remove semantic manifest

Author: gaurpulkit

src/datapilot/clients/altimate/constants.py

@@ -6,5 +6,4 @@
     "catalog",
     "run_results",
     "sources",
-    "semantic_manifest",
 }

src/datapilot/core/platforms/dbt/cli/cli.py

@@ -19,7 +19,6 @@
 from datapilot.core.platforms.dbt.utils import load_catalog
 from datapilot.core.platforms.dbt.utils import load_manifest
 from datapilot.core.platforms.dbt.utils import load_run_results
-from datapilot.core.platforms.dbt.utils import load_semantic_manifest
 from datapilot.core.platforms.dbt.utils import load_sources
 from datapilot.utils.formatting.utils import tabulate_data
 from datapilot.utils.utils import map_url_to_instance
@@ -160,7 +159,6 @@ def project_health(
 @click.option("--catalog-path", required=False, prompt=False, help="Path to the catalog file.")
 @click.option("--run-results-path", required=False, prompt=False, help="Path to the run_results.json file.")
 @click.option("--sources-path", required=False, prompt=False, help="Path to the sources.json file (source freshness results).")
-@click.option("--semantic-manifest-path", required=False, prompt=False, help="Path to the semantic_manifest.json file.")
 def onboard(
     token,
     instance_name,
@@ -172,7 +170,6 @@ def onboard(
     catalog_path,
     run_results_path,
     sources_path,
-    semantic_manifest_path,
 ):
     """Onboard a manifest file to DBT. You can specify either --dbt_integration_id or --dbt_integration_name."""
 
@@ -236,13 +233,6 @@ def onboard(
             click.echo(f"Error validating sources: {e}")
             return
 
-    if semantic_manifest_path:
-        try:
-            load_semantic_manifest(semantic_manifest_path)
-        except Exception as e:
-            click.echo(f"Error validating semantic_manifest: {e}")
-            return
-
     # Onboard manifest (required)
     response = onboard_file(token, instance_name, dbt_integration_id, dbt_integration_environment, "manifest", manifest_path, backend_url)
     if response["ok"]:
@@ -280,16 +270,6 @@ def onboard(
         else:
             click.echo(f"{response['message']}")
 
-    if semantic_manifest_path:
-        response = onboard_file(
-            token, instance_name, dbt_integration_id, dbt_integration_environment, "semantic_manifest", semantic_manifest_path, backend_url
-        )
-        if response["ok"]:
-            click.echo("Semantic manifest onboarded successfully!")
-            artifacts_uploaded.append("semantic_manifest")
-        else:
-            click.echo(f"{response['message']}")
-
     # Start ingestion
     response = start_dbt_ingestion(token, instance_name, dbt_integration_id, dbt_integration_environment, backend_url)
     if response["ok"]:

src/datapilot/core/platforms/dbt/schemas/semantic_manifest.py

@@ -23,7 +23,6 @@
 from datapilot.core.platforms.dbt.schemas.manifest import AltimateManifestTestNode
 from datapilot.core.platforms.dbt.schemas.manifest import Manifest
 from datapilot.core.platforms.dbt.schemas.run_results import RunResults
-from datapilot.core.platforms.dbt.schemas.semantic_manifest import SemanticManifest
 from datapilot.core.platforms.dbt.schemas.sources import Sources
 from datapilot.exceptions.exceptions import AltimateFileNotFoundError
 from datapilot.exceptions.exceptions import AltimateInvalidJSONError
@@ -33,7 +32,6 @@
 from datapilot.utils.utils import load_json
 from vendor.dbt_artifacts_parser.parser import parse_manifest
 from vendor.dbt_artifacts_parser.parser import parse_run_results
-from vendor.dbt_artifacts_parser.parser import parse_semantic_manifest
 from vendor.dbt_artifacts_parser.parser import parse_sources
 
 MODEL_TYPE_PATTERNS = {
@@ -132,22 +130,6 @@ def load_sources(sources_path: str) -> Sources:
     return sources
 
 
-def load_semantic_manifest(semantic_manifest_path: str) -> SemanticManifest:
-    try:
-        semantic_manifest_dict = load_json(semantic_manifest_path)
-    except FileNotFoundError as e:
-        raise AltimateFileNotFoundError(f"Semantic manifest file not found: {semantic_manifest_path}. Error: {e}") from e
-    except ValueError as e:
-        raise AltimateInvalidJSONError(f"Invalid JSON file: {semantic_manifest_path}. Error: {e}") from e
-
-    try:
-        semantic_manifest: SemanticManifest = parse_semantic_manifest(semantic_manifest_dict)
-    except ValueError as e:
-        raise AltimateInvalidManifestError(f"Invalid semantic manifest file: {semantic_manifest_path}. Error: {e}") from e
-
-    return semantic_manifest
-
-
 # TODO: Add tests!
 def get_table_name_from_source(source: AltimateManifestSourceNode) -> str:
     db = source.database

src/datapilot/core/platforms/dbt/utils.py

@@ -35,7 +35,6 @@
 from vendor.dbt_artifacts_parser.parsers.run_results.run_results_v4 import RunResultsV4
 from vendor.dbt_artifacts_parser.parsers.run_results.run_results_v5 import RunResultsV5
 from vendor.dbt_artifacts_parser.parsers.run_results.run_results_v6 import RunResultsV6
-from vendor.dbt_artifacts_parser.parsers.semantic_manifest.semantic_manifest_v1 import SemanticManifestV1
 from vendor.dbt_artifacts_parser.parsers.sources.sources_v1 import SourcesV1
 from vendor.dbt_artifacts_parser.parsers.sources.sources_v2 import SourcesV2
 from vendor.dbt_artifacts_parser.parsers.sources.sources_v3 import SourcesV3
@@ -346,20 +345,3 @@ def parse_sources_v3(sources: dict) -> SourcesV3:
     if dbt_schema_version == ArtifactTypes.SOURCES_V3.value.dbt_schema_version:
         return SourcesV3(**sources)
     raise ValueError("Not a sources.json v3")
-
-
-#
-# semantic-manifest
-#
-def parse_semantic_manifest(semantic_manifest: dict) -> SemanticManifestV1:
-    """Parse semantic_manifest.json
-
-    Args:
-        semantic_manifest: A dict of semantic_manifest.json
-
-    Returns:
-        SemanticManifestV1
-    """
-    # Semantic manifest uses a flexible schema, so we accept any valid dict
-    # The schema version check is optional since the format may vary
-    return SemanticManifestV1(**semantic_manifest)

src/vendor/dbt_artifacts_parser/parser.py

@@ -38,7 +38,6 @@
 from vendor.dbt_artifacts_parser.parsers.run_results.run_results_v4 import RunResultsV4
 from vendor.dbt_artifacts_parser.parsers.run_results.run_results_v5 import RunResultsV5
 from vendor.dbt_artifacts_parser.parsers.run_results.run_results_v6 import RunResultsV6
-from vendor.dbt_artifacts_parser.parsers.semantic_manifest.semantic_manifest_v1 import SemanticManifestV1
 from vendor.dbt_artifacts_parser.parsers.sources.sources_v1 import SourcesV1
 from vendor.dbt_artifacts_parser.parsers.sources.sources_v2 import SourcesV2
 from vendor.dbt_artifacts_parser.parsers.sources.sources_v3 import SourcesV3
@@ -79,5 +78,3 @@ class ArtifactTypes(Enum):
     SOURCES_V1 = ArtifactType("https://schemas.getdbt.com/dbt/sources/v1.json", SourcesV1)
     SOURCES_V2 = ArtifactType("https://schemas.getdbt.com/dbt/sources/v2.json", SourcesV2)
     SOURCES_V3 = ArtifactType("https://schemas.getdbt.com/dbt/sources/v3.json", SourcesV3)
-    # Semantic Manifest
-    SEMANTIC_MANIFEST_V1 = ArtifactType("https://schemas.getdbt.com/dbt/semantic-manifest/v1.json", SemanticManifestV1)

src/vendor/dbt_artifacts_parser/parsers/semantic_manifest/__init__.py

@@ -5,7 +5,7 @@
 class TestOnboardFile:
     def test_supported_artifact_types(self):
         """Test that all expected artifact types are supported."""
-        expected_types = {"manifest", "catalog", "run_results", "sources", "semantic_manifest"}
+        expected_types = {"manifest", "catalog", "run_results", "sources"}
         assert SUPPORTED_ARTIFACT_TYPES == expected_types
 
     def test_unsupported_file_type_returns_error(self):

src/vendor/dbt_artifacts_parser/parsers/semantic_manifest/semantic_manifest_v1.py

@@ -1,7 +1,6 @@
 import pytest
 
 from datapilot.core.platforms.dbt.utils import load_run_results
-from datapilot.core.platforms.dbt.utils import load_semantic_manifest
 from datapilot.core.platforms.dbt.utils import load_sources
 from datapilot.exceptions.exceptions import AltimateFileNotFoundError
 
@@ -35,19 +34,3 @@ def test_load_sources_v3(self):
     def test_load_sources_file_not_found(self):
         with pytest.raises(AltimateFileNotFoundError):
             load_sources("nonexistent_file.json")
-
-
-class TestLoadSemanticManifest:
-    def test_load_semantic_manifest_v1(self):
-        semantic_manifest_path = "tests/data/semantic_manifest_v1.json"
-        semantic_manifest = load_semantic_manifest(semantic_manifest_path)
-
-        assert semantic_manifest is not None
-        assert len(semantic_manifest.semantic_models) == 1
-        assert semantic_manifest.semantic_models[0]["name"] == "orders"
-        assert len(semantic_manifest.metrics) == 1
-        assert semantic_manifest.metrics[0]["name"] == "revenue"
-
-    def test_load_semantic_manifest_file_not_found(self):
-        with pytest.raises(AltimateFileNotFoundError):
-            load_semantic_manifest("nonexistent_file.json")