LpVariable dicts issue parsing indices of the form [(a1, b1, c1), (a2, b2, c2), ..., (an, bn, cn)]

[hdelrio]

Details for the issue

When creating a collection of LpVariables using the class method dicts, using an index that is a list of tuples in order to emulate a matrix it only work when the name does not contain a '%'

What did you do?

i1 = list(range(10))
i2 = list('abcdefghij')
i3 = 'jan feb mar apr may jun jul aug sep oct'.split()
indices = [(a, b, c) for a in i1 for b in i2 for c in i3]

P = pl.LpVariable.dicts('P', indices=indices)
P[0, 'a', 'jan']

What did you expect to see?

I was expecting to see: P_(0, 'a', 'jan')

What did you see instead?

P_(0,'a','jan')

Which is a weird looking name

[pchtsp]

I'm not sure if the formatting from github is changing the output. Because I see the same variable name...

[urbanophile]

I think their example is an issue with the github markdown formatting. The variable name is

In [7]: P[9,'j','oct'].name
Out[7]: "P_(9,_'j',_'oct')"

from this

In [1]: from pulp import *

In [2]: i1 = list(range(10))
   ...: i2 = list('abcdefghij')
   ...: i3 = 'jan feb mar apr may jun jul aug sep oct'.split()
   ...: indices = [(a, b, c) for a in i1 for b in i2 for c in i3]
   ...: 
   ...: P = LpVariable.dicts('P', indices=indices)

In [3]: P
Out[3]: 
{(0, 'a', 'jan'): P_(0,_'a',_'jan'),
 (0, 'a', 'feb'): P_(0,_'a',_'feb'),
 (0, 'a', 'mar'): P_(0,_'a',_'mar'),
 (0, 'a', 'apr'): P_(0,_'a',_'apr'),
 (0, 'a', 'may'): P_(0,_'a',_'may'),

Either a little more logic could be added to handle the construction of the variable names here :

pulp/pulp/pulp.py

Line 401 in 6af3801

name += "_%s" * len(indices)

Or the docstring could amend with more info. Happy to make a pull request.