Add original title to story text

africat/aa_create_dataset.py

@@ -105,6 +105,9 @@ def parse_and_extract(input_dir, verbose):
           text = list()
           lang = ""
           try:
+            for t in doc.find('./publisher_headline'):
+              if t.text is not None:
+                text.append(t.text)
             for p in doc.find('./body'):
               if p.text is not None:
                 text.append(p.text)
@@ -184,6 +187,7 @@ def main():
   articles = parse_and_extract(args.input, args.verbose)
 
   data = scrub_data(articles, args.verbose)
+  #print(data)
 
   write_csv(data, args.output)