Mercurial > repos > shellac > guppy_basecaller
view env/lib/python3.7/site-packages/galaxy/util/permutations.py @ 3:758bc20232e8 draft
"planemo upload commit 2a0fe2cc28b09e101d37293e53e82f61762262ec"
author | shellac |
---|---|
date | Thu, 14 May 2020 16:20:52 -0400 |
parents | 26e78fe6e8c4 |
children |
line wrap: on
line source
""" There is some shared logic between matching/multiplying inputs in workflows and tools. This module is meant to capture some general permutation logic that can be applicable for both cases but will only be used in the newer tools case first. Maybe this doesn't make sense and maybe much of this stuff could be replaced with itertools product and permutations. These are open questions. """ from collections import OrderedDict from galaxy.exceptions import MessageException from galaxy.util.bunch import Bunch input_classification = Bunch( SINGLE="single", MATCHED="matched", MULTIPLIED="multiplied", ) class InputMatchedException(MessageException): """ Indicates problem matching inputs while building up inputs permutations. """ def expand_multi_inputs(inputs, classifier, key_filter=None): key_filter = key_filter or (lambda x: True) single_inputs, matched_multi_inputs, multiplied_multi_inputs = __split_inputs( inputs, classifier, key_filter ) # Build up every combination of inputs to be run together. input_combos = __extend_with_matched_combos(single_inputs, matched_multi_inputs) input_combos = __extend_with_multiplied_combos(input_combos, multiplied_multi_inputs) return input_combos def __split_inputs(inputs, classifier, key_filter): key_filter = key_filter or (lambda x: True) single_inputs = OrderedDict() matched_multi_inputs = OrderedDict() multiplied_multi_inputs = OrderedDict() for input_key in filter(key_filter, inputs): input_type, expanded_val = classifier(input_key) if input_type == input_classification.SINGLE: single_inputs[input_key] = expanded_val elif input_type == input_classification.MATCHED: matched_multi_inputs[input_key] = expanded_val elif input_type == input_classification.MULTIPLIED: multiplied_multi_inputs[input_key] = expanded_val return (single_inputs, matched_multi_inputs, multiplied_multi_inputs) def __extend_with_matched_combos(single_inputs, multi_inputs): """ {a => 1, b => 2} and {c => {3, 4}, d => {5, 6}} Becomes [ {a => 1, b => 2, c => 3, d => 5}, {a => 1, b => 2, c => 4, d => 6}, ] """ if len(multi_inputs) == 0: return [single_inputs] matched_multi_inputs = [] first_multi_input_key = next(iter(multi_inputs.keys())) first_multi_value = multi_inputs.get(first_multi_input_key) for value in first_multi_value: new_inputs = __copy_and_extend_inputs(single_inputs, first_multi_input_key, value) matched_multi_inputs.append(new_inputs) for multi_input_key, multi_input_values in multi_inputs.items(): if multi_input_key == first_multi_input_key: continue if len(multi_input_values) != len(first_multi_value): raise InputMatchedException("Received %d inputs for '%s' and %d inputs for '%s', these should be of equal length" % (len(multi_input_values), multi_input_key, len(first_multi_value), first_multi_input_key)) for index, value in enumerate(multi_input_values): matched_multi_inputs[index][multi_input_key] = value return matched_multi_inputs def __extend_with_multiplied_combos(input_combos, multi_inputs): combos = input_combos for multi_input_key, multi_input_value in multi_inputs.items(): iter_combos = [] for combo in combos: for input_value in multi_input_value: iter_combo = __copy_and_extend_inputs(combo, multi_input_key, input_value) iter_combos.append(iter_combo) combos = iter_combos return combos def __copy_and_extend_inputs(inputs, key, value): new_inputs = dict(inputs) new_inputs[key] = value return new_inputs