A reactive notebook for Python — run reproducible experiments, query with SQL, execute as a script, deploy as an app, and version with git. Stored as pure Python. All in a modern, AI-native editor.
Python artificial-intelligencedagdata-sciencedata-visualizationdataflowdeveloper-toolsmachine-learningnotebookspipelinepythonreactivesqlweb-app
quote duckdb tables, schema with double quote (#8387)
Shahmir Varqhaa08c7d2a8cc9
diff --git a/marimo/_data/get_datasets.py b/marimo/_data/get_datasets.pyindex 503de7b94a..bd76ab81f4 100644--- a/marimo/_data/get_datasets.py+++ b/marimo/_data/get_datasets.py@@ -226,7 +226,7 @@ def _get_databases_from_duckdb_internal(and column_names[0] == CATALOG_TABLE_COLUMN_NAME)if catalog_table:- qualified_name = f"{database}.{schema}.{name}"+ qualified_name = f"{_quote_identifier(database)}.{_quote_identifier(schema)}.{_quote_identifier(name)}"columns = get_table_columns(connection, qualified_name)else:columns = [@@ -585,3 +585,15 @@ def _db_type_to_data_type(db_type: str) -> DataType:LOGGER.warning("Unknown DuckDB type: %s", db_type)return "unknown"+++def _quote_identifier(identifier: str) -> str:+ """+ Quote a DuckDB identifier with double quotes, escaping embedded double quotes.+ This prevents errors when the identifier contains special characters which need to be escaped.+ Eg. table.name -> "table.name"++ https://duckdb.org/docs/stable/sql/dialect/keywords_and_identifiers+ """+ escaped = identifier.replace('"', '""')+ return f'"{escaped}"'diff --git a/tests/_data/test_get_datasets.py b/tests/_data/test_get_datasets.pyindex b70af25572..182c49fad4 100644--- a/tests/_data/test_get_datasets.py+++ b/tests/_data/test_get_datasets.py@@ -5,9 +5,11 @@ from typing import Anyfrom unittest.mock import patchimport pytest+from inline_snapshot import snapshotfrom marimo._data.get_datasets import (_db_type_to_data_type,+ _quote_identifier,form_databases_from_dict,get_databases_from_duckdb,get_datasets_from_variables,@@ -751,3 +753,177 @@ def test_agg_query_does_not_backfill() -> None:mock_get_names.assert_not_called()connection.execute(cleanup_query)+++class TestQuoteIdentifier:+ @pytest.mark.parametrize(+ ("identifier", "expected"),+ [+ ("table", '"table"'),+ ("nested.namespace", '"nested.namespace"'),+ ("a.b.c.d", '"a.b.c.d"'),+ ('my"table', '"my""table"'),+ ("", '""'),+ ("my table", '"my table"'),+ ("schema-name", '"schema-name"'),+ ("name/with/slashes", '"name/with/slashes"'),+ ("has'single'quotes", "\"has'single'quotes\""),+ ("mixed.dots and spaces", '"mixed.dots and spaces"'),+ ("back`ticks", '"back`ticks"'),+ ("paren(theses)", '"paren(theses)"'),+ ('double""already', '"double""""already"'),+ ("unicode_ñoño", '"unicode_ñoño"'),+ ],+ )+ def test_quote_identifier(self, identifier: str, expected: str) -> None:+ assert _quote_identifier(identifier) == expected++[email protected]("duckdb")+class TestGetDatabasesNestedNamespace:+ """Tests for DuckDB catalog tables with special characters in the schema name.++ The catalog_table code path is triggered when SHOW ALL TABLES returns+ a table with a single column named "__" (the Iceberg catalog pattern).+ In that case, get_databases_from_duckdb falls back to DESCRIBE TABLE+ with a quoted qualified name.+ """++ def test_dotted_schema_with_catalog_table(self) -> None:+ """A table with a single '__' column in a dotted schema triggers+ the catalog_table path and must quote identifiers correctly."""+ import duckdb++ connection = duckdb.connect(":memory:")+ connection.execute('CREATE SCHEMA "nested.namespace"')+ connection.execute(+ 'CREATE TABLE "nested.namespace".my_table ("__" VARCHAR)'+ )++ result = get_databases_from_duckdb(connection=connection)+ assert result == snapshot(+ [+ Database(+ name="memory",+ dialect="duckdb",+ schemas=[+ Schema(+ name="nested.namespace",+ tables=[+ DataTable(+ name="my_table",+ source_type="duckdb",+ source="memory",+ num_rows=None,+ num_columns=1,+ variable_name=None,+ columns=[+ DataTableColumn(+ name="__",+ type="string",+ external_type="VARCHAR",+ sample_values=[],+ )+ ],+ )+ ],+ )+ ],+ engine=None,+ )+ ]+ )++ def test_deeply_dotted_schema_with_catalog_table(self) -> None:+ import duckdb++ connection = duckdb.connect(":memory:")+ connection.execute('CREATE SCHEMA "a.b.c.d"')+ connection.execute('CREATE TABLE "a.b.c.d".t ("__" INTEGER)')++ result = get_databases_from_duckdb(connection=connection)+ assert result == snapshot(+ [+ Database(+ name="memory",+ dialect="duckdb",+ schemas=[+ Schema(+ name="a.b.c.d",+ tables=[+ DataTable(+ source_type="duckdb",+ source="memory",+ name="t",+ num_rows=None,+ num_columns=1,+ variable_name=None,+ columns=[+ DataTableColumn(+ name="__",+ type="integer",+ external_type="INTEGER",+ sample_values=[],+ )+ ],+ engine=None,+ type="table",+ primary_keys=None,+ indexes=None,+ )+ ],+ )+ ],+ engine=None,+ )+ ]+ )++ @pytest.mark.parametrize(+ "schema_name",+ [+ "has-dashes",+ "has spaces",+ "has.dots.and-dashes",+ "slashes/in/name",+ ],+ )+ def test_special_char_schema_with_catalog_table(+ self, schema_name: str+ ) -> None:+ import duckdb++ connection = duckdb.connect(":memory:")+ connection.execute(f'CREATE SCHEMA "{schema_name}"')+ connection.execute(f'CREATE TABLE "{schema_name}".t ("__" VARCHAR)')++ result = get_databases_from_duckdb(connection=connection)+ assert result == [+ Database(+ name="memory",+ dialect="duckdb",+ schemas=[+ Schema(+ name=schema_name,+ tables=[+ DataTable(+ name="t",+ source_type="duckdb",+ source="memory",+ num_rows=None,+ num_columns=1,+ variable_name=None,+ columns=[+ DataTableColumn(+ name="__",+ type="string",+ external_type="VARCHAR",+ sample_values=[],+ )+ ],+ ),+ ],+ )+ ],+ ),+ ]