First pass at retrieving bounding boxes

main.py

@@ -11,15 +11,25 @@ class docTR(AddOn):
 
     def main(self):
         """The main add-on functionality goes here."""
-        model = ocr_predictor(pretrained=True)
+        model = ocr_predictor('db_resnet50_rotation', 'crnn_vgg16_bn', pretrained=True, assume_straight_pages=False)
         for document in self.get_documents():
-            pdf_name = f"'{document.title}.pdf'"
+            pdf_name = f"'{document.id}.pdf'"
             with open(pdf_name, "wb") as pdf:
                 pdf.write(document.pdf)
                 doc = DocumentFile.from_pdf(pdf_name)
                 result = model(doc)
                 json_export = result.export()
-                print(json_export)
+                for page in json_export['pages']:
+                    page_idx = page['page_idx']
+                    print(f"Page {page_idx}:")
+                    for block in page['blocks']:
+                        for line in block['lines']:
+                            for word in line['words']:
+                                word_value = word['value']
+                                word_bounding_box = word['geometry']
+                                print(f"Word: {word_value}")
+                                print(f"Bounding Box: {word_bounding_box}")
+
 
 if __name__ == "__main__":
     docTR().main()