Retrieving a schema from a table in polars

collect_schema

In addition to showing column data types in the dataframe, Polars allows you to get a dictionary of data types where the key is the column name and the value is the data type. This is especially helpful when you’re writing data to a database and want the schema to be maintained. Below is a dataframe with various data types.

shape: (5, 4)

date	sku	quantity	unit price
datetime[μs]	str	i64	f64
2014-07-10 17:09:39	"ZC-07383"	2	70.66
2013-12-21 23:59:55	"LK-02338"	8	43.46
2013-11-21 08:08:36	"QS-76400"	18	97.45
2014-06-08 01:33:20	"FT-50146"	20	66.51
2013-11-16 05:08:39	"XX-25746"	19	62.62

Get data types as dictionary

To get the data types as a dictionary, you can use the Polars method collect_schema. Here’s how to do it:

print(df
 .collect_schema()
)

Schema({'date': Datetime(time_unit='us', time_zone=None), 'sku': String, 'quantity': Int64, 'unit price': Float64})

Get names, keys from schema

The schema dictionary has properties that you can access. For instance, if you only want the table column names from the schema, you can get them like this:

(df
 .collect_schema()
 .names()
)

['date', 'sku', 'quantity', 'unit price']

Or, if you want the data types, you can get them like this:

(df
 .collect_schema()
 .keys()
)

odict_keys(['date', 'sku', 'quantity', 'unit price'])

My Polars course is accepting enrollments. I can’t wait to teach you!