website/tag-generator.py

#!/usr/bin/env python

"""
tag_generator.py

Copyright 2017 Long Qian
Contact: lqian8@jhu.edu

This script creates tags for your Jekyll blog hosted by Github page.
No plugins required.
"""

import glob
import os

post_dir = "_pages/"
tag_dir = "tag/"

filenames = glob.glob(post_dir + "*html")

total_tags = []
for filename in filenames:
    f = open(filename, "r")
    crawl = False
    for line in f:
        if crawl:
            current_tags = line.strip().split()
            if current_tags[0] == "tags:":
                total_tags.extend(current_tags[1:])
                crawl = False
                break
        if line.strip() == "---":
            if not crawl:
                crawl = True
            else:
                crawl = False
                break
    f.close()
total_tags = set(total_tags)

old_tags = glob.glob(tag_dir + "*.html")
for tag in old_tags:
    os.remove(tag)

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

for tag in total_tags:
    tag_filename = tag_dir + tag + ".html"
    f = open(tag_filename, "a")
    write_str = (
        "---\nlayout: tagpage\ntitle: meta.titles.tags."
        + tag
        + '\ntag: "'
        + tag
        + '"\ndescription: meta.descriptions.tags.'
        + tag
        + '"\nrobots: noindex\n---\n'
    )
    f.write(write_str)
    f.close()
print("Tags generated, count", total_tags.__len__())