pungi/pungi/util.py

# -*- coding: utf-8 -*-


# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; version 2 of the License.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU Library General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, see <https://gnu.org/licenses/>.

import argparse
import json
import subprocess
import os
import shutil
import string
import sys
import hashlib
import errno
import re
import contextlib
import traceback
import tempfile
import time
import functools
from six.moves import urllib, range, shlex_quote

import kobo.conf
from kobo.shortcuts import run, force_list
from kobo.threads import WorkerThread, ThreadPool
from productmd.common import get_major_version
from pungi.module_util import Modulemd

# Patterns that match all names of debuginfo packages
DEBUG_PATTERNS = ["*-debuginfo", "*-debuginfo-*", "*-debugsource"]
DEBUG_PATTERN_RE = re.compile(
    r"^(?:.*-debuginfo(?:-.*)?|.*-debuginfo-.*|.*-debugsource)$"
)


def _doRunCommand(
    command,
    logger,
    rundir="/tmp",
    output=subprocess.PIPE,
    error=subprocess.PIPE,
    env=None,
):
    """Run a command and log the output.  Error out if we get something on stderr"""

    logger.info("Running %s" % subprocess.list2cmdline(command))

    p1 = subprocess.Popen(
        command,
        cwd=rundir,
        stdout=output,
        stderr=error,
        universal_newlines=True,
        env=env,
        close_fds=True,
    )
    (out, err) = p1.communicate()

    if out:
        logger.debug(out)

    if p1.returncode != 0:
        logger.error("Got an error from %s" % command[0])
        logger.error(err)
        raise OSError(
            "Got an error (%d) from %s: %s" % (p1.returncode, command[0], err)
        )


def _link(local, target, logger, force=False):
    """Simple function to link or copy a package, removing target optionally."""

    if os.path.exists(target) and force:
        os.remove(target)

    # check for broken links
    if force and os.path.islink(target):
        if not os.path.exists(os.readlink(target)):
            os.remove(target)

    try:
        os.link(local, target)
    except OSError as e:
        if e.errno != 18:  # EXDEV
            logger.error("Got an error linking from cache: %s" % e)
            raise OSError(e)

        # Can't hardlink cross file systems
        shutil.copy2(local, target)


def _ensuredir(target, logger, force=False, clean=False):
    """Ensure that a directory exists, if it already exists, only continue
    if force is set."""

    # We have to check existence of a logger, as setting the logger could
    # itself cause an issue.
    def whoops(func, path, exc_info):
        message = "Could not remove %s" % path
        if logger:
            logger.error(message)
        else:
            sys.stderr(message)
        sys.exit(1)

    if os.path.exists(target) and not os.path.isdir(target):
        message = "%s exists but is not a directory." % target
        if logger:
            logger.error(message)
        else:
            sys.stderr(message)
        sys.exit(1)

    if not os.path.isdir(target):
        os.makedirs(target)
    elif force and clean:
        shutil.rmtree(target, onerror=whoops)
        os.makedirs(target)
    elif force:
        return
    else:
        message = "Directory %s already exists.  Use --force to overwrite." % target
        if logger:
            logger.error(message)
        else:
            sys.stderr(message)
        sys.exit(1)


def _doCheckSum(path, hash, logger):
    """Generate a checksum hash from a provided path.
    Return a string of type:hash"""

    # Try to figure out what hash we want to do
    try:
        sum = hashlib.new(hash)
    except ValueError:
        logger.error("Invalid hash type: %s" % hash)
        return False

    # Try to open the file, using binary flag.
    try:
        myfile = open(path, "rb")
    except IOError as e:
        logger.error("Could not open file %s: %s" % (path, e))
        return False

    # Loop through the file reading chunks at a time as to not
    # put the entire file in memory.  That would suck for DVDs
    while True:
        chunk = myfile.read(
            8192
        )  # magic number!  Taking suggestions for better blocksize
        if not chunk:
            break  # we're done with the file
        sum.update(chunk)
    myfile.close()

    return "%s:%s" % (hash, sum.hexdigest())


def makedirs(path, mode=0o775):
    try:
        os.makedirs(path, mode=mode)
    except OSError as ex:
        if ex.errno != errno.EEXIST:
            raise


def rmtree(path, ignore_errors=False, onerror=None):
    """shutil.rmtree ENOENT (ignoring no such file or directory) errors"""
    try:
        shutil.rmtree(path, ignore_errors, onerror)
    except OSError as ex:
        if ex.errno != errno.ENOENT:
            raise


def explode_rpm_package(pkg_path, target_dir):
    """Explode a rpm package into target_dir."""
    pkg_path = os.path.abspath(pkg_path)
    makedirs(target_dir)
    try:
        # rpm2archive writes to stdout only if reading from stdin, thus the redirect
        run(
            "rpm2archive - <%s | tar xfz - && chmod -R a+rX ." % shlex_quote(pkg_path),
            workdir=target_dir,
        )
    except RuntimeError:
        # Fall back to rpm2cpio in case rpm2archive failed (most likely due to
        # not being present on the system).
        run(
            "rpm2cpio %s | cpio -iuvmd && chmod -R a+rX ." % shlex_quote(pkg_path),
            workdir=target_dir,
        )


def pkg_is_rpm(pkg_obj):
    if pkg_is_srpm(pkg_obj):
        return False
    if pkg_is_debug(pkg_obj):
        return False
    return True


def pkg_is_srpm(pkg_obj):
    if isinstance(pkg_obj, str):
        # string, probably N.A, N-V-R.A, N-V-R.A.rpm
        for i in (".src", ".nosrc", ".src.rpm", ".nosrc.rpm"):
            if pkg_obj.endswith(i):
                return True
    else:
        # package object
        if pkg_obj.arch in ("src", "nosrc"):
            return True
    return False


def pkg_is_debug(pkg_obj):
    if pkg_is_srpm(pkg_obj):
        return False

    if isinstance(pkg_obj, str):
        # string
        name = pkg_obj
    else:
        name = pkg_obj.name

    return DEBUG_PATTERN_RE.match(name)


# format: [(variant_uid_regex, {arch|*: [data]})]
def get_arch_variant_data(conf, var_name, arch, variant, keys=None):
    result = []
    for conf_variant, conf_data in conf.get(var_name, []):
        if variant is not None and not re.match(conf_variant, variant.uid):
            continue
        for conf_arch in conf_data:
            if conf_arch != "*" and conf_arch != arch:
                continue
            if conf_arch == "*" and arch == "src":
                # src is excluded from '*' and needs to be explicitly
                # added to the mapping
                continue
            if keys is not None:
                keys.add(conf_variant)
            if isinstance(conf_data[conf_arch], list):
                result.extend(conf_data[conf_arch])
            else:
                result.append(conf_data[conf_arch])
    return result


def is_arch_multilib(conf, arch):
    """Check if at least one variant has multilib enabled on this variant."""
    return bool(get_arch_variant_data(conf, "multilib", arch, None))


def _get_git_ref(fragment):
    if fragment == "HEAD":
        return fragment
    if fragment.startswith("origin/"):
        branch = fragment.split("/", 1)[1]
        return "refs/heads/" + branch
    return None


class GitUrlResolveError(RuntimeError):
    pass


def resolve_git_ref(repourl, ref):
    """Resolve a reference in a Git repo to a commit.

    Raises RuntimeError if there was an error. Most likely cause is failure to
    run git command.
    """
    if re.match(r"^[a-f0-9]{40}$", ref):
        # This looks like a commit ID already.
        return ref
    try:
        _, output = git_ls_remote(repourl, ref)
    except RuntimeError as e:
        raise GitUrlResolveError(
            "ref does not exist in remote repo %s with the error %s %s"
            % (repourl, e, e.output)
        )

    lines = []
    for line in output.split("\n"):
        # Keep only lines that represent branches and tags, and also a line for
        # currently checked out HEAD. The leading tab is required to
        # distinguish it from HEADs that could exist in remotes.
        if line and ("refs/heads/" in line or "refs/tags/" in line or "\tHEAD" in line):
            lines.append(line)
    if len(lines) == 0:
        # Branch does not exist in remote repo
        raise GitUrlResolveError(
            "Failed to resolve %s: ref does not exist in remote repo" % repourl
        )
    if len(lines) != 1:
        # This should never happen. HEAD can not match multiple commits in a
        # single repo, and there can not be a repo without a HEAD.
        raise GitUrlResolveError("Failed to resolve %r in %s" % (ref, repourl))

    return lines[0].split()[0]


def resolve_git_url(url):
    """Given a url to a Git repo specifying HEAD or origin/<branch> as a ref,
    replace that specifier with actual SHA1 of the commit.

    Otherwise, the original URL will be returned.

    Raises RuntimeError if there was an error. Most likely cause is failure to
    run git command.
    """
    r = urllib.parse.urlsplit(url)
    ref = _get_git_ref(r.fragment)
    if not ref:
        return url

    # Remove git+ prefix from scheme if present. This is for resolving only,
    # the final result must use original scheme.
    scheme = r.scheme.replace("git+", "")

    baseurl = urllib.parse.urlunsplit((scheme, r.netloc, r.path, "", ""))
    fragment = resolve_git_ref(baseurl, ref)

    result = urllib.parse.urlunsplit((r.scheme, r.netloc, r.path, r.query, fragment))
    if "?#" in url:
        # The urllib library drops empty query string. This hack puts it back in.
        result = result.replace("#", "?#")
    return result


class GitUrlResolver(object):
    """A caching resolver for git references. As input it can either take repo
    URL with fragment describing reference, or url and refname. It will return
    either url with changed fragment or just resolved ref.
    """

    def __init__(self, offline=False):
        self.offline = offline
        self.cache = {}

    def __call__(self, url, branch=None):
        if self.offline:
            return branch or url
        key = (url, branch)
        if key not in self.cache:
            try:
                res = resolve_git_ref(url, branch) if branch else resolve_git_url(url)
                self.cache[key] = res
            except GitUrlResolveError as exc:
                self.cache[key] = exc
        if isinstance(self.cache[key], GitUrlResolveError):
            raise self.cache[key]
        return self.cache[key]


# format: {arch|*: [data]}
def get_arch_data(conf, var_name, arch):
    result = []
    for conf_arch, conf_data in conf.get(var_name, {}).items():
        if conf_arch != "*" and conf_arch != arch:
            continue
        if conf_arch == "*" and arch == "src":
            # src is excluded from '*' and needs to be explicitly added to the mapping
            continue
        if isinstance(conf_data, list):
            result.extend(conf_data)
        else:
            result.append(conf_data)
    return result


def get_variant_data(conf, var_name, variant, keys=None):
    """Get configuration for variant.

    Expected config format is a mapping from variant_uid regexes to lists of
    values.

    :param var_name: name of configuration key with which to work
    :param variant: Variant object for which to get configuration
    :param keys: A set to which a used pattern from config will be added (optional)
    :rtype: a list of values
    """
    result = []
    for conf_variant, conf_data in conf.get(var_name, {}).items():
        if not re.match(conf_variant, variant.uid):
            continue
        if keys is not None:
            keys.add(conf_variant)
        if isinstance(conf_data, list):
            result.extend(conf_data)
        else:
            result.append(conf_data)
    return result


def _apply_substitutions(compose, volid):
    substitutions = compose.conf["volume_id_substitutions"].items()
    # processing should start with the longest pattern, otherwise, we could
    # unexpectedly replace a substring of that longest pattern
    for k, v in sorted(substitutions, key=lambda x: len(x[0]), reverse=True):
        volid = volid.replace(k, v)
    return volid


def get_volid(compose, arch, variant=None, disc_type=False, formats=None, **kwargs):
    """Get ISO volume ID for arch and variant"""
    if variant and variant.type == "addon":
        # addons are part of parent variant media
        return None

    if variant and variant.type == "layered-product":
        release_short = variant.release_short
        release_version = variant.release_version
        release_is_layered = True
        base_product_short = compose.conf["release_short"]
        base_product_version = get_major_version(compose.conf["release_version"])
        variant_uid = variant.parent.uid
    else:
        release_short = compose.conf["release_short"]
        release_version = compose.conf["release_version"]
        release_is_layered = (
            True if compose.conf.get("base_product_name", "") else False
        )
        base_product_short = compose.conf.get("base_product_short", "")
        base_product_version = compose.conf.get("base_product_version", "")
        variant_uid = variant and variant.uid or None

    products = compose.conf["image_volid_formats"]
    layered_products = compose.conf["image_volid_layered_product_formats"]

    volid = None
    if release_is_layered:
        all_products = layered_products + products
    else:
        all_products = products
    formats = formats or all_products

    tried = set()
    for i in formats:
        if not variant_uid and "%(variant)s" in i:
            continue
        try:
            args = get_format_substs(
                compose,
                variant=variant_uid,
                release_short=release_short,
                version=release_version,
                arch=arch,
                disc_type=disc_type or "",
                base_product_short=base_product_short,
                base_product_version=base_product_version,
                **kwargs
            )
            volid = (i % args).format(**args)
        except KeyError as err:
            raise RuntimeError(
                "Failed to create volume id: unknown format element: %s" % err
            )
        volid = _apply_substitutions(compose, volid)
        if len(volid) <= 32:
            break
        tried.add(volid)

    if volid and len(volid) > 32:
        raise ValueError(
            "Could not create volume ID longer than 32 bytes, options are %r",
            sorted(tried, key=len),
        )

    if compose.conf["restricted_volid"]:
        # Replace all non-alphanumeric characters and non-underscores) with
        # dashes.
        volid = re.sub(r"\W", "-", volid)

    return volid


def get_mtime(path):
    return int(os.stat(path).st_mtime)


def get_file_size(path):
    return os.path.getsize(path)


def find_old_compose(
    old_compose_dirs,
    release_short,
    release_version,
    release_type_suffix,
    base_product_short=None,
    base_product_version=None,
    allowed_statuses=None,
):
    allowed_statuses = allowed_statuses or ("FINISHED", "FINISHED_INCOMPLETE", "DOOMED")
    composes = []

    def _sortable(compose_id):
        """Convert ID to tuple where respin is an integer for proper sorting."""
        try:
            prefix, respin = compose_id.rsplit(".", 1)
            return (prefix, int(respin))
        except Exception:
            return compose_id

    for compose_dir in force_list(old_compose_dirs):
        if not os.path.isdir(compose_dir):
            continue

        # get all finished composes
        for i in os.listdir(compose_dir):
            # TODO: read .composeinfo

            pattern = "%s-%s%s" % (release_short, release_version, release_type_suffix)
            if base_product_short:
                pattern += "-%s" % base_product_short
            if base_product_version:
                pattern += "-%s" % base_product_version

            if not i.startswith(pattern):
                continue

            suffix = i[len(pattern) :]
            if len(suffix) < 2 or not suffix[1].isdigit():
                # This covers the case where we are looking for -updates, but there
                # is an updates-testing as well.
                continue

            path = os.path.join(compose_dir, i)
            if not os.path.isdir(path):
                continue

            status_path = os.path.join(path, "STATUS")
            if not os.path.isfile(status_path):
                continue

            try:
                with open(status_path, "r") as f:
                    if f.read().strip() in allowed_statuses:
                        composes.append((_sortable(i), os.path.abspath(path)))
            except Exception:
                continue

    if not composes:
        return None

    return sorted(composes)[-1][1]


def process_args(fmt, args):
    """Given a list of arguments, format each value with the format string.

    >>> process_args('--opt=%s', ['foo', 'bar'])
    ['--opt=foo', '--opt=bar']
    """
    return [fmt % val for val in force_list(args or [])]


@contextlib.contextmanager
def failable(
    compose, can_fail, variant, arch, deliverable, subvariant=None, logger=None
):
    """If a deliverable can fail, log a message and go on as if it succeeded."""
    if not logger:
        logger = compose._logger
    if can_fail:
        compose.attempt_deliverable(variant, arch, deliverable, subvariant)
    else:
        compose.require_deliverable(variant, arch, deliverable, subvariant)
    try:
        yield
    except Exception as exc:
        if not can_fail:
            raise
        else:
            log_failed_task(
                compose, variant, arch, deliverable, subvariant, logger=logger, exc=exc
            )


def log_failed_task(
    compose, variant, arch, deliverable, subvariant, logger=None, exc=None
):
    logger = logger or compose._logger
    msg = deliverable.replace("-", " ").capitalize()
    compose.fail_deliverable(variant, arch, deliverable, subvariant)
    ident = "variant %s, arch %s" % (variant.uid if variant else "None", arch)
    if subvariant:
        ident += ", subvariant %s" % subvariant
    logger.error("[FAIL] %s (%s) failed, but going on anyway." % (msg, ident))
    if exc:
        logger.error(str(exc))
        tb = traceback.format_exc()
        logger.debug(tb)


def can_arch_fail(failable_arches, arch):
    """Check if `arch` is in `failable_arches` or `*` can fail."""
    return "*" in failable_arches or arch in failable_arches


def get_format_substs(compose, **kwargs):
    """Return a dict of basic format substitutions.

    Any kwargs will be added as well.
    """
    substs = {
        "compose_id": compose.compose_id,
        "release_short": compose.ci_base.release.short,
        "version": compose.ci_base.release.version,
        "date": compose.compose_date,
        "respin": compose.compose_respin,
        "type": compose.compose_type,
        "type_suffix": compose.compose_type_suffix,
        "label": compose.compose_label,
        "label_major_version": compose.compose_label_major_version,
    }
    substs.update(kwargs)
    return substs


def copy_all(src, dest):
    """
    Copy all files and directories within ``src`` to the ``dest`` directory.

    This is equivalent to running ``cp -r src/* dest``.

    :param src:
        Source directory to copy from.

    :param dest:
        Destination directory to copy to.

    :return:
        A list of relative paths to the files copied.

    Example:
        >>> _copy_all('/tmp/src/', '/tmp/dest/')
        ['file1', 'dir1/file2', 'dir1/subdir/file3']
    """
    contents = os.listdir(src)
    if not contents:
        raise RuntimeError("Source directory %s is empty." % src)
    makedirs(dest)
    for item in contents:
        source = os.path.join(src, item)
        destination = os.path.join(dest, item)
        if os.path.isdir(source):
            shutil.copytree(source, destination)
        else:
            if os.path.islink(source):
                # It's a symlink, we should preserve it instead of resolving.
                os.symlink(os.readlink(source), destination)
            else:
                shutil.copy2(source, destination)

    return recursive_file_list(src)


def move_all(src, dest, rm_src_dir=False):
    """
    Copy all files and directories within ``src`` to the ``dest`` directory.

    This is equivalent to running ``mv src/* dest``.

    :param src:
        Source directory to move from.

    :param dest:
        Destination directory to move to.

    :param rm_src_dir:
        If True, the `src` directory is removed once its content is moved.
    """
    contents = os.listdir(src)
    if not contents:
        raise RuntimeError("Source directory %s is empty." % src)
    makedirs(dest)
    for item in contents:
        source = os.path.join(src, item)
        destination = os.path.join(dest, item)
        shutil.move(source, destination)

    if rm_src_dir:
        os.rmdir(src)


def recursive_file_list(directory):
    """Return a list of files contained in ``directory``.

    The files are paths relative to ``directory``

    :param directory:
        Path to the directory to list.

    Example:
        >>> recursive_file_list('/some/dir')
        ['file1', 'subdir/file2']
    """
    file_list = []
    for root, dirs, files in os.walk(directory):
        file_list += [os.path.relpath(os.path.join(root, f), directory) for f in files]
    return file_list


def levenshtein(a, b):
    """Compute Levenshtein edit distance between two strings."""
    mat = [[0 for _ in range(len(a) + 1)] for _ in range(len(b) + 1)]

    for i in range(len(a) + 1):
        mat[0][i] = i

    for j in range(len(b) + 1):
        mat[j][0] = j

    for j in range(1, len(b) + 1):
        for i in range(1, len(a) + 1):
            cost = 0 if a[i - 1] == b[j - 1] else 1
            mat[j][i] = min(
                mat[j - 1][i] + 1, mat[j][i - 1] + 1, mat[j - 1][i - 1] + cost
            )

    return mat[len(b)][len(a)]


@contextlib.contextmanager
def temp_dir(log=None, *args, **kwargs):
    """Create a temporary directory and ensure it's deleted."""
    if kwargs.get("dir"):
        # If we are supposed to create the temp dir in a particular location,
        # ensure the location already exists.
        makedirs(kwargs["dir"])
    dir = tempfile.mkdtemp(*args, **kwargs)
    try:
        yield dir
    finally:
        try:
            shutil.rmtree(dir)
        except OSError as exc:
            # Okay, we failed to delete temporary dir.
            if log:
                log.warning("Error removing %s: %s", dir, exc.strerror)


def run_unmount_cmd(cmd, max_retries=10, path=None, logger=None):
    """Attempt to run the command to unmount an image.

    If the command fails and stderr complains about device being busy, try
    again. We will do up to ``max_retries`` attempts with increasing pauses.

    If both path and logger are specified, more debugging information will be
    printed in case of failure.
    """
    for i in range(max_retries):
        proc = subprocess.Popen(
            cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, universal_newlines=True
        )
        out, err = proc.communicate()
        if proc.returncode == 0:
            # We were successful
            return
        if "Device or resource busy" not in err:
            raise RuntimeError("Unhandled error when running %r: %r" % (cmd, err))
        time.sleep(i)
    # Still busy, there's something wrong.
    if path and logger:
        commands = [
            ["ls", "-lA", path],
            ["fuser", "-vm", path],
            ["lsof", "+D", path],
        ]
        for c in commands:
            try:
                proc = subprocess.Popen(
                    c,
                    stdout=subprocess.PIPE,
                    stderr=subprocess.STDOUT,
                    universal_newlines=True,
                )
                out, _ = proc.communicate()
                logger.debug(
                    "`%s` exited with %s and following output:\n%s",
                    " ".join(c),
                    proc.returncode,
                    out,
                )
            except OSError:
                logger.debug("`%s` command not available for debugging", " ".join(c))
    raise RuntimeError("Failed to run %r: Device or resource busy." % cmd)


def translate_path_raw(mapping, path):
    normpath = os.path.normpath(path)
    for prefix, newvalue in mapping:
        prefix = os.path.normpath(prefix)
        # Strip trailing slashes: the prefix has them stripped by `normpath`.
        newvalue = newvalue.rstrip("/")
        if normpath.startswith(prefix):
            # We can't call os.path.normpath on result since it is not actually
            # a path - http:// would get changed to  http:/ and so on.
            # Only the first occurrence should be replaced.
            return normpath.replace(prefix, newvalue, 1)
    return normpath


def translate_path(compose, path):
    """
    @param compose - required for access to config
    @param path
    """
    mapping = compose.conf["translate_paths"]
    return translate_path_raw(mapping, path)


def get_repo_url(compose, repo, arch="$basearch"):
    """
    Convert repo to repo URL.

    @param compose - required for access to variants
        special value compose==None determines that method is called during
        OSTreeInstaller phase where variant-type source repository is deprecated
    @param repo - string or a dict which at least contains 'baseurl' key
    @param arch - string to be used as arch in repo url
    """
    if isinstance(repo, dict):
        try:
            repo = repo["baseurl"]
        except KeyError:
            raise RuntimeError("Baseurl is required in repo dict %s" % str(repo))
    if repo.startswith("/"):
        # It's an absolute path, translate it and return it
        return translate_path(compose, repo)
    if "://" not in repo:
        # this is a variant name
        if compose is not None:
            v = compose.all_variants.get(repo)
            if not v:
                raise RuntimeError("There is no variant %s to get repo from." % repo)
        else:
            return None
        repo = translate_path(
            compose, compose.paths.compose.repository(arch, v, create_dir=False)
        )
    return repo


def get_repo_urls(compose, repos, arch="$basearch", logger=None):
    """
    Convert repos to a list of repo URLs.

    @param compose - required for access to variants
    @param repos - list of string or dict, if item is a dict, key 'baseurl' is required
    @param arch - string to be used as arch in repo url
    """
    urls = []
    for repo in repos:
        repo = get_repo_url(compose, repo, arch=arch)
        if repo is None:
            if logger:
                logger.log_warning(
                    "Variant-type source repository is deprecated and will "
                    "be ignored during 'OSTreeInstaller' phase: %s" % (repo)
                )
        else:
            urls.append(repo)
    return urls


def _translate_url_to_repo_id(url):
    """
    Translate url to valid repo id by replacing any invalid char to '_'.
    """
    _REPOID_CHARS = string.ascii_letters + string.digits + "-_.:"
    return "".join([s if s in list(_REPOID_CHARS) else "_" for s in url])


def get_repo_dict(repo):
    """
    Convert repo to a dict of repo options.

    If repo is a string that represents url, set it as 'baseurl' in result dict,
    also generate a repo id/name as 'name' key in result dict.
    If repo is a dict, and if 'name' key is missing in the dict, generate one for it.
    Repo (str or dict) that has not url format is no longer processed.

    @param repo - A string or dict, if it is a dict, key 'baseurl' is required
    """
    repo_dict = {}
    if isinstance(repo, dict):
        url = repo["baseurl"]
        name = repo.get("name", None)
        if "://" in url:
            if name is None:
                name = _translate_url_to_repo_id(url)
        else:
            # url is variant uid - this possibility is now discontinued
            return {}
        repo["name"] = name
        repo["baseurl"] = url
        return repo
    else:
        # repo is normal url or variant uid
        repo_dict = {}
        if "://" in repo:
            repo_dict["name"] = _translate_url_to_repo_id(repo)
            repo_dict["baseurl"] = repo
        else:
            return {}
    return repo_dict


def get_repo_dicts(repos, logger=None):
    """
    Convert repos to a list of repo dicts.

    @param repo - A list of string or dict, if item is a dict, key 'baseurl' is required
    """
    repo_dicts = []
    for repo in repos:
        repo_dict = get_repo_dict(repo)
        if repo_dict == {}:
            if logger:
                logger.log_warning(
                    "Variant-type source repository is deprecated and will "
                    "be ignored during 'OSTree' phase: %s" % (repo)
                )
        else:
            repo_dicts.append(repo_dict)
    return repo_dicts


def version_generator(compose, gen):
    """If ``gen`` is a known generator, create a value. Otherwise return
    the argument value unchanged.
    """
    if gen == "!OSTREE_VERSION_FROM_LABEL_DATE_TYPE_RESPIN":
        return "%s.%s" % (compose.image_version, compose.image_release)
    elif gen == "!RELEASE_FROM_LABEL_DATE_TYPE_RESPIN":
        return compose.image_release
    elif gen == "!RELEASE_FROM_DATE_RESPIN":
        return "%s.%s" % (compose.compose_date, compose.compose_respin)
    elif gen == "!VERSION_FROM_VERSION_DATE_RESPIN":
        return "%s.%s.%s" % (
            compose.ci_base.release.version,
            compose.compose_date,
            compose.compose_respin,
        )
    elif gen == "!VERSION_FROM_VERSION":
        return "%s" % (compose.ci_base.release.version)
    elif gen and gen[0] == "!":
        raise RuntimeError("Unknown version generator '%s'" % gen)
    return gen


def retry(timeout=120, interval=30, wait_on=Exception):
    """A decorator that allows to retry a section of code until success or
    timeout.
    """

    def wrapper(function):
        @functools.wraps(function)
        def inner(*args, **kwargs):
            start = time.time()
            while True:
                try:
                    return function(*args, **kwargs)
                except wait_on:
                    if (time.time() - start) >= timeout:
                        raise  # This re-raises the last exception.
                    time.sleep(interval)

        return inner

    return wrapper


@retry(wait_on=RuntimeError)
def git_ls_remote(baseurl, ref):
    return run(["git", "ls-remote", baseurl, ref], universal_newlines=True)


def get_tz_offset():
    """Return a string describing current local timezone offset."""
    is_dst = time.daylight and time.localtime().tm_isdst > 0
    # We need to negate the value: the values are in seconds west of UTC, but
    # ISO 8601 wants the offset to be negative for times behind UTC (i.e. to
    # the west).
    offset = -(time.altzone if is_dst else time.timezone)
    hours = offset / 3600
    minutes = (offset / 60) % 60
    return "%+03d:%02d" % (hours, minutes)


def parse_koji_event(event):
    """Process event specification. If event looks like a number, it will be
    used as is. If a string is given, it will be interpreted as a path to the
    topdir of another compose, from which an even it will be extracted.
    """
    try:
        return int(event)
    except ValueError:
        pass
    try:
        with open(os.path.join(event, "work/global/koji-event")) as f:
            return json.load(f)["id"]
    except (IOError, OSError, KeyError):
        raise argparse.ArgumentTypeError(
            "%s is not a number or path to compose with valid Koji event" % event
        )


def load_config(file_path, defaults={}):
    """Open and load configuration file form .conf or .json file."""
    conf = kobo.conf.PyConfigParser()
    conf.load_from_dict(defaults)
    if file_path.endswith(".json"):
        with open(file_path) as f:
            conf.load_from_dict(json.load(f))
        conf.opened_files = [file_path]
        conf._open_file = file_path
    else:
        conf.load_from_file(file_path)

    return conf


def _read_single_module_stream(
    file_or_string, compose=None, arch=None, build=None, is_file=True
):
    try:
        mod_index = Modulemd.ModuleIndex.new()
        if is_file:
            mod_index.update_from_file(file_or_string, True)
        else:
            mod_index.update_from_string(file_or_string, True)
        mod_names = mod_index.get_module_names()
        emit_warning = False
        if len(mod_names) > 1:
            emit_warning = True
        mod_streams = mod_index.get_module(mod_names[0]).get_all_streams()
        if len(mod_streams) > 1:
            emit_warning = True
        if emit_warning and compose:
            compose.log_warning(
                "Multiple modules/streams for arch: %s. Build: %s. "
                "Processing first module/stream only.",
                arch,
                build,
            )
        return mod_streams[0]
    except (KeyError, IndexError):
        # There is no modulemd for this arch. This could mean an arch was
        # added to the compose after the module was built. We don't want to
        # process this, let's skip this module.
        if compose:
            compose.log_info("Skipping arch: %s. Build: %s", arch, build)


def read_single_module_stream_from_file(*args, **kwargs):
    return _read_single_module_stream(*args, is_file=True, **kwargs)


def read_single_module_stream_from_string(*args, **kwargs):
    return _read_single_module_stream(*args, is_file=False, **kwargs)


@contextlib.contextmanager
def as_local_file(url):
    """If URL points to a file over HTTP, the file will be downloaded locally
    and a path to the local copy is yielded. For local files the original path
    is returned.
    """
    if url.startswith("http://") or url.startswith("https://"):
        local_filename, _ = urllib.request.urlretrieve(url)
        try:
            yield local_filename
        finally:
            os.remove(local_filename)
    elif url.startswith("file://"):
        yield url[7:]
    else:
        # Not a remote url, return unchanged.
        yield url


class PartialFuncWorkerThread(WorkerThread):
    """
    Worker thread executing partial_func and storing results
    in the PartialFuncThreadPool.
    """

    def process(self, partial_func, num):
        self.pool._results.append(partial_func())


class PartialFuncThreadPool(ThreadPool):
    """
    Thread pool for PartialFuncWorkerThread threads.

    Example:

        # Execute `pow` in one thread and print result.
        pool = PartialFuncThreadPool()
        pool.add(PartialFuncWorkerThread(pool))
        pool.queue_put(functools.partial(pow, 323, 1235))
        pool.start()
        pool.stop()
        print(pool.results)
    """

    def __init__(self, logger=None):
        ThreadPool.__init__(self, logger)
        self._results = []

    @property
    def results(self):
        return self._results


def read_json_file(file_path):
    """A helper function to read a JSON file."""
    with open(file_path) as f:
        return json.load(f)