#!/usr/bin/env python3

import glob, argparse, sys, os.path, tokenize

try:
    import enchant.checker
except ImportError:
    print('Please install python3-enchant.')
    sys.exit(1)

def recover_from_file(path):
    """
    Use tokenize to recover strings and comments
    """
    with open(path, 'rb') as checkme:
        print("File : %s" % path)
        return {i for i in tokenize.tokenize(checkme.readline) if i[0] in (tokenize.COMMENT, tokenize.STRING)}

def check_spelling(checker, tokens, interactive=False):
    """
    Use python3-enchant to check spelling.

    We remove some false positives (less than two letters, more than 12
    letters, etc) and use a custom word file for python specific words
    to check remaining words. 
    """
    for j in tokens:
        # Remove false positives and duplicates
        words = {i.strip() for i in j[1].split(' ') if all((2 < len(i) < 12,
            i.isalpha(), i[1:].islower()))}

        for word in words:
            if not checker.check(word):
                print("Error : %s" % word)
                # It's faster to type when the word does not exist...
                if interactive and input()=='':
                    add_to_pwl(word)

def add_to_pwl(word):
    """
    Add the word to our custom words.
    """
    with open('python_words.txt', 'a') as pwl:
        pwl.write('\n%s' % word)

interactive = True
for i in glob.iglob('Lib/*.py'):
    d = enchant.DictWithPWL('en_US', 'python_words.txt')
    check_spelling(d, recover_from_file(i), True)