indenter/indenter.py

#!/usr/bin/python
import re
import argparse
import functools
leading_whitespace = re.compile(r'^( +)(?! )')
#@functools.lru_cache(maxsize=None)
def get_whitespaces_re(line):
    return leading_whitespace.match(line)
def get_indent_frequency(lines):
    lens = sorted(match.span()[1] for match in filter(bool,map(get_whitespaces_re,lines)))
    freq = {}
    for i in range(len(lens)-1):
        p = lens[i]
        n = lens[i+1]
        diff = n - p
        if diff > 0:
            new_val = freq.get(diff,0) + 1
            freq[diff] = new_val
    return freq
def get_best_canitdate(lines):
    freqs = get_indent_frequency(lines)
    if freqs:
        return sorted(freqs.keys(),key = lambda k: freqs[k],reverse=True)[0]
def change_all(text,new_indent):
    lines = text.split('\n')
    i = get_best_canitdate(lines)
    for row,line in enumerate(lines):
        prev_whitespaces = get_whitespaces_re(line)
        if prev_whitespaces is not None:
            prev_len = prev_whitespaces.span()[1]
            n,remaining = divmod(prev_len,i)
            if remaining*2 >= i:
                n += 1
            stripped = line.lstrip(' ')
            lines[row]  = ' '*(new_indent*n) + stripped
    return '\n'.join(lines)
if __name__ == "__main__":
    parser = argparse.ArgumentParser()
    parser.add_argument('-o',help='output file')
    parser.add_argument('file')
    parser.add_argument('indent',type=int)
    args = parser.parse_args()
    with open(args.file) as file:
        text = file.read()
    new = change_all(text,args.indent)
    if args.o:
        with open(args.o,'w') as file:
            file.write(new)
    else:
        with open(args.file,'w') as file:
            file.write(new)