TaCle: Tabular Constraint Learner

TaCLe is a constraint learner designed for usage in spreadsheets and tabular data.

Installation

pip install tacle

Using TaCLe

Running TaCLe from command line

TaCLe can be used from command line to analyze a CSV file:

python -m tacle data/magic_ice_cream.csv

This returns a list of constraints in the file:

ALLDIFFERENT(T2[:, 1])
ALLDIFFERENT(T1[:, 4])
ALLDIFFERENT(T1[:, 5])
ALLDIFFERENT(T1[:, 6])
T1[:, 1] -> T2[:, 1]
T1[:, 6] = SUM(T1[:, 3:5], row)

Filter output

If you are interested in specific types of constraints, you can filter the output:

python -m tacle data/magic_ice_cream.csv -f "foreign-key"  # Report only foreign keys
python -m tacle data/magic_ice_cream.csv -f "<f>"  # Report only formulas -- use <c> for only constraints

Tables

To view the tables that TaCLe finds in a file you can run:

python -m tacle data/magic_ice_cream.csv -t

This returns the tables and blocks found by TaCLe:

Table T1, (1:9, 0:7)
Columns 0-2 (string), Columns 2-6 (int), Columns 6-7 (string)

Table T2, (11:15, 0:2)
Columns 0-1 (string), Columns 1-2 (int)

Papers

Read more about how TaCLe works in one of our two papers journal version, demo paper.