Possibility of only plotting lines or text boxes

src/python/def_classes.py

@@ -583,8 +583,8 @@ class Document:
                 
     def plot_XMLcorrect(self, range_pages = range(1), suffix_xml = '_data', 
                         flag_plot = 1, flag_save_figs = 0, name_outxml = '02_extractedxml',
-                        name_outcorrxml = '04_correctedxml', flag_compute = 0):
-        
+                        name_outcorrxml = '04_correctedxml', flag_compute = 0, flag_lines_textl = 1):
+        # flag_lines_textl, if 1, plots lines and textboxes, if 2, only lines, if 3, only textboxes
         if 'name_outxml' not in self.__dict__.keys():
             self.name_outxml = name_outxml
         if 'name_outcorrxml' not in self.__dict__.keys():
@@ -603,7 +603,7 @@ class Document:
             if flag_error:
                 print(str(ind_page) + ': non existing page!')
             else:            
-                im_met = plot_tools.plot_correctedXML(imarray, XML_enrich, bbox_page)
+                im_met = plot_tools.plot_correctedXML(imarray, XML_enrich, bbox_page, flag_lines_textl)
 
                 self._plot_save(im_met, 'XML corrected', 'XMLcorrect', ind_page, self.path_file,
                        flag_plot, flag_save_figs)

src/python/plot_tools.py

@@ -226,7 +226,8 @@ def plot_horzvertlines(img, coord_horz, coord_vert_def):
     
     return img_lines
 
-def plot_correctedXML(img, XML_enrich, bbox_page):
+def plot_correctedXML(img, XML_enrich, bbox_page, flag_lines_textl = 1):
+    # flag_lines_textl, if 1, plots lines and textboxes, if 2, only lines, if 3, only textboxes
     # Essentially plotting the corrected textboxes and the lines from the
     # final xml
     img_xml = np.copy(img)
@@ -234,10 +235,12 @@ def plot_correctedXML(img, XML_enrich, bbox_page):
         if XML_enrich[0][ind_el].tag == 'textbox':
             if 'bbox' in XML_enrich[0][ind_el].attrib:
                 coord_textbox = np.array(XML_enrich[0][ind_el].attrib['bbox'].split(',')).astype(np.float64)
-                if XML_enrich[0][ind_el].attrib['type_textbox'] == 'line':
-                    img_xml = highlight_text(img_xml, coord_textbox, bbox_page, color_vec = 'blue', alpha = True, filled = False, thick_line = 6) 
-                if XML_enrich[0][ind_el].attrib['type_textbox'] == 'text':
-                    img_xml = highlight_text(img_xml, coord_textbox, bbox_page, color_vec = 'red', alpha = True, filled = False, thick_line = 6) 
+                if flag_lines_textl < 3:
+                    if XML_enrich[0][ind_el].attrib['type_textbox'] == 'line':
+                        img_xml = highlight_text(img_xml, coord_textbox, bbox_page, color_vec = 'blue', alpha = True, filled = False, thick_line = 6)
+                if (flag_lines_textl == 1) or (flag_lines_textl == 3):
+                    if XML_enrich[0][ind_el].attrib['type_textbox'] == 'text':
+                        img_xml = highlight_text(img_xml, coord_textbox, bbox_page, color_vec = 'red', alpha = True, filled = False, thick_line = 6) 
     return img_xml
 
 def plot_save_parallel(folder_pickles):