blog/githublog.py

#!/usr/bin/env python

"""
File: githublog.py

This is a quick&dirty tool helping you generating a TOC file for blog repo based on Hugo.

Usage:

python githublog.py

IMPORTANT

Before starting, you should check that the working-directory is right at the root of
the hugo blog and managed by git (is preferred)

"""

import os
import string
import subprocess
from posixpath import join as urljoin

INTRO_NAME = '简介'
INTRO_DESC = '''本站所有原创、翻译、引用都会随着时间进行更新'''
POSTS_ROOT = 'content'      # hugo root dir of posts.
EXTS = ['.md', '.rst']      # post file extensions. (should be in lowercase)
README = 'README.md'        # filename to be generated.
GITREPO = 'https://git.wenlabs.org/joe/blog'      # repo address of blog code.
BRANCH = 'master'           # publish branch


def get_git_remote():
    """ Get git remote url  
        should be executed in git directory and with git installed.
    """
    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_git_branch():
    """ Get current git branch """
    pass


def get_category(dir):
    ''' 
    Get blog category by directory strucure
    '''
    parts = dir.split(os.sep)
    return os.sep.join(parts[1:])

# repo remote
remote = get_git_remote() if not GITREPO else GITREPO
if not remote:
    print('not a git repo, quit')
    exit(1)

# current git branch
branch = get_git_branch() if not BRANCH else BRANCH
if not branch:
    print('no branch?')
    exit(1)

# all posts, searching result 
# map[str][list]
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)
        meta_started = False
        metas = {}  # hugo post meta-info map[str]str
        with open(fullname, 'r', encoding='utf-8') as h:
            line = h.readline().strip()
            while line:
                if line == '---':
                    if not meta_started:
                        meta_started = True
                    else:
                        break   # from while
                else:
                    parts = line.split(': ')
                    if len(parts) != 2:
                        break   # from while
                    metas[parts[0].strip().lower()] = parts[1].strip().lower()
                line = h.readline().strip()
        if metas.get('draft') == 'false':
            metas['file'] = f
            posts[cate].append(metas)

## generate README.md
with open(README, 'wt', encoding='utf-8') as h:
    h.write(f'## {INTRO_NAME}\n\n')
    h.write(f'{INTRO_DESC}\n\n')

    for cate, metalist in posts.items():
        h.write('## {}\n\n'.format(cate))
        # sortint posts by published date.
        metalist = sorted(metalist, key=lambda metas: metas.get('date'), reverse=True)
        for metas in metalist:
            h.write('- [{}]({})\n'.format(metas.get('title').strip('"'), urljoin(remote, 'src', branch, POSTS_ROOT, cate, metas.get('file'))))
        h.write('\n')

print('done.')