temp save image to add lines to complete table borders

app.py

@@ -161,7 +161,7 @@ if 'page_count' in st.session_state:
         st.session_state.num_pages_to_extract = st.slider('Number of pages to extract:', min_value=1, max_value=st.session_state.page_count-1, value=st.session_state.num_pages_to_extract , key='num_pages_to_extract_slider')
 
     if 'num_pages_to_extract2'not in st.session_state:
-        st.session_state.num_pages_to_extract2 = st.slider('Number of pages to extract:', min_value=1, max_value=st.session_state.page_count-1, value=36, key='num_pages_to_extract_slider2')
+        st.session_state.num_pages_to_extract2 = st.slider('Number of pages to extract:', min_value=1, max_value=st.session_state.page_count-1, value=39, key='num_pages_to_extract_slider2')
     else:
         st.session_state.num_pages_to_extract2 = st.slider('Number of pages to extract:', min_value=1, max_value=st.session_state.page_count-1, value=st.session_state.num_pages_to_extract2 , key='num_pages_to_extract_slider2')
 
@@ -191,8 +191,23 @@ if 'page_count' in st.session_state:
              read_pdf_progress_bar.progress(progress_percentage)
         read_pdf_progress_bar.progress(0)
         for index, image in enumerate(st.session_state.color_image_list):
+            st.write("actual page = " + str(index + st.session_state.num_pages_to_extract))
             image_np = np.array(image)
-            st.session_state.gray_image_np_list.append(cv2.cvtColor(np.array(image_np), cv2.COLOR_BGR2GRAY))
+            gray_image_np=cv2.cvtColor(np.array(image_np), cv2.COLOR_BGR2GRAY)
+            cv2.imwrite(f"gray_image_{index}.png", gray_image_np)
+            # st.image(Image.fromarray(gray_image_np))
+            if index + st.session_state.num_pages_to_extract == 34:
+                cv2.line(gray_image_np, (223, 414), (223, 1185), 0, 2)
+                cv2.line(gray_image_np, (1527, 414), (1527, 1185), 0, 2)
+            if index + st.session_state.num_pages_to_extract == 35:
+                cv2.line(gray_image_np, (176, 248), (176, 1760), 0, 2)
+                cv2.line(gray_image_np, (1551, 248), (1551, 1760), 0, 2)
+            st.image(Image.fromarray(gray_image_np))
+            st.session_state.gray_image_np_list.append(gray_image_np)
+
+
+            # cv2.line(st.session_state.gray_image_np_list[38], (226,227),(226,1444), 0,3)
+            # cv2.line(st.session_state.gray_image_np_list[38], (1601,227),(1601,1444), 0,2)
             progress_percentage = (index) / len(st.session_state.color_image_list)
             read_pdf_progress_bar.progress(progress_percentage)
         st.session_state.extracted_text = ""
@@ -201,7 +216,6 @@ if 'page_count' in st.session_state:
         manage_temp_to_be_zipped_directory(temp_table_dir)
         manage_temp_to_be_zipped_directory(temp_textbox_dir)
         for index, gray_pdf_image_np in enumerate(st.session_state.gray_image_np_list):
-
           try:
               figures_image_list,tables_image_list,textbox_image_list,text=utils.gray_pdf_image_np_to_text(index,gray_pdf_image_np, debug=True)
               if textbox_image_list:

utils.py

@@ -324,6 +324,8 @@ def gray_pdf_image_np_to_text(image_index,gray_pdf_image_np, debug=False):
     if debug:
         bgr_image = cv2.cvtColor(gray_pdf_image_np, cv2.COLOR_GRAY2BGR)
         color_tuple = (0, 255, 0)
+        # print("bounding_boxes_list")
+        # print(bounding_boxes_list)
         draw_colored_boxes_on_image_np(bgr_image, bounding_boxes_list, color_tuple)
         # st.image(Image.fromarray(bgr_image)) #to_be_displayed