Agile blog recipe for Calibre

agile.recipe

from calibre.web.feeds.news import BasicNewsRecipe
from calibre.ebooks.BeautifulSoup import BeautifulSoup

class AgileRecipe(BasicNewsRecipe):

    title = u'Agile Blog'
    __author__ = 'Gerald Kim'

    publisher = 'agile.egloos.com'
    category = 'info'
    oldest_article = 9999
    max_articles_per_feed = 9999
    no_stylesheet = True
    auto_cleanup = True

    remove_javascript = True

    feeds = [
        u'http://agile.egloos.com/archives/2021/03',
        u'http://agile.egloos.com/archives/2020/05',
        u'http://agile.egloos.com/archives/2020/02',
        u'http://agile.egloos.com/archives/2019/09',
        u'http://agile.egloos.com/archives/2018/12',
        u'http://agile.egloos.com/archives/2018/04',
        u'http://agile.egloos.com/archives/2018/03',
        u'http://agile.egloos.com/archives/2018/02',
        u'http://agile.egloos.com/archives/2017/07',
        u'http://agile.egloos.com/archives/2017/05',
        u'http://agile.egloos.com/archives/2017/04',
        u'http://agile.egloos.com/archives/2017/03',
        u'http://agile.egloos.com/archives/2016/12',
        u'http://agile.egloos.com/archives/2016/07',
        u'http://agile.egloos.com/archives/2015/12',
        u'http://agile.egloos.com/archives/2015/11',
        u'http://agile.egloos.com/archives/2015/09',
        u'http://agile.egloos.com/archives/2015/08',
        u'http://agile.egloos.com/archives/2015/06',
        u'http://agile.egloos.com/archives/2015/05',
        u'http://agile.egloos.com/archives/2015/04',
        u'http://agile.egloos.com/archives/2015/03',
        u'http://agile.egloos.com/archives/2015/02',
        u'http://agile.egloos.com/archives/2014/12',
        u'http://agile.egloos.com/archives/2014/11',
        u'http://agile.egloos.com/archives/2014/09',
        u'http://agile.egloos.com/archives/2014/08',
        u'http://agile.egloos.com/archives/2014/03',
        u'http://agile.egloos.com/archives/2014/01',
        u'http://agile.egloos.com/archives/2013/12',
        u'http://agile.egloos.com/archives/2013/10',
        u'http://agile.egloos.com/archives/2013/08',
        u'http://agile.egloos.com/archives/2013/06',
        u'http://agile.egloos.com/archives/2013/05',
        u'http://agile.egloos.com/archives/2013/04',
        u'http://agile.egloos.com/archives/2013/02',
        u'http://agile.egloos.com/archives/2012/09',
        u'http://agile.egloos.com/archives/2012/08',
        u'http://agile.egloos.com/archives/2012/06',
        u'http://agile.egloos.com/archives/2012/05',
        u'http://agile.egloos.com/archives/2012/03',
        u'http://agile.egloos.com/archives/2011/12',
        u'http://agile.egloos.com/archives/2011/11',
        u'http://agile.egloos.com/archives/2011/10',
        u'http://agile.egloos.com/archives/2011/09',
        u'http://agile.egloos.com/archives/2011/04',
        u'http://agile.egloos.com/archives/2011/03',
        u'http://agile.egloos.com/archives/2011/02',
        u'http://agile.egloos.com/archives/2011/01',
        u'http://agile.egloos.com/archives/2010/12',
        u'http://agile.egloos.com/archives/2010/10',
        u'http://agile.egloos.com/archives/2010/09',
        u'http://agile.egloos.com/archives/2010/08',
        u'http://agile.egloos.com/archives/2010/07',
        u'http://agile.egloos.com/archives/2010/06',
        u'http://agile.egloos.com/archives/2010/05',
        u'http://agile.egloos.com/archives/2010/04',
        u'http://agile.egloos.com/archives/2010/03',
        u'http://agile.egloos.com/archives/2010/02',
        u'http://agile.egloos.com/archives/2010/01',
        u'http://agile.egloos.com/archives/2009/12',
        u'http://agile.egloos.com/archives/2009/11',
        u'http://agile.egloos.com/archives/2009/10',
        u'http://agile.egloos.com/archives/2009/09',
        u'http://agile.egloos.com/archives/2009/08',
        u'http://agile.egloos.com/archives/2009/07',
        u'http://agile.egloos.com/archives/2009/06',
        u'http://agile.egloos.com/archives/2009/05',
        u'http://agile.egloos.com/archives/2009/04',
        u'http://agile.egloos.com/archives/2009/03',
        u'http://agile.egloos.com/archives/2009/02',
        u'http://agile.egloos.com/archives/2009/01',
        u'http://agile.egloos.com/archives/2008/12',
        u'http://agile.egloos.com/archives/2008/11',
        u'http://agile.egloos.com/archives/2008/10',
        u'http://agile.egloos.com/archives/2008/09',
        u'http://agile.egloos.com/archives/2008/08',
        u'http://agile.egloos.com/archives/2008/07',
        u'http://agile.egloos.com/archives/2008/06',
        u'http://agile.egloos.com/archives/2008/05',
        u'http://agile.egloos.com/archives/2008/04',
        u'http://agile.egloos.com/archives/2008/03',
        u'http://agile.egloos.com/archives/2008/02',
        u'http://agile.egloos.com/archives/2008/01',
        u'http://agile.egloos.com/archives/2007/12',
        u'http://agile.egloos.com/archives/2007/11',
        u'http://agile.egloos.com/archives/2007/10',
        u'http://agile.egloos.com/archives/2007/09',
        u'http://agile.egloos.com/archives/2007/08',
        u'http://agile.egloos.com/archives/2007/07',
        u'http://agile.egloos.com/archives/2007/06',
        u'http://agile.egloos.com/archives/2007/05',
        u'http://agile.egloos.com/archives/2007/04',
        u'http://agile.egloos.com/archives/2007/03',
        u'http://agile.egloos.com/archives/2007/02',
        u'http://agile.egloos.com/archives/2007/01',
        u'http://agile.egloos.com/archives/2006/12',
        u'http://agile.egloos.com/archives/2006/11',
        u'http://agile.egloos.com/archives/2006/10',
        u'http://agile.egloos.com/archives/2006/09',
        u'http://agile.egloos.com/archives/2006/08',
        u'http://agile.egloos.com/archives/2006/07',
        u'http://agile.egloos.com/archives/2006/06',
        u'http://agile.egloos.com/archives/2006/05',
        u'http://agile.egloos.com/archives/2006/04',
        u'http://agile.egloos.com/archives/2006/03',
        u'http://agile.egloos.com/archives/2006/02'

    ]

    def parse_index(self):
        self.log.warning("parse_index")
        totalfeeds = []
        lfeeds = self.get_feeds()
        
        lfeeds.reverse()
        #lfeeds = [lfeeds[0]]
        
        for feedurl in lfeeds:
            articles = []
            
            self.report_progress(0, 'Fetching feed' + ' %s...' %
                                 (feedurl))

            soup = self.index_to_soup(feedurl)
            body = soup.find('div', attrs={'class': 'POST_BODY'})
            for item in body.findAll('a'):
                if item['href'].startswith('/archives'):
                    self.log.warning("skipping archives" + item['href'])
                    continue
                
                articles.append({
                    'url': 'http://agile.egloos.com' + item['href'],
                    #'url': 'http://localhost:8080' + item['href'],
                    'title': item.contents[0]
                })

            articles.reverse()
            totalfeeds.append((feedurl[-7:], articles))
            
        return totalfeeds
    
    
    # def preprocess_raw_html(self, raw_html, url):
    #     soup = BeautifulSoup(raw_html)

    #     post = soup.find('div', attrs={'class':'POST_BODY'})
    #     posttaglist = post.find('div', attrs={'class':'posttaglist'})
    #     if posttaglist:
    #         posttaglist.clear()

    #     soup = BeautifulSoup(u'''<html><body></body></html>''')
    #     soup.body.insert(0, post)

    #     return str(soup)

osx 가 아닌 다른 os 사용자는 calibre 설치 후 ebook-convert 파일을 찾아서 실행하시면 됩니다.

이걸해야하다니.. 우울하네요. 이글루스 밉다..

pdf 158MB 코드 안적어둔 이유가 있었군요.