fix(examples): set dataset UUID from YAML in generic loader

The generic data loader creates SqlaTable rows without the UUID from
the YAML config. When load_examples_from_configs() later tries to
import the same dataset via YAML, it looks up by UUID, misses the
existing row, and tries to INSERT a duplicate — hitting a UNIQUE
constraint on table_name.

Fix by reading the UUID from dataset YAML configs and setting it on
the SqlaTable during generic data loading, so the YAML import path
finds the existing dataset by UUID.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

Author: dpgaspar

superset/examples/data_loading.py

@@ -40,6 +40,7 @@ def get_dataset_config_from_yaml(example_dir: Path) -> Dict[str, Optional[str]]:
         "table_name": None,
         "schema": None,
         "data_file": None,
+        "uuid": None,
     }
     dataset_yaml = example_dir / "dataset.yaml"
     if dataset_yaml.exists():
@@ -48,6 +49,7 @@ def get_dataset_config_from_yaml(example_dir: Path) -> Dict[str, Optional[str]]:
                 config = yaml.safe_load(f)
                 result["table_name"] = config.get("table_name")
                 result["data_file"] = config.get("data_file")
+                result["uuid"] = config.get("uuid")
                 schema = config.get("schema")
                 # Treat SQLite's 'main' schema as null (use target database default)
                 result["schema"] = None if schema == "main" else schema
@@ -81,6 +83,7 @@ def _get_multi_dataset_config(
         with open(datasets_yaml) as f:
             yaml_config = yaml.safe_load(f)
             result["table_name"] = yaml_config.get("table_name") or dataset_name
+            result["uuid"] = yaml_config.get("uuid")
             raw_schema = yaml_config.get("schema")
             result["schema"] = None if raw_schema == "main" else raw_schema
 
@@ -142,6 +145,7 @@ def discover_datasets() -> Dict[str, Callable[..., None]]:
             table_name=table_name,
             schema=config["schema"],
             data_file=resolved_file,
+            uuid=config.get("uuid"),
         )
 
     # Discover multiple parquet files in data/ folders (complex examples)
@@ -160,6 +164,7 @@ def discover_datasets() -> Dict[str, Callable[..., None]]:
                 table_name=config["table_name"],
                 schema=config["schema"],
                 data_file=config["data_file"],
+                uuid=config.get("uuid"),
             )
 
     return loaders

superset/examples/generic_loader.py

@@ -57,6 +57,7 @@ def load_parquet_table(  # noqa: C901
     sample_rows: Optional[int] = None,
     data_file: Optional[Any] = None,
     schema: Optional[str] = None,
+    uuid: Optional[str] = None,
 ) -> SqlaTable:
     """Load a Parquet file into the example database.
 
@@ -175,6 +176,10 @@ def safe_serialize(x: Any, column_name: str) -> Optional[str]:
         # Set the database reference
         tbl.database = database
 
+    # Set UUID from YAML config so the YAML import path can find this dataset
+    if uuid and not tbl.uuid:
+        tbl.uuid = uuid
+
     if not only_metadata:
         # Ensure database reference is set before fetching metadata
         if not tbl.database:
@@ -194,6 +199,7 @@ def create_generic_loader(
     sample_rows: Optional[int] = None,
     data_file: Optional[Any] = None,
     schema: Optional[str] = None,
+    uuid: Optional[str] = None,
 ) -> Callable[[Database, SqlaTable], None]:
     """Create a loader function for a specific Parquet file.
 
@@ -230,6 +236,7 @@ def loader(
             sample_rows=rows,
             data_file=data_file,
             schema=schema,
+            uuid=uuid,
         )
 
         if description and tbl: