openmetadata-ingestion 1.8.10.0__py3-none-any.whl → 1.9.0.0__py3-none-any.whl

metadata/generated/schema/entity/data/spreadsheet.py

@@ -0,0 +1,209 @@
+# generated by datamodel-codegen:
+#   filename:  entity/data/spreadsheet.json
+#   timestamp: 2025-08-08T13:50:05+00:00
+
+from __future__ import annotations
+
+from enum import Enum
+from typing import List, Optional
+
+from pydantic import ConfigDict, Field
+from typing_extensions import Annotated
+
+from metadata.ingestion.models.custom_pydantic import BaseModel
+
+from ...type import (
+    assetCertification,
+    basic,
+    entityHistory,
+    entityReference,
+    entityReferenceList,
+    lifeCycle,
+    tagLabel,
+    usageDetails,
+    votes,
+)
+from ..services import driveService
+
+
+class SpreadsheetMimeType(Enum):
+    application_vnd_google_apps_spreadsheet = 'application/vnd.google-apps.spreadsheet'
+    application_vnd_openxmlformats_officedocument_spreadsheetml_sheet = (
+        'application/vnd.openxmlformats-officedocument.spreadsheetml.sheet'
+    )
+    application_vnd_ms_excel = 'application/vnd.ms-excel'
+    application_vnd_oasis_opendocument_spreadsheet = (
+        'application/vnd.oasis.opendocument.spreadsheet'
+    )
+    text_csv = 'text/csv'
+    text_tab_separated_values = 'text/tab-separated-values'
+    Other = 'Other'
+
+
+class Spreadsheet(BaseModel):
+    model_config = ConfigDict(
+        extra='forbid',
+    )
+    id: Annotated[
+        basic.Uuid,
+        Field(
+            description='Unique identifier that identifies this spreadsheet instance.'
+        ),
+    ]
+    name: Annotated[
+        basic.EntityName, Field(description='Name that identifies the spreadsheet.')
+    ]
+    fullyQualifiedName: Annotated[
+        Optional[basic.FullyQualifiedEntityName],
+        Field(
+            None,
+            description="Name that uniquely identifies a spreadsheet in the format 'ServiceName.DirectoryPath.SpreadsheetName'.",
+        ),
+    ]
+    displayName: Annotated[
+        Optional[str],
+        Field(None, description='Display Name that identifies this spreadsheet.'),
+    ]
+    description: Annotated[
+        Optional[basic.Markdown],
+        Field(None, description='Description of the spreadsheet instance.'),
+    ]
+    service: Annotated[
+        Optional[entityReference.EntityReference],
+        Field(
+            None,
+            description='Link to the drive service where this spreadsheet is hosted.',
+        ),
+    ]
+    serviceType: Annotated[
+        Optional[driveService.DriveServiceType],
+        Field(None, description='Service type where this spreadsheet is hosted.'),
+    ]
+    directory: Annotated[
+        Optional[entityReference.EntityReference],
+        Field(
+            None, description='Reference to Directory that contains this spreadsheet.'
+        ),
+    ]
+    worksheets: Annotated[
+        Optional[entityReferenceList.EntityReferenceList],
+        Field(None, description='References to worksheets/tabs in the spreadsheet.'),
+    ]
+    mimeType: Annotated[
+        Optional[SpreadsheetMimeType],
+        Field(None, description='MIME type of the spreadsheet file'),
+    ]
+    path: Annotated[
+        Optional[str], Field(None, description='Full path to the spreadsheet file')
+    ]
+    driveFileId: Annotated[
+        Optional[str], Field(None, description='Native file ID in the source system')
+    ]
+    size: Annotated[
+        Optional[int],
+        Field(
+            None,
+            description='File size in bytes (may be null for cloud-native files like Google Sheets)',
+        ),
+    ]
+    fileVersion: Annotated[
+        Optional[str], Field(None, description='File version information')
+    ]
+    createdTime: Annotated[
+        Optional[basic.Timestamp],
+        Field(None, description='Spreadsheet creation timestamp'),
+    ]
+    modifiedTime: Annotated[
+        Optional[basic.Timestamp],
+        Field(None, description='Last modification timestamp'),
+    ]
+    lastModifiedBy: Annotated[
+        Optional[entityReference.EntityReference],
+        Field(None, description='User who last modified the spreadsheet'),
+    ]
+    usageSummary: Annotated[
+        Optional[usageDetails.UsageDetails],
+        Field(None, description='Latest usage information for this spreadsheet.'),
+    ]
+    sourceUrl: Annotated[
+        Optional[basic.SourceUrl],
+        Field(
+            None,
+            description='Link to view/edit this spreadsheet in the source system (e.g., Google Sheets URL, SharePoint URL).',
+        ),
+    ]
+    href: Annotated[
+        Optional[basic.Href],
+        Field(
+            None, description='Link to the resource corresponding to this spreadsheet.'
+        ),
+    ]
+    owners: Annotated[
+        Optional[entityReferenceList.EntityReferenceList],
+        Field(None, description='Owners of this spreadsheet.'),
+    ]
+    followers: Annotated[
+        Optional[entityReferenceList.EntityReferenceList],
+        Field(None, description='Followers of this entity.'),
+    ]
+    tags: Annotated[
+        Optional[List[tagLabel.TagLabel]],
+        Field(None, description='Tags associated with this spreadsheet.'),
+    ]
+    version: Annotated[
+        Optional[entityHistory.EntityVersion],
+        Field(None, description='Metadata version of the entity.'),
+    ]
+    updatedAt: Annotated[
+        Optional[basic.Timestamp],
+        Field(
+            None,
+            description='Last update time corresponding to the new version of the entity in Unix epoch time milliseconds.',
+        ),
+    ]
+    updatedBy: Annotated[
+        Optional[str], Field(None, description='User who made the update.')
+    ]
+    changeDescription: Annotated[
+        Optional[entityHistory.ChangeDescription],
+        Field(None, description='Change that lead to this version of the entity.'),
+    ]
+    incrementalChangeDescription: Annotated[
+        Optional[entityHistory.ChangeDescription],
+        Field(None, description='Change that lead to this version of the entity.'),
+    ]
+    deleted: Annotated[
+        Optional[bool],
+        Field(
+            False, description='When `true` indicates the entity has been soft deleted.'
+        ),
+    ]
+    domains: Annotated[
+        Optional[entityReferenceList.EntityReferenceList],
+        Field(None, description='Domains the Spreadsheet belongs to.'),
+    ]
+    dataProducts: Annotated[
+        Optional[entityReferenceList.EntityReferenceList],
+        Field(None, description='List of data products this entity is part of.'),
+    ]
+    lifeCycle: Annotated[
+        Optional[lifeCycle.LifeCycle],
+        Field(None, description='Life Cycle of the entity'),
+    ]
+    sourceHash: Annotated[
+        Optional[str],
+        Field(
+            None, description='Source hash of the entity', max_length=32, min_length=1
+        ),
+    ]
+    extension: Annotated[
+        Optional[basic.EntityExtension],
+        Field(
+            None,
+            description='Entity extension data with custom attributes added to the entity.',
+        ),
+    ]
+    votes: Annotated[
+        Optional[votes.Votes], Field(None, description='Votes on the entity.')
+    ]
+    certification: Optional[assetCertification.AssetCertification] = None

metadata/generated/schema/entity/data/storedProcedure.py

@@ -1,6 +1,6 @@
 # generated by datamodel-codegen:
 #   filename:  entity/data/storedProcedure.json
-#   timestamp: 2025-08-05T07:46:01+00:00
+#   timestamp: 2025-08-08T13:50:05+00:00
 
 from __future__ import annotations
 
@@ -173,11 +173,11 @@ class StoredProcedure(BaseModel):
         Optional[basic.SourceUrl],
         Field(None, description='Source URL of database schema.'),
     ]
-    domain: Annotated[
-        Optional[entityReference.EntityReference],
+    domains: Annotated[
+        Optional[entityReferenceList.EntityReferenceList],
         Field(
             None,
-            description='Domain the Stored Procedure belongs to. When not set, the Stored Procedure inherits the domain from the database schemna it belongs to.',
+            description='Domains the Stored Procedure belongs to. When not set, the Stored Procedure inherits the domain from the database schemna it belongs to.',
         ),
     ]
     lifeCycle: Annotated[

metadata/generated/schema/entity/data/table.py

@@ -1,6 +1,6 @@
 # generated by datamodel-codegen:
 #   filename:  entity/data/table.json
-#   timestamp: 2025-08-05T07:46:01+00:00
+#   timestamp: 2025-08-08T13:50:05+00:00
 
 from __future__ import annotations
 
@@ -476,6 +476,45 @@ class PartitionProfilerConfig(BaseModel):
     ]
 
 
+class Partitioning(BaseModel):
+    model_config = ConfigDict(
+        extra='forbid',
+    )
+    partitionColumn: Annotated[
+        str,
+        Field(
+            description='Column to partition on. It should be a date, timestamp or integer column. It is important for the data to be reasonably equally distributed across the partitions.'
+        ),
+    ]
+    lowerBound: Annotated[
+        Optional[str],
+        Field(
+            None,
+            description='Lower bound of the partition range. If not provided, it will be fetched from the source.',
+        ),
+    ]
+    upperBound: Annotated[
+        Optional[str],
+        Field(
+            None,
+            description='Upper bound of the partition range. If not provided, it will be fetched from the source.',
+        ),
+    ]
+
+
+class SparkTableProfilerConfig(BaseModel):
+    model_config = ConfigDict(
+        extra='forbid',
+    )
+    partitioning: Annotated[
+        Optional[Partitioning],
+        Field(
+            None,
+            description='When reading big tables from sources, we optimize the reading by partitioning the data. This configuration is responsible for it.',
+        ),
+    ]
+
+
 class TableProfilerConfig(BaseModel):
     profileSampleType: Optional[ProfileSampleType] = ProfileSampleType.PERCENTAGE
     profileSample: Annotated[
@@ -536,6 +575,13 @@ class TableProfilerConfig(BaseModel):
             title='Compute Column Metrics',
         ),
     ]
+    sparkTableProfilerConfig: Annotated[
+        Optional[SparkTableProfilerConfig],
+        Field(
+            None,
+            description='Table Specific configuration for Profiling it with a Spark Engine. It is ignored for other engines.',
+        ),
+    ]
 
 
 class TableProfile(BaseModel):
@@ -911,11 +957,11 @@ class Table(BaseModel):
     sourceUrl: Annotated[
         Optional[basic.SourceUrl], Field(None, description='Source URL of table.')
     ]
-    domain: Annotated[
-        Optional[entityReference.EntityReference],
+    domains: Annotated[
+        Optional[entityReferenceList.EntityReferenceList],
         Field(
             None,
-            description='Domain the asset belongs to. When not set, the asset inherits the domain from the parent it belongs to.',
+            description='Domains the asset belongs to. When not set, the asset inherits the domain from the parent it belongs to.',
         ),
     ]
     dataProducts: Annotated[

metadata/generated/schema/entity/data/topic.py

@@ -1,6 +1,6 @@
 # generated by datamodel-codegen:
 #   filename:  entity/data/topic.json
-#   timestamp: 2025-08-05T07:46:01+00:00
+#   timestamp: 2025-08-08T13:50:05+00:00
 
 from __future__ import annotations
 
@@ -190,11 +190,11 @@ class Topic(BaseModel):
     sourceUrl: Annotated[
         Optional[basic.SourceUrl], Field(None, description='Source URL of topic.')
     ]
-    domain: Annotated[
-        Optional[entityReference.EntityReference],
+    domains: Annotated[
+        Optional[entityReferenceList.EntityReferenceList],
         Field(
             None,
-            description='Domain the Topic belongs to. When not set, the Topic inherits the domain from the messaging service it belongs to.',
+            description='Domains the Topic belongs to. When not set, the Topic inherits the domain from the messaging service it belongs to.',
         ),
     ]
     dataProducts: Annotated[

metadata/generated/schema/entity/data/worksheet.py

@@ -0,0 +1,168 @@
+# generated by datamodel-codegen:
+#   filename:  entity/data/worksheet.json
+#   timestamp: 2025-08-08T13:50:05+00:00
+
+from __future__ import annotations
+
+from typing import List, Optional
+
+from pydantic import ConfigDict, Field
+from typing_extensions import Annotated
+
+from metadata.ingestion.models.custom_pydantic import BaseModel
+
+from ...type import (
+    assetCertification,
+    basic,
+    entityHistory,
+    entityReference,
+    entityReferenceList,
+    lifeCycle,
+    tagLabel,
+    usageDetails,
+    votes,
+)
+from ..services import driveService
+from . import table
+
+
+class Worksheet(BaseModel):
+    model_config = ConfigDict(
+        extra='forbid',
+    )
+    id: Annotated[
+        basic.Uuid, Field(description='Unique identifier of this worksheet instance.')
+    ]
+    name: Annotated[basic.EntityName, Field(description='Name of the worksheet.')]
+    fullyQualifiedName: Annotated[
+        Optional[basic.FullyQualifiedEntityName],
+        Field(None, description='Fully qualified name of the worksheet.'),
+    ]
+    displayName: Annotated[
+        Optional[str],
+        Field(None, description='Display Name that identifies this worksheet.'),
+    ]
+    description: Annotated[
+        Optional[basic.Markdown],
+        Field(None, description='Description of the worksheet.'),
+    ]
+    spreadsheet: Annotated[
+        entityReference.EntityReference,
+        Field(
+            description='Reference to the parent File entity (with fileType=Spreadsheet)'
+        ),
+    ]
+    service: Annotated[
+        entityReference.EntityReference, Field(description='Link to the drive service')
+    ]
+    serviceType: Annotated[
+        Optional[driveService.DriveServiceType],
+        Field(None, description='Type of drive service'),
+    ]
+    worksheetId: Annotated[
+        Optional[str], Field(None, description='Native worksheet/tab ID')
+    ]
+    index: Annotated[
+        Optional[int],
+        Field(None, description='Position/index of the worksheet in the spreadsheet'),
+    ]
+    rowCount: Annotated[
+        Optional[int], Field(None, description='Number of rows in the worksheet')
+    ]
+    columnCount: Annotated[
+        Optional[int], Field(None, description='Number of columns in the worksheet')
+    ]
+    columns: Annotated[
+        Optional[List[table.Column]],
+        Field([], description='Column definitions if structured data'),
+    ]
+    sampleData: Annotated[
+        Optional[table.TableData],
+        Field(None, description='Sample data from the worksheet'),
+    ]
+    isHidden: Annotated[
+        Optional[bool], Field(False, description='Whether the worksheet is hidden')
+    ]
+    sourceUrl: Annotated[
+        Optional[basic.SourceUrl],
+        Field(None, description='Link to this worksheet in the source system.'),
+    ]
+    href: Annotated[
+        Optional[basic.Href],
+        Field(
+            None, description='Link to the resource corresponding to this worksheet.'
+        ),
+    ]
+    owners: Annotated[
+        Optional[entityReferenceList.EntityReferenceList],
+        Field(None, description='Owners of this worksheet.'),
+    ]
+    followers: Annotated[
+        Optional[entityReferenceList.EntityReferenceList],
+        Field(None, description='Followers of this entity.'),
+    ]
+    tags: Annotated[
+        Optional[List[tagLabel.TagLabel]],
+        Field(None, description='Tags associated with this worksheet.'),
+    ]
+    version: Annotated[
+        Optional[entityHistory.EntityVersion],
+        Field(None, description='Metadata version of the entity.'),
+    ]
+    updatedAt: Annotated[
+        Optional[basic.Timestamp],
+        Field(
+            None,
+            description='Last update time corresponding to the new version of the entity in Unix epoch time milliseconds.',
+        ),
+    ]
+    updatedBy: Annotated[
+        Optional[str], Field(None, description='User who made the update.')
+    ]
+    changeDescription: Annotated[
+        Optional[entityHistory.ChangeDescription],
+        Field(None, description='Change that lead to this version of the entity.'),
+    ]
+    incrementalChangeDescription: Annotated[
+        Optional[entityHistory.ChangeDescription],
+        Field(None, description='Change that lead to this version of the entity.'),
+    ]
+    deleted: Annotated[
+        Optional[bool],
+        Field(
+            False, description='When `true` indicates the entity has been soft deleted.'
+        ),
+    ]
+    domains: Annotated[
+        Optional[entityReferenceList.EntityReferenceList],
+        Field(None, description='Domains the Worksheet belongs to.'),
+    ]
+    dataProducts: Annotated[
+        Optional[entityReferenceList.EntityReferenceList],
+        Field(None, description='List of data products this entity is part of.'),
+    ]
+    lifeCycle: Annotated[
+        Optional[lifeCycle.LifeCycle],
+        Field(None, description='Life Cycle of the entity'),
+    ]
+    sourceHash: Annotated[
+        Optional[str],
+        Field(
+            None, description='Source hash of the entity', max_length=32, min_length=1
+        ),
+    ]
+    extension: Annotated[
+        Optional[basic.EntityExtension],
+        Field(
+            None,
+            description='Entity extension data with custom attributes added to the entity.',
+        ),
+    ]
+    votes: Annotated[
+        Optional[votes.Votes], Field(None, description='Votes on the entity.')
+    ]
+    certification: Optional[assetCertification.AssetCertification] = None
+    usageSummary: Annotated[
+        Optional[usageDetails.UsageDetails],
+        Field(None, description='Latest usage information for this worksheet.'),
+    ]

metadata/generated/schema/entity/datacontract/__init__.py

@@ -0,0 +1,3 @@
+# generated by datamodel-codegen:
+#   filename:  schema
+#   timestamp: 2025-08-08T13:50:05+00:00

metadata/generated/schema/entity/datacontract/dataContractResult.py

@@ -0,0 +1,74 @@
+# generated by datamodel-codegen:
+#   filename:  entity/datacontract/dataContractResult.json
+#   timestamp: 2025-08-08T13:50:05+00:00
+
+from __future__ import annotations
+
+from typing import Optional
+
+from pydantic import ConfigDict, Field
+from typing_extensions import Annotated
+
+from metadata.ingestion.models.custom_pydantic import BaseModel
+
+from ...type import basic, contractExecutionStatus
+from . import qualityValidation, schemaValidation, semanticsValidation, slaValidation
+
+
+class DataContractResult(BaseModel):
+    model_config = ConfigDict(
+        extra='forbid',
+    )
+    id: Annotated[
+        Optional[basic.Uuid],
+        Field(
+            None,
+            description='Unique Data Contract validation execution identifier of this data contract result instance.',
+        ),
+    ]
+    dataContractFQN: Annotated[
+        basic.FullyQualifiedEntityName,
+        Field(description='Fully qualified name of the data contract.'),
+    ]
+    timestamp: Annotated[
+        basic.Timestamp,
+        Field(description='Timestamp when the data contract was executed.'),
+    ]
+    contractExecutionStatus: Annotated[
+        contractExecutionStatus.ContractExecutionStatus,
+        Field(description='Overall status of the contract execution.'),
+    ]
+    result: Annotated[
+        Optional[str],
+        Field(None, description='Detailed result of the data contract execution.'),
+    ]
+    schemaValidation: Annotated[
+        Optional[schemaValidation.SchemaValidation],
+        Field(None, description='Schema validation details.'),
+    ]
+    semanticsValidation: Annotated[
+        Optional[semanticsValidation.SemanticsValidation],
+        Field(None, description='Semantics validation details.'),
+    ]
+    qualityValidation: Annotated[
+        Optional[qualityValidation.QualityValidation],
+        Field(None, description='Quality expectations validation details.'),
+    ]
+    slaValidation: Annotated[
+        Optional[slaValidation.SlaValidation],
+        Field(None, description='SLA validation details.'),
+    ]
+    incidentId: Annotated[
+        Optional[basic.Uuid],
+        Field(
+            None,
+            description='Incident ID if the contract execution failed and an incident was created.',
+        ),
+    ]
+    executionTime: Annotated[
+        Optional[basic.Timestamp],
+        Field(
+            None,
+            description='Time taken to execute the contract validation in milliseconds.',
+        ),
+    ]

metadata/generated/schema/entity/datacontract/qualityValidation.py

@@ -0,0 +1,31 @@
+# generated by datamodel-codegen:
+#   filename:  entity/datacontract/qualityValidation.json
+#   timestamp: 2025-08-08T13:50:05+00:00
+
+from __future__ import annotations
+
+from typing import Optional
+
+from pydantic import ConfigDict, Field
+from typing_extensions import Annotated
+
+from metadata.ingestion.models.custom_pydantic import BaseModel
+
+
+class QualityValidation(BaseModel):
+    model_config = ConfigDict(
+        extra='forbid',
+    )
+    passed: Annotated[
+        Optional[int], Field(None, description='Number of quality checks passed.')
+    ]
+    failed: Annotated[
+        Optional[int], Field(None, description='Number of quality checks failed.')
+    ]
+    total: Annotated[
+        Optional[int], Field(None, description='Total number of quality checks.')
+    ]
+    qualityScore: Annotated[
+        Optional[float],
+        Field(None, description='Overall quality score (0-100).', ge=0.0, le=100.0),
+    ]

metadata/generated/schema/entity/datacontract/schemaValidation.py

@@ -0,0 +1,31 @@
+# generated by datamodel-codegen:
+#   filename:  entity/datacontract/schemaValidation.json
+#   timestamp: 2025-08-08T13:50:05+00:00
+
+from __future__ import annotations
+
+from typing import List, Optional
+
+from pydantic import ConfigDict, Field
+from typing_extensions import Annotated
+
+from metadata.ingestion.models.custom_pydantic import BaseModel
+
+
+class SchemaValidation(BaseModel):
+    model_config = ConfigDict(
+        extra='forbid',
+    )
+    passed: Annotated[
+        Optional[int], Field(None, description='Number of schema checks passed.')
+    ]
+    failed: Annotated[
+        Optional[int], Field(None, description='Number of schema checks failed.')
+    ]
+    total: Annotated[
+        Optional[int], Field(None, description='Total number of schema checks.')
+    ]
+    failedFields: Annotated[
+        Optional[List[str]],
+        Field(None, description='List of fields that failed validation.'),
+    ]

metadata/generated/schema/entity/datacontract/semanticsValidation.py

@@ -0,0 +1,39 @@
+# generated by datamodel-codegen:
+#   filename:  entity/datacontract/semanticsValidation.json
+#   timestamp: 2025-08-08T13:50:05+00:00
+
+from __future__ import annotations
+
+from typing import List, Optional
+
+from pydantic import ConfigDict, Field
+from typing_extensions import Annotated
+
+from metadata.ingestion.models.custom_pydantic import BaseModel
+
+
+class FailedRule(BaseModel):
+    model_config = ConfigDict(
+        extra='forbid',
+    )
+    ruleName: Optional[str] = None
+    reason: Optional[str] = None
+
+
+class SemanticsValidation(BaseModel):
+    model_config = ConfigDict(
+        extra='forbid',
+    )
+    passed: Annotated[
+        Optional[int], Field(None, description='Number of semantics rules passed.')
+    ]
+    failed: Annotated[
+        Optional[int], Field(None, description='Number of semantics rules failed.')
+    ]
+    total: Annotated[
+        Optional[int], Field(None, description='Total number of semantics rules.')
+    ]
+    failedRules: Annotated[
+        Optional[List[FailedRule]],
+        Field(None, description='List of rules that failed validation.'),
+    ]