Generate true Reber grammar sequences.

reber.py

# This Python script can be used to generate true Reber grammar sequences.
# To generate false Reber grammar sequences, just change tuple values in 'traversal' at line 8.

import numpy as np

chars = 'BTSXPVE'

traversal = [[('T', 'P'), (1, 4)], [('S', 'X'), (1, 2)], [('S', 'X'), (3, 4)], [('E'), (-1,)], [('V', 'T'), (5, 4)], [('V', 'P'), (3, 2)]]

def generate_seq(min_length):
    while True:
        seq = ['B']
        node = 0

        while node is not -1:
            trans = traversal[node]
            index = np.random.randint(0, len(trans[0]))
            seq.append(trans[0][index])
            node = trans[1][index]

        if len(seq) > min_length:
            return ''.join(seq)

def check_seq(seq):
    seq = list(seq.upper())
    if seq[0] is not 'B':
        return False
    
    node = 0
    for s in seq[1:]:
        trans = traversal[node]
        try:
            index = trans[0].index(s)
            node = trans[1][index]
        except ValueError:
            return False

    return True

def to_one_hot(seq):
    seq = list(seq)
    in_seq = []

    for c in seq:
        encoded = np.zeros(7)
        encoded[chars.find(c)] = 1.
        in_seq.append(encoded)
    
    out_seq = in_seq[1:]
    in_seq = in_seq[0:-1]

    return in_seq, out_seq

def to_string(seq):
    reber_string = ''

    for s in seq:
        index = np.where(s==1.)[0][0]
        reber_string += chars[index]

    return reber_string