create_rss.py

import os
import pprint
from datetime import datetime, timedelta
import pytz
# https://pypi.python.org/pypi/feedgenerator
import feedgenerator

rss_max = 10
base_url = 'http://kamekokamekame.net/'
root_path = 'source/'


def get_rss_files(base_path):
    dirs = []
    for x in os.listdir(base_path):
        dd = os.path.join(base_path, x)
        if os.path.isdir(dd):
            if x[0] != '_':
                dirs.append(dd)
    dirs.sort(reverse=True)
    # print(dirs)
    rss = []
    num = 0
    for d in dirs:
        for ff in os.listdir(d):
            if ff[-3:] == '.md':
                rss.append(os.path.join(d, ff))
                num += 1
            if num > rss_max:
                break
        if num > rss_max:
            break

    rss.sort(reverse=True)

    # pprint.pprint(rss)
    return rss


def create_item_data(f_path):
    """itemタグ用の情報を取得."""
    item = dict()
    # link
    item['link'] = base_url + f_path.replace(root_path, '')[:-3] + '.html'
    item['unique_id'] = item['link']
    # pubDate
    b_name = os.path.basename(f_path)[:-3]
    jst = pytz.timezone('Asia/Tokyo')
    dt = datetime.strptime(b_name, '%Y_%m_%d').replace(tzinfo=jst)
    item['pubdate'] = dt + timedelta(hours=9)

    str_description = ''
    with open(f_path) as f:
        for idx, row in enumerate(f):
            if idx == 0:
                item['title'] = row.rstrip().replace('#','')
            elif idx == 1 and row[0] == '_':
                item['category'] = row.rstrip().replace('_', '')
            else:
                str_description += row.strip().replace('#','')

    item['description'] = str_description[:100]

    return item


def create_feed():
    title = "鉄は熱いうちに打て"
    link = base_url
    feed_url = "kamekokamekame.net/rss20.xml"
    description = "亀の歩みエンジニア"

    # フィードを生成
    return feedgenerator.Rss201rev2Feed(title=title,link=link,feed_url=feed_url,description=description,language="ja")


def main():
    feed = create_feed()
    files = get_rss_files(root_path)
    for ff in files:
        item_data = create_item_data(ff)
        feed.add_item(**item_data)
    with open('build/html/rss20.xml', 'w') as rf:
        rf.write(feed.writeString("utf-8"))
        
if __name__ == '__main__':
    main()