How to use wildcards in keyword wildcard_constraints

For example, I have the following wildcards.

dataset = ['A1', 'A2', 'A3', 'B1', 'B2', 'B3']
group = ['A', 'B']

I am trying to contrain my dataset with my group. for example, I want to create

A1/file.A.txt A2/file.A.txt A3/file.A.txt B1/file.B.txt …

I wrote a following rule hoping to make that possible

rule complex_conversion:
    input:
        "{dataset}/inputfile"
    output:
        "{dataset}/file.{group}.txt"
    wildcard_constraints:
        dataset = {group} + 'd+'
        #dataset = {wildcards.group} + 'd+'
    shell:
        "somecommand --group {wildcards.group}  < {input}  > {output}"

oops, I got the error

TypeError:unhashable type: 'list'
#NameError: name 'wildcards' is not defined

It seems like that the {group} is regarded as a list to pass in the keywords wildcard_constraints.

Are there any methods to use wildcards in wildcards_constrain or alternative to mapped the dataset to group.

Answer

This doesn’t answer your question but maybe it helps… If your list of output files is a combination of dataset and group, I would create that list first and then use it as list of output files:

dataset = ['A1', 'A2', 'A3', 'B1', 'B2', 'B3']
group = ['A', 'B']

# Use a for-loop or whatever to create this list:
datagrp = ['A1/file.A.txt','A2/file.A.txt', 'A3/file.A.txt', 'B1/file.B.txt']

wildcard_constraints:
    # This prevents wildcards to be interpreted as regexes
    dataset = '|'.join([x for x in dataset]),
    group = '|'.join([x for x in group])

rule all:
    input:
        datagrp,

rule complex_conversion:
    input:
        "{dataset}/inputfile"
    output:
        "{dataset}/file.{group}.txt"
    shell:
        "somecommand --group {wildcards.group}  < {input}  > {output}"