blog/githublog.py

#!/usr/bin/env python

"""
File: githublog.py

Hugo assistance to generate a special README file, including all posts title inside
as a TOC.

This will make blog repo to be a blog inplace.
"""
import os
import string
import subprocess
from posixpath import join as urljoin

POSTS_ROOT = 'content'
EXTS = ['.md', '.rst']
README = 'README.md'
GITREPO = 'http://git.wanbits.io/joe/blog'

def get_git_remote():
    proc = subprocess.Popen(['git', 'remote', '-v'], stdout=subprocess.PIPE, shell=True)
    (out, err) = proc.communicate()
    if err:
        return None
    lines = out.decode('utf-8').split('\n')
    if len(lines) < 2:
        return None
    parts = lines[1].split('\t')
    if len(parts) < 2:
        return None
    values = parts[1].split(' ')
    return None if len(values) < 1 else values[0]


def get_category(dir):
    ''' '''
    parts = dir.split(os.sep)
    return os.sep.join(parts[1:])

remote = get_git_remote() if not GITREPO else GITREPO

if not remote:
    print('not a git repo, quit')
    exit(1)

posts = {}
# find out all non-draft posts
for r, dl, fl in os.walk(POSTS_ROOT):
    if fl:
        cate = get_category(r)
        posts[cate] = []

    for f in fl:
        _, ext = os.path.splitext(f)
        if ext.lower() not in EXTS:
            continue
        fullname = os.path.join(r, f)
        comment_started = False
        metas = {}
        with open(fullname, 'r', encoding='utf-8') as h:
            line = h.readline().strip()
            while line:
                if line == '---':
                    if not comment_started:
                        comment_started = True
                    else:
                        break
                else:
                    parts = line.split(': ')
                    if len(parts) != 2:
                        break
                    metas[parts[0].strip().lower()] = parts[1].strip().lower()
                    metas['file'] = f

                line = h.readline().strip()
        if metas.get('draft') == 'false':
            posts[cate].append(metas)

# generate README.md
with open(README, 'wt', encoding='utf-8') as h:
    h.write('')
    for k, v in posts.items():
        h.write('## {}\n\n'.format(k))
        v = sorted(v, key=lambda metas: metas.get('date'), reverse=True)
        for metas in v:
            h.write('- [{}]({})\n'.format(metas.get('title'), urljoin(remote, 'src/master', POSTS_ROOT, k, metas.get('file'))))
        
        h.write('\n')

print('done.')