textbox zip file download implemented

app.py

@@ -12,6 +12,7 @@ import zipfile
 from io import BytesIO
 temp_figure_dir="pdf_figures/"
 temp_table_dir="pdf_tables/"
+temp_textbox_dir="pdf_textbox/"
 import time
 # get https://github.com/oschwartz10612/poppler-windows/releases/tag/v22.01.0-0
 # poppler-utils:
@@ -170,6 +171,7 @@ if 'page_count' in st.session_state:
         st.session_state.gray_image_np_list = []
         st.session_state.pdf_figures_image_list=[]
         st.session_state.pdf_tables_image_list = []
+        st.session_state.pdf_textbox_image_list=[]
         pdf_tables_image_list=[]
         st.session_state.pdf_text_list=[]
 
@@ -188,22 +190,33 @@ if 'page_count' in st.session_state:
 
         manage_temp_to_be_zipped_directory(temp_figure_dir)
         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):
-          print("index="+str(index))
 
-          figures_image_list,tables_image_list,text=utils.gray_pdf_image_np_to_text(index,gray_pdf_image_np, debug=True)
+
+          figures_image_list,tables_image_list,textbox_image_list,text=utils.gray_pdf_image_np_to_text(index,gray_pdf_image_np, debug=True)
+          print("index="+str(index)+" txt book " + str(len(textbox_image_list)))
           st.session_state.pdf_figures_image_list.append(figures_image_list)
           st.session_state.pdf_tables_image_list.append(tables_image_list)
+          st.session_state.pdf_textbox_image_list.append(textbox_image_list)
           if st.session_state.pdf_figures_image_list[index]:
               for pdf_figure_text_image in st.session_state.pdf_figures_image_list[index]:
-                  raw_image_file_name = f"page_{index}_{pdf_figure_text_image[0]}.png"
+                  raw_image_file_name = f"page_{index+1}_{pdf_figure_text_image[0]}.png"
                   cleaned_image_file_name = clean_filename(raw_image_file_name)
                   Image.fromarray(pdf_figure_text_image[1]).save(temp_figure_dir+cleaned_image_file_name)
           if st.session_state.pdf_tables_image_list:
               for pdf_table_text_image in st.session_state.pdf_tables_image_list[index]:
-                  raw_image_file_name = f"page_{index}_{pdf_table_text_image[0]}.png"
+                  raw_image_file_name = f"page_{index+1}_{pdf_table_text_image[0]}.png"
                   cleaned_image_file_name = clean_filename(raw_image_file_name)
                   Image.fromarray(pdf_table_text_image[1]).save(temp_table_dir + cleaned_image_file_name)
+          if st.session_state.pdf_textbox_image_list:
+              textbox_index = 1
+              for pdf_textbox_image in st.session_state.pdf_textbox_image_list[index]:
+                  raw_image_file_name = f"page_{index+1}_textbox_{textbox_index}.png"
+                  cleaned_image_file_name = clean_filename(raw_image_file_name)
+                  Image.fromarray(pdf_textbox_image).save(temp_textbox_dir + cleaned_image_file_name)
+                  textbox_index = textbox_index + 1
+
 
           st.session_state.pdf_text_list.append(text)
           st.session_state.extracted_text=st.session_state.extracted_text+f"<Page {index+1} start>\n" + text + f"\n<Page {index+1} end>\n>"
@@ -213,6 +226,7 @@ if 'page_count' in st.session_state:
           read_pdf_progress_bar.progress(progress_percentage)
         st.session_state.figure_zip_bytes=zip_directory(temp_figure_dir)
         st.session_state.table_zip_bytes = zip_directory(temp_table_dir)
+        st.session_state.textbox_zip_bytes = zip_directory(temp_textbox_dir)
         #add_animation_to_image()
         #st.session_state['video_generated'] = True
         st.rerun()
@@ -226,6 +240,7 @@ if 'page_count' in st.session_state:
 
         download_figure_zip_file_name = uploaded_locked_pdf_file.name.replace(".pdf", "_figures.zip")
         download_table_zip_file_name = uploaded_locked_pdf_file.name.replace(".pdf", "_tables.zip")
+        download_textbox_zip_file_name = uploaded_locked_pdf_file.name.replace(".pdf", "_textbox.zip")
         st.download_button(
             label="Download Figures ZIP",
             data=st.session_state.figure_zip_bytes,
@@ -239,6 +254,12 @@ if 'page_count' in st.session_state:
             file_name=download_table_zip_file_name,
             mime="application/zip"
         )
+        st.download_button(
+            label="Download Textbox ZIP",
+            data=st.session_state.textbox_zip_bytes,
+            file_name=download_textbox_zip_file_name,
+            mime="application/zip"
+        )
 
         #  st.image(Image.fromarray(bgr_image))
         # for index,pdf_text in enumerate(st.session_state.pdf_text_list):
@@ -257,7 +278,12 @@ if 'page_count' in st.session_state:
                 for pdf_table_text_image in st.session_state.pdf_tables_image_list[index]:
                     st.write(pdf_table_text_image[0])
                     st.image(Image.fromarray(pdf_table_text_image[1]))
-
+            if not st.session_state.pdf_textbox_image_list[index]:
+                st.write("no textbox")
+            else:
+                for text_box_index,pdf_textbox_image in enumerate(st.session_state.pdf_textbox_image_list[index]):
+                    st.write("text box "+str(text_box_index))
+                    st.image(Image.fromarray(pdf_textbox_image))
 
 
 

utils.py

@@ -87,8 +87,8 @@ def draw_colored_boxes_on_image_np(image, boxes_list,color_tuple):
 def is_filled_rectangle(image, rect, background_threshold=10, variance_threshold=0.1):
 
     x, y, w, h = rect
-    roi = image[y+1:y+h-1, x+1:x+w-1]
-
+    # roi = image[y+1:y+h-1, x+1:x+w-1]
+    roi = image[y + 10:y + h - 10, x + 10:x + w - 10]
     return np.all(roi == 0)
 def get_below_box(image_np, x, y,width,step=15):
   #print("x,y,width="+str(x)+","+str(y)+","+str(width))
@@ -140,7 +140,7 @@ def is_note_rectangle(image_np, rect):
   text = pytesseract.image_to_string(roi_converted)
   text = text.strip()
   note_str="note"
-  print("is note text box="+str(text.lower().startswith(note_str.lower())))
+  # print("is note text box="+str(text.lower().startswith(note_str.lower())))
   return text.lower().startswith(note_str.lower())
 def extract_bounding_boxes_from_image_np(image_np, bounding_boxes_list, above_check_offset, above_caption_offset, color_tuple):
 
@@ -183,15 +183,17 @@ def extract_bounding_boxes_from_image_np(image_np, bounding_boxes_list, above_ch
         figure_str ="Figure"
         table_str ="Table"
         if text.lower().startswith(figure_str.lower()):
-          print(text)
+          # print(text)
           figures_image_list.append((text,rect_content))
 
         elif  text.lower().startswith(table_str.lower()):
-          print(text)
+          # print(text)
           tables_image_list.append((text,rect_content))
         else:
+          print("*** text box have above content")
           above_rect_content_list.append((text, rect_content))
-        rect_content_list.append(rect_content)
+          rect_content_list.append(rect_content)
+
 
         cv2.rectangle(image_np_copy, (x, above_box_y), (x+w, y), color_tuple, cv2.FILLED)
       # above_rect_content = image_np[y-above_check_offset:y, x:x+w]
@@ -284,23 +286,23 @@ def extract_two_columns_text(image_index,image_np,debug):
       right_column_array_bgr_image = cv2.cvtColor(right_column_array, cv2.COLOR_GRAY2BGR)
       draw_edges(right_column_array_bgr_image)
       # imageio.imwrite("/content/gdrive/MyDrive/Avatar/demo_pdf_ingestion_steps/page_"+formatted_index_string + "step8_right_column.png", right_column_img)
-      if debug:
-          print("left column image start")
-          # display(left_column_img)
-          # st.image(Image.fromarray(left_column_array_bgr_image))  # to_be_displayed
-          print("left column image end")
-          print("right column image start")
-          # display(right_column_img)
-          # st.image(Image.fromarray(right_column_array_bgr_image))  # to_be_displayed
-          print("right column image end")
+      # if debug:
+          # print("left column image start")
+          # # display(left_column_img)
+          # # st.image(Image.fromarray(left_column_array_bgr_image))  # to_be_displayed
+          # print("left column image end")
+          # print("right column image start")
+          # # display(right_column_img)
+          # # st.image(Image.fromarray(right_column_array_bgr_image))  # to_be_displayed
+          # print("right column image end")
       left_text = pytesseract.image_to_string(left_column_img)
       # with open("/content/gdrive/MyDrive/Avatar/demo_pdf_ingestion_steps/page_"+formatted_index_string + "step9_left_column_text.txt", 'w') as file:
       #   file.write(left_text)
-      print("Extracted Text:\n", left_text)
+      # print("Extracted Text:\n", left_text)
       right_text = pytesseract.image_to_string(right_column_img)
       # with open("/content/gdrive/MyDrive/Avatar/demo_pdf_ingestion_steps/page_"+formatted_index_string + "step9_right_column_text.txt", 'w') as file:
       #   file.write(right_text)
-      print("Extracted Text:\n", right_text)
+      # print("Extracted Text:\n", right_text)
       return left_text + right_text
   else:
       return "error"
@@ -331,20 +333,20 @@ def gray_pdf_image_np_to_text(image_index,gray_pdf_image_np, debug=False):
     if debug:
         debug_text_box_index = 0
         for text_box, above_text_box in zip(text_box_list, above_test_box_list):
-            print("text box start")
-            if above_text_box is not None:
-                print(above_text_box[0])#to_be_displayed
+            # print("text box start")
+            # if above_text_box is not None:
+                # print(above_text_box[0])#to_be_displayed
                 # st.write(above_text_box[0])#to_be_displayed
                 # st.image(Image.fromarray(above_text_box[1]))#to_be_displayed
                 # st.write(text)
             # st.image(Image.fromarray(text_box))#to_be_displayed
             debug_text_box_index = debug_text_box_index + 1
-        for figure in figures_image_list:
-            print(figure[0])
+        # for figure in figures_image_list:
+            # print(figure[0])
             # st.write(figure[0])#to_be_displayed
             # st.image(Image.fromarray(figure[1]))#to_be_displayed
-        for table in tables_image_list:
-            print(table[0])
+        # for table in tables_image_list:
+            # print(table[0])
             # st.write(table[0])#to_be_displayed
             # st.image(Image.fromarray(table[1]))#to_be_displayed
         # st.image(Image.fromarray(cropped_image))#to_be_displayed
@@ -352,21 +354,21 @@ def gray_pdf_image_np_to_text(image_index,gray_pdf_image_np, debug=False):
     if found_hor_lines_list is not None:
         bgr_image = cv2.cvtColor(gray_pdf_image_np, cv2.COLOR_GRAY2BGR)
         draw_colored_lines_on_image_np(bgr_image, found_hor_lines_list, (0, 255, 0))
-        print("detected Lines start")
-        # st.image(Image.fromarray(bgr_image)) #to_be_displayed
-
-        print("detected lines end")
+        # print("detected Lines start")
+        # # st.image(Image.fromarray(bgr_image)) #to_be_displayed
+        #
+        # print("detected lines end")
         page_segment_np_list = segment_image_np(cropped_image, found_hor_lines_list)
         if debug:
             debug_page_segment_index = 0
             for element in page_segment_np_list:
-                print("element start")
+                # print("element start")
                 bgr_image = cv2.cvtColor(element, cv2.COLOR_GRAY2BGR)
                 draw_edges(bgr_image)
                 # st.image(Image.fromarray(bgr_image))#to_be_displayed
 
                 debug_page_segment_index = debug_page_segment_index + 1
-                print("element end")
+                # print("element end")
         min_height_filtered_page_segment_np_list = filter_segments_by_min_height(page_segment_np_list, 50)
         max_height_image = max(min_height_filtered_page_segment_np_list, key=lambda image: image.shape[0])
     else:
@@ -375,9 +377,9 @@ def gray_pdf_image_np_to_text(image_index,gray_pdf_image_np, debug=False):
     # print("max height image start")
     # st.image(Image.fromarray(max_height_image))#to_be_displayed
     # print("max height image end")
-    print("start text extraction")
+    # print("start text extraction")
     text=extract_two_columns_text(image_index,max_height_image,debug)
-    print("gray_pdf_image_np_to_text extracted text",text)
+    # print("gray_pdf_image_np_to_text extracted text",text)
     if text == "error":
         print("not two columns")
         max_height_image_converted = Image.fromarray(cv2.cvtColor(max_height_image, cv2.COLOR_BGR2RGB))
@@ -420,6 +422,6 @@ def gray_pdf_image_np_to_text(image_index,gray_pdf_image_np, debug=False):
           if text == "error":
             return("error")
           else:
-            return figures_image_list,tables_image_list,text
+            return figures_image_list,tables_image_list,text_box_list,text
     else:
-        return figures_image_list,tables_image_list,text
+        return figures_image_list,tables_image_list,text_box_list,text