def get_slot_values_BIO()

in main.py [0:0]

• 45 lines of code
• 16 McCabe index (conditional complexity)

def get_slot_values_BIO(tokens, labels):
    assert len(tokens) == len(labels), f'tokens:{tokens}\nlabels:{labels}'

    rectified_labels = []
    for l in labels:
        if l.startswith('I-'):
            if rectified_labels == []:
                rectified_labels.append('B-' + l[2:])
            elif rectified_labels[-1][2:] == l[2:]:
                rectified_labels.append(l)
            else:
                rectified_labels.append('B-' + l[2:])
        else:
            rectified_labels.append(l)
    labels = rectified_labels

    ans = dict()
    active_type = None
    active_value = ''
    for i in range(len(tokens)):
        if labels[i] == 'O':
            if active_type is not None:
                if active_type not in ans:
                    ans[active_type] = []
                ans[active_type].append(active_value)
                active_type = None
                active_value = ''
        elif labels[i].startswith('B-'):
            if active_type is not None:
                if active_type not in ans:
                    ans[active_type] = []
                ans[active_type].append(active_value)
            active_type = labels[i][2:]
            active_value = tokens[i]
        elif labels[i].startswith('I-'):
            if active_type == labels[i][2:]:
                active_value = active_value + ' ' + tokens[i]
            else:
                print(tokens, labels, i, active_type, labels[i][2:])
                assert False
        else:
            assert False

    if active_type is not None:
        if active_type not in ans:
            ans[active_type] = []
        ans[active_type].append(active_value)

    return ans