Skip to content

cmflib.cmfquery.CmfQuery

Bases: object

CMF Query communicates with the MLMD database and implements basic search and retrieval functionality.

This class has been designed to work with the CMF framework. CMF alters names of pipelines, stages and artifacts in various ways. This means that actual names in the MLMD database will be different from those originally provided by users via CMF API. When methods in this class accept name parameters, it is expected that values of these parameters are fully-qualified names of respective entities.

Parameters:

Name Type Description Default
filepath str

Path to the MLMD database file.

'mlmd'
Source code in cmflib/cmfquery.py
120
121
122
123
124
125
126
127
def __init__(self, filepath: str = "mlmd", is_server=False) -> None:
    temp_store: t.Union[PostgresStore, SqlliteStore]
    if is_server:
        config_dict = get_postgres_config()
        temp_store = PostgresStore(config_dict)
    else:
        temp_store = SqlliteStore({"filename": filepath})
    self.store = temp_store.connect()

get_pipeline_names()

Return names of all pipelines.

Returns:

Type Description
List[str]

List of all pipeline names.

Source code in cmflib/cmfquery.py
364
365
366
367
368
369
370
def get_pipeline_names(self) -> t.List[str]:
    """Return names of all pipelines.

    Returns:
        List of all pipeline names.
    """
    return [ctx.name for ctx in self._get_pipelines()]

get_pipeline_id(pipeline_name)

Return pipeline identifier for the pipeline names pipeline_name. Args: pipeline_name: Name of the pipeline. Returns: Pipeline identifier or -1 if one does not exist.

Source code in cmflib/cmfquery.py
372
373
374
375
376
377
378
379
380
def get_pipeline_id(self, pipeline_name: str) -> int:
    """Return pipeline identifier for the pipeline names `pipeline_name`.
    Args:
        pipeline_name: Name of the pipeline.
    Returns:
        Pipeline identifier or -1 if one does not exist.
    """
    pipeline: t.Optional[mlpb.Context] = self._get_pipeline(pipeline_name)  # type: ignore  # Context type not recognized by mypy, using ignore to bypass
    return -1 if not pipeline else pipeline.id

get_pipeline_stages(pipeline_name)

Return list of pipeline stages for the pipeline with the given name.

Parameters:

Name Type Description Default
pipeline_name str

Name of the pipeline for which stages need to be returned. In CMF, there are no different pipelines with the same name.

required

Returns: List of stage names associated with the given pipeline.

Source code in cmflib/cmfquery.py
382
383
384
385
386
387
388
389
390
391
392
393
394
def get_pipeline_stages(self, pipeline_name: str) -> t.List[str]:
    """Return list of pipeline stages for the pipeline with the given name.

    Args:
        pipeline_name: Name of the pipeline for which stages need to be returned. In CMF, there are no different
            pipelines with the same name.
    Returns:
        List of stage names associated with the given pipeline.
    """
    stages:t.List[str] = []
    for pipeline in self._get_pipelines(pipeline_name):
        stages.extend(stage.name for stage in self._get_stages(pipeline.id))
    return stages

get_all_exe_in_stage(stage_name)

Return list of all executions for the stage with the given name.

Parameters:

Name Type Description Default
stage_name str

Name of the stage. Before stages are recorded in MLMD, they are modified (e.g., pipeline name will become part of the stage name). So stage names from different pipelines will not collide.

required

Returns: List of executions for the given stage.

Source code in cmflib/cmfquery.py
396
397
398
399
400
401
402
403
404
405
406
407
408
409
def get_all_exe_in_stage(self, stage_name: str) -> t.List[mlpb.Execution]:  # type: ignore  # Execution type not recognized by mypy, using ignore to bypass
    """Return list of all executions for the stage with the given name.

    Args:
        stage_name: Name of the stage. Before stages are recorded in MLMD, they are modified (e.g., pipeline name
                    will become part of the stage name). So stage names from different pipelines will not collide.
    Returns:
        List of executions for the given stage.
    """
    for pipeline in self._get_pipelines():
        for stage in self._get_stages(pipeline.id):
            if stage.name == stage_name:
                return self.store.get_executions_by_context(stage.id)
    return []

get_all_executions_by_ids_list(exe_ids)

Return executions for given execution ids list as a pandas data frame.

Parameters:

Name Type Description Default
exe_ids List[int]

List of execution identifiers.

required

Returns:

Type Description
DataFrame

Data frame with all executions for the list of given execution identifiers.

Source code in cmflib/cmfquery.py
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
def get_all_executions_by_ids_list(self, exe_ids: t.List[int]) -> pd.DataFrame:
    """Return executions for given execution ids list as a pandas data frame.

    Args:
        exe_ids: List of execution identifiers.

    Returns:
        Data frame with all executions for the list of given execution identifiers.
    """

    df = pd.DataFrame()
    executions = self.store.get_executions_by_id(exe_ids)
    for exe in executions:
        d1 = self._transform_to_dataframe(exe)
        df = pd.concat([df, d1], sort=True, ignore_index=True)
    return df

get_all_artifacts_by_context(pipeline_name)

Return artifacts for given pipeline name as a pandas data frame.

Parameters:

Name Type Description Default
pipeline_name str

Name of the pipeline.

required

Returns:

Type Description
DataFrame

Data frame with all artifacts associated with given pipeline name.

Source code in cmflib/cmfquery.py
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
def get_all_artifacts_by_context(self, pipeline_name: str) -> pd.DataFrame:
    """Return artifacts for given pipeline name as a pandas data frame.

    Args:
        pipeline_name: Name of the pipeline.

    Returns:
        Data frame with all artifacts associated with given pipeline name.
    """
    df = pd.DataFrame()
    contexts = self.store.get_contexts_by_type("Parent_Context")
    context_id = self.get_pipeline_id(pipeline_name)
    for ctx in contexts:
        if ctx.id == context_id:
            child_contexts = self.store.get_children_contexts_by_context(ctx.id)
            for cc in child_contexts:
                artifacts = self.store.get_artifacts_by_context(cc.id)
                for art in artifacts:
                    d1 = self.get_artifact_df(art)
                    df = pd.concat([df, d1], sort=True, ignore_index=True)
    return df

get_all_artifacts_by_ids_list(artifact_ids)

Return all artifacts for the given artifact ids list.

Parameters:

Name Type Description Default
artifact_ids List[int]

List of artifact identifiers

required

Returns:

Type Description
DataFrame

Data frame with all artifacts for the given artifact ids list.

Source code in cmflib/cmfquery.py
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
def get_all_artifacts_by_ids_list(self, artifact_ids: t.List[int]) -> pd.DataFrame:
    """Return all artifacts for the given artifact ids list.

    Args:
        artifact_ids: List of artifact identifiers

    Returns:
        Data frame with all artifacts for the given artifact ids list.
    """
    df = pd.DataFrame()
    artifacts = self.store.get_artifacts_by_id(artifact_ids)
    for art in artifacts:
        d1 = self.get_artifact_df(art)
        df = pd.concat([df, d1], sort=True, ignore_index=True)
    return df

get_all_executions_in_stage(stage_name)

Return executions of the given stage as pandas data frame. Args: stage_name: Stage name. See doc strings for the prev method. Returns: Data frame with all executions associated with the given stage.

Source code in cmflib/cmfquery.py
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
def get_all_executions_in_stage(self, stage_name: str) -> pd.DataFrame:
    """Return executions of the given stage as pandas data frame.
    Args:
        stage_name: Stage name. See doc strings for the prev method.
    Returns:
        Data frame with all executions associated with the given stage.
    """
    df = pd.DataFrame()
    for pipeline in self._get_pipelines():
        for stage in self._get_stages(pipeline.id):
            if stage.name == stage_name:
                for execution in self._get_executions(stage.id):
                    ex_as_df: pd.DataFrame = self._transform_to_dataframe(
                        execution, {"id": execution.id, "name": execution.name}
                    )
                    df = pd.concat([df, ex_as_df], sort=True, ignore_index=True)
    return df

get_artifact_df(artifact, d=None)

Return artifact's data frame representation.

Parameters:

Name Type Description Default
artifact Artifact

MLMD entity representing artifact.

required
d Optional[Dict]

Optional initial content for data frame.

None

Returns: A data frame with the single row containing attributes of this artifact.

Source code in cmflib/cmfquery.py
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
def get_artifact_df(self, artifact: mlpb.Artifact, d: t.Optional[t.Dict] = None) -> pd.DataFrame:   # type: ignore  # Artifact type not recognized by mypy, using ignore to bypass
    """Return artifact's data frame representation.

    Args:
        artifact: MLMD entity representing artifact.
        d: Optional initial content for data frame.
    Returns:
        A data frame with the single row containing attributes of this artifact.
    """
    if d is None:
        d = {}
    d.update(
        {
            "id": artifact.id,
            "type": self.store.get_artifact_types_by_id([artifact.type_id])[0].name,
            "uri": artifact.uri,
            "name": artifact.name,
            "create_time_since_epoch": artifact.create_time_since_epoch,
            "last_update_time_since_epoch": artifact.last_update_time_since_epoch,
        }
    )
    return self._transform_to_dataframe(artifact, d)

get_all_artifacts()

Return names of all artifacts.

Returns:

Type Description
List[str]

List of all artifact names.

Source code in cmflib/cmfquery.py
507
508
509
510
511
512
513
def get_all_artifacts(self) -> t.List[str]:
    """Return names of all artifacts.

    Returns:
        List of all artifact names.
    """
    return [artifact.name for artifact in self.store.get_artifacts()]

get_artifact(name)

Return artifact's data frame representation using artifact name.

Parameters:

Name Type Description Default
name str

Artifact name.

required

Returns: Pandas data frame with one row containing attributes of this artifact.

Source code in cmflib/cmfquery.py
517
518
519
520
521
522
523
524
525
526
527
528
def get_artifact(self, name: str) -> t.Optional[pd.DataFrame]:
    """Return artifact's data frame representation using artifact name.

    Args:
        name: Artifact name.
    Returns:
        Pandas data frame with one row containing attributes of this artifact.
    """
    artifact: t.Optional[mlpb.Artifact] = self._get_artifact(name)  # type: ignore  # Artifact type not recognized by mypy, using ignore to bypass
    if artifact:
        return self.get_artifact_df(artifact)
    return None

get_all_artifacts_for_execution(execution_id)

Return input and output artifacts for the given execution.

Parameters:

Name Type Description Default
execution_id int

Execution identifier.

required

Return: Data frame containing input and output artifacts for the given execution, one artifact per row.

Source code in cmflib/cmfquery.py
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
def get_all_artifacts_for_execution(self, execution_id: int) -> pd.DataFrame:
    """Return input and output artifacts for the given execution.

    Args:
        execution_id: Execution identifier.
    Return:
        Data frame containing input and output artifacts for the given execution, one artifact per row.
    """
    df = pd.DataFrame()
    for event in self.store.get_events_by_execution_ids([execution_id]):
        event_type = "INPUT" if event.type == mlpb.Event.Type.INPUT else "OUTPUT"   # type: ignore  # Event type not recognized by mypy, using ignore to bypass
        for artifact in self.store.get_artifacts_by_id([event.artifact_id]):
            df = pd.concat(
                [df, self.get_artifact_df(artifact, {"event": event_type})], sort=True, ignore_index=True
            )
    return df

get_all_artifact_types()

Return names of all artifact types.

Returns:

Type Description
List[str]

List of all artifact types.

Source code in cmflib/cmfquery.py
547
548
549
550
551
552
553
554
555
def get_all_artifact_types(self) -> t.List[str]:
    """Return names of all artifact types.

    Returns:
        List of all artifact types.
    """
    artifact_list = self.store.get_artifact_types()
    types=[i.name for i in artifact_list]
    return types

get_all_executions_for_artifact(artifact_name)

Return executions that consumed and produced given artifact.

Parameters:

Name Type Description Default
artifact_name str

Artifact name.

required

Returns: Pandas data frame containing stage executions, one execution per row.

Source code in cmflib/cmfquery.py
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
def get_all_executions_for_artifact(self, artifact_name: str) -> pd.DataFrame:
    """Return executions that consumed and produced given artifact.

    Args:
        artifact_name: Artifact name.
    Returns:
        Pandas data frame containing stage executions, one execution per row.
    """
    df = pd.DataFrame()

    artifact: t.Optional[mlpb.Artifact] = self._get_artifact(artifact_name) # type: ignore  # Artifact type not recognized by mypy, using ignore to bypass
    if not artifact:
        return df

    for event in self.store.get_events_by_artifact_ids([artifact.id]):
        stage_ctx = self.store.get_contexts_by_execution(event.execution_id)[0]
        linked_execution = {
            "Type": "INPUT" if event.type == mlpb.Event.Type.INPUT else "OUTPUT",   # type: ignore  # Event type not recognized by mypy, using ignore to bypass
            "execution_id": event.execution_id,
            "execution_name": self.store.get_executions_by_id([event.execution_id])[0].name,
            "execution_type_name":self.store.get_executions_by_id([event.execution_id])[0].properties['Execution_type_name'],
            "stage": stage_ctx.name,
            "pipeline": self.store.get_parent_contexts_by_context(stage_ctx.id)[0].name,
        }
        d1 = pd.DataFrame(
            linked_execution,
            index=[
                0,
            ],
        )
        df = pd.concat([df, d1], sort=True, ignore_index=True)
    return df

get_one_hop_child_artifacts(artifact_name, pipeline_id=None)

Get artifacts produced by executions that consume given artifact.

Parameters:

Name Type Description Default
artifact name

Name of an artifact.

required

Return: Output artifacts of all executions that consumed given artifact.

Source code in cmflib/cmfquery.py
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
def get_one_hop_child_artifacts(self, artifact_name: str, pipeline_id: t.Optional[int] = None) -> pd.DataFrame:
    """Get artifacts produced by executions that consume given artifact.

    Args:
        artifact name: Name of an artifact.
    Return:
        Output artifacts of all executions that consumed given artifact.
    """
    artifact: t.Optional[mlpb.Artifact] = self._get_artifact(artifact_name)    # type: ignore  # Artifact type not recognized by mypy, using ignore to bypass
    if not artifact:
        return pd.DataFrame()

    # Get output artifacts of executions consumed the above artifact.
    artifacts_ids = self._get_output_artifacts(self._get_executions_by_input_artifact_id(artifact.id, pipeline_id))
    return self._as_pandas_df(
        self.store.get_artifacts_by_id(artifacts_ids), lambda _artifact: self.get_artifact_df(_artifact)
    )

get_all_child_artifacts(artifact_name)

Return all downstream artifacts starting from the given artifact.

Parameters:

Name Type Description Default
artifact_name str

Artifact name.

required

Returns: Data frame containing all child artifacts.

Source code in cmflib/cmfquery.py
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
def get_all_child_artifacts(self, artifact_name: str) -> pd.DataFrame:
    """Return all downstream artifacts starting from the given artifact.

    Args:
        artifact_name: Artifact name.
    Returns:
        Data frame containing all child artifacts.
    """
    df = pd.DataFrame()
    d1 = self.get_one_hop_child_artifacts(artifact_name)
    # df = df.append(d1, sort=True, ignore_index=True)
    df = pd.concat([df, d1], sort=True, ignore_index=True)
    for row in d1.itertuples():
        d1 = self.get_all_child_artifacts(str(row.name))    # Convert row.name to string to ensure compatibility with get_all_child_artifacts method
        # df = df.append(d1, sort=True, ignore_index=True)
        df = pd.concat([df, d1], sort=True, ignore_index=True)
    df = df.drop_duplicates(subset=None, keep="first", inplace=False)
    return df

get_one_hop_parent_artifacts(artifact_name)

Return input artifacts for the execution that produced the given artifact. Args: artifact_name: Artifact name. Returns: Data frame containing immediate parent artifactog of given artifact.

Source code in cmflib/cmfquery.py
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
def get_one_hop_parent_artifacts(self, artifact_name: str) -> pd.DataFrame:
    """Return input artifacts for the execution that produced the given artifact.
    Args:
        artifact_name: Artifact name.
    Returns:
        Data frame containing immediate parent artifactog of given artifact.
    """
    artifact: t.Optional[mlpb.Artifact] = self._get_artifact(artifact_name) # type: ignore  # Artifact type not recognized by mypy, using ignore to bypass
    if not artifact:
        return pd.DataFrame()

    artifact_ids: t.List[int] = self._get_input_artifacts(self._get_executions_by_output_artifact_id(artifact.id))

    return self._as_pandas_df(
        self.store.get_artifacts_by_id(artifact_ids), lambda _artifact: self.get_artifact_df(_artifact)
    )

get_all_parent_artifacts(artifact_name)

Return all upstream artifacts. Args: artifact_name: Artifact name. Returns: Data frame containing all parent artifacts.

Source code in cmflib/cmfquery.py
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
def get_all_parent_artifacts(self, artifact_name: str) -> pd.DataFrame:
    """Return all upstream artifacts.
    Args:
        artifact_name: Artifact name.
    Returns:
        Data frame containing all parent artifacts.
    """
    df = pd.DataFrame()
    d1 = self.get_one_hop_parent_artifacts(artifact_name)
    # df = df.append(d1, sort=True, ignore_index=True)
    df = pd.concat([df, d1], sort=True, ignore_index=True)
    for row in d1.itertuples():
        d1 = self.get_all_parent_artifacts(str(row.name))   # Convert row.name to string to ensure compatibility with get_all_parent_artifacts method
        # df = df.append(d1, sort=True, ignore_index=True)
        df = pd.concat([df, d1], sort=True, ignore_index=True)
    df = df.drop_duplicates(subset=None, keep="first", inplace=False)
    return df

get_all_parent_executions(artifact_name)

Return all executions that produced upstream artifacts for the given artifact. Args: artifact_name: Artifact name. Returns: Data frame containing all parent executions.

Source code in cmflib/cmfquery.py
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
def get_all_parent_executions(self, artifact_name: str) -> pd.DataFrame:
    """Return all executions that produced upstream artifacts for the given artifact.
    Args:
        artifact_name: Artifact name.
    Returns:
        Data frame containing all parent executions.
    """
    parent_artifacts: pd.DataFrame = self.get_all_parent_artifacts(artifact_name)
    if parent_artifacts.shape[0] == 0:
        # If it's empty, there's no `id` column and the code below raises an exception.
        return pd.DataFrame()

    execution_ids = set(
        event.execution_id
        for event in self.store.get_events_by_artifact_ids([int(id) for id in parent_artifacts.id.values.tolist()])
        if event.type == mlpb.Event.OUTPUT  # type: ignore  # Event type not recognized by mypy, using ignore to bypass
    )

    return self._as_pandas_df(
        self.store.get_executions_by_id(execution_ids),
        lambda _exec: self._transform_to_dataframe(_exec, {"id": _exec.id, "name": _exec.name}),
    )

get_metrics(metrics_name)

Return metric data frame. Args: metrics_name: Metrics name. Returns: Data frame containing all metrics.

Source code in cmflib/cmfquery.py
810
811
812
813
814
815
816
817
818
819
820
821
822
823
def get_metrics(self, metrics_name: str) -> t.Optional[pd.DataFrame]:
    """Return metric data frame.
    Args:
        metrics_name: Metrics name.
    Returns:
        Data frame containing all metrics.
    """
    for metric in self.store.get_artifacts_by_type("Step_Metrics"):
        if metric.name == metrics_name:
            name: t.Optional[str] = metric.custom_properties.get("Name", None)
            if name:
                return pd.read_parquet(name)
            break
    return None

dumptojson(pipeline_name, exec_uuid=None)

Return JSON-parsable string containing details about the given pipeline. Args: pipeline_name: Name of an AI pipelines. exec_uuid: Optional stage execution_uuid - filter stages by this execution_uuid. Returns: Pipeline in JSON format.

Source code in cmflib/cmfquery.py
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
def dumptojson(self, pipeline_name: str, exec_uuid: t.Optional[str] = None) -> t.Optional[str]:
    """Return JSON-parsable string containing details about the given pipeline.
    Args:
        pipeline_name: Name of an AI pipelines.
        exec_uuid: Optional stage execution_uuid - filter stages by this execution_uuid.
    Returns:
        Pipeline in JSON format.
    """
    def _get_node_attributes(_node: t.Union[mlpb.Context, mlpb.Execution, mlpb.Event], _attrs: t.Dict) -> t.Dict:   # type: ignore  # Context type not recognized by mypy, using ignore to bypass
        for attr in CONTEXT_LIST:
            #Artifacts getattr call on Type was giving empty string, which was overwriting 
            # the defined types such as Dataset, Metrics, Models
            if getattr(_node, attr, None) is not None and not getattr(_node, attr, None) == "":
                _attrs[attr] = getattr(_node, attr)

        if "properties" in _attrs:
            _attrs["properties"] = CmfQuery._copy(_attrs["properties"])
        if "custom_properties" in _attrs:
            # TODO: (sergey) why do we need to rename "type" to "user_type" if we just copy into a new dictionary?
            _attrs["custom_properties"] = CmfQuery._copy(
                _attrs["custom_properties"], key_mapper={"type": "user_type"}
            )
        return _attrs

    pipelines: t.List[t.Dict] = []
    for pipeline in self._get_pipelines(pipeline_name):
        pipeline_attrs = _get_node_attributes(pipeline, {"stages": []})
        for stage in self._get_stages(pipeline.id):
            stage_attrs = _get_node_attributes(stage, {"executions": []})
            for execution in self.get_all_executions_by_stage(stage.id, execution_uuid=exec_uuid):
                # name will be an empty string for executions that are created with
                # create new execution as true(default)
                # In other words name property will there only for execution
                # that are created with create new execution flag set to false(special case)
                exec_attrs = _get_node_attributes(
                    execution,
                    {
                        "type": self.store.get_execution_types_by_id([execution.type_id])[0].name,
                        "name": execution.name if execution.name != "" else "",
                        "events": [],
                    },
                )
                for event in self.store.get_events_by_execution_ids([execution.id]):
                    event_attrs = _get_node_attributes(event, {})
                    # An event has only a single Artifact associated with it. 
                    # For every artifact we create an event to link it to the execution.

                    artifacts =  self.store.get_artifacts_by_id([event.artifact_id])
                    artifact_attrs = _get_node_attributes(
                            artifacts[0], {"type": self.store.get_artifact_types_by_id([artifacts[0].type_id])[0].name}
                        )
                    event_attrs["artifact"] = artifact_attrs
                    exec_attrs["events"].append(event_attrs)
                stage_attrs["executions"].append(exec_attrs)
            pipeline_attrs["stages"].append(stage_attrs)
        pipelines.append(pipeline_attrs)

    return json.dumps({"Pipeline": pipelines})