything, whereas `python_version <= "3"` is
                # satisfied only by version 2.anything.
                #
                # We achieve the above by fiddling with the operator and version in the
                # marker.
                parsed_version = Version.parse(version)
                if parsed_version.precision < 3:
                    if op == "<=":
                        op = "<"
                    elif op == ">":
                        op = ">="

                if parsed_version.precision == 2:
                    version = parsed_version.next_minor().text

            elif op in ("in", "not in"):
                versions = []
                for v in SingleMarker.VALUE_SEPARATOR_RE.split(version):
                    split = v.split(".")
                    if len(split) in (1, 2):
                        split.append("*")
                        op_ = "" if op == "in" else "!="
                    else:
                        op_ = "==" if op == "in" else "!="

                    versions.append(op_ + ".".join(split))

                if versions:
                    glue = " || " if op == "in" else ", "
                    ands.append(glue.join(versions))

                continue

            ands.append(f"{op}{version}")

        ors.append(" ".join(ands))

    return " || ".join(ors)