#!/usr/bin/env python

import glob
import os
import re

post_dir = '_posts/'
tag_dir = 'tag/'

total_tags = []
for filename in glob.glob(post_dir + '*.md'):
    matcher = r'^tags:$'
    with open(filename, 'r') as fd:
        tagged_line = False
        for line in fd:
            if tagged_line:
                if line.startswith('---'):
                    tagged_line = False
                elif not line.startswith('-'):
                    tagged_line = False
                else:
                    total_tags.append(line[1:].strip())
            if re.match(matcher, line):
                tagged_line = True

total_tags = set(total_tags)

for tag in glob.glob(tag_dir + '*.md'):
    os.remove(tag)

if not os.path.exists(tag_dir):
    os.makedirs(tag_dir)

for tag in total_tags:
    td = os.path.sep.join([tag_dir, tag])
    if not os.path.exists(td):
        os.makedirs(td)
    with open(os.path.sep.join([td, 'index.md']), 'w+') as fd:
        fd.write('---\nlayout: tag_page\ntitle: \"Tag: ' + tag + '\"\ntag: ' + tag + '\nrobots: noindex\n---\n')
print("Tags generated, count", total_tags.__len__())