version1 (#5)

- fix table issue (77ca97ca240ccbde6bfc27dfd71ca451cedda871)

pdf_route.py

@@ -255,7 +255,7 @@ def create_markdown_file(result, output_file):
             table_cells = set()
             for _, element_type, element in elements:
                 if element_type == 'paragraph':
-                    if any(is_element_inside_table(element, get_table_max_polygon(table)) for table in result.tables):
+                    if any(is_element_inside_table(element, get_table_max_polygon(table)) for table in result.tables if table.bounding_regions[0].page_number == page.page_number):
                         continue
                     content = element.content.replace(":selected:", "").replace(":unselected:", "")
                     md_file.write(f"{content}\n\n")
@@ -321,7 +321,7 @@ def create_word_file(result, output_file):
         for _, element_type, element in elements:
             if element_type == 'paragraph':
                 # Skip lines that are part of a table
-                if any(is_element_inside_table(element, get_table_max_polygon(table)) for table in result.tables):
+                if any(is_element_inside_table(element, get_table_max_polygon(table)) for table in result.tables if table.bounding_regions[0].page_number == page.page_number):
                     continue
                 content = element.content.replace(":selected:", "").replace(":unselected:", "")
                 doc.add_paragraph(content)
@@ -349,12 +349,10 @@ def format_polygon(polygon):
 
 def get_table_max_polygon(table):
     # first coordination
-    first_cell = table.cells[0]
-    first_coordinate = first_cell.bounding_regions[0].polygon[0]
+    first_coordinate = table.bounding_regions[0].polygon[0]
 
     # last coordination
-    last_cell = table.cells[-1]
-    last_coordinate = last_cell.bounding_regions[0].polygon[2]
+    last_coordinate = table.bounding_regions[0].polygon[2]
 
     # return max polygon
     return [first_coordinate, last_coordinate]