fix repeated text rows

tsvtools/cli.py

@@ -125,11 +125,13 @@ def page2tsv(page_xml_file, tsv_out_file, purpose, image_url, ner_rest_endpoint,
                 for text_equiv in word.findall('./{%s}TextEquiv/{%s}Unicode' % (xmlns, xmlns)):
                     text = text_equiv.text
 
+                    points = []
+
                     for coords in word.findall('./{%s}Coords' % xmlns):
 
                         # transform OCR coordinates using `scale_factor` to derive
                         # correct coordinates for the web presentation image
-                        points = [int(scale_factor * float(pos))
+                        points += [int(scale_factor * float(pos))
                                   for p in coords.attrib['points'].split(' ') for pos in p.split(',')]
 
                     x_points, y_points = points[0::2], points[1::2]