acryl-datahub 1.0.0.2rc3__py3-none-any.whl → 1.0.0.2rc4__py3-none-any.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Potentially problematic release.
This version of acryl-datahub might be problematic. Click here for more details.
- {acryl_datahub-1.0.0.2rc3.dist-info → acryl_datahub-1.0.0.2rc4.dist-info}/METADATA +2536 -2536
- {acryl_datahub-1.0.0.2rc3.dist-info → acryl_datahub-1.0.0.2rc4.dist-info}/RECORD +46 -47
- datahub/_version.py +1 -1
- datahub/api/entities/datajob/dataflow.py +15 -0
- datahub/api/entities/dataprocess/dataprocess_instance.py +4 -0
- datahub/api/entities/dataset/dataset.py +2 -2
- datahub/api/entities/structuredproperties/structuredproperties.py +1 -1
- datahub/cli/migrate.py +6 -6
- datahub/configuration/common.py +1 -1
- datahub/ingestion/api/common.py +9 -0
- datahub/ingestion/api/source.py +4 -1
- datahub/ingestion/api/source_helpers.py +26 -1
- datahub/ingestion/run/pipeline.py +0 -6
- datahub/ingestion/source/aws/sagemaker_processors/models.py +4 -4
- datahub/ingestion/source/bigquery_v2/lineage.py +1 -1
- datahub/ingestion/source/dynamodb/dynamodb.py +1 -1
- datahub/ingestion/source/fivetran/fivetran.py +1 -0
- datahub/ingestion/source/fivetran/fivetran_log_api.py +1 -1
- datahub/ingestion/source/iceberg/iceberg.py +97 -9
- datahub/ingestion/source/kafka/kafka.py +1 -4
- datahub/ingestion/source/kafka_connect/sink_connectors.py +1 -1
- datahub/ingestion/source/kafka_connect/source_connectors.py +1 -1
- datahub/ingestion/source/looker/looker_source.py +2 -3
- datahub/ingestion/source/mlflow.py +3 -0
- datahub/ingestion/source/mode.py +2 -2
- datahub/ingestion/source/nifi.py +3 -3
- datahub/ingestion/source/openapi.py +3 -3
- datahub/ingestion/source/openapi_parser.py +8 -8
- datahub/ingestion/source/powerbi/config.py +1 -1
- datahub/ingestion/source/powerbi/powerbi.py +2 -2
- datahub/ingestion/source/redshift/profile.py +2 -2
- datahub/ingestion/source/snowflake/snowflake_utils.py +1 -1
- datahub/ingestion/source/tableau/tableau.py +4 -4
- datahub/ingestion/source/tableau/tableau_common.py +2 -2
- datahub/ingestion/source/unity/source.py +1 -1
- datahub/ingestion/transformer/add_dataset_dataproduct.py +1 -1
- datahub/ingestion/transformer/add_dataset_ownership.py +1 -1
- datahub/ingestion/transformer/dataset_domain.py +1 -1
- datahub/lite/lite_util.py +2 -2
- datahub/testing/mcp_diff.py +1 -1
- datahub/utilities/file_backed_collections.py +6 -6
- datahub/utilities/hive_schema_to_avro.py +2 -2
- datahub/ingestion/transformer/system_metadata_transformer.py +0 -45
- {acryl_datahub-1.0.0.2rc3.dist-info → acryl_datahub-1.0.0.2rc4.dist-info}/WHEEL +0 -0
- {acryl_datahub-1.0.0.2rc3.dist-info → acryl_datahub-1.0.0.2rc4.dist-info}/entry_points.txt +0 -0
- {acryl_datahub-1.0.0.2rc3.dist-info → acryl_datahub-1.0.0.2rc4.dist-info}/licenses/LICENSE +0 -0
- {acryl_datahub-1.0.0.2rc3.dist-info → acryl_datahub-1.0.0.2rc4.dist-info}/top_level.txt +0 -0
|
@@ -1,45 +0,0 @@
|
|
|
1
|
-
import functools
|
|
2
|
-
from typing import Iterable
|
|
3
|
-
|
|
4
|
-
from datahub.emitter.mce_builder import get_sys_time
|
|
5
|
-
from datahub.ingestion.api.common import PipelineContext, RecordEnvelope
|
|
6
|
-
from datahub.ingestion.api.transform import Transformer
|
|
7
|
-
from datahub.ingestion.api.workunit import MetadataWorkUnit
|
|
8
|
-
from datahub.ingestion.transformer.auto_helper_transformer import AutoHelperTransformer
|
|
9
|
-
from datahub.metadata.schema_classes import SystemMetadataClass
|
|
10
|
-
|
|
11
|
-
|
|
12
|
-
def auto_system_metadata(
|
|
13
|
-
ctx: PipelineContext,
|
|
14
|
-
stream: Iterable[MetadataWorkUnit],
|
|
15
|
-
) -> Iterable[MetadataWorkUnit]:
|
|
16
|
-
if not ctx.pipeline_config:
|
|
17
|
-
raise ValueError("Pipeline config is required for system metadata")
|
|
18
|
-
set_system_metadata = ctx.pipeline_config.flags.set_system_metadata
|
|
19
|
-
set_pipeline_name = ctx.pipeline_config.flags.set_system_metadata_pipeline_name
|
|
20
|
-
|
|
21
|
-
for workunit in stream:
|
|
22
|
-
if set_system_metadata:
|
|
23
|
-
workunit.metadata.systemMetadata = SystemMetadataClass(
|
|
24
|
-
lastObserved=get_sys_time(), runId=ctx.run_id
|
|
25
|
-
)
|
|
26
|
-
if set_pipeline_name:
|
|
27
|
-
workunit.metadata.systemMetadata.pipelineName = ctx.pipeline_name
|
|
28
|
-
|
|
29
|
-
yield workunit
|
|
30
|
-
|
|
31
|
-
|
|
32
|
-
class SystemMetadataTransformer(Transformer):
|
|
33
|
-
def __init__(self, ctx: PipelineContext):
|
|
34
|
-
self._inner_transformer = AutoHelperTransformer(
|
|
35
|
-
functools.partial(auto_system_metadata, ctx)
|
|
36
|
-
)
|
|
37
|
-
|
|
38
|
-
def transform(
|
|
39
|
-
self, record_envelopes: Iterable[RecordEnvelope]
|
|
40
|
-
) -> Iterable[RecordEnvelope]:
|
|
41
|
-
yield from self._inner_transformer.transform(record_envelopes)
|
|
42
|
-
|
|
43
|
-
@classmethod
|
|
44
|
-
def create(cls, config_dict: dict, ctx: PipelineContext) -> Transformer:
|
|
45
|
-
raise NotImplementedError(f"{cls.__name__} cannot be created from config")
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|