Various Korean news sources by Hoje Lee

Also update various exisitng Korean news sources.

Merge branch 'master' of https://github.com/hojel/calibre

recipes/chosun.recipe

@@ -0,0 +1,28 @@
+# -*- coding: utf-8 -*-
+__license__   = 'GPL v3'
+__copyright__ = '2015, Hoje Lee <hojelei at gmail.com>'
+'''
+Profile to download Chosun.com
+'''
+
+from calibre.web.feeds.news import BasicNewsRecipe
+
+class ChosunDotcom(BasicNewsRecipe):
+    language    = 'ko'
+    title       = u'조선일보'
+    description = u'조선닷컴 기사'
+    __author__	= 'Hoje Lee'
+    oldest_article = 7
+    max_articles_per_feed = 10
+    auto_cleanup   = True
+
+    feeds = [
+	(u'정치',    'http://www.chosun.com/site/data/rss/politics.xml'),
+	(u'조선비즈','http://biz.chosun.com/site/data/rss/rss.xml'),
+	(u'사회',    'http://www.chosun.com/site/data/rss/national.xml'),
+	(u'문화',    'http://www.chosun.com/site/data/rss/culture.xml'),
+	(u'국제',    'http://www.chosun.com/site/data/rss/international.xml'),
+	(u'오피니언','http://www.chosun.com/site/data/rss/editorials.xml'),
+	(u'스포츠',  'http://www.chosun.com/site/data/rss/sports.xml'),
+	(u'연예',    'http://www.chosun.com/site/data/rss/ent.xml'),
+           ]

recipes/hankyoreh.recipe

@@ -1,3 +1,4 @@
+# -*- coding: utf-8 -*-
 __license__   = 'GPL v3'
 __copyright__ = '2011, Seongkyoun Yoo <seongkyoun.yoo at gmail.com>'
 '''
@@ -6,42 +7,34 @@ Profile to download The Hankyoreh
 from calibre.web.feeds.news import BasicNewsRecipe
 
 class Hankyoreh(BasicNewsRecipe):
-    title          = u'Hankyoreh'
     language = 'ko'
+    title       = u'한겨례'
     description = u'The Hankyoreh News articles'
     __author__  = 'Seongkyoun Yoo'
-    oldest_article = 5
+    oldest_article = 7
-    recursions = 1
+    max_articles_per_feed = 10
-    max_articles_per_feed = 5
     no_stylesheets        = True
+    remove_javascript     = True
+
     keep_only_tags    = [
-                        dict(name='tr', attrs={'height':['60px']}),
+                        dict(name='div', attrs ={'class':['article-head']}),
-                        dict(id=['fontSzArea'])
+                        dict(name='div', attrs ={'class':['article-text']}),
                         ]
     remove_tags   = [
-       dict(target='_blank'),
+                    dict(name='p', attrs={'class':['category']}),
-       dict(name='td', attrs={'style':['padding: 10px 8px 5px 8px;']}),
-       dict(name='iframe', attrs={'width':['590']}),
-                  ]
-    remove_tags_after  = [
-                          dict(target='_top')
                     ]
+    remove_tags_after  = dict(id={'ad_box01'})
+
     feeds = [
-    ('All News','http://www.hani.co.kr/rss/'),
+        #(u'전체기사', 'http://www.hani.co.kr/ilram/rss/hkr_news_list_all.xml'),
-    ('Politics','http://www.hani.co.kr/rss/politics/'),
+        (u'정치',     'http://www.hani.co.kr/ilram/rss/hkr_news_list_politics.xml'),
-    ('Economy','http://www.hani.co.kr/rss/economy/'),
+        #(u'경제',     'http://www.hani.co.kr/rss/economy/'),
-    ('Society','http://www.hani.co.kr/rss/society/'),
+        (u'사회',     'http://www.hani.co.kr/ilram/rss/hkr_news_list_society.xml'),
-    ('International','http://www.hani.co.kr/rss/international/'),
+        #(u'국제',     'http://www.hani.co.kr/rss/international/'),
-    ('Culture','http://www.hani.co.kr/rss/culture/'),
+        (u'문화',     'http://www.hani.co.kr/ilram/rss/hkr_news_list_culture.xml'), 
-    ('Sports','http://www.hani.co.kr/rss/sports/'),
+        (u'스포츠',   'http://www.hani.co.kr/ilram/rss/hkr_news_list_sports.xml'),
-    ('Science','http://www.hani.co.kr/rss/science/'),
+        #(u'과학',     'http://www.hani.co.kr/rss/science/'),
-    ('Opinion','http://www.hani.co.kr/rss/opinion/'),
+        (u'사설·칼럼','http://www.hani.co.kr/ilram/rss/hkr_news_list_opinion.xml'),
-    ('Cartoon','http://www.hani.co.kr/rss/cartoon/'),
+        (u'만화만평', 'http://www.hani.co.kr/ilram/rss/hkr_news_list_cartoon.xml'),
-    ('English Edition','http://www.hani.co.kr/rss/english_edition/'),
+        #(u'한겨례섹션','http://www.hani.co.kr/rss/specialsection/'),
-    ('Specialsection','http://www.hani.co.kr/rss/specialsection/'),
-    ('Hanionly','http://www.hani.co.kr/rss/hanionly/'),
-    ('Hkronly','http://www.hani.co.kr/rss/hkronly/'),
-    ('Multihani','http://www.hani.co.kr/rss/multihani/'),
-    ('Lead','http://www.hani.co.kr/rss/lead/'),
-    ('Newsrank','http://www.hani.co.kr/rss/newsrank/'),
            ]

recipes/hankyoreh21.recipe

@@ -10,16 +10,29 @@ class Hankyoreh21(BasicNewsRecipe):
     language = 'ko'
     description = u'The Hankyoreh21 Magazine articles'
     __author__  = 'Seongkyoun Yoo'
-    oldest_article = 20
+    oldest_article = 30
-    recursions = 1
+    max_articles_per_feed = 10
-    max_articles_per_feed = 120
     no_stylesheets        = True
     remove_javascript     = True
     keep_only_tags    = [
-						dict(name='font', attrs={'class':'t18bk'}),
+                        dict(name='header', attrs ={'class':['article_head']}),
-						dict(id=['fontSzArea'])
+                        dict(name='div', attrs ={'class':['article_body']}),
+                        ]
+    remove_tags   = [
+                    dict(name='div', attrs ={'class':['article_tools']}),
                     ]
 
     feeds = [
-	('Hani21','http://h21.hani.co.kr/rss/ '),
+        #('전체기사',  'http://h21.hani.co.kr/rss/'),
+        ('표지이야기','http://h21.hani.co.kr/rss/cover/'),
+        ('특집',      'http://h21.hani.co.kr/rss/special/'),
+        ('정치',      'http://h21.hani.co.kr/rss/politics/'),
+        ('경제',      'http://h21.hani.co.kr/rss/economy/'),
+        ('사회',      'http://h21.hani.co.kr/rss/society/'),
+        ('세계',      'http://h21.hani.co.kr/rss/world/'),
+        ('문화',      'http://h21.hani.co.kr/rss/culture/'),
            ]
+
+    def get_article_url(self, article):
+        org_url = BasicNewsRecipe.get_article_url(self, article)
+        return "http://h21.hani.co.kr"+org_url if org_url[0]=='/' else org_url

recipes/joongang.recipe

@@ -0,0 +1,35 @@
+# -*- coding: utf-8 -*-
+__license__   = 'GPL v3'
+__copyright__ = '2015, Hoje Lee <hojelei at gmail.com>'
+'''
+Profile to download Joongang Ilbo
+'''
+
+from calibre.web.feeds.news import BasicNewsRecipe
+
+class JoongangIlbo(BasicNewsRecipe):
+    language = 'ko'
+    title       = u'중앙일보'
+    description = u'중앙일보 신문 기사'
+    __author__	= 'Hoje Lee'
+    oldest_article = 7
+    max_articles_per_feed = 5
+    auto_cleanup   = True
+
+    feeds = [
+        #(u'전체기사', 'http://rss.joins.com/joins_news_list.xml'),
+        (u'주요기사', 'http://rss.joins.com/joins_homenews_list.xml'),
+        #(u'경제',     'http://rss.joins.com/joins_money_list.xml'),
+        #(u'사회',     'http://rss.joins.com/joins_life_list.xml'),
+        #(u'정치',     'http://rss.joins.com/joins_politics_list.xml'),
+        ### 많이 본 뉴스
+        (u'전체기사', 'http://rss.joins.com/sonagi/joins_sonagi_total_list.xml'),
+        (u'경제',     'http://rss.joins.com/sonagi/joins_sonagi_money_list.xml'),
+        (u'스포츠',   'http://rss.joins.com/sonagi/joins_sonagi_sports_list.xml'),
+        (u'연예',     'http://rss.joins.com/sonagi/joins_sonagi_star_list.xml'),
+        (u'사회',     'http://rss.joins.com/sonagi/joins_sonagi_life_list.xml'),
+        (u'정치',     'http://rss.joins.com/sonagi/joins_sonagi_politics_list.xml'),
+        (u'지구촌',   'http://rss.joins.com/sonagi/joins_sonagi_world_list.xml'),
+        (u'IT과학',   'http://rss.joins.com/sonagi/joins_sonagi_it_list.xml'),
+        (u'사설',     'http://rss.joins.com/sonagi/joins_sonagi_opinion_list.xml'),
+           ]

recipes/kyungyhang.recipe

@@ -1,3 +1,4 @@
+# -*- coding: utf-8 -*-
 __license__   = 'GPL v3'
 __copyright__ = '2011, Seongkyoun Yoo <seongkyoun.yoo at gmail.com>'
 '''
@@ -5,26 +6,34 @@ Profile to download The Kyungyhang
 '''
 
 from calibre.web.feeds.news import BasicNewsRecipe
+import re
 
 class Kyungyhang(BasicNewsRecipe):
     title          = u'Kyungyhang'
     language = 'ko'
     description = u'The Kyungyhang Shinmun articles'
     __author__	= 'Seongkyoun Yoo'
-    oldest_article = 20
+    oldest_article = 7
-    recursions = 2
+    max_articles_per_feed = 10
-    max_articles_per_feed = 20
     no_stylesheets        = True
     remove_javascript     = True
 
+    preprocess_regexps = [
+			(re.compile("<div class='ad_movFocus'.*</html>", re.DOTALL|re.IGNORECASE), lambda match: '</html>'),
+			]
+
     keep_only_tags    = [
                         dict(name='div', attrs ={'class':['article_title_wrap']}),
+                        dict(name='div', attrs ={'class':['viewHeader']}),
                         dict(name='span', attrs ={'class':['article_txt']})
                         ]
 
     remove_tags_after = dict(id={'sub_bottom'})
 
     remove_tags = [
+       dict(name='div', attrs={'class':['widget_top_dable']}),
+       dict(name='div', attrs={'class':['article_bottom_ad']}),
+       dict(name='div', attrs={'class':['article_date']}),
        dict(name='iframe'),
        dict(id={'TdHot'}),
        dict(name='div', attrs={'class':['btn_list','bline','linebottom','bestArticle']}),
@@ -33,5 +42,14 @@ class Kyungyhang(BasicNewsRecipe):
        ]
 
     feeds = [
-	('All News','http://www.khan.co.kr/rss/rssdata/total_news.xml'),
+	#(u'전체기사','http://www.khan.co.kr/rss/rssdata/total_news.xml'),
+	(u'정치',    'http://www.khan.co.kr/rss/rssdata/politic_news.xml'),
+	(u'경제',    'http://www.khan.co.kr/rss/rssdata/economy_news.xml'),
+	(u'사회',    'http://www.khan.co.kr/rss/rssdata/society_news.xml'),
+	(u'세계',    'http://www.khan.co.kr/rss/rssdata/kh_world.xml'),
+	(u'스포츠',  'http://www.khan.co.kr/rss/rssdata/kh_sports.xml'),
+	(u'문화',    'http://www.khan.co.kr/rss/rssdata/culture_news.xml'),
+	(u'연예',    'http://www.khan.co.kr/rss/rssdata/kh_entertainment.xml'),
+	(u'IT',      'http://www.khan.co.kr/rss/rssdata/it_news.xml'),
+	(u'오피니언','http://www.khan.co.kr/rss/rssdata/opinion_news.xml'),
            ]

recipes/maekyung.recipe

@@ -0,0 +1,34 @@
+# -*- coding: utf-8 -*-
+__license__   = 'GPL v3'
+__copyright__ = '2015, Hoje Lee <hojelei at gmail.com>'
+'''
+Profile to download Maeil Business
+'''
+
+from calibre.web.feeds.news import BasicNewsRecipe
+
+class MaeilBusiness(BasicNewsRecipe):
+    language = 'ko'
+    title       = u'매일경제'
+    description = u'매일경제 신문 기사'
+    __author__	= 'Hoje Lee'
+    oldest_article = 7
+    max_articles_per_feed = 10
+    auto_cleanup   = True
+
+    feeds = [
+        (u'헤드라인',  'http://file.mk.co.kr/news/rss/rss_30000001.xml'),
+        #(u'전체뉴스',  'http://file.mk.co.kr/news/rss/rss_40300001.xml'),
+        (u'경제',      'http://file.mk.co.kr/news/rss/rss_30100041.xml'),
+        (u'정치',      'http://file.mk.co.kr/news/rss/rss_30200030.xml'),
+        (u'사회',      'http://file.mk.co.kr/news/rss/rss_50400012.xml'),
+        (u'국제',      'http://file.mk.co.kr/news/rss/rss_30300018.xml'),
+        (u'기업ㆍ경영','http://file.mk.co.kr/news/rss/rss_50100032.xml'),
+        (u'증권',      'http://file.mk.co.kr/news/rss/rss_50200011.xml'),
+        (u'부동산',    'http://file.mk.co.kr/news/rss/rss_50300009.xml'),
+        (u'문화ㆍ연예','http://file.mk.co.kr/news/rss/rss_30000023.xml'),
+        (u'패션',      'http://file.mk.co.kr/news/rss/rss_72000001.xml'),
+        (u'스포츠',    'http://file.mk.co.kr/news/rss/rss_71000001.xml'),
+        (u'게임',      'http://file.mk.co.kr/news/rss/rss_50700001.xml'),
+        (u'오피니언',  'http://file.mk.co.kr/news/rss/rss_30500041.xml'),
+           ]

recipes/sisainlive.recipe

@@ -0,0 +1,48 @@
+# -*- coding: utf-8 -*-
+__license__   = 'GPL v3'
+__copyright__ = '2015, Hoje Lee <hojelei at gmail.com>'
+'''
+Profile to download SisaIN Live
+'''
+from calibre.web.feeds.news import BasicNewsRecipe
+
+class SisaINLive(BasicNewsRecipe):
+    language = 'ko'
+    title       = u'시사인 라이브'
+    description = u'시사인 라이브 기사'
+    __author__  = 'Hoje Lee'
+    oldest_article = 30
+    max_articles_per_feed = 10
+    auto_cleanup   = True
+    """
+    # manual cleanup
+    no_stylesheets    = True
+    remove_javascript = True
+
+    keep_only_tags    = [
+                        dict(name='div', attrs ={'class':['View_Title']}),
+                        dict(name='div', attrs ={'class':['View_Info']}),
+                        dict(name='div', attrs ={'class':['View_Time']}),
+                        dict(id='articleBody'),
+                        ]
+    remove_tags   = [
+                    dict(name='table', attrs ={'width':['320'], 'height':['265']}),
+                    ]
+    """
+
+    feeds = [
+        #(u'전체기사',  'http://www.sisainlive.com/rss.xml'),
+        (u'인기기사',  'http://www.sisainlive.com/rss/clickTop.xml'),
+        (u'커버스토리','http://www.sisainlive.com/rss/SRN121.xml'),
+        (u'특집',      'http://www.sisainlive.com/rss/SRN122.xml'),
+        (u'정치',      'http://www.sisainlive.com/rss/S1N15.xml'),
+        (u'경제',      'http://www.sisainlive.com/rss/S1N16.xml'),
+        (u'사회',      'http://www.sisainlive.com/rss/S1N17.xml'),
+        (u'문화',      'http://www.sisainlive.com/rss/S1N18.xml'),
+        (u'국제.한반도','http://www.sisainlive.com/rss/S1N4.xml'),
+        (u'실용.과학', 'http://www.sisainlive.com/rss/S1N6.xml'),
+        (u'휴먼&휴',   'http://www.sisainlive.com/rss/S1N19.xml'),
+        (u'인터뷰.오피니언','http://www.sisainlive.com/rss/S1N5.xml'),
+        (u'사진.만화', 'http://www.sisainlive.com/rss/S1N7.xml'),
+        (u'별책부록',  'http://www.sisainlive.com/rss/S1N14.xml'),
+           ]