Misc Tables¶

A miscellaneous (misc) table links labels to an index. The index has no restrictions and can contain an arbitrary number of columns (called levels), but should not contain duplicated entries. In the CSV file of the misc table, the index columns are stored in front of the label columns. An empty label means that no label has been assigned (yet).

audformat implementation¶

Create an index with the levels "file" and "other":

import audformat
import audformat.testing
import pandas as pd


index = pd.MultiIndex.from_tuples(
    [
        ("f1", "f2"),
        ("f1", "f3"),
        ("f2", "f3"),
    ],
    names=["file", "other"],
)
index


file	other
f1	f2
f1	f3
f2	f3

Create database and add misc table with the index:

db = audformat.testing.create_db(minimal=True)
db["misc"] = audformat.MiscTable(index)
db["misc"]["values"] = audformat.Column()
db.misc_tables["misc"]

levels: {file: object, other: object}
columns:
  values: {}

Assign labels to a table:

values_list = [0, 1, 0]
values_dict = {"values": values_list}
db["misc"].set(values_dict)

Access labels as pandas.DataFrame:

db["misc"].get()

		values
file	other
f1	f2	0
f1	f3	1
f2	f3	0

Assign labels to a column:

db["misc"]["values"].set(values_list)

Access labels as pandas.Series

db["misc"]["values"].get()

		values
file	other
f1	f2	0
f1	f3	1
f2	f3	0

Access labels from a misc table with an index:

db["misc"].get(index[:2])

		values
file	other
f1	f2	0
f1	f3	1