fix dw

Author: duty-machine

run_test.rb

@@ -23,7 +23,7 @@
 p fetch_article('https://www.dw.com/zh/%E4%B8%AD%E5%9B%BD%E7%BB%AD%E6%89%93%E5%8E%8B%E6%96%B0%E5%86%A0%E8%A8%80%E8%AE%BA-%E5%8C%97%E4%BA%AC%E4%B8%89%E5%BF%97%E6%84%BF%E8%80%85%E9%81%AD%E6%8D%95/a-53255271')
 
 # scmp
-p fetch_article('https://www.scmp.com/news/china/politics/article/3081569/chinese-activists-detained-after-sharing-censored-coronavirus')[:content]
+p fetch_article('https://www.scmp.com/news/china/politics/article/3081569/chinese-activists-detained-after-sharing-censored-coronavirus')
 
 # default article
 p fetch_article('http://www.rfi.fr/cn/%E4%B8%AD%E5%9B%BD/20200426-%E7%96%91%E5%9B%A0%E5%A4%87%E4%BB%BD%E9%81%AD%E5%88%A0%E9%99%A4%E6%96%B0%E5%86%A0%E7%96%AB%E6%83%85%E6%8A%A5%E9%81%93-%E5%8C%97%E4%BA%AC3%E5%90%8D90%E5%90%8E2%E4%BA%BA%E9%81%AD%E6%8B%98%E6%8A%BC1%E4%BA%BA%E5%A4%B1%E8%81%94')

websites/dw.rb

@@ -8,11 +8,17 @@
     document = Nokogiri::HTML(html)
     title = document.css('#bodyContent h1').first.content
     author = ''
-    content = document.css('#bodyContent').first
+    content = document.css('#bodyContent .longText').first
 
-    content.css('h1').first.remove
-    content.css('.artikel').first.remove
-    content.css('#sharing-bar').first.remove
+    content.css('.picBox').each do |el|
+      if el.css('a').first
+        img = el.css('img').first
+        el.css('a').first.replace(img)
+      else
+      end
+    end
+
+    content.css('script').each(&:remove)
 
     {
       title: title,

websites/scmp.rb

@@ -7,6 +7,7 @@
   process: -> (html) {
     document = Nokogiri::HTML(html)
     title = document.css('h1.info__headline.headline').first.content.strip
+    subheadline = document.css('.info__subHeadline').first.to_html
     author = document.css('.main-info__name a').first.content
     apollo = document.css('script').find{|x| x.content.start_with?('window.__APOLLO_STATE__=') }.content
     data = apollo.split('\"more-on-this\"')
@@ -49,6 +50,7 @@
       end
     }
     content = recur.(arr)
+    content = "#{subheadline}#{content}"
 
     {
       title: title,