print a word as it should appear if it is a node label (including tip label) in a newick tree

newick_label.py

#!/usr/bin/env python
import sys
import os
import re
_SCRIPT_NAME = os.path.split(sys.argv[0])[-1]
_FORBIDDEN = re.compile(r'[^- 0-9a-zA-Z`~@#$%^&*()_+={}|\\\[\]:;"\'<,>.?/]')
_NEEDS_QU_PUNC_STR = r'[\[\]():,;]'
_NEEDS_QUOTES_PATTERN = re.compile(r'(\s|' + _NEEDS_QU_PUNC_STR + ')')
_NEEDS_QUOTES_PUNC_PATTERN = re.compile(_NEEDS_QU_PUNC_STR)
_SINGLE_QUOTE = "'"
_TWO_QUOTES = "''"
def print_quoted(word):
    # reject labels any label with a character other than:
    #   spaces, Roman alphabet letters, and US keyboard punctuation
    if _FORBIDDEN.search(word):
        msg = 'The input word "' + word + '" a forbidden character!\n'
        sys.stderr.write(_SCRIPT_NAME + ': ' + msg)
        return False
    if _SINGLE_QUOTE in word:
        # in newick and NEXUS a label with ' has to have 
        #  ' on the outside and change every ' to '' (two single quotes)
        split_on_quote = word.split(_SINGLE_QUOTE)
        doubled_single_quotes = _TWO_QUOTES.join(split_on_quote)
        print _SINGLE_QUOTE + doubled_single_quotes + _SINGLE_QUOTE
    elif _NEEDS_QUOTES_PATTERN.search(word):
        if not _NEEDS_QUOTES_PUNC_PATTERN.search(word):
            # space is the only special char.
            #   we can use the _ for space trick that users prefer...
            print '_'.join(word.split(' '))
        else:
            print _SINGLE_QUOTE + word + _SINGLE_QUOTE
    else:
        print word
    return True
if __name__ == '__main__':
    if len(sys.argv) > 1:
        inp = sys.argv[1:]
        strip_newline = False
    else:
        inp = iter(sys.stdin)
        strip_newline = True
    n_failed = 0
    for word in inp:
        if strip_newline:
            word = word[:-1]
        if not print_quoted(word):
            n_failed += 1
    if n_failed > 0:
        sys.stderr.write('{s}: {f} words rejected.\n'.format(s=_SCRIPT_NAME, f=n_failed))
        sys.exit(1)