sequences.py

import numpy as np...
import numpy as np
import pandas as pd
import re

def get_patterns(ids, debug=False):
    max_char = len(str(np.max(ids))) # getting max length

    ids = list(map(str, ids)) # changing to chars

    # padding any zeros to the left
    ids = [number.zfill(max_char) for number in ids]

    # converting to all chars
    ids = [list(id) for id in ids]
    ids = np.array(ids)

    counts = pd.DataFrame(columns=['i', 'dict', 'len'])

    for i in range(0, max_char):
        unique, cnt = np.unique(ids[:, i], return_counts=True)
        mydict = (dict(zip(unique, cnt)))

        counts = counts.append({
            'i': i,
            'dict': mydict,
            'len': len(mydict)
        }, ignore_index=True)

    # we only want cols where the char is the same for all rows
    counts = counts[counts['len'] == 1]

    counts['i2'] = (counts['i'].shift(-1))
    counts['diff'] = (counts['i2'] - counts['i'])

    # we want sequences in the col indexes
    sequences = counts[counts['diff'] != 1].copy()
    n_sequences = len(counts) - len(sequences)

    debug and print(counts)
    debug and print(sequences)
    debug and print("Number of sequences: ", n_sequences)

    indexes = sequences[sequences['diff'] != 1].index.values.astype(int)
    debug and print(indexes)

    prev = 0
    seqs = []

    # apparently when the sequence is at the beginning, slicing fails
    switch_i0 = len(counts[counts['i'] == 0]) == 1

    for i in indexes:
        # todo -> improve and double check
        to_i = i + 1 if prev == 0 and switch_i0 else i

        myseq = [''.join(list(mydict))
                 for mydict in counts[prev:to_i]['dict']]
        seqs.append(''.join(myseq))
        
        prev = to_i

    return seqs

def get_suffixes(ids):
    ids_string = list(map(str, ids))
    suffixes = []

    for pattern in get_patterns(ids):
        print("^.*{}(.*)$".format(pattern))
        r = re.compile("^.*{}(.*)$".format(pattern))
        suffixes.append([r.match(line).group(1)
                    for line in ids_string if r.match(line)])
    
    return suffixes


ids = np.array([417118.0, 417119.0, 417213.0, 417166.0, 417167.0,
 417168.0, 417210.0, 517066.0, 517004.0, 517005.0, 517059.0])

get_patterns(ids)
# ^.*17(.*)$
# ^.*.0(.*)$

get_suffixes(ids)
# [['118.0',
#   '119.0',
#   '213.0',
#   '166.0',
#   '167.0',
#   '168.0',
#   '210.0',
#   '066.0',
#   '004.0',
#   '005.0',
#   '059.0'],
#  ['', '', '', '', '', '', '', '', '', '', '']]


ids = np.array([41117118.10, 41117119.10, 41117213.10, 41117166.10, 41117167.10,
                41117168.10, 41117210.10, 51117066.10, 51117004.10, 51117005.10,
                51117059.10])
print(get_patterns(ids))
# ['1117', '.1']

ids = np.array([12341117118.10, 12341117119.10, 12341117213.10, 12341117166.10, 12341117167.10, 12341117168.10, 12341117210.10, 12351117066.10, 12351117004.10, 12351117005.10, 12351117059.10])
print(get_patterns(ids))
# ['123', '1117', '.1']

ids = np.array([123, 345, 678])
get_patterns(ids)
get_suffixes(ids)