Extend leappdb for leapp-inspector

This commit combines multiple additions to the leapp database.

First addition is tracking of entity metadata. The `Metadata` model
stores the metadata of entities such as `Actor` or `Workflow`. This data
is stored in a new table `metadata` of the `leapp.db` file.

  1. metadata of *discovered* actors.

     For an actor, the metadata stored contain:

	`class_name` 	- the name of the actor class
	`name`		- the name given to the actor
	`description`	- the actor's description
	`phase` 	- phase of execution of the actor
	`tags`		- names of any tags associated with an actor
	`consumes` 	- list of all messages the actor consumes
	`produces`	- list of all messages the actor produces
	`path`		- the path to the actor source file

  2. workflow metadata.

     For a workflow, the metadata stored contain:

       `name`		- name of the workflow
       `short_name` 	- short name of the workflow
       `tag`		- workflow tag
       `description`	- workflow description
       `phases` 	- all phases associated with the workflow

Next addition is tracking of dialog question. Previously leapp was not
able to detect the actual question asked from the user as it could be
generated dynamically when actor is called and depend on the
configuration of the user's system.

Last addition includes storing the actor exit status. Exit status is now
saved as an audit event `actor-exit-status`. Exit status 0 represents
successful execution or `StopActorExecution`/`StopActorExecutionError`,
while 1 indicates an unexpected and unhandled exception.

These changes collectively improve the metadata handling capabilities
of, ensuring accurate storage and retrieval of essential information for
various entities.

leapp/actors/__init__.py

@@ -10,6 +10,7 @@
 from leapp.models.error_severity import ErrorSeverity
 from leapp.tags import Tag
 from leapp.utils import get_api_models, path
+from leapp.utils.audit import store_dialog
 from leapp.utils.i18n import install_translation_for_actor
 from leapp.utils.meta import get_flattened_subclasses
 from leapp.workflows.api import WorkflowAPI
@@ -122,12 +123,17 @@ def get_answers(self, dialog):
         :return: dictionary with the requested answers, None if not a defined dialog
         """
         self._messaging.register_dialog(dialog, self)
+        answer = None
         if dialog in type(self).dialogs:
             if self.skip_dialogs:
                 # non-interactive mode of operation
-                return self._messaging.get_answers(dialog)
-            return self._messaging.request_answers(dialog)
-        return None
+                answer = self._messaging.get_answers(dialog)
+            else:
+                answer = self._messaging.request_answers(dialog)
+
+        store_dialog(dialog, answer)
+
+        return answer
 
     def show_message(self, message):
         """
@@ -285,6 +291,7 @@ def get_actor_tool_path(self, name):
     def run(self, *args):
         """ Runs the actor calling the method :py:func:`process`. """
         os.environ['LEAPP_CURRENT_ACTOR'] = self.name
+
         try:
             self.process(*args)
         except StopActorExecution:

leapp/dialogs/dialog.py

@@ -114,4 +114,5 @@ def request_answers(self, store, renderer):
             self._store = store
             renderer.render(self)
             self._store = None
+
         return store.get(self.scope, {})

leapp/messaging/answerstore.py

@@ -117,6 +117,9 @@ def get(self, scope, fallback=None):
         # NOTE(ivasilev) self.storage.get() will return a DictProxy. To avoid TypeError during later
         # JSON serialization a copy() should be invoked to get a shallow copy of data
         answer = self._storage.get(scope, fallback).copy()
+
+        # NOTE(dkubek): It is possible that we do not need to save the 'answer'
+        # here as it is being stored with dialog question right after query
         create_audit_entry('dialog-answer', {'scope': scope, 'fallback': fallback, 'answer': answer})
         return answer
 

leapp/utils/audit/__init__.py

@@ -221,6 +221,49 @@ def do_store(self, connection):
         self._data_source_id = cursor.fetchone()[0]
 
 
+class Metadata(Host):
+    """
+    Metadata of an entity (e.g. actor, workflow)
+    """
+
+    def __init__(self, context=None, hostname=None, kind=None, metadata=None, name=None):
+        """
+        :param context: The execution context
+        :type context: str
+        :param hostname: Hostname of the system that produced the entry
+        :type hostname: str
+        :param kind: Kind of the entity for which metadata is stored
+        :type kind: str
+        :param metadata: Actual metadata
+        :type metadata: dict
+        :param name: Name of the entity
+        :type name: str
+        """
+        super(Metadata, self).__init__(context=context, hostname=hostname)
+        self.kind = kind
+        self.name = name
+        self.metadata = metadata
+        self._metadata_id = None
+
+    @property
+    def metadata_id(self):
+        """
+        Returns the id of the entry, which is only set when already stored.
+        :return: Integer id or None
+        """
+        return self._metadata_id
+
+    def do_store(self, connection):
+        super(Metadata, self).do_store(connection)
+        connection.execute(
+            'INSERT OR IGNORE INTO metadata (context, kind, name, metadata) VALUES(?, ?, ?, ?)',
+            (self.context, self.kind, self.name, json.dumps(self.metadata)))
+        cursor = connection.execute(
+            'SELECT id FROM metadata WHERE context = ? AND kind = ? AND name = ?',
+            (self.context, self.kind, self.name))
+        self._metadata_id = cursor.fetchone()[0]
+
+
 class Message(DataSource):
     def __init__(self, stamp=None, msg_type=None, topic=None, data=None, actor=None, phase=None,
                  hostname=None, context=None):
@@ -267,6 +310,47 @@ def do_store(self, connection):
         self._message_id = cursor.lastrowid
 
 
+class Dialog(DataSource):
+    """
+    Stores information about dialog questions and their answers
+    """
+
+    def __init__(self, scope=None, data=None, actor=None, phase=None, hostname=None, context=None):
+        """
+        :param scope: Dialog scope
+        :type scope: str
+        :param data: Payload data
+        :type data: dict
+        :param actor: Name of the actor that triggered the entry
+        :type actor: str
+        :param phase: In which phase of the workflow execution the dialog was triggered
+        :type phase: str
+        :param hostname: Hostname of the system that produced the message
+        :type hostname: str
+        :param context: The execution context
+        :type context: str
+        """
+        super(Dialog, self).__init__(actor=actor, phase=phase, hostname=hostname, context=context)
+        self.scope = scope or ''
+        self.data = data
+        self._dialog_id = None
+
+    @property
+    def dialog_id(self):
+        """
+        Returns the id of the entry, which is only set when already stored.
+        :return: Integer id or None
+        """
+        return self._dialog_id
+
+    def do_store(self, connection):
+        super(Dialog, self).do_store(connection)
+        cursor = connection.execute(
+            'INSERT OR IGNORE INTO dialog (context, scope, data, data_source_id) VALUES(?, ?, ?, ?)',
+            (self.context, self.scope, json.dumps(self.data), self.data_source_id))
+        self._dialog_id = cursor.lastrowid
+
+
 def create_audit_entry(event, data, message=None):
     """
     Create an audit entry
@@ -291,10 +375,10 @@ def get_audit_entry(event, context):
     """
     Retrieve audit entries stored in the database for the given context
 
-    :param context: The execution context
-    :type context: str
     :param event: Event type identifier
     :type event: str
+    :param context: The execution context
+    :type context: str
     :return: list of dicts with id, time stamp, actor and phase fields
     """
     with get_connection(None) as conn:
@@ -470,3 +554,81 @@ def get_checkpoints(context):
         ''', (context, _AUDIT_CHECKPOINT_EVENT))
         cursor.row_factory = dict_factory
         return cursor.fetchall()
+
+
+def store_dialog(dialog, answer):
+    """
+    Store ``dialog`` with accompanying ``answer``.
+
+    :param dialog: instance of a workflow to store.
+    :type dialog: :py:class:`leapp.dialogs.Dialog`
+    :param answer: Answer to for each component of the dialog
+    :type answer: dict
+    """
+
+    component_keys = ('key', 'label', 'description', 'default', 'value', 'reason')
+    dialog_keys = ('title', 'reason')  # + 'components'
+
+    tmp = dialog.serialize()
+    data = {
+        'components': [dict((key, component[key]) for key in component_keys) for component in tmp['components']],
+
+        # NOTE(dkubek): Storing answer here is redundant as it is already
+        # being stored in audit when we query from the answerstore, however,
+        # this keeps the information coupled with the question more closely
+        'answer': answer
+    }
+    data.update((key, tmp[key]) for key in dialog_keys)
+
+    e = Dialog(
+        scope=dialog.scope,
+        data=data,
+        context=os.environ['LEAPP_EXECUTION_ID'],
+        actor=os.environ['LEAPP_CURRENT_ACTOR'],
+        phase=os.environ['LEAPP_CURRENT_PHASE'],
+        hostname=os.environ['LEAPP_HOSTNAME'],
+    )
+    e.store()
+
+    return e
+
+
+def store_workflow_metadata(workflow):
+    """
+    Store the metadata of the given ``workflow`` into the database.
+
+    :param workflow: Workflow to store.
+    :type workflow: :py:class:`leapp.workflows.Workflow`
+    """
+
+    metadata = type(workflow).serialize()
+    md = Metadata(kind='workflow', name=workflow.name, context=os.environ['LEAPP_EXECUTION_ID'],
+                  hostname=os.environ['LEAPP_HOSTNAME'], metadata=metadata)
+    md.store()
+
+
+def store_actor_metadata(actor_definition, phase):
+    """
+    Store the metadata of the given actor given as an ``actor_definition``
+    object into the database.
+
+    :param actor_definition: Actor to store
+    :type actor_definition: :py:class:`leapp.repository.actor_definition.ActorDefinition`
+    """
+
+    metadata = dict(actor_definition.discover())
+    metadata.update({
+        'consumes': [model.__name__ for model in metadata.get('consumes', ())],
+        'produces': [model.__name__ for model in metadata.get('produces', ())],
+        'tags': [tag.__name__ for tag in metadata.get('tags', ())],
+    })
+    metadata['phase'] = phase
+
+    actor_metadata_fields = (
+        'class_name', 'name', 'description', 'phase', 'tags', 'consumes', 'produces', 'path'
+    )
+    md = Metadata(kind='actor', name=actor_definition.name,
+                  context=os.environ['LEAPP_EXECUTION_ID'],
+                  hostname=os.environ['LEAPP_HOSTNAME'],
+                  metadata={field: metadata[field] for field in actor_metadata_fields})
+    md.store()

leapp/workflows/__init__.py

@@ -11,7 +11,7 @@
 from leapp.messaging.commands import SkipPhasesUntilCommand
 from leapp.tags import ExperimentalTag
 from leapp.utils import reboot_system
-from leapp.utils.audit import checkpoint, get_errors
+from leapp.utils.audit import checkpoint, get_errors, create_audit_entry, store_workflow_metadata, store_actor_metadata
 from leapp.utils.meta import with_metaclass, get_flattened_subclasses
 from leapp.utils.output import display_status_current_phase, display_status_current_actor
 from leapp.workflows.phases import Phase
@@ -165,7 +165,7 @@ def __init__(self, logger=None, auto_reboot=False):
         self.description = self.description or type(self).__doc__
 
         for phase in self.phases:
-            phase.filter.tags += (self.tag,)
+            phase.filter.tags += (self.tag,) if self.tag not in phase.filter.tags else ()
             self._phase_actors.append((
                 phase,
                 # filters all actors with the give tags
@@ -279,6 +279,8 @@ def run(self, context=None, until_phase=None, until_actor=None, skip_phases_unti
         self.log.info('Starting workflow execution: {name} - ID: {id}'.format(
             name=self.name, id=os.environ['LEAPP_EXECUTION_ID']))
 
+        store_workflow_metadata(self)
+
         skip_phases_until = (skip_phases_until or '').lower()
         needle_phase = until_phase or ''
         needle_stage = None
@@ -295,6 +297,12 @@ def run(self, context=None, until_phase=None, until_actor=None, skip_phases_unti
             if phase and not self.is_valid_phase(phase):
                 raise CommandError('Phase {phase} does not exist in the workflow'.format(phase=phase))
 
+        # Save metadata of all discovered actors
+        for phase in self._phase_actors:
+            for stage in phase[1:]:
+                for actor in stage.actors:
+                    store_actor_metadata(actor, phase[0].name)
+
         self._stop_after_phase_requested = False
         for phase in self._phase_actors:
             os.environ['LEAPP_CURRENT_PHASE'] = phase[0].name
@@ -332,10 +340,12 @@ def run(self, context=None, until_phase=None, until_actor=None, skip_phases_unti
                     display_status_current_actor(actor, designation=designation)
                     current_logger.info("Executing actor {actor} {designation}".format(designation=designation,
                                                                                        actor=actor.name))
+
                     messaging = InProcessMessaging(config_model=config_model, answer_store=self._answer_store)
                     messaging.load(actor.consumes)
                     instance = actor(logger=current_logger, messaging=messaging,
                                      config_model=config_model, skip_dialogs=skip_dialogs)
+
                     try:
                         instance.run()
                     except BaseException as exc:
@@ -346,6 +356,14 @@ def run(self, context=None, until_phase=None, until_actor=None, skip_phases_unti
                         current_logger.error('Actor {actor} has crashed: {trace}'.format(actor=actor.name,
                                                                                          trace=exc.exception_info))
                         raise
+                    finally:
+                        # Set and unset the enviromental variable so that audit
+                        # associates the entry with the correct data source
+                        os.environ['LEAPP_CURRENT_ACTOR'] = actor.name
+                        create_audit_entry(
+                            event='actor-exit-status',
+                            data={'exit_status': 1 if self._unhandled_exception else 0})
+                        os.environ.pop('LEAPP_CURRENT_ACTOR')
 
                     self._stop_after_phase_requested = messaging.stop_after_phase or self._stop_after_phase_requested
 

res/schema/audit-layout.sql

@@ -1,6 +1,6 @@
 BEGIN;
 
-PRAGMA user_version = 2;
+PRAGMA user_version = 3;
 
 CREATE TABLE IF NOT EXISTS execution (
   id            INTEGER PRIMARY KEY NOT NULL,
@@ -43,6 +43,24 @@ CREATE TABLE IF NOT EXISTS message (
 );
 
 
+CREATE TABLE IF NOT EXISTS metadata (
+  id                INTEGER PRIMARY KEY NOT NULL,
+  context           VARCHAR(36)         NOT NULL REFERENCES execution (context),
+  kind              VARCHAR(256)        NOT NULL DEFAULT '',
+  name              VARCHAR(1024)       NOT NULL DEFAULT '',
+  metadata          TEXT                         DEFAULT NULL,
+  UNIQUE (context, kind, name)
+);
+
+CREATE TABLE IF NOT EXISTS dialog (
+  id                INTEGER PRIMARY KEY NOT NULL,
+  context           VARCHAR(36)         NOT NULL REFERENCES execution (context),
+  scope             VARCHAR(1024)       NOT NULL DEFAULT '',
+  data              TEXT                         DEFAULT NULL,
+  data_source_id    INTEGER             NOT NULL REFERENCES data_source (id)
+);
+
+
 CREATE TABLE IF NOT EXISTS audit (
   id             INTEGER PRIMARY KEY NOT NULL,
   event          VARCHAR(256)        NOT NULL REFERENCES execution (context),
@@ -74,4 +92,4 @@ CREATE VIEW IF NOT EXISTS messages_data AS
     host                     ON host.id                   = data_source.host_id
 ;
 
-COMMIT;
+COMMIT;

res/schema/migrations/0002-add-metadata-dialog-tables.sql

@@ -0,0 +1,22 @@
+BEGIN;
+
+CREATE TABLE metadata (
+  id                INTEGER PRIMARY KEY NOT NULL,
+  context           VARCHAR(36)         NOT NULL REFERENCES execution (context),
+  kind              VARCHAR(256)        NOT NULL DEFAULT '',
+  name              VARCHAR(1024)       NOT NULL DEFAULT '',
+  metadata          TEXT                         DEFAULT NULL,
+  UNIQUE (context, kind, name)
+);
+
+CREATE TABLE dialog (
+  id                INTEGER PRIMARY KEY NOT NULL,
+  context           VARCHAR(36)         NOT NULL REFERENCES execution (context),
+  scope             VARCHAR(1024)       NOT NULL DEFAULT '',
+  data              TEXT                         DEFAULT NULL,
+  data_source_id    INTEGER             NOT NULL REFERENCES data_source (id)
+);
+
+PRAGMA user_version = 3;
+
+COMMIT;

tests/data/leappdb-tests/.leapp/info

@@ -0,0 +1 @@
+{"name": "workflow-tests", "id": "07005707-67bc-46e5-9732-a10fb13d4e7d"}

tests/data/leappdb-tests/.leapp/leapp.conf

@@ -0,0 +1,6 @@
+
+[repositories]
+repo_path=${repository:root_dir}
+
+[database]
+path=${repository:state_dir}/leapp.db