python3-plainbox 0.25-1 / usr / lib / python3 / dist-packages / plainbox / abc.py

# This file is part of Checkbox.
#
# Copyright 2012-2014 Canonical Ltd.
# Written by:
#   Zygmunt Krynicki <zygmunt.krynicki@canonical.com>
#
# Checkbox is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License version 3,
# as published by the Free Software Foundation.
#
# Checkbox 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 General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with Checkbox.  If not, see <http://www.gnu.org/licenses/>.
"""
:mod:`plainbox.abc` -- abstract base classes
============================================

Those classes are actually implemented in the plainbox.impl package. This
module is here so that the essential API concepts are in a single spot and are
easier to understand (by not being mixed with additional source code).

.. note::

    This module has API stability guarantees. We are not going to break or
    introduce backwards incompatible interfaces here without following our API
    deprecation policy. All existing features will be retained for at least
    three releases. All deprecated symbols will warn when they will cease to be
    available.
"""
from abc import ABCMeta, abstractproperty, abstractmethod


class ITextSource(metaclass=ABCMeta):
    """
    An abstract source of text.

    Concrete instances of this class are used by
    :class:`plainbox.impl.rfc822.Origin` to keep track of where each entry of a
    RFC822-like document came from.
    """

    @abstractmethod
    def relative_to(self, path):
        """
        Compute a new text source of the same kind with any filesystem
        references translated to be relative from the specified base directory.

        :param base_dir:
            A base directory name
        :returns:
            A new text source
        """


class IJobDefinition(metaclass=ABCMeta):
    """
    Job definition that contains a mixture of meta-data and executable
    information that can be consumed by the job runner to produce results.
    """

    # XXX: All IO methods to save/load this would be in a helper class/function
    # that would also handle format detection, serialization and validation.

    @abstractproperty
    def plugin(self):
        """
        Name of the job interpreter.

        Various interpreters are provided by the job runner.
        """

    @abstractproperty
    def name(self):
        """
        Name of the job
        """

    @abstractproperty
    def summary(self):
        """
        Short (one line) description of the job
        """

    @abstractproperty
    def id(self):
        """
        Unique job identifier

        The identifier is unique within the provider a job belongs to
        """

    @abstractproperty
    def requires(self):
        """
        List of expressions that need to be true for this job to be available

        This value can be None
        """

    @abstractproperty
    def shell(self):
        """
        Shell that is used to interpret the command

        Defaults to 'bash' for checkbox compatibility.
        """

    @abstractproperty
    def command(self):
        """
        The shell command to execute to perform the job.

        The return code, standard output and standard error streams are
        automatically recorded and processed, depending on the plugin type.

        This value can be None
        """

    @abstractproperty
    def description(self):
        """
        Human-readable description of the job.

        This field is typically used to include execution and verification
        steps for manual and human-assisted tests.

        This value can be None
        """

    @abstractproperty
    def depends(self):
        """
        Comma-delimited dependency expression

        This field can be used to express job dependencies. If a job depends on
        another job it can only start if the other job had ran and succeeded.

        This is the original data as provided when constructed. Use
        get_direct_dependencies() to obtain the parsed equivalent.

        This value can be None
        """

    @abstractproperty
    def provider(self):
        """
        The provider this job definition belongs to

        .. note::
            Technically this still can be None (a provider-less job may exist)
            but it can only happen in testing. This mode is discouraged and
            will be eventually forbidden. All job definition units must belong
            to a provider.
        """

    @abstractproperty
    def purpose(self):
        """
        Human readable purpose of the test.
        """

    @abstractproperty
    def steps(self):
        """
        Human readable instruction what actions should user perform while
        performing test
        """

    @abstractproperty
    def verification(self):
        """
        Human readable instruction how to verify outcome of a test
        """


class IJobResult(metaclass=ABCMeta):
    """
    Class for representing results from a single job
    """

    # XXX: We could also store stuff like job duration and other meta-data but
    # I wanted to avoid polluting this proposal with mundane details

    # The outcome of a job is a one-word classification how how it ran.  There
    # are several values that were not used in the original implementation but
    # their existence helps to organize and implement plainbox. They are
    # discussed below to make their intended meaning more detailed than is
    # possible from the variable name alone.
    #
    # The None outcome - a job that basically did not run at all.
    OUTCOME_NONE = None
    # The pass and fail outcomes are the two most essential, and externally
    # visible, job outcomes. They can be provided by either automated or manual
    # "classifier" - a script or a person that clicks a "pass" or "fail"
    # button.
    OUTCOME_PASS = 'pass'
    OUTCOME_FAIL = 'fail'
    # The skip outcome is used when the operator selected a job but then
    # skipped it. This is typically used for a manual job that is tedious or
    # was selected by accident.
    OUTCOME_SKIP = 'skip'
    # The not supported outcome is used when a job was about to run but a
    # dependency or resource requirement prevent it from running.  XXX: perhaps
    # this should be called "not available", not supported has the "unsupported
    # code" feeling associated with it.
    OUTCOME_NOT_SUPPORTED = 'not-supported'
    # A temporary state that should be removed later on, used to indicate that
    # job runner is not implemented but the job "ran" so to speak.
    OUTCOME_NOT_IMPLEMENTED = 'not-implemented'
    # A temporary state before the user decides on the outcome of a manual
    # job or any other job that requires manual verification
    OUTCOME_UNDECIDED = 'undecided'
    # A kind of failed that indicates the underlying test misbehaved. Currently
    # it is only used when the test program is killed by a signal.
    OUTCOME_CRASH = 'crash'

    @abstractproperty
    def outcome(self):
        """
        Outcome of the test.

        The result of either automatic or manual verification. Depending on the
        plugin (test type). Available values are defined as class properties
        above.
        """

    @abstractproperty
    def comments(self):
        """
        The comment that was added by the user, if any
        """

    @abstractproperty
    def io_log(self):
        """
        A sequence of tuples (delay, stream-name, data) where delay is the
        delay since the previous message seconds (typically a fractional
        number), stream name is either 'stdout' or 'stderr' and data is the
        bytes object that was obtained from that stream.
        """
        # XXX: it could also encode 'stdin' if the user was presented with a
        # console to type in and we sent that to the process.

        # XXX: This interface is low-level but captures everything that has
        # occurred and is text-safe. You can call an utility function to
        # convert that to a text string that most closely represents what a
        # user would see, having ran this command in the terminal.

    @abstractproperty
    def return_code(self):
        """
        Command return code.

        This is the return code of the process started to execute the command
        from the job definition. It can also encode the signal that the
        process was killed with, if any.
        """

    @abstractmethod
    def get_io_log(self):
        """
        Compute and return the sequence of IOLogRecord objects.

        :returns:
            A sequence of tuples (delay, stream-name, data) where delay is the
            delay since the previous message seconds (typically a fractional
            number), stream name is either 'stdout' or 'stderr' and data is the
            bytes object that was obtained from that stream.
        """

    @abstractproperty
    def is_hollow(self):
        """
        flag that indicates if the result is hollow

        Hollow results may have been created but hold no data at all.
        Hollow results are also tentatively deprecated, once we have some
        time to re-factor SessionState and specifically the job_state_map
        code we will remove the need to have hollow results.

        Hollow results are not saved, beginning with
        :class:`plainbox.impl.session.suspend.SessionSuspendHelper4`.
        """


class IJobQualifier(metaclass=ABCMeta):
    """
    An opaque qualifier for a job definition.

    This is an abstraction for matching jobs definitions to names, patterns and
    other means of selecting jobs.

    There are two ways to use a qualifier object. The naive, direct, old API
    can simply check if a qualifier designates a particular job (if it selects
    it and marks for subsequent execution). This API works fine for certain
    tasks but it was found that it is insufficient to implement so-called
    whitelist ordering, where the order of jobs in a whitelist is preserved
    when selecting that whitelist for execution. This spawned the second,
    lower-level API, that gives portable visibility into composite qualifiers
    (such as a whitelist) and distinct select, deselect vote so that full range
    of current expressiveness can be preserved.

    :attr VOTE_EXCLUDE:
        (0) vote indicating that a job should *not* be included for
        selection. It overwrites any other votes.

    :attr VOTE_INCLUDE:
        (1) vote indicating that a job should be included for selection. It is
        overridden by VOTE_EXCLUDE.

    :attr VOTE_IGNORE:
        (2) vote indicating that a job should neither be included nor excluded
        for selection. This is a neutral value overridden by all other votes.
    """

    # NOTE: VOTE_xxx are sorted by priority, lowest being the most important
    # one.  When multiple votes are cast, the one with lowest value (highest
    # priority) takes precedence. When adding additional votes keep this in
    # mind.
    VOTE_EXCLUDE = 0
    VOTE_INCLUDE = 1
    VOTE_IGNORE = 2

    @abstractmethod
    def get_vote(self, job):
        """
        Get one of the :attr:`VOTE_IGNORE`, :attr:`VOTE_INCLUDE`,
        :attr:`VOTE_EXCLUDE` votes that this qualifier associated with the
        specified job.

        :param job:
            A IJobDefinition instance that is to be visited
        :returns:
            one of the ``VOTE_xxx`` constants

        .. versionadded: 0.5
        """

    @abstractmethod
    def get_primitive_qualifiers(self):
        """
        Return a list of primitives that constitute this qualifier.

        :returns:
            A list of IJobQualifier objects that each is the smallest,
            indivisible entity.

        When each vote cast by those qualifiers is applied sequentially to
        a given job then the result is the same as the return value of the
        :meth:`designates()` method. The resulting list has more structure
        and this structure may matter to job ordering when a list of jobs
        is matched against a list of qualifiers. The resulting sets are
        identical but ordering of results is more accurately reflected by
        iterating over the fine structure of each qualifier.

        .. versionadded: 0.5
        """

    @abstractproperty
    def is_primitive(self):
        """
        property indicating that a qualifier is not divisible by calling
        :meth:`get_primitive_qualifiers()`.

        If a qualifier is not primitive it can be replaced with a list of
        qualifiers it produces by the call to the aforementioned method.

        .. versionadded: 0.5
        """

    @abstractmethod
    def designates(self, job):
        """
        Check if this qualifier designates the specified
        :class:`plainbox.abc.IJobDefinition`

        :returns True: if the qualifier designates the specified job
        :returns False: otherwise
        """

    @abstractproperty
    def origin(self):
        """
        Origin of this qualifier

        :raises NonPrimitiveQualifierOrigin:
            If the :meth:`is_primitive` property is True and this qualifier
            has no well-defined origin of itself.

        This property can be used to trace the origin of a qualifier back to
        its definition point. Note that it may not always be available when
        it doesn't make sense to say that a composite object came from any one
        particular place.
        """


class IJobRunner(metaclass=ABCMeta):
    """
    Something that can run a job definition and produce results.

    You can run many jobs with one runner, each time you'll get additional
    result object. Typically you will need to connect the runner to a user
    interface but headless mode is also possible.
    """

    @abstractmethod
    def run_job(self, job, config=None, ui=None):
        """
        Run the specified job.

        Calling this method may block for arbitrary amount of time. User
        interfaces should ensure that it runs in a separate thread.

        The return value is a JobResult object that contains all the data that
        was captured during the execution of the job. Some jobs may not return
        a JobResult value.
        """
        # XXX: threads suck, could we make this fully asynchronous? The only
        # thing that we really want is to know when the command has stopped
        # executing. We could expose the underlying process mechanics so that
        # QT/GTK applications could tie that directly into their event loop.

    @abstractmethod
    def get_warm_up_sequence(self, job_list):
        """
        Determine if authentication warm-up may be needed.

        :param job_lits:
            A list of jobs that may be executed
        :returns:
            A list of methods to call to complete the warm-up step.

        Authentication warm-up is related to the plainbox-secure-launcher-1
        program that can be 'warmed-up' to perhaps cache the security
        credentials. This is usually done early in the testing process so that
        we can prompt for passwords before doing anything that takes an
        extended amount of time.
        """


class IJobRunnerUI(metaclass=ABCMeta):
    """
    User interface (textual) related to running a single job
    """

    @abstractmethod
    def considering_job(self, job, job_state):
        """
        Method called as the runner is considering the specified job can
        run or not.
        """

    @abstractmethod
    def about_to_start_running(self, job, job_state):
        """
        Method called as the runner has decided to run the job and is
        getting ready to start.
        """

    @abstractmethod
    def wait_for_interaction_prompt(self, job):
        """
        Method called only for user-interact and user-interact-verify jobs
        that should instruct the user to read the description (that needs
        to be displayed somehow) and confirm before the test is actually
        started. The user should be one of the few choices listed below:

        quit:
            save and quit
        run:
            run the job
        skip:
            skip and continue

        :returns:
            The action selected by the user.
        """

    @abstractmethod
    def started_running(self, job, job_state):
        """
        Method called immediately before the runner starts to run the job
        """

    @abstractmethod
    def about_to_execute_program(self, args, kwargs):
        """
        Method called just prior to execute an external program.

        :param args:
            Same as for subprocess.call
        :param kwargs:
            Same as for subprocess.call
        """

    @abstractmethod
    def got_program_output(self, stream_name, line):
        """
        Method called on every line of output from an external program

        :param stream_name:
            either 'stdin' or 'stdout'
        :param line:
            the full text of the intercepted line
        """

    @abstractmethod
    def finished_executing_program(self, returncode):
        """
        Method called just after running an external program

        :param returncode:
            The return code of the external program
        """

    @abstractmethod
    def notify_about_description(self, job):
        """
        Method called prior to user interactions that might require familiarity
        of the job description. Gets called only if purpose/steps/verification
        fields are missing.
        """

    @abstractmethod
    def notify_about_purpose(self, job):
        """
        Method called after starting a job
        """

    @abstractmethod
    def notify_about_steps(self, job):
        """
        Method called prior to user interactions that might require familiarity
        with interaction steps
        """

    @abstractmethod
    def notify_about_verification(self, job):
        """
        Method called before user selects outcome of a job
        """

    @abstractmethod
    def finished_running(self, job, job_state, job_result):
        """
        Method called immediately after the runner finishes to run the job
        """

    @abstractmethod
    def job_cannot_start(self, job, job_state, job_result):
        """
        Method called when a job cannot be started
        """

    @abstractmethod
    def finished(self, job, job_state, job_result):
        """
        Method called at the end of the process, regardless if the job was
        actually started or not
        """

    @abstractmethod
    def pick_action_cmd(self, action_list, prompt=None):
        """
        Present a list of actions and let the user pick one

        :param action_list:
            A list of 3-tuples (accel, label, cmd)
        :prompt:
            An optional prompt string
        :returns:
            cmd of the selected action or None
        """

    @abstractmethod
    def noreturn_job(self):
        """
        Method called when job that is about to run has 'noreturn' flag
        (plainbox will suspend operation after running that job).
        """


class IUserInterfaceIO(metaclass=ABCMeta):
    """
    Base class that allows job runner to interact with the user interface.
    """

    @abstractmethod
    def get_manual_verification_outcome(self):
        """
        Get the outcome of the manual verification, as according to the user
        May raise NotImplementedError if the user interface cannot provide this
        answer.
        """


class IProviderBackend1(metaclass=ABCMeta):
    """
    Provider for the current type of tests.

    This class provides the APIs required by the internal implementation
    that are not considered normal public APIs. The only consumer of the
    those methods and properties are internal to plainbox.
    """

    @abstractproperty
    def CHECKBOX_SHARE(self):
        """
        Return the required value of CHECKBOX_SHARE environment variable.

        .. note::
            This variable is only required by one script.
            It would be nice to remove this later on.
        """

    @abstractproperty
    def extra_PYTHONPATH(self):
        """
        Return additional entry for PYTHONPATH, if needed.

        This entry is required for CheckBox scripts to import the correct
        CheckBox python libraries.

        .. note::
            The result may be None
        """

    @abstractproperty
    def bin_dir(self):
        """
        directory where all the executables needed by this provider reside
        """

    @abstractproperty
    def jobs_dir(self):
        """
        Return an absolute path of the jobs directory
        """

    @abstractproperty
    def whitelists_dir(self):
        """
        Return an absolute path of the whitelist directory
        """

    @abstractproperty
    def data_dir(self):
        """
        absolute path of the data directory
        """

    @abstractproperty
    def base_dir(self):
        """
        absolute path of the base directory with other directories.

        May be None
        """

    @abstractproperty
    def units_dir(self):
        """
        absolute path of the units directory

        May be None
        """

    @abstractproperty
    def secure(self):
        """
        flag indicating that this provider was loaded from the secure portion
        of PROVIDERPATH and thus can be used with the
        plainbox-trusted-launcher-1.
        """


class IProvider1(IProviderBackend1):
    """
    Provider for the current type of tests

    Also known as the 'checkbox-like' provider.
    """

    @abstractproperty
    def name(self):
        """
        name of this provider

        This name should be dbus-friendly. It should not be localizable.
        """

    @abstractproperty
    def namespace(self):
        """
        namespace component of the provider name

        This property defines the namespace in which all provider jobs are
        defined in. Jobs within one namespace do not need to be fully qualified
        by prefixing their partial identifier with provider namespace (so all
        stays 'as-is'). Jobs that need to interact with other provider
        namespaces need to use the fully qualified job identifier instead.

        The identifier is defined as the part of the provider name, up to the
        colon. This effectively gives organizations flat namespace within one
        year-domain pair and allows to create private namespaces by using
        sub-domains.
        """

    @abstractproperty
    def version(self):
        """
        version of this provider
        """

    @abstractproperty
    def description(self):
        """
        description of this provider
        """

    @abstractproperty
    def gettext_domain(self):
        """
        the name of the gettext domain associated with this provider

        This value may be empty, in such case provider data cannot be localized
        for the user environment.
        """

    @abstractproperty
    def locale_dir(self):
        """
        absolute path of the directory with locale data

        The value is applicable as argument bindtextdomain()
        """

    @abstractproperty
    def unit_list(self):
        """
        List of loaded units.
        """

    @abstractproperty
    def job_list(self):
        """
        List of loaded job definition units.
        """

    @abstractproperty
    def executable_list(self):
        """
        List of all the executables
        """

    @abstractproperty
    def whitelist_list(self):
        """
        List of loaded whitelists.

        .. warning::
            :class:`WhiteList` is currently deprecated. You should never need
            to access them in any new code.  They are entirely replaced by
            :class:`TestPlan`. This property is provided for completeness and
            it will be **removed** once whitelists classes are no longer used.
        """

    @abstractproperty
    def problem_list(self):
        """
        list of problems encountered by the loading process
        """

    @abstractproperty
    def id_map(self):
        """
        A mapping from unit identifier to list of units with that identifier.

        .. note::
            Typically the list will be one element long but invalid providers
            may break that guarantee. Code defensively if you can.
        """

    @abstractproperty
    def path_map(self):
        """
        A mapping from filename path to a list of units stored in that file.

        .. note::
            For ``.pxu`` files this will enumerate all units stored there. For
            other things it will typically be just the FileUnit.
        """


class ISessionStateController(metaclass=ABCMeta):
    """
    Interface for session state controller classes.

    Session state controller classes cooperate with
    :class:`~plainbox.impl.session.state.SessionState` and
    :class:`~plainbox.impl.depmgr.DependencySolver` classes and implement
    knowledge unique to particular job semantics. Before execution the
    controller can influence job runnability (by setting inhibitors). After
    execution the controller can observe the result and influence session state
    """

    @abstractmethod
    def get_dependency_set(self, job):
        """
        Get the set of direct dependencies of a particular job.

        :param job:
            A IJobDefinition instance that is to be visited
        :returns:
            set of pairs (dep_type, job_name)

        Returns a set of pairs (dep_type, job_name) that describe all
        dependencies of the specified job. The first element in the pair,
        dep_type, is either DEP_TYPE_DIRECT, DEP_TYPE_ORDERING or
        DEP_TYPE_RESOURCE. The second element is the name of the job.
        """

    @abstractmethod
    def get_inhibitor_list(self, session_state, job):
        """
        Get a list of readiness inhibitors that inhibit a particular job.

        :param session_state:
            A SessionState instance that is used to interrogate the
            state of the session where it matters for a particular
            job. Currently this is used to access resources and job
            results.
        :param job:
            A JobDefinition instance
        :returns:
            List of JobReadinessInhibitor
        """

    @abstractmethod
    def observe_result(self, session_state, job, result):
        """
        Notice the specified test result and update readiness state.

        :param session_state:
            A SessionState object
        :param job:
            A JobDefinition object
        :param result:
            A IJobResult object

        This function updates the internal result collection with the data from
        the specified test result. Results can safely override older results.
        Results also change the ready map (jobs that can run) because of
        dependency relations.
        """


class IExecutionController(metaclass=ABCMeta):
    """
    Interface for job execution controller clases.

    Execution controllers encapsulate knowledge on how to run command
    associated with a particular job. Some executors might run the command
    directly, others might delegate the task to a helper program or perform
    some special-cased customization to the execution environment.
    """

    @abstractmethod
    def execute_job(self, job, job_state, config, session_dir, extcmd_popen):
        """
        Execute the specified job using the specified subprocess-like object

        :param job:
            The JobDefinition to execute
        :param job_state:
            The JobState associated to the job to execute.
        :param config:
            A PlainBoxConfig instance which can be used to load missing
            environment definitions that apply to all jobs. It is used to
            provide values for missing environment variables that are required
            by the job (as expressed by the environ key in the job definition
            file).
        :param session_dir:
            Base directory of the session this job will execute in.
            This directory is used to co-locate some data that is unique to
            this execution as well as data that is shared by all executions.
        :param extcmd_popen:
            A subprocess.Popen like object
        :returns:
            The return code of the command, as returned by subprocess.call()
        """

    @abstractmethod
    def get_score(self, job):
        """
        Compute how applicable this controller is for the specified job.

        :returns:
            A numeric score, or None if the controller cannot run this job.
            The higher the value, the more applicable this controller is.
        """

    @abstractmethod
    def get_warm_up_for_job(self, job):
        """
        Get a warm-up function that should be called before running this job.

        :returns:
            A callable (without arguments) or None, depending on needs of a
            particular job.

        The warm-up function is an optional advisory interface to improve the
        testing experience for the user. A job may not require any warm-up. In
        such case the return value is None. Note that even if this function is
        not called the testing process should perform the same way (correctly)
        but the user may be prompted for additional steps mid-way.
        """


class IBuildSystem(metaclass=ABCMeta):
    """
    A pluggable build system definition

    PlainBox uses build systems to assist provider authors in
    building additional executables from source code. To facilitate
    support for a specific language or toolkit a build system may
    detect it and offer proper commands without the test developer
    having to copy/paste those commands from provider to provider.

    PlainBox discovers providers from the ``plainbox.buildsystem``
    entry point. Each entry point there must be a class implementing
    this interface.
    """

    @abstractmethod
    def probe(self, src_dir: str) -> int:
        """
        Look at the source directory and determine how applicable this build
        system is.

        :param src_dir:
            absolute path of the directory with source code
        :returns:
            the suitability value, 0 if the build system doesn't support
            sources of the particular kind, all values greater than zero
            indicate some level of suitability. The largest return value wins.

        The return value is a number. Values closer to zero mean that the build
        system is not suitable, values closer to one mean the build system is
        more suitable. Value of 0 mean that the build system is totally
        unsuitable and will not be applied, even if no other choices are
        available.

        The idea is that multiple build systems may recognize a source
        directory but since the system is extensible, other people may come up
        with more suitable build system that spots additional files and returns
        a score better than the average.
        """

    @abstractmethod
    def get_build_command(self, src_dir: str, build_dir: str) -> str:
        """
        Get shell command to build the sources.

        :param src_dir:
            absolute path of the source directory
        :param build_dir:
            absolute path of the build directory
        :returns:
            shell command to execute

        With the given source and build directory, come up with a piece of
        shell that knows how to build stuff so that it ends up in the build
        directory.

        .. note::
            The command will be executed in build_dir.
        """


class ISessionStateExporter(metaclass=ABCMeta):
    """
    Interface for classes that export session state to a byte stream.

    Exporters write out the state of the session after all jobs have finished
    running, in a user-selected format. The intent is not to preserve
    everything that the session may hold but instead to present it to the user
    in the best format possible.

    Each exporter can support a set of options that can alter the way it
    operates. Options can either be set boolean-like, or they can be assigned a
    value (a string). If an option contains a "=", the part of the string on
    the right of the equal sign will be assigned as the option's value;
    otherwise they operate in a boolean fashion.

    It's best to keep the list of exporter options under control to keep the
    user interface from becoming annoying.
    """

    @abstractmethod
    def __init__(self, option_list=None, exporter_unit=None):
        """
        Initialize the session exporter object

        :param option_list:
            a list of option to enable
        :param exporter_unit:
            an ExporterUnit object
        """

    @abstractmethod
    # @classproperty -- this is a class-level property
    def supported_option_list(cls):
        """
        Return the list of supported options
        """

    @abstractmethod
    def get_session_data_subset(self, session_manager):
        """
        Compute a subset of session data.

        The subset of the data that should be saved may depend on a particular
        saver class and options selected by the user.

        Must return a collection that can be handled by :meth:`dump()`.
        Special care must be taken when processing io_log (and in the future,
        attachments) as those can be arbitrarily large.
        """

    @abstractmethod
    def dump(self, data, stream):
        """
        Dump data to stream.

        This method operates on data that was returned by
        :meth:`get_session_data_subset()`. It may not really process bytes or
        simple collections. Instead, for efficiency, anything is required.

        As in get_session_data_subset() it's essential to safely save
        arbitrarily large data sets (or actually, only where it matters the
        most, like in io_log).

        Data is a text stream suitable for writing.
        """
        # TODO: Add a way for the stream to be binary as well.

    @abstractmethod
    def dump_from_session_manager(self, session_manager, stream):
        """
        Dump session information pulled from session manager to stream.

        This method takes session manager instance, extracts session
        information from it, and dumps it to a stream.
        """


class ISessionStateTransport(metaclass=ABCMeta):
    """
    Interface for transports that send test data somewhere.

    They handle just the transmission portion of data sending; exporters are
    expected to produce data in the proper format (e.g. json, xml).

    Each transport can have specific parameters that are required for the
    other end to properly process received information (like system
    identification, authorization data and so on), and that don't semantically
    belong in the test data as produced by the exporter. Additionally
    each transport needs to be told *where* to send test data. This is
    transport-dependent; things like a HTTP endpoint, IP address, port
    are good examples.
    """

    @abstractmethod
    def __init__(self, where, option_string):
        """
        Initialize the transport object

        :param where:
            a string encoding the destination location
        :param option_string:
            a string of additonal configuration for the transport
        :raises ValueError:
            if any of the arguments are somehow invalid
        """

    @abstractmethod
    def send(self, data, config=None, session_state=None):
        """
        Send data somewhere.

        :param data:
            a stream-like object of data to send (read only)
        :param config:
            a PlainBoxConfig object (optional)
        :param session_state:
            the session for which this transport is associated with
            the data being sent (optional)
        :raises ValueError:
            if any of the arguments are somehow invalid
        :raises TransportError:
            if any transport-specific problem arises
        :returns:
            a dictionary with additional items, see notes below

        .. note::
            The return value is especially vague specifically to allow various
            transports to express whatever they may need to express for a
            particular vertical use case yet still to allow most of the code to
            just work with all transports.

            It is expected that certain keys in the returned dictionary will
            gain special semantics that can be further standardized. As of this
            writing there are no standard keys.
        """