Generate tag directories for Jekyll

tag-generator.py

#!/usr/local/bin/python3

import re
import glob
import os

TAG_TEMPLATE = '''---
layout: tag_pages
title: "{tag}"
tag: {tag}
---
'''


def get_tags(line):
    tag_list = re.search("\[.*\]", line) \
            .group(0).replace('[', '').replace(']', '')
    return list(filter(
        lambda x: x, map(lambda x: x.strip(), tag_list.split(','))
        ))


def extract_tags(file_name):
    f = open(file_name, 'r', encoding='utf8')
    front_matter = '---'
    fm_count = 0
    tags = []
    for line in f.readlines():
        if fm_count == 2:
            break
        if line == front_matter:
            fm_count += 1
        if line[:4] == 'tags':
            tags.extend(get_tags(line))
    return tags


def _summary(op, file_names):
    print(op)
    print("----")
    for f in file_names:
        print(f)
    else:
        print('None')
    print("----\n")

def update_tag_files():
    # create tag directory
    tag_dir = 'tag'
    if not os.path.exists(tag_dir):
        os.makedirs(tag_dir)

    # collect all the tags
    tags = []
    file_names = glob.glob('_posts/*')
    for f in file_names:
        tags.extend(extract_tags(f))
    tags = set(tags)

    removed = []
    skipped = []
    created = []
    # remove non-existing tags
    tag_files = glob.glob('{}/*'.format(tag_dir))
    for file_name in tag_files:
         # let's hack :)
         # it's always {tag_dir}/{tag}{.md}
        tag = file_name[len(tag_dir) + 1:-3]
        if tag not in tags:
            # outdated tag file, remove
            os.remove(file_name)
            removed.append(file_name)

    for tag in tags:
        file_name = '{}/{}.md'.format(tag_dir, tag)
        if os.path.exists(file_name):
            skipped.append(file_name)
        else:
            f = open(file_name, 'a')
            f.write(TAG_TEMPLATE.format(tag=tag))
            f.close()
            created.append(file_name)
    _summary('Created', created)
    _summary('Skipped', skipped)
    _summary('Removed', removed)


if __name__ == "__main__":
    update_tag_files()