#!/usr/bin/env python
# -*- coding: utf-8 -*-

import sys, os, re
import html

END_OF_ARTICLE = '--------'
END_OF_BODY = '-----'

class Post(object):
template = '''\
---
layout: post
title: {title}
tags:
{tag_list}
---
{RAW}
{body}
{ENDRAW}
'''

def __init__(self):
self.author = None
self.title = None
self.basename = None
self.status = None
self.date = None
self.tags = []
self.body = None
self.comments = []
self.rest_metadata = {}

def render_tag_list(self):
out = ''
for tag in self.tags:
out += '- {}\n'.format(tag)
return out

def render_body(self):
return self.template.format(
title=self.title,
tag_list=self.render_tag_list(),
body=self.body,
RAW='', # jekyll の命令が format に間違えて解釈されるのを防ぐ
ENDRAW=''
)

def out_path(self):
return self.basename.replace('/', '-') + '.html'

def add_metadata(self, name, value):
if name == 'TITLE':
self.title = html.unescape(value).replace(':', ':')
elif name in ['AUTHOR', 'BASENAME', 'STATUS', 'DATE']:
self.__setattr__(name.lower(), value)
elif name == 'CATEGORY':
self.tags.append(value)
else:
self.rest_metadata[name] = value


# f から一記事分だけ読み取る
def read_post(f):
post = Post()
while True:
line = f.readline()
if line == '': raise EOFError() # EOF
line = line.strip()
if line == END_OF_ARTICLE:
return post
elif line == END_OF_BODY or line == '':
pass
elif line == 'BODY:':
post.body = read_body(f)
elif line == 'COMMENT:':
post.comments.append(read_comment(f))
else:
# HOGE: FUGA 形式のメタデータ
name, value = line.split(': ', maxsplit=1)
post.add_metadata(name, value)
return post

# f から記事の body を読み取る
def read_body(f):
body = ''
link_to_self = re.compile(r'https?://inkar-us-i.hatenablog.com/entry/(?P<y>\d+)/(?P<m>\d+)/(?P<d>\d+)/(?P<base>\d+)')
  while True:
  line = f.readline()
  if line.strip() == END_OF_BODY:
  return body
  else:
  body += link_to_self.sub(r'/\g<y>/\g<m>/\g<d>/\g<base>.html', line)
  return body

  # f から記事のコメントを読み取る
  def read_comment(f):
  # 今のところコメントは使わないので、ただの文字列として読み取る
  # 実際は AUTHOR: とかのメタデータがあったりするので、そのへんも考慮する必要がある
  return read_body(f)

  def main():
  try:
  src_file = sys.argv[1]
  dst_dir = sys.argv[2]
  except IndexError:
  sys.stderr.write('Usage: python mt_to_md.py SRC_FILE DST_DIR\n')
  sys.exit(1)

  os.makedirs(dst_dir, exist_ok=True)

  with open(src_file) as f:
  while True:
  try:
  post = read_post(f)
  out_path = os.path.join(dst_dir, post.out_path())
  if post.status == 'Publish':
  with open(out_path, 'w') as out:
  out.write(post.render_body())
  print('wrote: {}'.format(out_path))
  else:
  print('skipped: {}'.format(out_path))
  except EOFError:
  return # たぶん終わり

  print('done.')

  if __name__ == '__main__':
  main()