feat(ingest): new hex connector - part 2 (datahub-project#12985)

Author: sgomezvillamor

metadata-ingestion/docs/sources/hex/README.md

@@ -20,4 +20,12 @@ Currently, the [Hex API](https://learn.hex.tech/docs/api/api-reference) has some
 
 2. **Metadata Access**: There is no direct method to retrieve metadata for Collections, Status, or Categories. This information is only available indirectly through references within Projects and Components.
 
-Please keep these limitations in mind when working with the Hex connector.
+Please keep these limitations in mind when working with the Hex connector.
+
+For the Dataset - Hex Project lineage, the connector relies on the 
+[_Hex query metadata_](https://learn.hex.tech/docs/explore-data/cells/sql-cells/sql-cells-introduction#query-metadata) feature.
+Therefore, in order to extract lineage information, the required setup must include:
+
+- A separated warehouse ingestor (_eg_ BigQuery, Snowflake, Redshift, ...) with `use_queries_v2` enabled in order to fetch Queries.
+  This will ingest the queries into DataHub as `Query` entities and the ones triggered by Hex will include the corresponding _Hex query metadata_.
+- A DataHub server with version >= SaaS `0.3.10` or > OSS `1.0.0` so the `Query` entities are properly indexed by source (Hex in this case) and so fetched and processed by the Hex ingestor in order to emit the Dataset - Project lineage.

metadata-ingestion/src/datahub/ingestion/source/hex/constants.py

@@ -1,3 +1,8 @@
+from datahub.metadata.urns import DataPlatformUrn
+
 HEX_PLATFORM_NAME = "hex"
+HEX_PLATFORM_URN = DataPlatformUrn(platform_name=HEX_PLATFORM_NAME)
 HEX_API_BASE_URL_DEFAULT = "https://app.hex.tech/api/v1"
 HEX_API_PAGE_SIZE_DEFAULT = 100
+
+DATAHUB_API_PAGE_SIZE_DEFAULT = 100

metadata-ingestion/src/datahub/ingestion/source/hex/hex.py

@@ -1,9 +1,12 @@
+from dataclasses import dataclass
+from datetime import datetime, timedelta, timezone
 from typing import Any, Dict, Iterable, List, Optional
 
-from pydantic import Field, SecretStr
+from pydantic import Field, SecretStr, root_validator
 from typing_extensions import assert_never
 
 from datahub.configuration.common import AllowDenyPattern
+from datahub.configuration.datetimes import parse_user_datetime
 from datahub.configuration.source_common import (
     EnvConfigMixin,
     PlatformInstanceConfigMixin,
@@ -21,22 +24,28 @@
 from datahub.ingestion.api.workunit import MetadataWorkUnit
 from datahub.ingestion.source.hex.api import HexApi, HexApiReport
 from datahub.ingestion.source.hex.constants import (
+    DATAHUB_API_PAGE_SIZE_DEFAULT,
     HEX_API_BASE_URL_DEFAULT,
     HEX_API_PAGE_SIZE_DEFAULT,
     HEX_PLATFORM_NAME,
 )
 from datahub.ingestion.source.hex.mapper import Mapper
 from datahub.ingestion.source.hex.model import Component, Project
+from datahub.ingestion.source.hex.query_fetcher import (
+    HexQueryFetcher,
+    HexQueryFetcherReport,
+)
 from datahub.ingestion.source.state.stale_entity_removal_handler import (
     StaleEntityRemovalHandler,
     StaleEntityRemovalSourceReport,
     StatefulStaleMetadataRemovalConfig,
 )
 from datahub.ingestion.source.state.stateful_ingestion_base import (
     StatefulIngestionConfigBase,
-    StatefulIngestionReport,
     StatefulIngestionSourceBase,
 )
+from datahub.ingestion.source_report.ingestion_stage import IngestionStageReport
+from datahub.sdk.main_client import DataHubClient
 
 
 class HexSourceConfig(
@@ -93,9 +102,73 @@ class HexSourceConfig(
         default=True,
         description="Set ownership identity from owner/creator email",
     )
+    include_lineage: bool = Field(
+        default=True,
+        description='Include Hex lineage, being fetched from DataHub. See "Limitations" section in the docs for more details about the limitations of this feature.',
+    )
+    lineage_start_time: Optional[datetime] = Field(
+        default=None,
+        description="Earliest date of lineage to consider. Default: 1 day before lineage end time. You can specify absolute time like '2023-01-01' or relative time like '-7 days' or '-7d'.",
+    )
+    lineage_end_time: Optional[datetime] = Field(
+        default=None,
+        description="Latest date of lineage to consider. Default: Current time in UTC. You can specify absolute time like '2023-01-01' or relative time like '-1 day' or '-1d'.",
+    )
+    datahub_page_size: int = Field(
+        default=DATAHUB_API_PAGE_SIZE_DEFAULT,
+        description="Number of items to fetch per DataHub API call.",
+    )
+
+    @root_validator(pre=True)
+    def validate_lineage_times(cls, data: Dict[str, Any]) -> Dict[str, Any]:
+        # lineage_end_time default = now
+        if "lineage_end_time" not in data or data["lineage_end_time"] is None:
+            data["lineage_end_time"] = datetime.now(tz=timezone.utc)
+        # if string is given, parse it
+        if isinstance(data["lineage_end_time"], str):
+            data["lineage_end_time"] = parse_user_datetime(data["lineage_end_time"])
+        # if no timezone is given, assume UTC
+        if data["lineage_end_time"].tzinfo is None:
+            data["lineage_end_time"] = data["lineage_end_time"].replace(
+                tzinfo=timezone.utc
+            )
+        # at this point, we ensure there is a non null datetime with UTC timezone for lineage_end_time
+        assert (
+            data["lineage_end_time"]
+            and isinstance(data["lineage_end_time"], datetime)
+            and data["lineage_end_time"].tzinfo is not None
+            and data["lineage_end_time"].tzinfo == timezone.utc
+        )
+
+        # lineage_start_time default = lineage_end_time - 1 day
+        if "lineage_start_time" not in data or data["lineage_start_time"] is None:
+            data["lineage_start_time"] = data["lineage_end_time"] - timedelta(days=1)
+        # if string is given, parse it
+        if isinstance(data["lineage_start_time"], str):
+            data["lineage_start_time"] = parse_user_datetime(data["lineage_start_time"])
+        # if no timezone is given, assume UTC
+        if data["lineage_start_time"].tzinfo is None:
+            data["lineage_start_time"] = data["lineage_start_time"].replace(
+                tzinfo=timezone.utc
+            )
+        # at this point, we ensure there is a non null datetime with UTC timezone for lineage_start_time
+        assert (
+            data["lineage_start_time"]
+            and isinstance(data["lineage_start_time"], datetime)
+            and data["lineage_start_time"].tzinfo is not None
+            and data["lineage_start_time"].tzinfo == timezone.utc
+        )
+
+        return data
 
 
-class HexReport(StaleEntityRemovalSourceReport, HexApiReport):
+@dataclass
+class HexReport(
+    StaleEntityRemovalSourceReport,
+    HexApiReport,
+    IngestionStageReport,
+    HexQueryFetcherReport,
+):
     pass
 
 
@@ -110,7 +183,7 @@ class HexSource(StatefulIngestionSourceBase):
     def __init__(self, config: HexSourceConfig, ctx: PipelineContext):
         super().__init__(config, ctx)
         self.source_config = config
-        self.report = HexReport()
+        self.report: HexReport = HexReport()
         self.platform = HEX_PLATFORM_NAME
         self.hex_api = HexApi(
             report=self.report,
@@ -129,6 +202,28 @@ def __init__(self, config: HexSourceConfig, ctx: PipelineContext):
             categories_as_tags=self.source_config.categories_as_tags,
             set_ownership_from_email=self.source_config.set_ownership_from_email,
         )
+        self.project_registry: Dict[str, Project] = {}
+        self.component_registry: Dict[str, Component] = {}
+
+        self.datahub_client: Optional[DataHubClient] = None
+        self.query_fetcher: Optional[HexQueryFetcher] = None
+        if self.source_config.include_lineage:
+            graph = ctx.require_graph("Lineage")
+            assert self.source_config.lineage_start_time and isinstance(
+                self.source_config.lineage_start_time, datetime
+            )
+            assert self.source_config.lineage_end_time and isinstance(
+                self.source_config.lineage_end_time, datetime
+            )
+            self.datahub_client = DataHubClient(graph=graph)
+            self.query_fetcher = HexQueryFetcher(
+                datahub_client=self.datahub_client,
+                workspace_name=self.source_config.workspace_name,
+                start_datetime=self.source_config.lineage_start_time,
+                end_datetime=self.source_config.lineage_end_time,
+                report=self.report,
+                page_size=self.source_config.datahub_page_size,
+            )
 
     @classmethod
     def create(cls, config_dict: Dict[str, Any], ctx: PipelineContext) -> "HexSource":
@@ -143,25 +238,58 @@ def get_workunit_processors(self) -> List[Optional[MetadataWorkUnitProcessor]]:
             ).workunit_processor,
         ]
 
-    def get_report(self) -> StatefulIngestionReport:
+    def get_report(self) -> HexReport:
         return self.report
 
     def get_workunits_internal(self) -> Iterable[MetadataWorkUnit]:
-        yield from self.mapper.map_workspace()
-
-        for project_or_component in self.hex_api.fetch_projects():
-            if isinstance(project_or_component, Project):
-                if self.source_config.project_title_pattern.allowed(
-                    project_or_component.title
-                ):
-                    yield from self.mapper.map_project(project=project_or_component)
-            elif isinstance(project_or_component, Component):
-                if (
-                    self.source_config.include_components
-                    and self.source_config.component_title_pattern.allowed(
+        with self.report.new_stage("Fetch Hex assets from Hex API"):
+            for project_or_component in self.hex_api.fetch_projects():
+                if isinstance(project_or_component, Project):
+                    if self.source_config.project_title_pattern.allowed(
                         project_or_component.title
-                    )
-                ):
-                    yield from self.mapper.map_component(component=project_or_component)
-            else:
-                assert_never(project_or_component)
+                    ):
+                        self.project_registry[project_or_component.id] = (
+                            project_or_component
+                        )
+                elif isinstance(project_or_component, Component):
+                    if (
+                        self.source_config.include_components
+                        and self.source_config.component_title_pattern.allowed(
+                            project_or_component.title
+                        )
+                    ):
+                        self.component_registry[project_or_component.id] = (
+                            project_or_component
+                        )
+                else:
+                    assert_never(project_or_component)
+
+        if self.source_config.include_lineage:
+            assert self.datahub_client and self.query_fetcher
+
+            with self.report.new_stage(
+                "Fetch Hex lineage from existing Queries in DataHub"
+            ):
+                for query_metadata in self.query_fetcher.fetch():
+                    project = self.project_registry.get(query_metadata.hex_project_id)
+                    if project:
+                        project.upstream_datasets.extend(
+                            query_metadata.dataset_subjects
+                        )
+                        project.upstream_schema_fields.extend(
+                            query_metadata.schema_field_subjects
+                        )
+                    else:
+                        self.report.report_warning(
+                            title="Missing project for lineage",
+                            message="Lineage missed because missed project, likely due to filter patterns or deleted project.",
+                            context=str(query_metadata),
+                        )
+
+        with self.report.new_stage("Emit"):
+            yield from self.mapper.map_workspace()
+
+            for project in self.project_registry.values():
+                yield from self.mapper.map_project(project=project)
+            for component in self.component_registry.values():
+                yield from self.mapper.map_component(component=component)

metadata-ingestion/src/datahub/ingestion/source/hex/mapper.py

@@ -1,6 +1,6 @@
 import logging
 from datetime import datetime
-from typing import Iterable, List, Optional, Tuple
+from typing import Iterable, List, Optional, Tuple, Union
 
 from datahub._codegen.aspect import (
     _Aspect,  # TODO: is there a better import than this one?
@@ -46,14 +46,22 @@
     DashboardInfoClass,
     DashboardUsageStatisticsClass,
     DataPlatformInstanceClass,
+    EdgeClass,
     GlobalTagsClass,
     OwnerClass,
     OwnershipClass,
     SubTypesClass,
     TagAssociationClass,
     TimeWindowSizeClass,
 )
-from datahub.metadata.urns import ContainerUrn, CorpUserUrn, DashboardUrn, Urn
+from datahub.metadata.urns import (
+    ContainerUrn,
+    CorpUserUrn,
+    DashboardUrn,
+    DatasetUrn,
+    SchemaFieldUrn,
+    Urn,
+)
 
 logger = logging.getLogger(__name__)
 
@@ -116,6 +124,8 @@ def map_project(self, project: Project) -> Iterable[MetadataWorkUnit]:
             ),
             externalUrl=f"{self._base_url}/{self._workspace_name}/hex/{project.id}",
             customProperties=dict(id=project.id),
+            datasetEdges=self._dataset_edges(project.upstream_datasets),
+            # TODO: support schema field upstream, maybe InputFields?
         )
 
         subtypes = SubTypesClass(
@@ -343,6 +353,22 @@ def _platform_instance_aspect(self) -> DataPlatformInstanceClass:
             else None,
         )
 
+    def _dataset_edges(
+        self, upstream: List[Union[DatasetUrn, SchemaFieldUrn]]
+    ) -> Optional[List[EdgeClass]]:
+        # TBC: is there support for CLL in Dashboards? for the moment, skip SchemaFieldUrns
+        return (
+            [
+                EdgeClass(
+                    destinationUrn=upstream_urn.urn(),
+                )
+                for upstream_urn in upstream
+                if isinstance(upstream_urn, DatasetUrn)
+            ]
+            if upstream
+            else None
+        )
+
     def _yield_mcps(
         self, entity_urn: Urn, aspects: List[Optional[_Aspect]]
     ) -> Iterable[MetadataWorkUnit]:

metadata-ingestion/src/datahub/ingestion/source/hex/model.py

@@ -1,6 +1,8 @@
-from dataclasses import dataclass
+from dataclasses import dataclass, field
 from datetime import datetime
-from typing import List, Optional
+from typing import List, Optional, Union
+
+from datahub.metadata.urns import DatasetUrn, SchemaFieldUrn
 
 
 @dataclass
@@ -51,6 +53,12 @@ class Project:
     creator: Optional[Owner] = None
     owner: Optional[Owner] = None
     analytics: Optional[Analytics] = None
+    upstream_datasets: List[Union[DatasetUrn, SchemaFieldUrn]] = field(
+        default_factory=list
+    )
+    upstream_schema_fields: List[Union[DatasetUrn, SchemaFieldUrn]] = field(
+        default_factory=list
+    )
 
 
 @dataclass