Catalog View Reader

This example shows how to query the generated SQL views through CatalogViewReader and work with the returned pandas DataFrame objects.

import pandas as pd

Setup

import pandera.pandas as pa

from pandera_catalog import CatalogViewReader, PanderaCatalog, SqlCatalogBackend

Create a SQL-backed catalog

backend = SqlCatalogBackend(url="sqlite:///catalog.db")
catalog = PanderaCatalog(backend=backend)

catalog.register(
    "sales",
    pa.DataFrameSchema(
        {
            "id": pa.Column(int),
            "value": pa.Column(float, pa.Check.greater_than(0)),
        }
    ),
    tags=["finance"],
    overwrite=True  # for repeated runs of the example
)
catalog.register_projection(
    "sales_reporting",
    [{"schema": "sales", "kind": "columns", "names": ["id", "value"]}],
    overwrite=True  # for repeated runs of the example
)

Query the generated views

reader = CatalogViewReader(catalog)
schema_catalog: pd.DataFrame = reader.read_schema_catalog()
projection_steps: pd.DataFrame = reader.read_projection_steps()
schema_columns: pd.DataFrame = reader.read_schema_columns()

schema_catalog

	schema_name	description	tags_json	is_projected	source_projection_name
0	sales	None	["finance"]	0	NaN
1	sales_reporting	None	[]	1	sales_reporting

projection_steps

	projection_name	step_index	kind	source_schema_name	name_ordinal	selected_name
0	sales_reporting	1	columns	sales	1	id
1	sales_reporting	1	columns	sales	2	value

schema_columns

	schema_name	column_name	ordinal	dtype	nullable	required	unique_value	coerce	regex	checks_json	is_projected	source_projection_name	source_schema_name	source_column_name
0	sales	id	1	int64	None	None	None	None	None	[]	0	NaN	sales	id
1	sales	value	2	float64	None	None	None	None	None	[{"greater_than": 0}]	0	NaN	sales	value
2	sales_reporting	id	1	int64	None	None	None	None	None	[]	1	sales_reporting	sales	id
3	sales_reporting	value	2	float64	None	None	None	None	None	[{"greater_than": 0}]	1	sales_reporting	sales	value