/lib/python2.7/site-packages/sos/plugins/ Current File: //lib/python2.7/site-packages/sos/plugins/__init__.py
# Copyright (C) 2006 Steve Conklin <sconklin@redhat.com>

# This file is part of the sos project: https://github.com/sosreport/sos
#
# This copyrighted material is made available to anyone wishing to use,
# modify, copy, or redistribute it subject to the terms and conditions of
# version 2 of the GNU General Public License.
#
# See the LICENSE file in the source distribution for further information.

""" This exports methods available for use by plugins for sos """

from __future__ import with_statement

from sos.utilities import (sos_get_command_output, import_module, grep,
                           fileobj, tail, is_executable)
import os
import glob
import re
import stat
from time import time
import logging
import fnmatch
import errno

# PYCOMPAT
import six
from six.moves import zip, filter
from datetime import datetime

# FileNotFoundError does not exist in 2.7, so map it to IOError
try:
    FileNotFoundError
except NameError:
    FileNotFoundError = IOError


def _to_u(s):
    if not isinstance(s, six.text_type):
        # Workaround python.six mishandling of strings ending in '\' by
        # adding a single space following any '\' at end-of-line.
        # See Six issue #60.
        if s.endswith('\\'):
            s += " "
        s = six.u(s)
    return s


def regex_findall(regex, fname):
    """Return a list of all non overlapping matches in the string(s)"""
    try:
        with fileobj(fname) as f:
            return re.findall(regex, f.read(), re.MULTILINE)
    except AttributeError:
        return []


def _mangle_command(command, name_max):
    mangledname = re.sub(r"^/(usr/|)(bin|sbin)/", "", command)
    mangledname = re.sub(r"[^\w\-\.\/]+", "_", mangledname)
    mangledname = re.sub(r"/", ".", mangledname).strip(" ._-")
    mangledname = mangledname[0:name_max]
    return mangledname


def _path_in_path_list(path, path_list):
    return any(p == path for p in path_list)


def _node_type(st):
    """ return a string indicating the type of special node represented by
    the stat buffer st (block, character, fifo, socket).
    """
    _types = [
        (stat.S_ISBLK, "block device"),
        (stat.S_ISCHR, "character device"),
        (stat.S_ISFIFO, "named pipe"),
        (stat.S_ISSOCK, "socket")
    ]
    for t in _types:
        if t[0](st.st_mode):
            return t[1]


def _file_is_compressed(path):
    """Check if a file appears to be compressed

    Return True if the file specified by path appears to be compressed,
    or False otherwise by testing the file name extension against a
    list of known file compression extentions.
    """
    return path.endswith(('.gz', '.xz', '.bz', '.bz2'))


_certmatch = re.compile("-*BEGIN.*?-*END", re.DOTALL)
_cert_replace = "-----SCRUBBED"


class SoSPredicate(object):
    """A class to implement collection predicates.

        A predicate gates the collection of data by an sos plugin. For any
        `add_cmd_output()`, `add_copy_spec()` or `add_journal()` call, the
        passed predicate will be evaulated and collection will proceed if
        the result is `True`, and not otherwise.

        Predicates may be used to control conditional data collection
        without the need for explicit conditional blocks in plugins.
    """
    #: The plugin that owns this predicate
    _owner = None

    #: Skip all collection?
    dry_run = False

    #: Kernel module enablement list
    kmods = []

    #: Services enablement list
    services = []

    #: Package presence list
    packages = []

    # Command output inclusion pairs {'cmd': 'foo --help', 'output': 'bar'}
    cmd_outputs = []

    #: Allowed architecture(s) of the system
    arch = []

    def __str(self, quote=False, prefix="", suffix=""):
        """Return a string representation of this SoSPredicate with
            optional prefix, suffix and value quoting.
        """
        quotes = '"%s"'
        pstr = "dry_run=%s, " % self.dry_run

        kmods = self.kmods
        kmods = [quotes % k for k in kmods] if quote else kmods
        pstr += "kmods=[%s], " % (",".join(kmods))

        services = self.services
        services = [quotes % s for s in services] if quote else services
        pstr += "services=[%s], " % (",".join(services))

        pkgs = self.packages
        pkgs = [quotes % p for p in pkgs] if quote else pkgs
        pstr += "packages=[%s], " % (",".join(pkgs))

        cmdoutputs = [
            "{ %s: %s, %s: %s }" % (quotes % "cmd",
                                    quotes % cmdoutput['cmd'],
                                    quotes % "output",
                                    quotes % cmdoutput['output'])
            for cmdoutput in self.cmd_outputs
        ]
        pstr += "cmdoutputs=[%s], " % (",".join(cmdoutputs))

        arches = self.arch
        arches = [quotes % a for a in arches] if quote else arches
        pstr += "arches=[%s]" % (",".join(arches))

        return prefix + pstr + suffix

    def __str__(self):
        """Return a string representation of this SoSPredicate.

            "dry_run=False, kmods=[], services=[], cmdoutputs=[]"
        """
        return self.__str()

    def __repr__(self):
        """Return a machine readable string representation of this
            SoSPredicate.

            "SoSPredicate(dry_run=False, kmods=[], services=[], cmdoutputs=[])"
        """
        return self.__str(quote=True, prefix="SoSPredicate(", suffix=")")

    def _check_required_state(self, items, required):
        """Helper to simplify checking the state of the predicate's evaluations
        against the setting of the required state of that evaluation
        """
        if required == 'any':
            return any(items)
        elif required == 'all':
            return all(items)
        elif required == 'none':
            return not any(items)

    def _failed_or_forbidden(self, test, item):
        """Helper to direct failed predicates to provide the proper messaging
        based on the required check type

            :param test:      The type of check we're doing, e.g. kmods, arch
            :param item:      The string of what failed
        """
        _req = self.required[test]
        if _req != 'none':
            self._failed[test].append(item)
        else:
            self._forbidden[test].append(item)

    def _eval_kmods(self):
        if not self.kmods or self._owner.get_option('allow_system_changes'):
            return True

        _kmods = []
        # Are kernel modules loaded?
        for kmod in self.kmods:
            res = self._owner.is_module_loaded(kmod)
            _kmods.append(res)
            if not res:
                self._failed_or_forbidden('kmods', kmod)

        return self._check_required_state(_kmods, self.required['kmods'])

    def _eval_services(self):
        if not self.services:
            return True

        _svcs = []
        for svc in self.services:
            res = self._owner.is_service_running(svc)
            _svcs.append(res)
            if not res:
                self._failed_or_forbidden('services', svc)

        return self._check_required_state(_svcs, self.required['services'])

    def _eval_packages(self):
        if not self.packages:
            return True

        _pkgs = []
        for pkg in self.packages:
            res = self._owner.is_installed(pkg)
            _pkgs.append(res)
            if not res:
                self._failed_or_forbidden('packages', pkg)

        return self._check_required_state(_pkgs, self.required['packages'])

    def _eval_cmd_output(self, cmd_output):
        """Does 'cmd' output contain string 'output'?"""
        if 'cmd' not in cmd_output or 'output' not in cmd_output:
            return False
        result = sos_get_command_output(cmd_output['cmd'])
        if result['status'] != 0:
            return False
        for line in result['output'].splitlines():
            if cmd_output['output'] in line:
                return True
        return False

    def _eval_cmd_outputs(self):
        if not self.cmd_outputs:
            return True

        _cmds = []
        for cmd in self.cmd_outputs:
            res = self._eval_cmd_output(cmd)
            _cmds.append(res)
            if not res:
                self._failed_or_forbidden(
                    'cmd_outputs',
                    "%s: %s" % (cmd['cmd'], cmd['output'])
                )
        return self._check_required_state(_cmds, self.required['cmd_outputs'])

    def _eval_arch(self):
        if not self.arch:
            return True

        # a test for 'all' against arch does not make sense, so only test to
        # see if the system's reported architecture is in the last of 'allowed'
        # arches requested by the predicate
        _arch = self._owner.policy.get_arch()
        regex = '(?:%s)' % '|'.join(self.arch)
        if self.required['arch'] == 'none':
            if re.match(regex, _arch):
                self._forbidden['architecture'].append(_arch)
                return False
            return True
        if re.match(regex, _arch):
            return True
        self._failed['architecture'].append(_arch)
        return False

    def _report_failed(self):
        """Return a string informing user what caused the predicate to fail
        evaluation
        """
        msg = ''
        _substr = "required %s missing: %s."
        for key, val in self._failed.items():
            if not val:
                continue
            val = set(val)
            msg += _substr % (key, ', '.join(v for v in val))
        return msg

    def _report_forbidden(self):
        """Return a string informing the user that a forbidden condition exists
        which caused the predicate to fail
        """
        msg = ''
        _substr = "forbidden %s '%s' found."
        for key, val in self._forbidden.items():
            if not val:
                continue
            val = set(val)
            msg += _substr % (key, ', '.join(v for v in val))
        return msg

    def report_failure(self):
        """Used by `Plugin()` to obtain the error string based on if the reason
        was a failed check or a forbidden check
        """
        msg = [self._report_failed(), self._report_forbidden()]
        return " ".join(msg).lstrip()

    def __nonzero__(self):
        """Predicate evaluation hook.
        """

        # Null predicate?
        if not any([self.kmods, self.services, self.packages, self.cmd_outputs,
                    self.arch, self.dry_run]):
            return True

        return ((self._eval_kmods() and self._eval_services() and
                 self._eval_packages() and self._eval_cmd_outputs() and
                 self._eval_arch())
                and not self.dry_run)

    def __bool__(self):
        # Py3 evaluation ends in a __bool__() call where py2 ends in a call
        # to __nonzero__(). Wrap the latter here, to support both versions
        return self.__nonzero__()

    def __init__(self, owner, dry_run=False, kmods=[], services=[],
                 packages=[], cmd_outputs=[], arch=[], required={}):
        """Initialise a new SoSPredicate object.
        """
        self._owner = owner
        self.kmods = list(kmods)
        self.services = list(services)
        self.packages = list(packages)
        self.arch = list(arch)
        if not isinstance(cmd_outputs, list):
            cmd_outputs = [cmd_outputs]
        self.cmd_outputs = cmd_outputs
        self.dry_run = dry_run | self._owner.commons['cmdlineopts'].dry_run
        self.required = {'kmods': 'any', 'services': 'any', 'packages': 'any',
                         'cmd_outputs': 'any', 'arch': 'any'}
        self.required.update({
            k: v for k, v in required.items() if
            required[k] != self.required[k]
        })
        #: Dict holding failed evaluations
        self._failed = {
            'kmods': [], 'services': [], 'packages': [], 'cmd_outputs': [],
            'architecture': []
        }
        self._forbidden = {
            'kmods': [], 'services': [], 'packages': [], 'cmd_outputs': [],
            'architecture': []
        }


class SoSCommand(object):
    """A class to represent a command to be collected.

    A SoSCommand() object is instantiated for each command handed to an
    _add_cmd_output() call, so that we no longer need to pass around a very
    long tuple to handle the parameters.

    Any option supported by _add_cmd_output() is passed to the SoSCommand
    object and converted to an attribute. SoSCommand.__dict__ is then passed to
    _get_command_output_now() for each command to be collected.
    """

    def __init__(self, **kwargs):
        self.__dict__.update(kwargs)

    def __str__(self):
        """Return a human readable string representation of this SoSCommand
        """
        return ', '.join("%s=%r" % (param, val) for (param, val) in
                         sorted(self.__dict__.items()))


class Plugin(object):
    """ This is the base class for sosreport plugins. Plugins should subclass
    this and set the class variables where applicable.

    plugin_name is a string returned by plugin.name(). If this is set to None
    (the default) class\\_.__name__.tolower() will be returned. Be sure to set
    this if you are defining multiple plugins that do the same thing on
    different platforms.

    requires_root is a boolean that specifies whether or not sosreport should
    execute this plugin as a super user.

    version is a string representing the version of the plugin. This can be
    useful for post-collection tooling.

    packages (files) is an iterable of the names of packages (the paths
    of files) to check for before running this plugin. If any of these packages
    or files is found on the system, the default implementation of
    check_enabled will return True.

    profiles is an iterable of profile names that this plugin belongs to.
    Whenever any of the profiles is selected on the command line the plugin
    will be enabled (subject to normal check_enabled tests).
    """

    plugin_name = None
    requires_root = True
    version = 'unversioned'
    packages = ()
    files = ()
    commands = ()
    kernel_mods = ()
    services = ()
    architectures = None
    archive = None
    profiles = ()
    sysroot = '/'
    plugin_timeout = 300
    cmd_timeout = 300
    _timeout_hit = False

    # Default predicates
    predicate = None
    cmd_predicate = None
    _default_plug_opts = [
        ('timeout', 'Timeout in seconds for plugin', 'fast', -1),
        ('postproc', 'Enable post-processing collected plugin data', 'fast',
         True)
    ]

    def __init__(self, commons):
        if not getattr(self, "option_list", False):
            self.option_list = []

        self.copied_files = []
        self.executed_commands = []
        self._env_vars = set()
        self.alerts = []
        self.custom_text = ""
        self.opt_names = []
        self.opt_parms = []
        self.commons = commons
        self.forbidden_paths = []
        self.copy_paths = set()
        self.copy_strings = []
        self.collect_cmds = []
        self.sysroot = commons['sysroot']
        self.policy = commons['policy']

        self.soslog = self.commons['soslog'] if 'soslog' in self.commons \
            else logging.getLogger('sos')

        # add the default plugin opts
        self.option_list.extend(self._default_plug_opts)

        # get the option list into a dictionary
        for opt in self.option_list:
            self.opt_names.append(opt[0])
            self.opt_parms.append({'desc': opt[1], 'speed': opt[2],
                                   'enabled': opt[3]})

        # Initialise the default --dry-run predicate
        self.set_predicate(SoSPredicate(self))

    @property
    def timeout(self):
        """Returns either the default plugin timeout value, the value as
        provided on the commandline via -k plugin.timeout=value, or the value
        of the global --plugin-timeout option.
        """
        _timeout = None
        try:
            opt_timeout = self.get_option('plugin_timeout')
            own_timeout = int(self.get_option('timeout'))
            if opt_timeout is None:
                _timeout = own_timeout
            elif opt_timeout is not None and own_timeout == -1:
                _timeout = int(opt_timeout)
            elif opt_timeout is not None and own_timeout > -1:
                _timeout = own_timeout
            else:
                return None
        except ValueError:
            return self.plugin_timeout  # Default to known safe value
        if _timeout is not None and _timeout > -1:
            return _timeout
        return self.plugin_timeout

    def check_timeout(self):
        """
        Checks to see if the plugin has hit its timeout.

        This is set when the sos.collect_plugin() method hits a timeout and
        terminates the thread. From there, a Popen() call can still continue to
        run, and we need to manually terminate it. Thus, check_timeout() should
        only be called in sos_get_command_output().

        Since sos_get_command_output() is not plugin aware, this method is
        handed to that call to use as a polling method, to avoid passing the
        entire plugin object.

        Returns True if timeout has been hit, else False.

        """
        return self._timeout_hit

    @classmethod
    def name(cls):
        """Returns the plugin's name as a string. This should return a
        lowercase string.
        """
        if cls.plugin_name:
            return cls.plugin_name
        return cls.__name__.lower()

    def _format_msg(self, msg):
        return "[plugin:%s] %s" % (self.name(), msg)

    def _log_error(self, msg):
        self.soslog.error(self._format_msg(msg))

    def _log_warn(self, msg):
        self.soslog.warning(self._format_msg(msg))

    def _log_info(self, msg):
        self.soslog.info(self._format_msg(msg))

    def _log_debug(self, msg):
        self.soslog.debug(self._format_msg(msg))

    def join_sysroot(self, path):
        if path[0] == os.sep:
            path = path[1:]
        return os.path.join(self.sysroot, path)

    def strip_sysroot(self, path):
        if not self.use_sysroot():
            return path
        if path.startswith(self.sysroot):
            return path[len(self.sysroot):]
        return path

    def use_sysroot(self):
        return self.sysroot != os.path.abspath(os.sep)

    def tmp_in_sysroot(self):
        paths = [self.sysroot, self.archive.get_tmp_dir()]
        return os.path.commonprefix(paths) == self.sysroot

    def is_installed(self, package_name):
        """Is the package $package_name installed?"""
        return self.policy.pkg_by_name(package_name) is not None

    def is_service(self, name):
        """Does the service $name exist on the system?"""
        return self.policy.init_system.is_service(name)

    def is_service_enabled(self, name):
        """Is the service $name enabled?"""
        return self.policy.init_system.is_enabled(name)

    def is_service_disabled(self, name):
        """Is the service $name disabled?"""
        return self.policy.init_system.is_disabled(name)

    def is_service_running(self, name):
        """Is the service $name currently running?"""
        return self.policy.init_system.is_running(name)

    def get_service_status(self, name):
        """Return the reported status for service $name"""
        return self.policy.init_system.get_service_status(name)['status']

    def get_service_names(self, regex):
        """Get all service names matching regex"""
        return self.policy.init_system.get_service_names(regex)

    def set_predicate(self, pred):
        """Set or clear the default predicate for this plugin.
        """
        self.predicate = pred

    def set_cmd_predicate(self, pred):
        """Set or clear the default predicate for command collection
            for this plugin. If set, this predecate takes precedence
            over the `Plugin` default predicate for command and journal
            data collection.
        """
        self.cmd_predicate = pred

    def get_predicate(self, cmd=False, pred=None):
        """Get the current default `Plugin` or command predicate. If the
            `cmd` argument is `True`, the current command predicate is
            returned if set, otherwise the default `Plugin` predicate
            will be returned (which may be `None`).

            If no default predicate is set and a `pred` value is passed
            it will be returned.
        """
        if pred is not None:
            return pred
        if cmd and self.cmd_predicate is not None:
            return self.cmd_predicate
        return self.predicate

    def test_predicate(self, cmd=False, pred=None):
        """Test the current predicate and return its value.

            :param cmd: ``True`` if the predicate is gating a command or
                        ``False`` otherwise.
            :param pred: An optional predicate to override the current
                         ``Plugin`` or command predicate.
        """
        pred = self.get_predicate(cmd=cmd, pred=pred)
        if pred is not None:
            return bool(pred)
        return False

    def log_skipped_cmd(self, pred, cmd, kmods=False, services=False,
                        changes=False):
        """Log that a command was skipped due to predicate evaluation.

            Emit a warning message indicating that a command was skipped due
            to predicate evaluation. If ``kmods`` or ``services`` are ``True``
            then the list of expected kernel modules or services will be
            included in the log message. If ``allow_changes`` is ``True`` a
            message indicating that the missing data can be collected by using
            the "--allow-system-changes" command line option will be included.
        """
        msg = "skipped command '%s': %s" % (cmd, pred.report_failure())

        if changes:
            msg += " Use '--allow-system-changes' to enable collection."

        self._log_warn(msg)

    def do_cmd_private_sub(self, cmd, desc=""):
        """Remove certificate and key output archived by sosreport. cmd
        is the command name from which output is collected (i.e. exlcuding
        parameters). Any matching instances are replaced with: '-----SCRUBBED'
        and this function does not take a regexp or substituting string.

        This function returns the number of replacements made.
        """
        if not self.executed_commands:
            return 0

        self._log_debug("Scrubbing certs and keys for commands matching %s"
                        % (cmd))

        replace = "%s %s" % (_cert_replace, desc) if desc else _cert_replace

        return self.do_cmd_output_sub(cmd, _certmatch, replace)

    def do_cmd_output_sub(self, cmd, regexp, subst):
        """Apply a regexp substitution to command output archived by sosreport.
        cmd is the command name from which output is collected (i.e. excluding
        parameters). The regexp can be a string or a compiled re object. The
        substitution string, subst, is a string that replaces each occurrence
        of regexp in each file collected from cmd. Internally 'cmd' is treated
        as a glob with a leading and trailing '*' and each matching file from
        the current module's command list is subjected to the replacement.

        This function returns the number of replacements made.
        """
        globstr = '*' + cmd + '*'
        pattern = regexp.pattern if hasattr(regexp, "pattern") else regexp
        self._log_debug("substituting '%s' for '%s' in commands matching '%s'"
                        % (subst, pattern, globstr))

        if not self.executed_commands:
            return 0

        replacements = None
        try:
            for called in self.executed_commands:
                # was anything collected?
                if called['file'] is None:
                    continue
                if called['binary'] == 'yes':
                    self._log_warn("Cannot apply regex substitution to binary"
                                   " output: '%s'" % called['exe'])
                    continue
                if fnmatch.fnmatch(called['cmd'], globstr):
                    path = os.path.join(self.commons['cmddir'], called['file'])
                    self._log_debug("applying substitution to '%s'" % path)
                    readable = self.archive.open_file(path)
                    result, replacements = re.subn(
                        regexp, subst, readable.read())
                    if replacements:
                        self.archive.add_string(result, path)

        except Exception as e:
            msg = "regex substitution failed for '%s' with: '%s'"
            self._log_error(msg % (called['exe'], e))
            replacements = None
        return replacements

    def do_file_private_sub(self, pathregex, desc=""):
        """Scrub certificate/key/etc information from files collected by sos.

        Files matching the provided pathregex are searched for content that
        resembles certificate, ssh keys, or similar information. Any matches
        are replaced with "-----SCRUBBED $desc" where `desc` is a description
        of the specific type of content being replaced, e.g.
        "-----SCRUBBED RSA PRIVATE KEY" so that support representatives can
        at least be informed of what type of content it was originally.

        Positional arguments:
            :param pathregex: A string or regex of a filename to match against
            :param desc: A description of the replaced content
        """
        self._log_debug("Scrubbing certs and keys for paths matching %s"
                        % pathregex)
        match = re.compile(pathregex).match
        replace = "%s %s" % (_cert_replace, desc) if desc else _cert_replace
        file_list = [f for f in self.copied_files if match(f['srcpath'])]
        for i in file_list:
            path = i['dstpath']
            if not path:
                continue
            self.do_file_sub(path, _certmatch, replace)

    def do_file_sub(self, srcpath, regexp, subst):
        """Apply a regexp substitution to a file archived by sosreport.
        srcpath is the path in the archive where the file can be found.  regexp
        can be a regexp string or a compiled re object.  subst is a string to
        replace each occurance of regexp in the content of srcpath.

        This function returns the number of replacements made.
        """
        try:
            path = self._get_dest_for_srcpath(srcpath)
            pattern = regexp.pattern if hasattr(regexp, "pattern") else regexp
            self._log_debug("substituting scrpath '%s'" % srcpath)
            self._log_debug("substituting '%s' for '%s' in '%s'"
                            % (subst, pattern, path))
            if not path:
                return 0
            readable = self.archive.open_file(path)
            content = readable.read()
            if not isinstance(content, six.string_types):
                content = content.decode('utf8', 'ignore')
            result, replacements = re.subn(regexp, subst, content)
            if replacements:
                self.archive.add_string(result, srcpath)
            else:
                replacements = 0
        except (OSError, IOError) as e:
            # if trying to regexp a nonexisting file, dont log it as an
            # error to stdout
            if e.errno == errno.ENOENT:
                msg = "file '%s' not collected, substitution skipped"
                self._log_debug(msg % path)
            else:
                msg = "regex substitution failed for '%s' with: '%s'"
                self._log_error(msg % (path, e))
            replacements = 0
        return replacements

    def do_path_regex_sub(self, pathexp, regexp, subst):
        """Apply a regexp substituation to a set of files archived by
        sos. The set of files to be substituted is generated by matching
        collected file pathnames against pathexp which may be a regular
        expression string or compiled re object. The portion of the file
        to be replaced is specified via regexp and the replacement string
        is passed in subst."""
        if not hasattr(pathexp, "match"):
            pathexp = re.compile(pathexp)
        match = pathexp.match
        file_list = [f for f in self.copied_files if match(f['srcpath'])]
        for file in file_list:
            self.do_file_sub(file['srcpath'], regexp, subst)

    def do_regex_find_all(self, regex, fname):
        return regex_findall(regex, fname)

    def _copy_symlink(self, srcpath):
        # the target stored in the original symlink
        linkdest = os.readlink(srcpath)
        dest = os.path.join(os.path.dirname(srcpath), linkdest)
        # Absolute path to the link target. If SYSROOT != '/' this path
        # is relative to the host root file system.
        absdest = os.path.normpath(dest)
        # adjust the target used inside the report to always be relative
        if os.path.isabs(linkdest):
            # Canonicalize the link target path to avoid additional levels
            # of symbolic links (that would affect the path nesting level).
            realdir = os.path.realpath(os.path.dirname(srcpath))
            reldest = os.path.relpath(linkdest, start=realdir)
            # trim leading /sysroot
            if self.use_sysroot():
                reldest = reldest[len(os.sep + os.pardir):]
            self._log_debug("made link target '%s' relative as '%s'"
                            % (linkdest, reldest))
        else:
            reldest = linkdest

        self._log_debug("copying link '%s' pointing to '%s' with isdir=%s"
                        % (srcpath, linkdest, os.path.isdir(absdest)))

        dstpath = self.strip_sysroot(srcpath)
        # use the relative target path in the tarball
        self.archive.add_link(reldest, dstpath)

        if os.path.isdir(absdest):
            self._log_debug("link '%s' is a directory, skipping..." % linkdest)
            return

        self.copied_files.append({'srcpath': srcpath,
                                  'dstpath': dstpath,
                                  'symlink': "yes",
                                  'pointsto': linkdest})

        # Check for indirect symlink loops by stat()ing the next step
        # in the link chain.
        try:
            os.stat(absdest)
        except OSError as e:
            if e.errno == 40:
                self._log_debug("link '%s' is part of a file system "
                                "loop, skipping target..." % dstpath)
                return

        # copy the symlink target translating relative targets
        # to absolute paths to pass to _do_copy_path.
        self._log_debug("normalized link target '%s' as '%s'"
                        % (linkdest, absdest))

        # skip recursive copying of symlink pointing to itself.
        if (absdest != srcpath):
            self._do_copy_path(absdest)
        else:
            self._log_debug("link '%s' points to itself, skipping target..."
                            % linkdest)

    def _copy_dir(self, srcpath):
        try:
            for name in os.listdir(srcpath):
                self._log_debug("recursively adding '%s' from '%s'"
                                % (name, srcpath))
                path = os.path.join(srcpath, name)
                self._do_copy_path(path)
        except OSError as e:
            if e.errno == errno.EPERM or errno.EACCES:
                msg = "Permission denied"
                self._log_warn("_copy_dir: '%s' %s" % (srcpath, msg))
                return
            if e.errno == errno.ELOOP:
                msg = "Too many levels of symbolic links copying"
                self._log_error("_copy_dir: %s '%s'" % (msg, srcpath))
                return
            raise

    def _get_dest_for_srcpath(self, srcpath):
        if self.use_sysroot():
            srcpath = self.join_sysroot(srcpath)
        for copied in self.copied_files:
            if srcpath == copied["srcpath"]:
                return copied["dstpath"]
        return None

    def _is_forbidden_path(self, path):
        return _path_in_path_list(path, self.forbidden_paths)

    def _copy_node(self, path, st):
        dev_maj = os.major(st.st_rdev)
        dev_min = os.minor(st.st_rdev)
        mode = st.st_mode
        self.archive.add_node(path, mode, os.makedev(dev_maj, dev_min))

    # Methods for copying files and shelling out
    def _do_copy_path(self, srcpath, dest=None):
        """Copy file or directory to the destination tree. If a directory, then
        everything below it is recursively copied. A list of copied files are
        saved for use later in preparing a report.
        """
        if self._timeout_hit:
            return

        if self._is_forbidden_path(srcpath):
            self._log_debug("skipping forbidden path '%s'" % srcpath)
            return ''

        if not dest:
            dest = srcpath

        if self.use_sysroot():
            dest = self.strip_sysroot(dest)

        try:
            st = os.lstat(srcpath)
        except (OSError, IOError):
            self._log_info("failed to stat '%s'" % srcpath)
            return

        if stat.S_ISLNK(st.st_mode):
            self._copy_symlink(srcpath)
            return
        else:
            if stat.S_ISDIR(st.st_mode) and os.access(srcpath, os.R_OK):
                # copy empty directory
                if not os.listdir(srcpath):
                    self.archive.add_dir(dest)
                    return
                self._copy_dir(srcpath)
                return

        # handle special nodes (block, char, fifo, socket)
        if not (stat.S_ISREG(st.st_mode) or stat.S_ISDIR(st.st_mode)):
            ntype = _node_type(st)
            self._log_debug("creating %s node at archive:'%s'"
                            % (ntype, dest))
            self._copy_node(dest, st)
            return

        # if we get here, it's definitely a regular file (not a symlink or dir)
        self._log_debug("copying path '%s' to archive:'%s'" % (srcpath, dest))

        # if not readable(srcpath)
        if not st.st_mode & 0o444:
            # FIXME: reflect permissions in archive
            self.archive.add_string("", dest)
        else:
            self.archive.add_file(srcpath, dest)

        self.copied_files.append({
            'srcpath': srcpath,
            'dstpath': dest,
            'symlink': "no"
        })

    def add_forbidden_path(self, forbidden):
        """Specify a path, or list of paths, to not copy, even if it's
            part of a copy_specs[] entry.
        """
        if isinstance(forbidden, six.string_types):
            forbidden = [forbidden]

        if self.use_sysroot():
            forbidden = [self.join_sysroot(f) for f in forbidden]

        for forbid in forbidden:
            self._log_info("adding forbidden path '%s'" % forbid)
            for path in glob.glob(forbid):
                self.forbidden_paths.append(path)

    def get_all_options(self):
        """return a list of all options selected"""
        return (self.opt_names, self.opt_parms)

    def set_option(self, optionname, value):
        """Set the named option to value. Ensure the original type
           of the option value is preserved.
        """
        for name, parms in zip(self.opt_names, self.opt_parms):
            if name == optionname:
                # FIXME: ensure that the resulting type of the set option
                # matches that of the default value. This prevents a string
                # option from being coerced to int simply because it holds
                # a numeric value (e.g. a password).
                # See PR #1526 and Issue #1597
                defaulttype = type(parms['enabled'])
                if defaulttype != type(value) and defaulttype != type(None):
                    value = (defaulttype)(value)
                parms['enabled'] = value
                return True
        return False

    def get_option(self, optionname, default=0):
        """Returns the first value that matches 'optionname' in parameters
        passed in via the command line or set via set_option or via the
        global_plugin_options dictionary, in that order.

        optionaname may be iterable, in which case the first option that
        matches any of the option names is returned.
        """

        global_options = (
            'all_logs', 'allow_system_changes', 'log_size', 'plugin_timeout',
            'since', 'verify'
        )

        if optionname in global_options:
            return getattr(self.commons['cmdlineopts'], optionname)

        for name, parms in zip(self.opt_names, self.opt_parms):
            if name == optionname:
                val = parms['enabled']
                if val is not None:
                    return val

        return default

    def get_option_as_list(self, optionname, delimiter=",", default=None):
        """Will try to return the option as a list separated by the
        delimiter.
        """
        option = self.get_option(optionname)
        try:
            opt_list = [opt.strip() for opt in option.split(delimiter)]
            return list(filter(None, opt_list))
        except Exception:
            return default

    def _add_copy_paths(self, copy_paths):
        self.copy_paths.update(copy_paths)

    def add_copy_spec(self, copyspecs, sizelimit=None, maxage=None,
                      tailit=True, pred=None):
        """Add a file or glob but limit it to sizelimit megabytes. Collect
        files with mtime not older than maxage hours.
        If fname is a single file the file will be tailed to meet sizelimit.
        If the first file in a glob is too large it will be tailed to meet
        the sizelimit.
        """
        since = None
        if self.get_option('since'):
            since = self.get_option('since')

        logarchive_pattern = re.compile(r'.*((\.(zip|gz|bz2|xz))|[-.][\d]+)$')
        configfile_pattern = re.compile(r"^%s/*" % self.join_sysroot("etc"))

        if not self.test_predicate(pred=pred):
            self._log_info("skipped copy spec '%s' due to predicate (%s)" %
                           (copyspecs, self.get_predicate(pred=pred)))
            return

        if sizelimit is None:
            sizelimit = self.get_option("log_size")

        if self.get_option('all_logs'):
            sizelimit = None

        if sizelimit:
            sizelimit *= 1024 * 1024  # in MB

        if not copyspecs:
            return False

        if isinstance(copyspecs, six.string_types):
            copyspecs = [copyspecs]

        for copyspec in copyspecs:
            if not (copyspec and len(copyspec)):
                return False

            if self.use_sysroot():
                copyspec = self.join_sysroot(copyspec)

            files = self._expand_copy_spec(copyspec)

            if len(files) == 0:
                continue

            # Files hould be sorted in most-recently-modified order, so that
            # we collect the newest data first before reaching the limit.
            def getmtime(path):
                try:
                    return os.path.getmtime(path)
                except OSError:
                    return 0

            def time_filter(path):
                """ When --since is passed, or maxage is coming from the
                plugin, we need to filter out older files """

                # skip config files or not-logarchive files from the filter
                if ((logarchive_pattern.search(path) is None) or
                   (configfile_pattern.search(path) is not None)):
                    return True
                filetime = datetime.fromtimestamp(getmtime(path))
                if ((since and filetime < since) or
                   (maxage and (time()-filetime < maxage*3600))):
                    return False
                return True

            if since or maxage:
                files = list(filter(lambda f: time_filter(f), files))

            files.sort(key=getmtime, reverse=True)
            current_size = 0
            limit_reached = False

            for _file in files:
                if _file in self.copy_paths:
                    self._log_debug("skipping redundant file '%s'" % _file)
                    continue
                if self._is_forbidden_path(_file):
                    self._log_debug("skipping forbidden path '%s'" % _file)
                    continue
                if limit_reached:
                    self._log_info("skipping '%s' over size limit" % _file)
                    continue

                try:
                    filestat = os.stat(_file)
                except OSError:
                    self._log_info("failed to stat '%s'" % _file)
                    continue
                current_size += filestat[stat.ST_SIZE]

                if sizelimit and current_size > sizelimit:
                    limit_reached = True

                    if tailit and not _file_is_compressed(_file):
                        self._log_info("collecting tail of '%s' due to size "
                                       "limit" % _file)
                        file_name = _file
                        if file_name[0] == os.sep:
                            file_name = file_name.lstrip(os.sep)
                        strfile = (
                            file_name.replace(os.path.sep, ".") + ".tailed"
                        )
                        self.add_string_as_file(tail(_file, sizelimit),
                                                strfile)
                        rel_path = os.path.relpath('/', os.path.dirname(_file))
                        link_path = os.path.join(rel_path, 'sos_strings',
                                                 self.name(), strfile)
                        self.archive.add_link(link_path, _file)
                    else:
                        self._log_info("skipping '%s' over size limit" % _file)
                else:
                    # size limit not hit, copy the file
                    self._add_copy_paths([_file])

    def _add_cmd_output(self, **kwargs):
        """Internal helper to add a single command to the collection list."""
        pred = kwargs.pop('pred') if 'pred' in kwargs else None
        soscmd = SoSCommand(**kwargs)
        self._log_debug("packed command: " + soscmd.__str__())
        if self.test_predicate(cmd=True, pred=pred):
            self.collect_cmds.append(soscmd)
            self._log_info("added cmd output '%s'" % soscmd.cmd)
        else:
            self.log_skipped_cmd(pred, soscmd.cmd, kmods=bool(pred.kmods),
                                 services=bool(pred.services),
                                 changes=soscmd.changes)

    def add_cmd_output(self, cmds, suggest_filename=None,
                       root_symlink=None, timeout=cmd_timeout, stderr=True,
                       chroot=True, runat=None, env=None, binary=False,
                       sizelimit=None, pred=None, subdir=None,
                       changes=False, foreground=False):
        """Run a program or a list of programs and collect the output"""
        if isinstance(cmds, six.string_types):
            cmds = [cmds]
        if len(cmds) > 1 and (suggest_filename or root_symlink):
            self._log_warn("ambiguous filename or symlink for command list")
        if sizelimit is None:
            sizelimit = self.get_option("log_size")
        if pred is None:
            pred = self.get_predicate(cmd=True)
        for cmd in cmds:
            self._add_cmd_output(cmd=cmd, suggest_filename=suggest_filename,
                                 root_symlink=root_symlink, timeout=timeout,
                                 stderr=stderr, chroot=chroot, runat=runat,
                                 env=env, binary=binary, sizelimit=sizelimit,
                                 pred=pred, subdir=subdir,
                                 changes=changes, foreground=foreground)

    def get_cmd_output_path(self, name=None, make=True):
        """Return a path into which this module should store collected
        command output
        """
        cmd_output_path = os.path.join(self.archive.get_tmp_dir(),
                                       'sos_commands', self.name())
        if name:
            cmd_output_path = os.path.join(cmd_output_path, name)
        if make:
            os.makedirs(cmd_output_path)

        return cmd_output_path

    def file_grep(self, regexp, *fnames):
        """Returns lines matched in fnames, where fnames can either be
        pathnames to files to grep through or open file objects to grep through
        line by line.
        """
        return grep(regexp, *fnames)

    def _mangle_command(self, exe):
        name_max = self.archive.name_max()
        return _mangle_command(exe, name_max)

    def _make_command_filename(self, exe, subdir=None):
        """The internal function to build up a filename based on a command."""

        plugin_dir = self.name()
        if subdir:
            # only allow a single level of subdir to be created
            plugin_dir += "/%s" % subdir.split('/')[0]
        outdir = os.path.join(self.commons['cmddir'], plugin_dir)
        outfn = self._mangle_command(exe)

        # check for collisions
        if os.path.exists(os.path.join(self.archive.get_tmp_dir(),
                                       outdir, outfn)):
            inc = 1
            name_max = self.archive.name_max()
            while True:
                suffix = ".%d" % inc
                newfn = outfn
                if name_max < len(newfn)+len(suffix):
                    newfn = newfn[:(name_max-len(newfn)-len(suffix))]
                newfn = newfn + suffix
                if not os.path.exists(os.path.join(self.archive.get_tmp_dir(),
                                                   outdir, newfn)):
                    outfn = newfn
                    break
                inc += 1

        return os.path.join(outdir, outfn)

    def add_env_var(self, name):
        """Add an environment variable to the list of to-be-collected env vars.

        Accepts either a single variable name or a list of names. Any value
        given will be added as provided to the method, as well as an upper-
        and lower- cased version.
        """
        if not isinstance(name, list):
            name = [name]
        for env in name:
            # get both upper and lower cased vars since a common support issue
            # is setting the env vars to the wrong case, and if the plugin
            # adds a mixed case variable name, still get that as well
            self._env_vars.update([env, env.upper(), env.lower()])

    def add_string_as_file(self, content, filename, pred=None):
        """Add a string to the archive as a file named `filename`"""

        # Generate summary string for logging
        summary = content.splitlines()[0] if content else ''
        if not isinstance(summary, six.string_types):
            summary = content.decode('utf8', 'ignore')

        if not self.test_predicate(cmd=False, pred=pred):
            self._log_info("skipped string ...'%s' due to predicate (%s)" %
                           (summary, self.get_predicate(pred=pred)))
            return

        self.copy_strings.append((content, filename))
        self._log_debug("added string ...'%s' as '%s'" % (summary, filename))

    def _collect_cmd_output(self, cmd, suggest_filename=None,
                            root_symlink=False, timeout=cmd_timeout,
                            stderr=True, chroot=True, runat=None, env=None,
                            binary=False, sizelimit=None, subdir=None,
                            changes=False, foreground=False):
        """Execute a command and save the output to a file for inclusion in the
        report.

        Positional Arguments:
            :param cmd:                 The command to run

        Keyword Arguments:
            :param suggest_filename:    Filename to use when writing to the
                                        archive
            :param root_symlink:        Create a symlink in the archive root
            :param timeout:             Time in seconds to allow a cmd to run
            :param stderr:              Write stderr to stdout?
            :param chroot:              Perform chroot before running cmd?
            :param runat:               Run the command from this location,
                                        overriding chroot
            :param env:                 Dict of env vars to set for the cmd
            :param binary:              Is the output in binary?
            :param sizelimit:           Maximum size in MB of output to save
            :param subdir:              Subdir in plugin directory to save to
            :param changes:             Does this cmd potentially make a change
                                        on the system?

        :returns:       dict containing status, output, and filename in the
                        archive for the executed cmd

        """
        if self._timeout_hit:
            return

        if chroot or self.commons['cmdlineopts'].chroot == 'always':
            root = self.sysroot
        else:
            root = None

        start = time()

        result = sos_get_command_output(
            cmd, timeout=timeout, stderr=stderr, chroot=root,
            chdir=runat, env=env, binary=binary, sizelimit=sizelimit,
            poller=self.check_timeout, foreground=foreground
        )

        if result['status'] == 124:
            self._log_warn(
                "command '%s' timed out after %ds" % (cmd, timeout)
            )

        # command not found or not runnable
        if result['status'] == 126 or result['status'] == 127:
            # automatically retry chroot'ed commands in the host namespace
            if root and root != '/':
                if self.commons['cmdlineopts'].chroot != 'always':
                    self._log_info("command '%s' not found in %s - "
                                   "re-trying in host root"
                                   % (cmd.split()[0], root))
                    result = sos_get_command_output(
                        cmd, timeout=timeout, chroot=False, chdir=runat,
                        env=env, binary=binary, sizelimit=sizelimit,
                        poller=self.check_timeout
                    )
            self._log_debug("could not run '%s': command not found" % cmd)
            # Exit here if the command was not found in the chroot check above
            # as otherwise we will create a blank file in the archive
            if result['status'] in [126, 127]:
                return result

        self._log_debug("collected output of '%s' in %s (changes=%s)"
                        % (cmd.split()[0], time() - start, changes))

        if suggest_filename:
            outfn = self._make_command_filename(suggest_filename, subdir)
        else:
            outfn = self._make_command_filename(cmd, subdir)

        outfn_strip = outfn[len(self.commons['cmddir'])+1:]

        if binary:
            self.archive.add_binary(result['output'], outfn)
        else:
            self.archive.add_string(result['output'], outfn)
        if root_symlink:
            self.archive.add_link(outfn, root_symlink)

        # save info for later
        self.executed_commands.append({'cmd': cmd, 'file': outfn_strip,
                                       'binary': 'yes' if binary else 'no'})

        result['filename'] = (
            os.path.join(self.archive.get_archive_path(), outfn) if outfn else
            ''
        )
        return result

    def collect_cmd_output(self, cmd, suggest_filename=None,
                           root_symlink=False, timeout=cmd_timeout,
                           stderr=True, chroot=True, runat=None, env=None,
                           binary=False, sizelimit=None, pred=None,
                           subdir=None):
        """Execute a command and save the output to a file for inclusion in the
        report.
        """
        if not self.test_predicate(cmd=True, pred=pred):
            self._log_info("skipped cmd output '%s' due to predicate (%s)" %
                           (cmd, self.get_predicate(cmd=True, pred=pred)))
            return {
                'status': None,  # don't match on if result['status'] checks
                'output': '',
                'filename': ''
            }

        return self._collect_cmd_output(
            cmd, suggest_filename=suggest_filename, root_symlink=root_symlink,
            timeout=timeout, stderr=stderr, chroot=chroot, runat=runat,
            env=env, binary=binary, sizelimit=sizelimit, subdir=subdir
        )

    def exec_cmd(self, cmd, timeout=cmd_timeout, stderr=True, chroot=True,
                 runat=None, env=None, binary=False, pred=None,
                 foreground=False):
        """Execute a command right now and return the output and status, but
        do not save the output within the archive.

        Use this method in a plugin's setup() if command output is needed to
        build subsequent commands added to a report via add_cmd_output().
        """
        if not self.test_predicate(cmd=True, pred=pred):
            return {
                'status': None,
                'output': ''
            }

        if chroot or self.commons['cmdlineopts'].chroot == 'always':
            root = self.sysroot
        else:
            root = None

        return sos_get_command_output(cmd, timeout=timeout, chroot=root,
                                      chdir=runat, binary=binary, env=env,
                                      foreground=foreground)

    def is_module_loaded(self, module_name):
        """Return whether specified module as module_name is loaded or not"""
        return module_name in self.policy.kernel_mods

    # For adding output
    def add_alert(self, alertstring):
        """Add an alert to the collection of alerts for this plugin. These
        will be displayed in the report
        """
        self.alerts.append(alertstring)

    def add_custom_text(self, text):
        """Append text to the custom text that is included in the report. This
        is freeform and can include html.
        """
        self.custom_text += text

    def add_service_status(self, services, **kwargs):
        """Collect service status information based on the InitSystem used.

        :param services: A string, or list of strings, specifying the services
                          to collect
        :param kwargs    Optional arguments to pass to _add_cmd_output
                         (timeout, predicate, suggest_filename,..)
        """
        if isinstance(services, six.string_types):
            services = [services]

        query = self.policy.init_system.query_cmd
        if not query:
            # No policy defined InitSystem, cannot use add_service_status
            self._log_debug('Cannot add service output, policy does not define'
                            ' an InitSystem to use')
            return

        for service in services:
            self._add_cmd_output(cmd="%s %s" % (query, service), **kwargs)

    def add_journal(self, units=None, boot=None, since=None, until=None,
                    lines=None, allfields=False, output=None,
                    timeout=cmd_timeout, identifier=None, catalog=None,
                    sizelimit=None, pred=None):
        """Collect journald logs from one of more units.

        :param units: A string, or list of strings specifying the
                       systemd units for which journal entries will be
                       collected.

        :param boot: A string selecting a boot index using the
                      journalctl syntax. The special values 'this' and
                      'last' are also accepted.

        :param since: A string representation of the start time for
                       journal messages.

        :param until: A string representation of the end time for
                       journal messages.

        :param lines: The maximum number of lines to be collected.

        :param allfields: A bool. Include all journal fields
                           regardless of size or non-printable
                           characters.

        :param output: A journalctl output control string, for
                        example "verbose".

        :param timeout: An optional timeout in seconds.
        :param identifier: An optional message identifier.
        :param catalog: Bool. If True, augment lines with descriptions
                        from the system catalog.
        :param sizelimit: Limit to the size of output returned in MB.
                          Defaults to the value of --log-size.
        """
        journal_cmd = "journalctl --no-pager "
        unit_opt = " --unit %s"
        boot_opt = " --boot %s"
        since_opt = " --since '%s'"
        until_opt = " --until %s"
        lines_opt = " --lines %s"
        output_opt = " --output %s"
        identifier_opt = " --identifier %s"
        catalog_opt = " --catalog"

        journal_size = 100
        all_logs = self.get_option("all_logs")
        log_size = sizelimit or self.get_option("log_size")
        log_size = max(log_size, journal_size) if not all_logs else 0

        if isinstance(units, six.string_types):
            units = [units]

        if units:
            for unit in units:
                journal_cmd += unit_opt % unit

        if identifier:
            journal_cmd += identifier_opt % identifier

        if catalog:
            journal_cmd += catalog_opt

        if allfields:
            journal_cmd += " --all"

        if boot:
            if boot == "this":
                boot = ""
            if boot == "last":
                boot = "-1"
            journal_cmd += boot_opt % boot

        if since:
            journal_cmd += since_opt % since

        if until:
            journal_cmd += until_opt % until

        if lines:
            journal_cmd += lines_opt % lines

        if output:
            journal_cmd += output_opt % output

        self._log_debug("collecting journal: %s" % journal_cmd)
        self._add_cmd_output(cmd=journal_cmd, timeout=timeout,
                             sizelimit=log_size, pred=pred)

    def add_udev_info(self, device, attrs=False):
        """Collect udevadm info output for a given device

        :param device: A string or list of strings of device names or sysfs
                       paths. E.G. either '/sys/class/scsi_host/host0' or
                       '/dev/sda' is valid.
        :param attrs: If True, run udevadm with the --attribute-walk option.
        """
        udev_cmd = 'udevadm info'
        if attrs:
            udev_cmd += ' -a'

        if isinstance(device, six.string_types):
            device = [device]

        for dev in device:
            self._log_debug("collecting udev info for: %s" % dev)
            self._add_cmd_output(cmd='%s %s' % (udev_cmd, dev))

    def _expand_copy_spec(self, copyspec):
        return glob.glob(copyspec)

    def _collect_copy_specs(self):
        for path in self.copy_paths:
            self._log_info("collecting path '%s'" % path)
            self._do_copy_path(path)

    def _collect_cmds(self):
        for soscmd in self.collect_cmds:
            self._log_debug("unpacked command: " + soscmd.__str__())
            self._log_info("collecting output of '%s'" % soscmd.cmd)
            self._collect_cmd_output(**soscmd.__dict__)

    def _collect_strings(self):
        for string, file_name in self.copy_strings:
            if self._timeout_hit:
                return
            content = ''
            if string:
                content = string.splitlines()[0]
                if not isinstance(content, six.string_types):
                    content = content.decode('utf8', 'ignore')
            self._log_info("collecting string ...'%s' as '%s'"
                           % (content, file_name))
            try:
                self.archive.add_string(string,
                                        os.path.join('sos_strings',
                                                     self.name(),
                                                     file_name))
            except Exception as e:
                self._log_debug("could not add string '%s': %s"
                                % (file_name, e))

    def collect(self):
        """Collect the data for a plugin."""
        start = time()
        self._collect_copy_specs()
        self._collect_cmds()
        self._collect_strings()
        fields = (self.name(), time() - start)
        self._log_debug("collected plugin '%s' in %s" % fields)

    def get_description(self):
        """ This function will return the description for the plugin"""
        try:
            if hasattr(self, '__doc__') and self.__doc__:
                return self.__doc__.strip()
            return super(self.__class__, self).__doc__.strip()
        except Exception:
            return "<no description available>"

    def check_enabled(self):
        """This method will be used to verify that a plugin should execute
        given the condition of the underlying environment.

        The default implementation will return True if none of class.files,
        class.packages, nor class.commands is specified. If any of these is
        specified the plugin will check for the existence of any of the
        corresponding paths, packages or commands and return True if any
        are present.

        For SCLPlugin subclasses, it will check whether the plugin can be run
        for any of installed SCLs. If so, it will store names of these SCLs
        on the plugin class in addition to returning True.

        For plugins with more complex enablement checks this method may be
        overridden.
        """
        # some files or packages have been specified for this package
        if any([self.files, self.packages, self.commands, self.kernel_mods,
                self.services, self.architectures]):
            if isinstance(self.files, six.string_types):
                self.files = [self.files]

            if isinstance(self.packages, six.string_types):
                self.packages = [self.packages]

            if isinstance(self.commands, six.string_types):
                self.commands = [self.commands]

            if isinstance(self.kernel_mods, six.string_types):
                self.kernel_mods = [self.kernel_mods]

            if isinstance(self.services, six.string_types):
                self.services = [self.services]

            if isinstance(self, SCLPlugin):
                # save SCLs that match files or packages
                type(self)._scls_matched = []
                for scl in self._get_scls():
                    files = [f % {"scl_name": scl} for f in self.files]
                    packages = [p % {"scl_name": scl} for p in self.packages]
                    commands = [c % {"scl_name": scl} for c in self.commands]
                    services = [s % {"scl_name": scl} for s in self.services]
                    if self._check_plugin_triggers(files,
                                                   packages,
                                                   commands,
                                                   services):
                        type(self)._scls_matched.append(scl)
                return len(type(self)._scls_matched) > 0

            return self._check_plugin_triggers(self.files,
                                               self.packages,
                                               self.commands,
                                               self.services)

        if isinstance(self, SCLPlugin):
            # if files and packages weren't specified, we take all SCLs
            type(self)._scls_matched = self._get_scls()

        return True

    def _check_plugin_triggers(self, files, packages, commands, services):
        return ((any(os.path.exists(fname) for fname in files) or
                any(self.is_installed(pkg) for pkg in packages) or
                any(is_executable(cmd) for cmd in commands) or
                any(self.is_module_loaded(mod) for mod in self.kernel_mods) or
                any(self.is_service(svc) for svc in services)) and
                self.check_is_architecture())

    def check_is_architecture(self):
        """Checks whether or not the system is running on an architecture that
        the plugin allows. If not architecture is set, assume plugin can run
        on all arches.
        """
        if self.architectures is None:
            return True
        regex = '(?:%s)' % '|'.join(self.architectures)
        return re.match(regex, self.policy.get_arch())

    def default_enabled(self):
        """This decides whether a plugin should be automatically loaded or
        only if manually specified in the command line."""
        return True

    def add_default_collections(self):
        """Based on the class attrs defined for plugin enablement, add a
        standardized set of collections before we call the plugin's own setup()
        method.
        """
        # For any service used for enablement checks, collect its current
        # status if it exists
        for service in self.services:
            if self.is_service(service):
                self.add_service_status(service)

    def setup(self):
        """Collect the list of files declared by the plugin. This method
        may be overridden to add further copy_specs, forbidden_paths, and
        external programs if required.
        """
        self.add_copy_spec(list(self.files))

    def setup_verify(self):
        if not hasattr(self, "verify_packages") or not self.verify_packages:
            if hasattr(self, "packages") and self.packages:
                # Limit automatic verification to only the named packages
                self.verify_packages = [p + "$" for p in self.packages]
            else:
                return

        pm = self.policy.package_manager
        verify_cmd = pm.build_verify_command(self.verify_packages)
        if verify_cmd:
            self.add_cmd_output(verify_cmd)

    def postproc(self):
        """Perform any postprocessing. To be replaced by a plugin if required.
        """
        pass

    def check_process_by_name(self, process):
        """Checks if a named process is found in /proc/[0-9]*/cmdline.
        Returns either True or False."""
        status = False
        cmd_line_glob = "/proc/[0-9]*/cmdline"
        try:
            cmd_line_paths = glob.glob(cmd_line_glob)
            for path in cmd_line_paths:
                f = open(path, 'r')
                cmd_line = f.read().strip()
                if process in cmd_line:
                    status = True
        except IOError as e:
            return False
        return status

    def get_process_pids(self, process):
        """Returns PIDs of all processes with process name.
        If the process doesn't exist, returns an empty list"""
        pids = []
        cmd_line_glob = "/proc/[0-9]*/cmdline"
        cmd_line_paths = glob.glob(cmd_line_glob)
        for path in cmd_line_paths:
            try:
                with open(path, 'r') as f:
                    cmd_line = f.read().strip()
                    if process in cmd_line:
                        pids.append(path.split("/")[2])
            except IOError as e:
                continue
        return pids


class RedHatPlugin(object):
    """Tagging class for Red Hat's Linux distributions"""
    pass


class SCLPlugin(RedHatPlugin):
    """Superclass for plugins operating on Software Collections (SCLs).

    Subclasses of this plugin class can specify class.files and class.packages
    using "%(scl_name)s" interpolation. The plugin invoking mechanism will try
    to match these against all found SCLs on the system. SCLs that do match
    class.files or class.packages are then accessible via self.scls_matched
    when the plugin is invoked.

    Additionally, this plugin class provides "add_cmd_output_scl" (run
    a command in context of given SCL), and "add_copy_spec_scl" and
    "add_copy_spec_limit_scl" (copy package from file system of given SCL).

    For example, you can implement a plugin that will list all global npm
    packages in every SCL that contains "npm" package:

    class SCLNpmPlugin(Plugin, SCLPlugin):
        packages = ("%(scl_name)s-npm",)

        def setup(self):
            for scl in self.scls_matched:
                self.add_cmd_output_scl(scl, "npm ls -g --json")
    """

    @property
    def scls_matched(self):
        if not hasattr(type(self), '_scls_matched'):
            type(self)._scls_matched = []
        return type(self)._scls_matched

    def _get_scls(self):
        output = sos_get_command_output("scl -l")["output"]
        return [scl.strip() for scl in output.splitlines()]

    def convert_cmd_scl(self, scl, cmd):
        """wrapping command in "scl enable" call and adds proper PATH
        """
        # load default SCL prefix to PATH
        prefix = self.policy.get_default_scl_prefix()
        # read prefix from /etc/scl/prefixes/${scl} and strip trailing '\n'
        try:
            prefix = open('/etc/scl/prefixes/%s' % scl, 'r').read()\
                     .rstrip('\n')
        except Exception as e:
            self._log_error("Failed to find prefix for SCL %s, using %s"
                            % (scl, prefix))

        # expand PATH by equivalent prefixes under the SCL tree
        path = os.environ["PATH"]
        for p in path.split(':'):
            path = '%s/%s%s:%s' % (prefix, scl, p, path)

        scl_cmd = "scl enable %s \"PATH=%s %s\"" % (scl, path, cmd)
        return scl_cmd

    def add_cmd_output_scl(self, scl, cmds, **kwargs):
        """Same as add_cmd_output, except that it wraps command in
        "scl enable" call and sets proper PATH.
        """
        if isinstance(cmds, six.string_types):
            cmds = [cmds]
        scl_cmds = []
        for cmd in cmds:
            scl_cmds.append(self.convert_cmd_scl(scl, cmd))
        self.add_cmd_output(scl_cmds, **kwargs)

    # config files for Software Collections are under /etc/${prefix}/${scl} and
    # var files are under /var/${prefix}/${scl} where the ${prefix} is distro
    # specific path. So we need to insert the paths after the appropriate root
    # dir.
    def convert_copyspec_scl(self, scl, copyspec):
        scl_prefix = self.policy.get_default_scl_prefix()
        for rootdir in ['etc', 'var']:
            p = re.compile('^/%s/' % rootdir)
            copyspec = p.sub('/%s/%s/%s/' % (rootdir, scl_prefix, scl),
                             copyspec)
        return copyspec

    def add_copy_spec_scl(self, scl, copyspecs):
        """Same as add_copy_spec, except that it prepends path to SCL root
        to "copyspecs".
        """
        if isinstance(copyspecs, six.string_types):
            copyspecs = [copyspecs]
        scl_copyspecs = []
        for copyspec in copyspecs:
            scl_copyspecs.append(self.convert_copyspec_scl(scl, copyspec))
        self.add_copy_spec(scl_copyspecs)

    def add_copy_spec_limit_scl(self, scl, copyspec, **kwargs):
        """Same as add_copy_spec_limit, except that it prepends path to SCL
        root to "copyspec".
        """
        self.add_copy_spec_limit(
            self.convert_copyspec_scl(scl, copyspec),
            **kwargs
        )


class PowerKVMPlugin(RedHatPlugin):
    """Tagging class for IBM PowerKVM Linux"""
    pass


class ZKVMPlugin(RedHatPlugin):
    """Tagging class for IBM ZKVM Linux"""
    pass


class UbuntuPlugin(object):
    """Tagging class for Ubuntu Linux"""
    pass


class DebianPlugin(object):
    """Tagging class for Debian Linux"""
    pass


class SuSEPlugin(object):
    """Tagging class for SuSE Linux distributions"""
    pass


class IndependentPlugin(object):
    """Tagging class for plugins that can run on any platform"""
    pass


class ExperimentalPlugin(object):
    """Tagging class that indicates that this plugin is experimental"""
    pass


def import_plugin(name, superclasses=None):
    """Import name as a module and return a list of all classes defined in that
    module. superclasses should be a tuple of valid superclasses to import,
    this defaults to (Plugin,).
    """
    plugin_fqname = "sos.plugins.%s" % name
    if not superclasses:
        superclasses = (Plugin,)
    return import_module(plugin_fqname, superclasses)

# vim: set et ts=4 sw=4 :