Update files/functions.py

files/functions.py

@@ -346,7 +346,7 @@ def pdf_to_images(uploaded_pdf):
         except PdfReadError:
             path_to_file = pdf_blank
             filename = path_to_file.replace(examples_dir,"")
-            msg = "invalid PDF file."
+            msg = "Invalid PDF file."
             images = [Image.open(image_blank)]
         else:
             try:
@@ -463,8 +463,21 @@ def prepare_inference_features(example):
     # we want sorted lists from top to bottom of the image
     boxes, texts = sort_data_wo_labels(normalize_bboxes_par, texts)
 
+    bboxes_unique_list, texts_blocks = list(), list()
+    bbox_prev = [-100, -100, -100, -100]
+    for bbox, text in zip(boxes, texts):
+      if bbox != bbox_prev and bbox != cls_box:
+        bboxes_unique_list.append(bbox)
+        texts_block = text
+      else:
+        if bbox != cls_box:
+          texts_block += '\n' + text
+        else:
+          texts_blocks.append(texts_block)
+      bbox_prev = bbox
+
     count = 0
-    for box, text in zip(boxes, texts):
+    for box, text in zip(bboxes_unique_list, texts_blocks):
       tokens = tokenizer.tokenize(text)
       num_tokens = len(tokens) # get number of tokens
       tokens_list.extend(tokens)