Quelle  util.py   Sprache: Python

#
# Copyright (C) 2012-2023 The Python Software Foundation.
# See LICENSE.txt and CONTRIBUTORS.txt.
#
import codecs
from collections import deque
import contextlib
import csv
from glob import iglob as std_iglob
import io
import json
import logging
import os
import py_compile
import re
import socket
try:
    import ssl
except ImportError:  # pragma: no cover
    ssl = None
import subprocess
import sys
import tarfile
import tempfile
import textwrap

try:
    import threading
except ImportError:  # pragma: no cover
    import dummy_threading as threading
import time

from . import DistlibException
from .compat import (string_types, text_type, shutil, raw_input, StringIO,
                     cache_from_source, urlopen, urljoin, httplib, xmlrpclib,
                     HTTPHandler, BaseConfigurator, valid_ident,
                     Container, configparser, URLError, ZipFile, fsdecode,
                     unquote, urlparse)

logger = logging.getLogger(__name__)

#
# Requirement parsing code as per PEP 508
#

IDENTIFIER = re.compile(r'^([\w\.-]+)\s*')
VERSION_IDENTIFIER = re.compile(r'^([\w\.*+-]+)\s*')
COMPARE_OP = re.compile(r'^(<=?|>=?|={2,3}|[~!]=)\s*')
MARKER_OP = re.compile(r'^((<=?)|(>=?)|={2,3}|[~!]=|in|not\s+in)\s*')
OR = re.compile(r'^or\b\s*')
AND = re.compile(r'^and\b\s*')
NON_SPACE = re.compile(r'(\S+)\s*')
STRING_CHUNK = re.compile(r'([\s\w\.{}()*+#:;,/?!~`@$%^&=|<>\[\]-]+)')


def parse_marker(marker_string):
    """
    Parse a marker string and return a dictionary containing a marker expression.

    The dictionary will contain keys "op", "lhs" and "rhs" for non-terminals in
    the expression grammar, or strings. A string contained in quotes is to be
    interpreted as a literal string, and a string not contained in quotes is a
    variable (such as os_name).
    """

    def marker_var(remaining):
        # either identifier, or literal string
        m = IDENTIFIER.match(remaining)
        if m:
            result = m.groups()[0]
            remaining = remaining[m.end():]
        elif not remaining:
            raise SyntaxError('unexpected end of input')
        else:
            q = remaining[0]
            if q not in '\'"':
                raise SyntaxError('invalid expression: %s' % remaining)
            oq = '\'"'.replace(q, '')
            remaining = remaining[1:]
            parts = [q]
            while remaining:
                # either a string chunk, or oq, or q to terminate
                if remaining[0] == q:
                    break
                elif remaining[0] == oq:
                    parts.append(oq)
                    remaining = remaining[1:]
                else:
                    m = STRING_CHUNK.match(remaining)
                    if not m:
                        raise SyntaxError('error in string literal: %s' %
                                          remaining)
                    parts.append(m.groups()[0])
                    remaining = remaining[m.end():]
            else:
                s = ''.join(parts)
                raise SyntaxError('unterminated string: %s' % s)
            parts.append(q)
            result = ''.join(parts)
            remaining = remaining[1:].lstrip()  # skip past closing quote
        return result, remaining

    def marker_expr(remaining):
        if remaining and remaining[0] == '(':
            result, remaining = marker(remaining[1:].lstrip())
            if remaining[0] != ')':
                raise SyntaxError('unterminated parenthesis: %s' % remaining)
            remaining = remaining[1:].lstrip()
        else:
            lhs, remaining = marker_var(remaining)
            while remaining:
                m = MARKER_OP.match(remaining)
                if not m:
                    break
                op = m.groups()[0]
                remaining = remaining[m.end():]
                rhs, remaining = marker_var(remaining)
                lhs = {'op': op, 'lhs': lhs, 'rhs': rhs}
            result = lhs
        return result, remaining

    def marker_and(remaining):
        lhs, remaining = marker_expr(remaining)
        while remaining:
            m = AND.match(remaining)
            if not m:
                break
            remaining = remaining[m.end():]
            rhs, remaining = marker_expr(remaining)
            lhs = {'op': 'and', 'lhs': lhs, 'rhs': rhs}
        return lhs, remaining

    def marker(remaining):
        lhs, remaining = marker_and(remaining)
        while remaining:
            m = OR.match(remaining)
            if not m:
                break
            remaining = remaining[m.end():]
            rhs, remaining = marker_and(remaining)
            lhs = {'op': 'or', 'lhs': lhs, 'rhs': rhs}
        return lhs, remaining

    return marker(marker_string)


def parse_requirement(req):
    """
    Parse a requirement passed in as a string. Return a Container
    whose attributes contain the various parts of the requirement.
    """
    remaining = req.strip()
    if not remaining or remaining.startswith('#'):
        return None
    m = IDENTIFIER.match(remaining)
    if not m:
        raise SyntaxError('name expected: %s' % remaining)
    distname = m.groups()[0]
    remaining = remaining[m.end():]
    extras = mark_expr = versions = uri = None
    if remaining and remaining[0] == '[':
        i = remaining.find(']', 1)
        if i < 0:
            raise SyntaxError('unterminated extra: %s' % remaining)
        s = remaining[1:i]
        remaining = remaining[i + 1:].lstrip()
        extras = []
        while s:
            m = IDENTIFIER.match(s)
            if not m:
                raise SyntaxError('malformed extra: %s' % s)
            extras.append(m.groups()[0])
            s = s[m.end():]
            if not s:
                break
            if s[0] != ',':
                raise SyntaxError('comma expected in extras: %s' % s)
            s = s[1:].lstrip()
        if not extras:
            extras = None
    if remaining:
        if remaining[0] == '@':
            # it's a URI
            remaining = remaining[1:].lstrip()
            m = NON_SPACE.match(remaining)
            if not m:
                raise SyntaxError('invalid URI: %s' % remaining)
            uri = m.groups()[0]
            t = urlparse(uri)
            # there are issues with Python and URL parsing, so this test
            # is a bit crude. See bpo-20271, bpo-23505. Python doesn't
            # always parse invalid URLs correctly - it should raise
            # exceptions for malformed URLs
            if not (t.scheme and t.netloc):
                raise SyntaxError('Invalid URL: %s' % uri)
            remaining = remaining[m.end():].lstrip()
        else:

            def get_versions(ver_remaining):
                """
                Return a list of operator, version tuples if any are
                specified, else None.
                """
                m = COMPARE_OP.match(ver_remaining)
                versions = None
                if m:
                    versions = []
                    while True:
                        op = m.groups()[0]
                        ver_remaining = ver_remaining[m.end():]
                        m = VERSION_IDENTIFIER.match(ver_remaining)
                        if not m:
                            raise SyntaxError('invalid version: %s' %
                                              ver_remaining)
                        v = m.groups()[0]
                        versions.append((op, v))
                        ver_remaining = ver_remaining[m.end():]
                        if not ver_remaining or ver_remaining[0] != ',':
                            break
                        ver_remaining = ver_remaining[1:].lstrip()
                        # Some packages have a trailing comma which would break things
                        # See issue #148
                        if not ver_remaining:
                            break
                        m = COMPARE_OP.match(ver_remaining)
                        if not m:
                            raise SyntaxError('invalid constraint: %s' %
                                              ver_remaining)
                    if not versions:
                        versions = None
                return versions, ver_remaining

            if remaining[0] != '(':
                versions, remaining = get_versions(remaining)
            else:
                i = remaining.find(')', 1)
                if i < 0:
                    raise SyntaxError('unterminated parenthesis: %s' %
                                      remaining)
                s = remaining[1:i]
                remaining = remaining[i + 1:].lstrip()
                # As a special diversion from PEP 508, allow a version number
                # a.b.c in parentheses as a synonym for ~= a.b.c (because this
                # is allowed in earlier PEPs)
                if COMPARE_OP.match(s):
                    versions, _ = get_versions(s)
                else:
                    m = VERSION_IDENTIFIER.match(s)
                    if not m:
                        raise SyntaxError('invalid constraint: %s' % s)
                    v = m.groups()[0]
                    s = s[m.end():].lstrip()
                    if s:
                        raise SyntaxError('invalid constraint: %s' % s)
                    versions = [('~=', v)]

    if remaining:
        if remaining[0] != ';':
            raise SyntaxError('invalid requirement: %s' % remaining)
        remaining = remaining[1:].lstrip()

        mark_expr, remaining = parse_marker(remaining)

    if remaining and remaining[0] != '#':
        raise SyntaxError('unexpected trailing data: %s' % remaining)

    if not versions:
        rs = distname
    else:
        rs = '%s %s' % (distname, ', '.join(
            ['%s %s' % con for con in versions]))
    return Container(name=distname,
                     extras=extras,
                     constraints=versions,
                     marker=mark_expr,
                     url=uri,
                     requirement=rs)


def get_resources_dests(resources_root, rules):
    """Find destinations for resources files"""

    def get_rel_path(root, path):
        # normalizes and returns a lstripped-/-separated path
        root = root.replace(os.path.sep, '/')
        path = path.replace(os.path.sep, '/')
        assert path.startswith(root)
        return path[len(root):].lstrip('/')

    destinations = {}
    for base, suffix, dest in rules:
        prefix = os.path.join(resources_root, base)
        for abs_base in iglob(prefix):
            abs_glob = os.path.join(abs_base, suffix)
            for abs_path in iglob(abs_glob):
                resource_file = get_rel_path(resources_root, abs_path)
                if dest is None:  # remove the entry if it was here
                    destinations.pop(resource_file, None)
                else:
                    rel_path = get_rel_path(abs_base, abs_path)
                    rel_dest = dest.replace(os.path.sep, '/').rstrip('/')
                    destinations[resource_file] = rel_dest + '/' + rel_path
    return destinations


def in_venv():
    if hasattr(sys, 'real_prefix'):
        # virtualenv venvs
        result = True
    else:
        # PEP 405 venvs
        result = sys.prefix != getattr(sys, 'base_prefix', sys.prefix)
    return result


def get_executable():
    # The __PYVENV_LAUNCHER__ dance is apparently no longer needed, as
    # changes to the stub launcher mean that sys.executable always points
    # to the stub on OS X
    #    if sys.platform == 'darwin' and ('__PYVENV_LAUNCHER__'
    #                                     in os.environ):
    #        result =  os.environ['__PYVENV_LAUNCHER__']
    #    else:
    #        result = sys.executable
    #    return result
    # Avoid normcasing: see issue #143
    # result = os.path.normcase(sys.executable)
    result = sys.executable
    if not isinstance(result, text_type):
        result = fsdecode(result)
    return result


def proceed(prompt, allowed_chars, error_prompt=None, default=None):
    p = prompt
    while True:
        s = raw_input(p)
        p = prompt
        if not s and default:
            s = default
        if s:
            c = s[0].lower()
            if c in allowed_chars:
                break
            if error_prompt:
                p = '%c: %s\n%s' % (c, error_prompt, prompt)
    return c


def extract_by_key(d, keys):
    if isinstance(keys, string_types):
        keys = keys.split()
    result = {}
    for key in keys:
        if key in d:
            result[key] = d[key]
    return result


def read_exports(stream):
    if sys.version_info[0] >= 3:
        # needs to be a text stream
        stream = codecs.getreader('utf-8')(stream)
    # Try to load as JSON, falling back on legacy format
    data = stream.read()
    stream = StringIO(data)
    try:
        jdata = json.load(stream)
        result = jdata['extensions']['python.exports']['exports']
        for group, entries in result.items():
            for k, v in entries.items():
                s = '%s = %s' % (k, v)
                entry = get_export_entry(s)
                assert entry is not None
                entries[k] = entry
        return result
    except Exception:
        stream.seek(0, 0)

    def read_stream(cp, stream):
        if hasattr(cp, 'read_file'):
            cp.read_file(stream)
        else:
            cp.readfp(stream)

    cp = configparser.ConfigParser()
    try:
        read_stream(cp, stream)
    except configparser.MissingSectionHeaderError:
        stream.close()
        data = textwrap.dedent(data)
        stream = StringIO(data)
        read_stream(cp, stream)

    result = {}
    for key in cp.sections():
        result[key] = entries = {}
        for name, value in cp.items(key):
            s = '%s = %s' % (name, value)
            entry = get_export_entry(s)
            assert entry is not None
            # entry.dist = self
            entries[name] = entry
    return result


def write_exports(exports, stream):
    if sys.version_info[0] >= 3:
        # needs to be a text stream
        stream = codecs.getwriter('utf-8')(stream)
    cp = configparser.ConfigParser()
    for k, v in exports.items():
        # TODO check k, v for valid values
        cp.add_section(k)
        for entry in v.values():
            if entry.suffix is None:
                s = entry.prefix
            else:
                s = '%s:%s' % (entry.prefix, entry.suffix)
            if entry.flags:
                s = '%s [%s]' % (s, ', '.join(entry.flags))
            cp.set(k, entry.name, s)
    cp.write(stream)


@contextlib.contextmanager
def tempdir():
    td = tempfile.mkdtemp()
    try:
        yield td
    finally:
        shutil.rmtree(td)


@contextlib.contextmanager
def chdir(d):
    cwd = os.getcwd()
    try:
        os.chdir(d)
        yield
    finally:
        os.chdir(cwd)


@contextlib.contextmanager
def socket_timeout(seconds=15):
    cto = socket.getdefaulttimeout()
    try:
        socket.setdefaulttimeout(seconds)
        yield
    finally:
        socket.setdefaulttimeout(cto)


class cached_property(object):

    def __init__(self, func):
        self.func = func
        # for attr in ('__name__', '__module__', '__doc__'):
        #     setattr(self, attr, getattr(func, attr, None))

    def __get__(self, obj, cls=None):
        if obj is None:
            return self
        value = self.func(obj)
        object.__setattr__(obj, self.func.__name__, value)
        # obj.__dict__[self.func.__name__] = value = self.func(obj)
        return value


def convert_path(pathname):
    """Return 'pathname' as a name that will work on the native filesystem.

    The path is split on '/' and put back together again using the current
    directory separator.  Needed because filenames in the setup script are
    always supplied in Unix style, and have to be converted to the local
    convention before we can actually use them in the filesystem.  Raises
    ValueError on non-Unix-ish systems if 'pathname' either starts or
    ends with a slash.
    """
    if os.sep == '/':
        return pathname
    if not pathname:
        return pathname
    if pathname[0] == '/':
        raise ValueError("path '%s' cannot be absolute" % pathname)
    if pathname[-1] == '/':
        raise ValueError("path '%s' cannot end with '/'" % pathname)

    paths = pathname.split('/')
    while os.curdir in paths:
        paths.remove(os.curdir)
    if not paths:
        return os.curdir
    return os.path.join(*paths)


class FileOperator(object):

    def __init__(self, dry_run=False):
        self.dry_run = dry_run
        self.ensured = set()
        self._init_record()

    def _init_record(self):
        self.record = False
        self.files_written = set()
        self.dirs_created = set()

    def record_as_written(self, path):
        if self.record:
            self.files_written.add(path)

    def newer(self, source, target):
        """Tell if the target is newer than the source.

        Returns true if 'source' exists and is more recently modified than
        'target', or if 'source' exists and 'target' doesn't.

        Returns false if both exist and 'target' is the same age or younger
        than 'source'. Raise PackagingFileError if 'source' does not exist.

        Note that this test is not very accurate: files created in the same
        second will have the same "age".
        """
        if not os.path.exists(source):
            raise DistlibException("file '%r' does not exist" %
                                   os.path.abspath(source))
        if not os.path.exists(target):
            return True

        return os.stat(source).st_mtime > os.stat(target).st_mtime

    def copy_file(self, infile, outfile, check=True):
        """Copy a file respecting dry-run and force flags.
        """
        self.ensure_dir(os.path.dirname(outfile))
        logger.info('Copying %s to %s', infile, outfile)
        if not self.dry_run:
            msg = None
            if check:
                if os.path.islink(outfile):
                    msg = '%s is a symlink' % outfile
                elif os.path.exists(outfile) and not os.path.isfile(outfile):
                    msg = '%s is a non-regular file' % outfile
            if msg:
                raise ValueError(msg + ' which would be overwritten')
            shutil.copyfile(infile, outfile)
        self.record_as_written(outfile)

    def copy_stream(self, instream, outfile, encoding=None):
        assert not os.path.isdir(outfile)
        self.ensure_dir(os.path.dirname(outfile))
        logger.info('Copying stream %s to %s', instream, outfile)
        if not self.dry_run:
            if encoding is None:
                outstream = open(outfile, 'wb')
            else:
                outstream = codecs.open(outfile, 'w', encoding=encoding)
            try:
                shutil.copyfileobj(instream, outstream)
            finally:
                outstream.close()
        self.record_as_written(outfile)

    def write_binary_file(self, path, data):
        self.ensure_dir(os.path.dirname(path))
        if not self.dry_run:
            if os.path.exists(path):
                os.remove(path)
            with open(path, 'wb') as f:
                f.write(data)
        self.record_as_written(path)

    def write_text_file(self, path, data, encoding):
        self.write_binary_file(path, data.encode(encoding))

    def set_mode(self, bits, mask, files):
        if os.name == 'posix' or (os.name == 'java' and os._name == 'posix'):
            # Set the executable bits (owner, group, and world) on
            # all the files specified.
            for f in files:
                if self.dry_run:
                    logger.info("changing mode of %s", f)
                else:
                    mode = (os.stat(f).st_mode | bits) & mask
                    logger.info("changing mode of %s to %o", f, mode)
                    os.chmod(f, mode)

    set_executable_mode = lambda s, f: s.set_mode(0o555, 0o7777, f)

    def ensure_dir(self, path):
        path = os.path.abspath(path)
        if path not in self.ensured and not os.path.exists(path):
            self.ensured.add(path)
            d, f = os.path.split(path)
            self.ensure_dir(d)
            logger.info('Creating %s' % path)
            if not self.dry_run:
                os.mkdir(path)
            if self.record:
                self.dirs_created.add(path)

    def byte_compile(self,
                     path,
                     optimize=False,
                     force=False,
                     prefix=None,
                     hashed_invalidation=False):
        dpath = cache_from_source(path, not optimize)
        logger.info('Byte-compiling %s to %s', path, dpath)
        if not self.dry_run:
            if force or self.newer(path, dpath):
                if not prefix:
                    diagpath = None
                else:
                    assert path.startswith(prefix)
                    diagpath = path[len(prefix):]
            compile_kwargs = {}
            if hashed_invalidation and hasattr(py_compile,
                                               'PycInvalidationMode'):
                compile_kwargs[
                    'invalidation_mode'] = py_compile.PycInvalidationMode.CHECKED_HASH
            py_compile.compile(path, dpath, diagpath, True,
                               **compile_kwargs)  # raise error
        self.record_as_written(dpath)
        return dpath

    def ensure_removed(self, path):
        if os.path.exists(path):
            if os.path.isdir(path) and not os.path.islink(path):
                logger.debug('Removing directory tree at %s', path)
                if not self.dry_run:
                    shutil.rmtree(path)
                if self.record:
                    if path in self.dirs_created:
                        self.dirs_created.remove(path)
            else:
                if os.path.islink(path):
                    s = 'link'
                else:
                    s = 'file'
                logger.debug('Removing %s %s', s, path)
                if not self.dry_run:
                    os.remove(path)
                if self.record:
                    if path in self.files_written:
                        self.files_written.remove(path)

    def is_writable(self, path):
        result = False
        while not result:
            if os.path.exists(path):
                result = os.access(path, os.W_OK)
                break
            parent = os.path.dirname(path)
            if parent == path:
                break
            path = parent
        return result

    def commit(self):
        """
        Commit recorded changes, turn off recording, return
        changes.
        """
        assert self.record
        result = self.files_written, self.dirs_created
        self._init_record()
        return result

    def rollback(self):
        if not self.dry_run:
            for f in list(self.files_written):
                if os.path.exists(f):
                    os.remove(f)
            # dirs should all be empty now, except perhaps for
            # __pycache__ subdirs
            # reverse so that subdirs appear before their parents
            dirs = sorted(self.dirs_created, reverse=True)
            for d in dirs:
                flist = os.listdir(d)
                if flist:
                    assert flist == ['__pycache__']
                    sd = os.path.join(d, flist[0])
                    os.rmdir(sd)
                os.rmdir(d)  # should fail if non-empty
        self._init_record()


def resolve(module_name, dotted_path):
    if module_name in sys.modules:
        mod = sys.modules[module_name]
    else:
        mod = __import__(module_name)
    if dotted_path is None:
        result = mod
    else:
        parts = dotted_path.split('.')
        result = getattr(mod, parts.pop(0))
        for p in parts:
            result = getattr(result, p)
    return result


class ExportEntry(object):

    def __init__(self, name, prefix, suffix, flags):
        self.name = name
        self.prefix = prefix
        self.suffix = suffix
        self.flags = flags

    @cached_property
    def value(self):
        return resolve(self.prefix, self.suffix)

    def __repr__(self):  # pragma: no cover
        return '' % (self.name, self.prefix,
                                                self.suffix, self.flags)

    def __eq__(self, other):
        if not isinstance(other, ExportEntry):
            result = False
        else:
            result = (self.name == other.name and self.prefix == other.prefix
                      and self.suffix == other.suffix
                      and self.flags == other.flags)
        return result

    __hash__ = object.__hash__


ENTRY_RE = re.compile(
    r'''(?P([^\[]\S*))
                      \s*=\s*(?P<callable>(\w+)([:\.]\w+)*)
                      \s*(\[\s*(?P<flags>[\w-]+(=\w+)?(,\s*\w+(=\w+)?)*)\s*\])?
                      ''', re.VERBOSE)


def get_export_entry(specification):
    m = ENTRY_RE.search(specification)
    if not m:
        result = None
        if '[' in specification or ']' in specification:
            raise DistlibException("Invalid specification "
                                   "'%s'" % specification)
    else:
        d = m.groupdict()
        name = d['name']
        path = d['callable']
        colons = path.count(':')
        if colons == 0:
            prefix, suffix = path, None
        else:
            if colons != 1:
                raise DistlibException("Invalid specification "
                                       "'%s'" % specification)
            prefix, suffix = path.split(':')
        flags = d['flags']
        if flags is None:
            if '[' in specification or ']' in specification:
                raise DistlibException("Invalid specification "
                                       "'%s'" % specification)
            flags = []
        else:
            flags = [f.strip() for f in flags.split(',')]
        result = ExportEntry(name, prefix, suffix, flags)
    return result


def get_cache_base(suffix=None):
    """
    Return the default base location for distlib caches. If the directory does
    not exist, it is created. Use the suffix provided for the base directory,
    and default to '.distlib' if it isn't provided.

    On Windows, if LOCALAPPDATA is defined in the environment, then it is
    assumed to be a directory, and will be the parent directory of the result.
    On POSIX, and on Windows if LOCALAPPDATA is not defined, the user's home
    directory - using os.expanduser('~') - will be the parent directory of
    the result.

    The result is just the directory '.distlib' in the parent directory as
    determined above, or with the name specified with ``suffix``.
    """
    if suffix is None:
        suffix = '.distlib'
    if os.name == 'nt' and 'LOCALAPPDATA' in os.environ:
        result = os.path.expandvars('$localappdata')
    else:
        # Assume posix, or old Windows
        result = os.path.expanduser('~')
    # we use 'isdir' instead of 'exists', because we want to
    # fail if there's a file with that name
    if os.path.isdir(result):
        usable = os.access(result, os.W_OK)
        if not usable:
            logger.warning('Directory exists but is not writable: %s', result)
    else:
        try:
            os.makedirs(result)
            usable = True
        except OSError:
            logger.warning('Unable to create %s', result, exc_info=True)
            usable = False
    if not usable:
        result = tempfile.mkdtemp()
        logger.warning('Default location unusable, using %s', result)
    return os.path.join(result, suffix)


def path_to_cache_dir(path):
    """
    Convert an absolute path to a directory name for use in a cache.

    The algorithm used is:

    #. On Windows, any ``':'`` in the drive is replaced with ``'---'``.
    #. Any occurrence of ``os.sep`` is replaced with ``'--'``.
    #. ``'.cache'`` is appended.
    """
    d, p = os.path.splitdrive(os.path.abspath(path))
    if d:
        d = d.replace(':', '---')
    p = p.replace(os.sep, '--')
    return d + p + '.cache'


def ensure_slash(s):
    if not s.endswith('/'):
        return s + '/'
    return s


def parse_credentials(netloc):
    username = password = None
    if '@' in netloc:
        prefix, netloc = netloc.rsplit('@', 1)
        if ':' not in prefix:
            username = prefix
        else:
            username, password = prefix.split(':', 1)
    if username:
        username = unquote(username)
    if password:
        password = unquote(password)
    return username, password, netloc


def get_process_umask():
    result = os.umask(0o22)
    os.umask(result)
    return result


def is_string_sequence(seq):
    result = True
    i = None
    for i, s in enumerate(seq):
        if not isinstance(s, string_types):
            result = False
            break
    assert i is not None
    return result


PROJECT_NAME_AND_VERSION = re.compile(
    '([a-z0-9_]+([.-][a-z_][a-z0-9_]*)*)-'
    '([a-z0-9_.+-]+)', re.I)
PYTHON_VERSION = re.compile(r'-py(\d\.?\d?)')


def split_filename(filename, project_name=None):
    """
    Extract name, version, python version from a filename (no extension)

    Return name, version, pyver or None
    """
    result = None
    pyver = None
    filename = unquote(filename).replace(' ', '-')
    m = PYTHON_VERSION.search(filename)
    if m:
        pyver = m.group(1)
        filename = filename[:m.start()]
    if project_name and len(filename) > len(project_name) + 1:
        m = re.match(re.escape(project_name) + r'\b', filename)
        if m:
            n = m.end()
            result = filename[:n], filename[n + 1:], pyver
    if result is None:
        m = PROJECT_NAME_AND_VERSION.match(filename)
        if m:
            result = m.group(1), m.group(3), pyver
    return result


# Allow spaces in name because of legacy dists like "Twisted Core"
NAME_VERSION_RE = re.compile(r'(?P[\w .-]+)\s*'
                             r'\(\s*(?P[^\s)]+)\)$')


def parse_name_and_version(p):
    """
    A utility method used to get name and version from a string.

    From e.g. a Provides-Dist value.

    :param p: A value in a form 'foo (1.0)'
    :return: The name and version as a tuple.
    """
    m = NAME_VERSION_RE.match(p)
    if not m:
        raise DistlibException('Ill-formed name/version string: \'%s\'' % p)
    d = m.groupdict()
    return d['name'].strip().lower(), d['ver']


def get_extras(requested, available):
    result = set()
    requested = set(requested or [])
    available = set(available or [])
    if '*' in requested:
        requested.remove('*')
        result |= available
    for r in requested:
        if r == '-':
            result.add(r)
        elif r.startswith('-'):
            unwanted = r[1:]
            if unwanted not in available:
                logger.warning('undeclared extra: %s' % unwanted)
            if unwanted in result:
                result.remove(unwanted)
        else:
            if r not in available:
                logger.warning('undeclared extra: %s' % r)
            result.add(r)
    return result


#
# Extended metadata functionality
#


def _get_external_data(url):
    result = {}
    try:
        # urlopen might fail if it runs into redirections,
        # because of Python issue #13696. Fixed in locators
        # using a custom redirect handler.
        resp = urlopen(url)
        headers = resp.info()
        ct = headers.get('Content-Type')
        if not ct.startswith('application/json'):
            logger.debug('Unexpected response for JSON request: %s', ct)
        else:
            reader = codecs.getreader('utf-8')(resp)
            # data = reader.read().decode('utf-8')
            # result = json.loads(data)
            result = json.load(reader)
    except Exception as e:
        logger.exception('Failed to get external data for %s: %s', url, e)
    return result


_external_data_base_url = 'https://www.red-dove.com/pypi/projects/'


def get_project_data(name):
    url = '%s/%s/project.json' % (name[0].upper(), name)
    url = urljoin(_external_data_base_url, url)
    result = _get_external_data(url)
    return result


def get_package_data(name, version):
    url = '%s/%s/package-%s.json' % (name[0].upper(), name, version)
    url = urljoin(_external_data_base_url, url)
    return _get_external_data(url)


class Cache(object):
    """
    A class implementing a cache for resources that need to live in the file system
    e.g. shared libraries. This class was moved from resources to here because it
    could be used by other modules, e.g. the wheel module.
    """

    def __init__(self, base):
        """
        Initialise an instance.

        :param base: The base directory where the cache should be located.
        """
        # we use 'isdir' instead of 'exists', because we want to
        # fail if there's a file with that name
        if not os.path.isdir(base):  # pragma: no cover
            os.makedirs(base)
        if (os.stat(base).st_mode & 0o77) != 0:
            logger.warning('Directory \'%s\' is not private', base)
        self.base = os.path.abspath(os.path.normpath(base))

    def prefix_to_dir(self, prefix):
        """
        Converts a resource prefix to a directory name in the cache.
        """
        return path_to_cache_dir(prefix)

    def clear(self):
        """
        Clear the cache.
        """
        not_removed = []
        for fn in os.listdir(self.base):
            fn = os.path.join(self.base, fn)
            try:
                if os.path.islink(fn) or os.path.isfile(fn):
                    os.remove(fn)
                elif os.path.isdir(fn):
                    shutil.rmtree(fn)
            except Exception:
                not_removed.append(fn)
        return not_removed


class EventMixin(object):
    """
    A very simple publish/subscribe system.
    """

    def __init__(self):
        self._subscribers = {}

    def add(self, event, subscriber, append=True):
        """
        Add a subscriber for an event.

        :param event: The name of an event.
        :param subscriber: The subscriber to be added (and called when the
                           event is published).
        :param append: Whether to append or prepend the subscriber to an
                       existing subscriber list for the event.
        """
        subs = self._subscribers
        if event not in subs:
            subs[event] = deque([subscriber])
        else:
            sq = subs[event]
            if append:
                sq.append(subscriber)
            else:
                sq.appendleft(subscriber)

    def remove(self, event, subscriber):
        """
        Remove a subscriber for an event.

        :param event: The name of an event.
        :param subscriber: The subscriber to be removed.
        """
        subs = self._subscribers
        if event not in subs:
            raise ValueError('No subscribers: %r' % event)
        subs[event].remove(subscriber)

    def get_subscribers(self, event):
        """
        Return an iterator for the subscribers for an event.
        :param event: The event to return subscribers for.
        """
        return iter(self._subscribers.get(event, ()))

    def publish(self, event, *args, **kwargs):
        """
        Publish a event and return a list of values returned by its
        subscribers.

        :param event: The event to publish.
        :param args: The positional arguments to pass to the event's
                     subscribers.
        :param kwargs: The keyword arguments to pass to the event's
                       subscribers.
        """
        result = []
        for subscriber in self.get_subscribers(event):
            try:
                value = subscriber(event, *args, **kwargs)
            except Exception:
                logger.exception('Exception during event publication')
                value = None
            result.append(value)
        logger.debug('publish %s: args = %s, kwargs = %s, result = %s', event,
                     args, kwargs, result)
        return result


#
# Simple sequencing
#
class Sequencer(object):

    def __init__(self):
        self._preds = {}
        self._succs = {}
        self._nodes = set()  # nodes with no preds/succs

    def add_node(self, node):
        self._nodes.add(node)

    def remove_node(self, node, edges=False):
        if node in self._nodes:
            self._nodes.remove(node)
        if edges:
            for p in set(self._preds.get(node, ())):
                self.remove(p, node)
            for s in set(self._succs.get(node, ())):
                self.remove(node, s)
            # Remove empties
            for k, v in list(self._preds.items()):
                if not v:
                    del self._preds[k]
            for k, v in list(self._succs.items()):
                if not v:
                    del self._succs[k]

    def add(self, pred, succ):
        assert pred != succ
        self._preds.setdefault(succ, set()).add(pred)
        self._succs.setdefault(pred, set()).add(succ)

    def remove(self, pred, succ):
        assert pred != succ
        try:
            preds = self._preds[succ]
            succs = self._succs[pred]
        except KeyError:  # pragma: no cover
            raise ValueError('%r not a successor of anything' % succ)
        try:
            preds.remove(pred)
            succs.remove(succ)
        except KeyError:  # pragma: no cover
            raise ValueError('%r not a successor of %r' % (succ, pred))

    def is_step(self, step):
        return (step in self._preds or step in self._succs
                or step in self._nodes)

    def get_steps(self, final):
        if not self.is_step(final):
            raise ValueError('Unknown: %r' % final)
        result = []
        todo = []
        seen = set()
        todo.append(final)
        while todo:
            step = todo.pop(0)
            if step in seen:
                # if a step was already seen,
                # move it to the end (so it will appear earlier
                # when reversed on return) ... but not for the
                # final step, as that would be confusing for
                # users
                if step != final:
                    result.remove(step)
                    result.append(step)
            else:
                seen.add(step)
                result.append(step)
                preds = self._preds.get(step, ())
                todo.extend(preds)
        return reversed(result)

    @property
    def strong_connections(self):
        # http://en.wikipedia.org/wiki/Tarjan%27s_strongly_connected_components_algorithm
        index_counter = [0]
        stack = []
        lowlinks = {}
        index = {}
        result = []

        graph = self._succs

        def strongconnect(node):
            # set the depth index for this node to the smallest unused index
            index[node] = index_counter[0]
            lowlinks[node] = index_counter[0]
            index_counter[0] += 1
            stack.append(node)

            # Consider successors
            try:
                successors = graph[node]
            except Exception:
                successors = []
            for successor in successors:
                if successor not in lowlinks:
                    # Successor has not yet been visited
                    strongconnect(successor)
                    lowlinks[node] = min(lowlinks[node], lowlinks[successor])
                elif successor in stack:
                    # the successor is in the stack and hence in the current
                    # strongly connected component (SCC)
                    lowlinks[node] = min(lowlinks[node], index[successor])

            # If `node` is a root node, pop the stack and generate an SCC
            if lowlinks[node] == index[node]:
                connected_component = []

                while True:
                    successor = stack.pop()
                    connected_component.append(successor)
                    if successor == node:
                        break
                component = tuple(connected_component)
                # storing the result
                result.append(component)

        for node in graph:
            if node not in lowlinks:
                strongconnect(node)

        return result

    @property
    def dot(self):
        result = ['digraph G {']
        for succ in self._preds:
            preds = self._preds[succ]
            for pred in preds:
                result.append(' %s -> %s;' % (pred, succ))
        for node in self._nodes:
            result.append(' %s;' % node)
        result.append('}')
        return '\n'.join(result)


#
# Unarchiving functionality for zip, tar, tgz, tbz, whl
#

ARCHIVE_EXTENSIONS = ('.tar.gz', '.tar.bz2', '.tar', '.zip', '.tgz', '.tbz',
                      '.whl')


def unarchive(archive_filename, dest_dir, format=None, check=True):

    def check_path(path):
        if not isinstance(path, text_type):
            path = path.decode('utf-8')
        p = os.path.abspath(os.path.join(dest_dir, path))
        if not p.startswith(dest_dir) or p[plen] != os.sep:
            raise ValueError('path outside destination: %r' % p)

    dest_dir = os.path.abspath(dest_dir)
    plen = len(dest_dir)
    archive = None
    if format is None:
        if archive_filename.endswith(('.zip', '.whl')):
            format = 'zip'
        elif archive_filename.endswith(('.tar.gz', '.tgz')):
            format = 'tgz'
            mode = 'r:gz'
        elif archive_filename.endswith(('.tar.bz2', '.tbz')):
            format = 'tbz'
            mode = 'r:bz2'
        elif archive_filename.endswith('.tar'):
            format = 'tar'
            mode = 'r'
        else:  # pragma: no cover
            raise ValueError('Unknown format for %r' % archive_filename)
    try:
        if format == 'zip':
            archive = ZipFile(archive_filename, 'r')
            if check:
                names = archive.namelist()
                for name in names:
                    check_path(name)
        else:
            archive = tarfile.open(archive_filename, mode)
            if check:
                names = archive.getnames()
                for name in names:
                    check_path(name)
        if format != 'zip' and sys.version_info[0] < 3:
            # See Python issue 17153. If the dest path contains Unicode,
            # tarfile extraction fails on Python 2.x if a member path name
            # contains non-ASCII characters - it leads to an implicit
            # bytes -> unicode conversion using ASCII to decode.
            for tarinfo in archive.getmembers():
                if not isinstance(tarinfo.name, text_type):
                    tarinfo.name = tarinfo.name.decode('utf-8')

        # Limit extraction of dangerous items, if this Python
        # allows it easily. If not, just trust the input.
        # See: https://docs.python.org/3/library/tarfile.html#extraction-filters
        def extraction_filter(member, path):
            """Run tarfile.tar_filter, but raise the expected ValueError"""
            # This is only called if the current Python has tarfile filters
            try:
                return tarfile.tar_filter(member, path)
            except tarfile.FilterError as exc:
                raise ValueError(str(exc))

        archive.extraction_filter = extraction_filter

        archive.extractall(dest_dir)

    finally:
        if archive:
            archive.close()


def zip_dir(directory):
    """zip a directory tree into a BytesIO object"""
    result = io.BytesIO()
    dlen = len(directory)
    with ZipFile(result, "w") as zf:
        for root, dirs, files in os.walk(directory):
            for name in files:
                full = os.path.join(root, name)
                rel = root[dlen:]
                dest = os.path.join(rel, name)
                zf.write(full, dest)
    return result


#
# Simple progress bar
#

UNITS = ('', 'K', 'M', 'G', 'T', 'P')


class Progress(object):
    unknown = 'UNKNOWN'

    def __init__(self, minval=0, maxval=100):
        assert maxval is None or maxval >= minval
        self.min = self.cur = minval
        self.max = maxval
        self.started = None
        self.elapsed = 0
        self.done = False

    def update(self, curval):
        assert self.min <= curval
        assert self.max is None or curval <= self.max
        self.cur = curval
        now = time.time()
        if self.started is None:
            self.started = now
        else:
            self.elapsed = now - self.started

    def increment(self, incr):
        assert incr >= 0
        self.update(self.cur + incr)

    def start(self):
        self.update(self.min)
        return self

    def stop(self):
        if self.max is not None:
            self.update(self.max)
        self.done = True

    @property
    def maximum(self):
        return self.unknown if self.max is None else self.max

    @property
    def percentage(self):
        if self.done:
            result = '100 %'
        elif self.max is None:
            result = ' ?? %'
        else:
            v = 100.0 * (self.cur - self.min) / (self.max - self.min)
            result = '%3d %%' % v
        return result

    def format_duration(self, duration):
        if (duration <= 0) and self.max is None or self.cur == self.min:
            result = '??:??:??'
        # elif duration < 1:
        #     result = '--:--:--'
        else:
            result = time.strftime('%H:%M:%S', time.gmtime(duration))
        return result

    @property
    def ETA(self):
        if self.done:
            prefix = 'Done'
            t = self.elapsed
            # import pdb; pdb.set_trace()
        else:
            prefix = 'ETA '
            if self.max is None:
                t = -1
            elif self.elapsed == 0 or (self.cur == self.min):
                t = 0
            else:
                # import pdb; pdb.set_trace()
                t = float(self.max - self.min)
                t /= self.cur - self.min
                t = (t - 1) * self.elapsed
        return '%s: %s' % (prefix, self.format_duration(t))

    @property
    def speed(self):
        if self.elapsed == 0:
            result = 0.0
        else:
            result = (self.cur - self.min) / self.elapsed
        for unit in UNITS:
            if result < 1000:
                break
            result /= 1000.0
        return '%d %sB/s' % (result, unit)


#
# Glob functionality
#

RICH_GLOB = re.compile(r'\{([^}]*)\}')
_CHECK_RECURSIVE_GLOB = re.compile(r'[^/\\,{]\*\*|\*\*[^/\\,}]')
_CHECK_MISMATCH_SET = re.compile(r'^[^{]*\}|\{[^}]*$')


def iglob(path_glob):
    """Extended globbing function that supports ** and {opt1,opt2,opt3}."""
    if _CHECK_RECURSIVE_GLOB.search(path_glob):
        msg = """invalid glob %r: recursive glob "**" must be used alone"""
        raise ValueError(msg % path_glob)
    if _CHECK_MISMATCH_SET.search(path_glob):
        msg = """invalid glob %r: mismatching set marker '{' or '}'"""
        raise ValueError(msg % path_glob)
    return _iglob(path_glob)


def _iglob(path_glob):
    rich_path_glob = RICH_GLOB.split(path_glob, 1)
    if len(rich_path_glob) > 1:
        assert len(rich_path_glob) == 3, rich_path_glob
        prefix, set, suffix = rich_path_glob
        for item in set.split(','):
            for path in _iglob(''.join((prefix, item, suffix))):
                yield path
    else:
        if '**' not in path_glob:
            for item in std_iglob(path_glob):
                yield item
        else:
            prefix, radical = path_glob.split('**', 1)
            if prefix == '':
                prefix = '.'
            if radical == '':
                radical = '*'
            else:
                # we support both
                radical = radical.lstrip('/')
                radical = radical.lstrip('\\')
            for path, dir, files in os.walk(prefix):
                path = os.path.normpath(path)
                for fn in _iglob(os.path.join(path, radical)):
                    yield fn


if ssl:
    from .compat import (HTTPSHandler as BaseHTTPSHandler, match_hostname,
                         CertificateError)

    #
    # HTTPSConnection which verifies certificates/matches domains
    #

    class HTTPSConnection(httplib.HTTPSConnection):
        ca_certs = None  # set this to the path to the certs file (.pem)
        check_domain = True  # only used if ca_certs is not None

        # noinspection PyPropertyAccess
        def connect(self):
            sock = socket.create_connection((self.host, self.port),
                                            self.timeout)
            if getattr(self, '_tunnel_host', False):
                self.sock = sock
                self._tunnel()

            context = ssl.SSLContext(ssl.PROTOCOL_SSLv23)
            if hasattr(ssl, 'OP_NO_SSLv2'):
                context.options |= ssl.OP_NO_SSLv2
            if getattr(self, 'cert_file', None):
                context.load_cert_chain(self.cert_file, self.key_file)
            kwargs = {}
            if self.ca_certs:
                context.verify_mode = ssl.CERT_REQUIRED
                context.load_verify_locations(cafile=self.ca_certs)
                if getattr(ssl, 'HAS_SNI', False):
                    kwargs['server_hostname'] = self.host

            self.sock = context.wrap_socket(sock, **kwargs)
            if self.ca_certs and self.check_domain:
                try:
                    match_hostname(self.sock.getpeercert(), self.host)
                    logger.debug('Host verified: %s', self.host)
                except CertificateError:  # pragma: no cover
                    self.sock.shutdown(socket.SHUT_RDWR)
                    self.sock.close()
                    raise

    class HTTPSHandler(BaseHTTPSHandler):

        def __init__(self, ca_certs, check_domain=True):
            BaseHTTPSHandler.__init__(self)
            self.ca_certs = ca_certs
            self.check_domain = check_domain

        def _conn_maker(self, *args, **kwargs):
            """
            This is called to create a connection instance. Normally you'd
            pass a connection class to do_open, but it doesn't actually check for
            a class, and just expects a callable. As long as we behave just as a
            constructor would have, we should be OK. If it ever changes so that
            we *must* pass a class, we'll create an UnsafeHTTPSConnection class
            which just sets check_domain to False in the class definition, and
            choose which one to pass to do_open.
            """
            result = HTTPSConnection(*args, **kwargs)
            if self.ca_certs:
                result.ca_certs = self.ca_certs
                result.check_domain = self.check_domain
            return result

        def https_open(self, req):
            try:
                return self.do_open(self._conn_maker, req)
            except URLError as e:
                if 'certificate verify failed' in str(e.reason):
                    raise CertificateError(
                        'Unable to verify server certificate '
                        'for %s' % req.host)
                else:
                    raise

    #
    # To prevent against mixing HTTP traffic with HTTPS (examples: A Man-In-The-
    # Middle proxy using HTTP listens on port 443, or an index mistakenly serves
    # HTML containing a http://xyz link when it should be https://xyz),
    # you can use the following handler class, which does not allow HTTP traffic.
    #
    # It works by inheriting from HTTPHandler - so build_opener won't add a
    # handler for HTTP itself.
    #
    class HTTPSOnlyHandler(HTTPSHandler, HTTPHandler):

        def http_open(self, req):
            raise URLError(
                'Unexpected HTTP request on what should be a secure '
                'connection: %s' % req)


#
# XML-RPC with timeouts
#
class Transport(xmlrpclib.Transport):

    def __init__(self, timeout, use_datetime=0):
        self.timeout = timeout
        xmlrpclib.Transport.__init__(self, use_datetime)

    def make_connection(self, host):
        h, eh, x509 = self.get_host_info(host)
        if not self._connection or host != self._connection[0]:
            self._extra_headers = eh
            self._connection = host, httplib.HTTPConnection(h)
        return self._connection[1]


if ssl:

    class SafeTransport(xmlrpclib.SafeTransport):

        def __init__(self, timeout, use_datetime=0):
            self.timeout = timeout
            xmlrpclib.SafeTransport.__init__(self, use_datetime)

        def make_connection(self, host):
            h, eh, kwargs = self.get_host_info(host)
            if not kwargs:
                kwargs = {}
            kwargs['timeout'] = self.timeout
            if not self._connection or host != self._connection[0]:
                self._extra_headers = eh
                self._connection = host, httplib.HTTPSConnection(
                    h, None, **kwargs)
            return self._connection[1]


class ServerProxy(xmlrpclib.ServerProxy):

    def __init__(self, uri, **kwargs):
        self.timeout = timeout = kwargs.pop('timeout', None)
        # The above classes only come into play if a timeout
        # is specified
        if timeout is not None:
            # scheme = splittype(uri)  # deprecated as of Python 3.8
            scheme = urlparse(uri)[0]
            use_datetime = kwargs.get('use_datetime', 0)
            if scheme == 'https':
                tcls = SafeTransport
            else:
                tcls = Transport
            kwargs['transport'] = t = tcls(timeout, use_datetime=use_datetime)
            self.transport = t
        xmlrpclib.ServerProxy.__init__(self, uri, **kwargs)


#
# CSV functionality. This is provided because on 2.x, the csv module can't
# handle Unicode. However, we need to deal with Unicode in e.g. RECORD files.
#


def _csv_open(fn, mode, **kwargs):
    if sys.version_info[0] < 3:
        mode += 'b'
    else:
        kwargs['newline'] = ''
        # Python 3 determines encoding from locale. Force 'utf-8'
        # file encoding to match other forced utf-8 encoding
        kwargs['encoding'] = 'utf-8'
    return open(fn, mode, **kwargs)


class CSVBase(object):
    defaults = {
        'delimiter': str(','),  # The strs are used because we need native
        'quotechar': str('"'),  # str in the csv API (2.x won't take
        'lineterminator': str('\n')  # Unicode)
    }

    def __enter__(self):
        return self

    def __exit__(self, *exc_info):
        self.stream.close()


class CSVReader(CSVBase):

    def __init__(self, **kwargs):
        if 'stream' in kwargs:
            stream = kwargs['stream']
            if sys.version_info[0] >= 3:
                # needs to be a text stream
                stream = codecs.getreader('utf-8')(stream)
            self.stream = stream
        else:
            self.stream = _csv_open(kwargs['path'], 'r')
        self.reader = csv.reader(self.stream, **self.defaults)

    def __iter__(self):
        return self

    def next(self):
        result = next(self.reader)
        if sys.version_info[0] < 3:
            for i, item in enumerate(result):
                if not isinstance(item, text_type):
                    result[i] = item.decode('utf-8')
        return result

    __next__ = next


class CSVWriter(CSVBase):

    def __init__(self, fn, **kwargs):
        self.stream = _csv_open(fn, 'w')
        self.writer = csv.writer(self.stream, **self.defaults)

    def writerow(self, row):
        if sys.version_info[0] < 3:
            r = []
            for item in row:
                if isinstance(item, text_type):
                    item = item.encode('utf-8')
                r.append(item)
            row = r
        self.writer.writerow(row)


#
#   Configurator functionality
#


class Configurator(BaseConfigurator):

    value_converters = dict(BaseConfigurator.value_converters)
    value_converters['inc'] = 'inc_convert'

    def __init__(self, config, base=None):
        super(Configurator, self).__init__(config)
        self.base = base or os.getcwd()

    def configure_custom(self, config):

        def convert(o):
            if isinstance(o, (list, tuple)):
                result = type(o)([convert(i) for i in o])
            elif isinstance(o, dict):
                if '()' in o:
                    result = self.configure_custom(o)
                else:
                    result = {}
                    for k in o:
                        result[k] = convert(o[k])
            else:
                result = self.convert(o)
            return result

        c = config.pop('()')
        if not callable(c):
            c = self.resolve(c)
        props = config.pop('.', None)
        # Check for valid identifiers
        args = config.pop('[]', ())
        if args:
            args = tuple([convert(o) for o in args])
        items = [(k, convert(config[k])) for k in config if valid_ident(k)]
        kwargs = dict(items)
        result = c(*args, **kwargs)
        if props:
            for n, v in props.items():
                setattr(result, n, convert(v))
        return result

    def __getitem__(self, key):
        result = self.config[key]
        if isinstance(result, dict) and '()' in result:
            self.config[key] = result = self.configure_custom(result)
        return result

    def inc_convert(self, value):
        """Default converter for the inc:// protocol."""
        if not os.path.isabs(value):
            value = os.path.join(self.base, value)
        with codecs.open(value, 'r', encoding='utf-8') as f:
            result = json.load(f)
        return result


class SubprocessMixin(object):
    """
    Mixin for running subprocesses and capturing their output
    """

    def __init__(self, verbose=False, progress=None):
        self.verbose = verbose
        self.progress = progress

    def reader(self, stream, context):
        """
        Read lines from a subprocess' output stream and either pass to a progress
        callable (if specified) or write progress information to sys.stderr.
        """
        progress = self.progress
        verbose = self.verbose
        while True:
            s = stream.readline()
            if not s:
                break
            if progress is not None:
                progress(s, context)
            else:
                if not verbose:
                    sys.stderr.write('.')
                else:
                    sys.stderr.write(s.decode('utf-8'))
                sys.stderr.flush()
        stream.close()

    def run_command(self, cmd, **kwargs):
        p = subprocess.Popen(cmd,
                             stdout=subprocess.PIPE,
                             stderr=subprocess.PIPE,
                             **kwargs)
        t1 = threading.Thread(target=self.reader, args=(p.stdout, 'stdout'))
        t1.start()
        t2 = threading.Thread(target=self.reader, args=(p.stderr, 'stderr'))
        t2.start()
        p.wait()
        t1.join()
        t2.join()
        if self.progress is not None:
            self.progress('done.', 'main')
        elif self.verbose:
            sys.stderr.write('done.\n')
        return p


def normalize_name(name):
    """Normalize a python package name a la PEP 503"""
    # https://www.python.org/dev/peps/pep-0503/#normalized-names
    return re.sub('[-_.]+', '-', name).lower()


# def _get_pypirc_command():
# """
# Get the distutils command for interacting with PyPI configurations.
# :return: the command.
# """
# from distutils.core import Distribution
# from distutils.config import PyPIRCCommand
# d = Distribution()
# return PyPIRCCommand(d)


class PyPIRCFile(object):

    DEFAULT_REPOSITORY = 'https://upload.pypi.org/legacy/'
    DEFAULT_REALM = 'pypi'

    def __init__(self, fn=None, url=None):
        if fn is None:
            fn = os.path.join(os.path.expanduser('~'), '.pypirc')
        self.filename = fn
        self.url = url

    def read(self):
        result = {}

        if os.path.exists(self.filename):
            repository = self.url or self.DEFAULT_REPOSITORY

            config = configparser.RawConfigParser()
            config.read(self.filename)
            sections = config.sections()
            if 'distutils' in sections:
                # let's get the list of servers
                index_servers = config.get('distutils', 'index-servers')
                _servers = [
                    server.strip() for server in index_servers.split('\n')
                    if server.strip() != ''
                ]
                if _servers == []:
                    # nothing set, let's try to get the default pypi
                    if 'pypi' in sections:
                        _servers = ['pypi']
                else:
                    for server in _servers:
                        result = {'server': server}
                        result['username'] = config.get(server, 'username')

                        # optional params
                        for key, default in (('repository',
                                              self.DEFAULT_REPOSITORY),
                                             ('realm', self.DEFAULT_REALM),
                                             ('password', None)):
                            if config.has_option(server, key):
                                result[key] = config.get(server, key)
                            else:
                                result[key] = default

                        # work around people having "repository" for the "pypi"
                        # section of their config set to the HTTP (rather than
                        # HTTPS) URL
                        if (server == 'pypi' and repository
                                in (self.DEFAULT_REPOSITORY, 'pypi')):
                            result['repository'] = self.DEFAULT_REPOSITORY
                        elif (result['server'] != repository
                              and result['repository'] != repository):
                            result = {}
            elif 'server-login' in sections:
                # old format
                server = 'server-login'
                if config.has_option(server, 'repository'):
                    repository = config.get(server, 'repository')
                else:
                    repository = self.DEFAULT_REPOSITORY
                result = {
                    'username': config.get(server, 'username'),
                    'password': config.get(server, 'password'),
                    'repository': repository,
                    'server': server,
                    'realm': self.DEFAULT_REALM
                }
        return result

    def update(self, username, password):
        # import pdb; pdb.set_trace()
        config = configparser.RawConfigParser()
        fn = self.filename
        config.read(fn)
        if not config.has_section('pypi'):
            config.add_section('pypi')
        config.set('pypi', 'username', username)
        config.set('pypi', 'password', password)
        with open(fn, 'w') as f:
            config.write(f)


def _load_pypirc(index):
    """
    Read the PyPI access configuration as supported by distutils.
    """
    return PyPIRCFile(url=index.url).read()


def _store_pypirc(index):
    PyPIRCFile().update(index.username, index.password)


#
# get_platform()/get_host_platform() copied from Python 3.10.a0 source, with some minor
# tweaks
#


def get_host_platform():
    """Return a string that identifies the current platform. This is used mainly to
    distinguish platform-specific build directories and platform-specific built
    distributions.  Typically includes the OS name and version and the
    architecture (as supplied by 'os.uname()'), although the exact information
    included depends on the OS; eg. on Linux, the kernel version isn't
    particularly important.

    Examples of returned values:
       linux-i586
       linux-alpha (?)
       solaris-2.6-sun4u

    Windows will return one of:
       win-amd64 (64bit Windows on AMD64 (aka x86_64, Intel64, EM64T, etc)
       win32 (all others - specifically, sys.platform is returned)

    For other non-POSIX platforms, currently just returns 'sys.platform'.

    """
    if os.name == 'nt':
        if 'amd64' in sys.version.lower():
            return 'win-amd64'
        if '(arm)' in sys.version.lower():
            return 'win-arm32'
        if '(arm64)' in sys.version.lower():
            return 'win-arm64'
        return sys.platform

    # Set for cross builds explicitly
    if "_PYTHON_HOST_PLATFORM" in os.environ:
        return os.environ["_PYTHON_HOST_PLATFORM"]

    if os.name != 'posix' or not hasattr(os, 'uname'):
        # XXX what about the architecture? NT is Intel or Alpha,
        # Mac OS is M68k or PPC, etc.
        return sys.platform

    # Try to distinguish various flavours of Unix

    (osname, host, release, version, machine) = os.uname()

--> --------------------

--> maximum size reached

--> --------------------