case: when only get content

j3nnn1 · Aug 7, 2013 · 1666993 · 1666993
1 parent 1765807
commit 1666993
Showing 1 changed file with 4 additions and 3 deletions.
diff --git a/textmining/remove_html_tags/remove_html.rb b/textmining/remove_html_tags/remove_html.rb
@@ -2,7 +2,7 @@
 
 require 'rubygems'
 require 'sanitize'
-require 'CSV'
+require 'csv'
 
 terms = Hash.new
 terms = {'\xe1' => 'á', '\xe9' => 'é', '\xed' => 'í', '\xfa' => 'ú', '\xf3' => 'ó'} 
@@ -14,15 +14,16 @@ def removeaccent(word, terms)
    return word
 end
 
-file_clean = File.open("lanacion.com.ar.csv.data.clean", "w")
+file_clean = File.open("infobae_finanza.csv.data.clean", "w")
 
-CSV.foreach("lanacion.com.ar.csv.data", encoding: 'UTF-8' ) do |row|
+CSV.foreach("infobae_finanza.csv", encoding: 'UTF-8' ) do |row|
     #title
     title = removeaccent(Sanitize.clean(row[1]).force_encoding('UTF-8'), terms)
     #content
     content = removeaccent(Sanitize.clean(row[0]).force_encoding('UTF-8'), terms)
     #csv
     csv_string = [title, content].to_csv
+    #csv_string = [content].to_csv
     file_clean.write(csv_string)
 end