# 44
# azure 26 - 8core
# desktop 8 - 8core
# dell 10 - 8core
timeOut = 2
get_headline_num = 10 # 링크에서 담는 숫자
site_name = [
# "fnnews" # 1
# , "asiae_stock" # 2
# , "yakup" # 3
"newsis" # 4
, "ebn__" # 5
#
, 'news.mtn' # 6
# , "moneys" # 7
# , "hankyung" # 8
# ,"newspim" #9
# ,"sentv_stock" #10
#
# ,"news1"
# ,"sentv_industy"
# ,"etoday_stock"
# ,"mdtoday_stock"
# ,"news.mt" #15
#
# , "moneytoday_stock"
# ,"etoday_biz"
# ,"topstar"
# "viva100"
,"kmib_" #20
, "mk___1"
, "mk___2"
, "mk___3"
, "mk___4"
, "mk___5"
, "mk___6"
, "mk___7"
, "mk___8"
,"edaily"
,"asiae_IT"
,"asiae_social"
,"biospec" #25
# ,"fntimes"
# ,"econovill"
,"signalm"
,'einfomax'
,'asiae_politics' #30
,'dailyhankook'
,'ftoday'
,"thebell"
,"fetv_"
,"yna__" #35
,"chosun"
,"dailypharm"
,"etnews"
,"inthe"
,"sisaj" #40
,"aitimes"
,"seoulwire"
,"mk_dandok"
,"herald"
,"KBS"
,"paxnet"
,"sedaily"
,"news.tf"
,"newsway"
,"inews24"
,"newstown"
, "getnews"
, "consumuch"
# ,"medipana" #45
, "nspna"
,"naver" #46
] # 46
url_str = [
# "https://www.fnnews.com/newsflash/" # fnnews
# , "https://www.asiae.co.kr/list/stock-all" # asiae_stock
# , "http://www.yakup.com/news/index.html?cat=all" # yakup
"https://www.newsis.com/realnews/" # newsis
, "http://www.ebn.co.kr/news/lists?kind=&key" # ebn__ #5
#
, "https://news.mtn.co.kr/newscenter/index.mtn" # newsmt
# , "https://moneys.mt.co.kr/news/mwList.php?code=w0000&code2=w0100" # moneys
# , "https://www.hankyung.com/all-news/" # hankyung
# ,"http://www.newspim.com/news/lists/?category_cd=1" #newspim
# ,"http://www.sentv.co.kr/news/10" #sentv_stock #10
#
# ,"https://www.news1.kr/search_front/search.php" #news1
# ,"http://www.sentv.co.kr/news/40" #sentv_industy
# ,"http://www.etoday.co.kr/news/section/?MID=1200" #etoday
# ,"http://www.mdtoday.co.kr/mdtoday/index.html?cate=3" #mdtoday
# ,"https://news.mt.co.kr/newsList.html?pDepth1=politics&pDepth2=Ptotal" #mtn__ #15
#
#
# , "https://news.mt.co.kr/newsList.html?comd=7&pDepth=stock&pDepth1=sNews&pDepth2=Ftotal" # moneytoday_stock
# ,"http://www.etoday.co.kr/news/section/?MID=1300" #etoday_biz
# ,"http://www.topstarnews.net/news/articleList.html?sc_section_code=S1N3&view_type=sm" #topstar
# "http://www.viva100.com/main/list_news.php?ncid=N01_01" #viva100
,"http://news.kmib.co.kr/article/list.asp?sid1=all&sid2=&st=all" # kmib #20
,"https://www.mk.co.kr/news/economy/" #mk__
, "https://www.mk.co.kr/news/business/" # mk__
, "https://www.mk.co.kr/news/society/" # mk__
, "https://www.mk.co.kr/news/world/" # mk__
, "https://www.mk.co.kr/news/stock/" # mk__
, "https://www.mk.co.kr/news/politics/" # mk__
, "https://www.mk.co.kr/news/it/" # mk__
, "https://www.mk.co.kr/news/culture/" # mk__
# _
,"https://www.edaily.co.kr/news/realtimenews?tab=0" #edaily_SME
,"https://www.asiae.co.kr/list/industry-IT-all" #asiae_IT
,"https://www.asiae.co.kr/list/social-general" #asiae_social
,"http://www.biospectator.com/section/section_list.php?MID=11100" #biospec #25
# ,"https://www.fntimes.com/html/list.php?ct=g1101" # fntimes
# ,"http://www.econovill.com/news/articleList.html?sc_section_code=S1N8" #econovill
,"https://signalm.sedaily.com/Main/Content/SubMain?NClass=GX11" #signalm
, 'http://news.einfomax.co.kr/news/articleList.html?sc_section_code=S1N7' #einfomax
, 'https://www.asiae.co.kr/list/politics-all' #asiae_politics #30
, 'http://daily.hankooki.com/Article/ArticleList.php?section=industry&subsection=all&type=sum&page=1' #dailyhankook
,'http://www.ftoday.co.kr/news/articleList.html?view_type=sm' #ftdoay
,"http://www.thebell.co.kr/free/content/article.asp?svccode=00" #thebell
,"https://www.fetv.co.kr/news/section.html?sec_no=3" #fetv_
,"https://www.yna.co.kr/news" #yna__ #35
,"https://biz.chosun.com/svc/list_in/list.html" #chosun
,"http://www.dailypharm.com/Users/News/NewsList.html?dpsearch=" #dailypharm
,"https://news.etnews.com/" #etnews
,"http://inthenews.co.kr/category/finance-economy/economy/stock/?_page=2" #inthe
,"http://www.sisajournal.com/news/articleList.html?sc_section_code=S1N47&view_type=sm" #sisaj #40
,"http://www.aitimes.com/news/articleList.html?sc_section_code=S1N43&view_type=tm" #aitimes
,"http://www.seoulwire.com/news/articleList.html" #seoulwire
, "http://find.mk.co.kr/new/search.php?pageNum=1&cat=&cat1=&media_eco=&pageSize=20&sub=news&dispFlag=OFF&page=news&s_kwd=%B4%DC%B5%B6&s_page=total&go_page=&ord=1&ord1=1&ord2=0&s_keyword=%B4%DC%B5%B6&y1=1991&m1=01&d1=01&y2=2020&m2=04&d2=21&area=ttbd" #mk_dandok
, "http://biz.heraldcorp.com/list.php?ct=010000000000" #herald
, "http://news.kbs.co.kr/common/main.html?ref=pLogo" #KBS
,"https://paxnetnews.com/categories" #paxnet
,"https://m.sedaily.com/v/RankAll/" #sedaily
,"http://news.tf.co.kr/list/all" #news.tf
,"http://www.newsway.co.kr/news/lists" #newsway
,"http://www.inews24.com/list/inews" #inews24
,"http://www.newstown.co.kr/news/articleList.html?view_type=sm" #newstown
, "http://www.getnews.co.kr/list.php?ct=g0000" # getnews
, "http://www.consumuch.com/news/articleList.html?sc_section_code=S1N8" # consumuch
# ,"http://medipana.com/news/news_list_new.asp?MainKind=A&NewsKind=103&vCount=15&vKind=1" #medipana
,"http://www.nspna.com/news_issue/?issue_id=37" #nspna
,"https://news.naver.com/main/list.nhn?mode=LSD&mid=sec&sid1=001" # naver #45
]
tag_str = [
# "li > strong > a" # fnnews
# , "#container > div.content > div.cont_listarea > div.cont_list > div > h3 > a" # asiae_stock
# , "#content > div.listBoxType_2.tm_10.clear > ul > div > dl > dd a" # yakup
"#content > div.lst_p6.mgt21 > ul > li > div.area > strong > a" # newsis
, "#wrap > div.container > div.con_s_l > div > h4 > a" # ebn__ #5
#
, "#newsListArea > ul > li > h3 > a" # newsmtn
# , "#content > div > ul > li > a > div > strong" # moneys
# , "#container div.daily_article > div > ul > li > div > h3 > a" # hankyung
# ,"#wrap > div > div.section_wrap > div.listgroup > article > h2 > a > strong" # newspim
# ,"#content > div > section > div.article_list_cont > div.article_list.cont1 > div > a > div.txt_box > div.tit.ellipsis_tp1_ib" # sentv_stock #10
#
# ,"#content > div.search_detail > div.listType1 > ul > li > dl > dt > a" #news1
# ,"#content > div > section > div.article_list_cont > div.article_list.cont1 > div > a > div.txt_box > div.tit.ellipsis_tp1_ib" #sentv_industy
# ,"body div.mt45 > div > div > a > div.cluster_text_headline" # etoday_stock
# ,"#contents > p > a > b" # mdtoday
# ,"#content > ul > li > div > strong > a" # news.mt 머니투데이 #15
#
# , "#content > ul > li > div > strong > a" # moneytoday_stock
# ,"body div.mt45 > div > div > a > div.cluster_text_headline" #etoday_biz
# ,"#article-control-2 > div > section > article > div.article-list > section > div > div.list-titles > a" #topstar
# "#container > div.N_left > div > p.allnews_tit > a" #viva100
,"#sub > div > div.NwsCon > div.nws_list > div > dl > dt > a" # kmib #20
, "#container_left dt > a" #mk1
, "#container_left dt > a" #2
, "#container_left dt > a" #3
, "#container_left dt > a" #4
, "#container_left dt > a" #5
, "#container_left dt > a" #6
, "#container_left dt > a" #7
, "#container_left dt > a" #8
,"#sticker02 dl > dd > a > span" #edaily
,"#container > div.content > div.cont_listarea > div.cont_list > div > h3 > a" # asiae_IT
,"#container > div.content > div.cont_listarea > div.cont_list > div > h3 > a" #asiae_social
,"#container > div > div > div.contents > div.article_list > ul > li > strong > a" # biospec #25
# ,"body > div.con.mt30 div.lcon_dv a > span > span.w1 > span" #fntimes
# ,"body > table td > div > div > strong.article-titles > a" #econovill
,"#div_NewsLitAjax > div.m_newsCon > div > a > span.cateText > strong > span" #signalm
,"#user-container section > div > div.list-titles.table-cell > a > strong" #einfomax
,"#container > div.content > div.cont_listarea > div.cont_list > div > h3 > a" #asiae_politics #30
,"#sect790 > div.listgroup > div.gs-list-sum > div.dan > ul > li.tit.w600 > a" #dailyhankook
,"#section-list > ul > li > h4 > a" #ftdoay
,"#contents > div.contentSection > div > div.newsBox > div.newsList > div.listBox > ul > li > dl > a > dt" #thebell
,"#container > div.column.col73.pb00 > div:nth-child(1) > div > div.arl_053 > ul > li > a > h2" #fetv_
,"#container > div > div > div.section01 > section > div.list-type038 > ul > li > div > div.news-con > a > strong" #yna__ #35
,"#contents div.list_cont_wrap > div.list_content > dl > dt > a" #chosun
,"body > div.wrap > div.NewsCenterSide > div > ul > li > a > div.listHead" #dailypharm
,"#container > div > strong > a" #etnews
,"#main > div.category-grid-1 h4 > a" #inthe
,"#custom-article-area > div > a > div.contents > div > strong" #siaj #40
,"#user-container div.article-list > section > div > div > div > a > div.list-image > span" #aitimes
,"#section-list > ul > li > h4 > a"
,"body td:nth-child(1) > div > span.art_tit > a" #mk_dandok
,"body ul > li> a > div > div.list_t1.ellipsis" #herald
,"#content div.fl.col-box.col-recent > ul > li > a > span > em" #KBS
,"div.list > div > div > h1 > a" #paxnet
,"#DataAjax > li > a > div > h2" #sedaily
, "div.txtList > ul:nth-child(1) > li > a" #news.tf
,"#warp div.totalList > ul > li > div > a > strong" #newsway
,"body > main > article > ol > li > a" #inews24
, "#user-container div > div.list-titles > a > strong" #newstown
,"body ul > li > a > span.txt > span.w1.elip1" #getnews
,"#article-control-list div > div.list-titles.table-cell > a > strong" #consumuch
# ,"body div.totalNews > ul > li > a > span.tit" #medipana
,"#news_list > div > a > div > div.subject" #nspna
,"#main_content li > dl > dt:nth-child(2) > a" # naver #45
]
link_str = [
# "https://www.fnnews.com" # fnnews
# , "" # asiae_stock
# , "http://www.yakup.com" # yakup
"https://www.newsis.com/" # newsis
, "http://www.ebn.co.kr/" # ebn #5
#
, "https://news.mtn.co.kr/" # newsmtn o
# , "" # moneys x
# , "" # hankyung o
# , "" #newspim
# , "" #sentv_stock #10
#
# , "" #news1
# , "" #sentv_indu
# , "" #etoday_stock
# , "" #mdtoday
# , "" #mt #15
#
# , "" # moneytoday_stock
# , "" #etoday_biz
# , "http://www.topstarnews.net/" #topstar
# "http://www.viva100.com/main/" #viva100
,"http://news.kmib.co.kr/article/" #kmib #20
, "" # mk1
, "" # mk2
, "" # mk3
, "" # mk4
, "" # mk5
, "" # mk6
, "" # mk7
, "" # mk8
, "https://www.edaily.co.kr/news/realtimenews?tab=0" #edaily
, "" #asiae_IT
, "" #asiae_social
, "http://www.biospectator.com/" #biospec #25
# , "" #fntimes
# , "http://www.econovill.com/news/" #econovill
, "" # signalm
, "" #einfomax
, "" #asiae_politics #30
, "http://daily.hankooki.com/" #dailyhankook
, "http://www.ftoday.co.kr/" #ftoday
, "" #thebell
, "" #fetv
, "" #yna__ #35
, "" #chosun
, "" #dailypharm
, "" #etnews
, "" #inthe
, "" #sisaj #40
, "" #aitimes
,"http://www.seoulwire.com/" #seoulwire
,"" #mk_dandok
,"" #herald
,"" #KBS
, "https://paxnetnews.com/" #paxnet
,"" #sedaily
,"" #news.tf
, "" #newsway
, "http://www.inews24.com/" #inews24
, "" #newstown
, "" # getnews
, "" # consumuch
# , "" #medipana
,"http://www.nspna.com/news_issue/?issue_id=37" #nspna
, "" # naver #45
]
댓글