Comprehend now uses custom spacy recognisers on top of defaults. Added zoom functionality to annotator. Fixed some pdf mediabox issues and redacted image output issues.

app.py

@@ -13,9 +13,10 @@ from tools.helper_functions import ensure_output_folder_exists, add_folder_to_pa
 from tools.aws_functions import upload_file_to_s3, download_file_from_s3, RUN_AWS_FUNCTIONS, bucket_name
 from tools.file_redaction import choose_and_run_redactor
 from tools.file_conversion import prepare_image_or_pdf, get_input_file_names
-from tools.redaction_review import apply_redactions, crop, get_boxes_json, modify_existing_page_redactions, decrease_page, increase_page, update_annotator
+from tools.redaction_review import apply_redactions, crop, get_boxes_json, modify_existing_page_redactions, decrease_page, increase_page, update_annotator, update_zoom
 from tools.data_anonymise import anonymise_data_files
 from tools.auth import authenticate_user
+from tools.load_spacy_model_custom_recognisers import custom_entities
 
 
 today_rev = datetime.now().strftime("%Y%m%d")
@@ -29,6 +30,10 @@ chosen_comprehend_entities = ['BANK_ACCOUNT_NUMBER','BANK_ROUTING','CREDIT_DEBIT
 
 full_comprehend_entity_list = ['BANK_ACCOUNT_NUMBER','BANK_ROUTING','CREDIT_DEBIT_NUMBER','CREDIT_DEBIT_CVV','CREDIT_DEBIT_EXPIRY','PIN','EMAIL','ADDRESS','NAME','PHONE','SSN','DATE_TIME','PASSPORT_NUMBER','DRIVER_ID','URL','AGE','USERNAME','PASSWORD','AWS_ACCESS_KEY','AWS_SECRET_KEY','IP_ADDRESS','MAC_ADDRESS','ALL','LICENSE_PLATE','VEHICLE_IDENTIFICATION_NUMBER','UK_NATIONAL_INSURANCE_NUMBER','CA_SOCIAL_INSURANCE_NUMBER','US_INDIVIDUAL_TAX_IDENTIFICATION_NUMBER','UK_UNIQUE_TAXPAYER_REFERENCE_NUMBER','IN_PERMANENT_ACCOUNT_NUMBER','IN_NREGA','INTERNATIONAL_BANK_ACCOUNT_NUMBER','SWIFT_CODE','UK_NATIONAL_HEALTH_SERVICE_NUMBER','CA_HEALTH_NUMBER','IN_AADHAAR','IN_VOTER_NUMBER']
 
+# Add custom spacy recognisers to the Comprehend list, so that local Spacy model can be used to pick up e.g. titles, streetnames, UK postcodes that are sometimes missed by comprehend
+chosen_comprehend_entities.extend(custom_entities)
+full_comprehend_entity_list.extend(custom_entities)
+
 chosen_redact_entities = ["TITLES", "PERSON", "PHONE_NUMBER", "EMAIL_ADDRESS", "STREETNAME", "UKPOSTCODE"]
 
 full_entity_list = ["TITLES", "PERSON", "PHONE_NUMBER", "EMAIL_ADDRESS", "STREETNAME", "UKPOSTCODE", 'CREDIT_CARD', 'CRYPTO', 'DATE_TIME', 'IBAN_CODE', 'IP_ADDRESS', 'NRP', 'LOCATION', 'MEDICAL_LICENSE', 'URL', 'UK_NHS']
@@ -117,6 +122,12 @@ with app:
     default_allow_list_output_folder_location = gr.Textbox(label = "Output default allow list location", value=default_allow_list_loc, visible=False)
 
 
+    ## Annotator zoom value
+    annotator_zoom_number = gr.Number(label = "Current annotator zoom level", value=100, precision=0, visible=False)
+    zoom_true_bool = gr.State(True)
+    zoom_false_bool = gr.State(False)
+
+
     ###
     # UI DESIGN
     ###
@@ -164,6 +175,9 @@ with app:
             annotate_current_page = gr.Number(value=1, label="Page (press enter to change)", precision=0, scale = 2)
             annotate_max_pages = gr.Number(value=1, label="Total pages", precision=0, interactive=False, scale = 1)
             annotation_next_page_button = gr.Button("Next page", scale = 3)
+        with gr.Row():
+            annotate_zoom_in = gr.Button("Zoom in")
+            annotate_zoom_out = gr.Button("Zoom out")
 
         annotation_button_apply = gr.Button("Apply revised redactions", variant="primary")
 
@@ -238,9 +252,9 @@ with app:
                     in_allow_list_text = gr.Textbox(label="Custom allow list load status")
 
             with gr.Accordion("Add or remove entity types to redact", open = False):
-                in_redact_entities = gr.Dropdown(value=chosen_redact_entities, choices=full_entity_list, multiselect=True, label="Entities to redact - local PII identification model (click close to down arrow for full list)")
-                
                 in_redact_comprehend_entities = gr.Dropdown(value=chosen_comprehend_entities, choices=full_comprehend_entity_list, multiselect=True, label="Entities to redact - AWS Comprehend PII identification model (click close to down arrow for full list)")
+
+                in_redact_entities = gr.Dropdown(value=chosen_redact_entities, choices=full_entity_list, multiselect=True, label="Entities to redact - local PII identification model (click close to down arrow for full list)")
             
             handwrite_signature_checkbox = gr.CheckboxGroup(label="AWS Textract settings", choices=["Redact all identified handwriting", "Redact all identified signatures"], value=["Redact all identified handwriting", "Redact all identified signatures"])
             #with gr.Row():
@@ -260,18 +274,19 @@ with app:
     ###
     in_doc_files.upload(fn=get_input_file_names, inputs=[in_doc_files], outputs=[doc_file_name_textbox, doc_file_name_with_extension_textbox])
 
-    document_redact_btn.click(fn = reset_state_vars, outputs=[pdf_doc_state, all_image_annotations_state, all_line_level_ocr_results_df_state, all_decision_process_table_state, comprehend_query_number, textract_metadata_textbox]).\
+    document_redact_btn.click(fn = reset_state_vars, outputs=[pdf_doc_state, all_image_annotations_state, all_line_level_ocr_results_df_state, all_decision_process_table_state, comprehend_query_number, textract_metadata_textbox, annotator]).\
     then(fn = prepare_image_or_pdf, inputs=[in_doc_files, in_redaction_method, in_allow_list, latest_file_completed_text, output_summary, first_loop_state, annotate_max_pages, current_loop_page_number], outputs=[output_summary, prepared_pdf_state, images_pdf_state, annotate_max_pages, annotate_max_pages_bottom, pdf_doc_state], api_name="prepare_doc").\
     then(fn = choose_and_run_redactor, inputs=[in_doc_files, prepared_pdf_state, images_pdf_state, in_redact_language, in_redact_entities, in_redact_comprehend_entities, in_redaction_method, in_allow_list_state, latest_file_completed_text, output_summary, output_file_list_state, log_files_output_list_state, first_loop_state, page_min, page_max, estimated_time_taken_number, handwrite_signature_checkbox, textract_metadata_textbox, all_image_annotations_state, all_line_level_ocr_results_df_state, all_decision_process_table_state, pdf_doc_state, current_loop_page_number, page_break_return, pii_identification_method_drop, comprehend_query_number],
-                    outputs=[output_summary, output_file, output_file_list_state, latest_file_completed_text, log_files_output, log_files_output_list_state, estimated_time_taken_number, textract_metadata_textbox, pdf_doc_state, all_image_annotations_state, current_loop_page_number, page_break_return, all_line_level_ocr_results_df_state, all_decision_process_table_state, comprehend_query_number], api_name="redact_doc")#.\
-                    #then(fn=update_annotator, inputs=[all_image_annotations_state, page_min], outputs=[annotator, annotate_current_page])
+                    outputs=[output_summary, output_file, output_file_list_state, latest_file_completed_text, log_files_output, log_files_output_list_state, estimated_time_taken_number, textract_metadata_textbox, pdf_doc_state, all_image_annotations_state, current_loop_page_number, page_break_return, all_line_level_ocr_results_df_state, all_decision_process_table_state, comprehend_query_number], api_name="redact_doc").\
+                    then(fn=update_annotator, inputs=[all_image_annotations_state, page_min, annotator_zoom_number], outputs=[annotator, annotate_current_page, annotate_current_page_bottom])
     
     # If the app has completed a batch of pages, it will run this until the end of all pages in the document
     current_loop_page_number.change(fn = choose_and_run_redactor, inputs=[in_doc_files, prepared_pdf_state, images_pdf_state, in_redact_language, in_redact_entities, in_redact_comprehend_entities, in_redaction_method, in_allow_list_state, latest_file_completed_text, output_summary, output_file_list_state, log_files_output_list_state, second_loop_state, page_min, page_max, estimated_time_taken_number, handwrite_signature_checkbox, textract_metadata_textbox, all_image_annotations_state, all_line_level_ocr_results_df_state, all_decision_process_table_state, pdf_doc_state, current_loop_page_number, page_break_return, pii_identification_method_drop, comprehend_query_number],
-                    outputs=[output_summary, output_file, output_file_list_state, latest_file_completed_text, log_files_output, log_files_output_list_state, estimated_time_taken_number, textract_metadata_textbox, pdf_doc_state, all_image_annotations_state, current_loop_page_number, page_break_return, all_line_level_ocr_results_df_state, all_decision_process_table_state, comprehend_query_number])
+                    outputs=[output_summary, output_file, output_file_list_state, latest_file_completed_text, log_files_output, log_files_output_list_state, estimated_time_taken_number, textract_metadata_textbox, pdf_doc_state, all_image_annotations_state, current_loop_page_number, page_break_return, all_line_level_ocr_results_df_state, all_decision_process_table_state, comprehend_query_number]).\
+                    then(fn=update_annotator, inputs=[all_image_annotations_state, page_min, annotator_zoom_number], outputs=[annotator, annotate_current_page, annotate_current_page_bottom])
     
     # If a file has been completed, the function will continue onto the next document
-    latest_file_completed_text.change(fn=update_annotator, inputs=[all_image_annotations_state, page_min], outputs=[annotator, annotate_current_page, annotate_current_page_bottom]).\
+    latest_file_completed_text.change(fn=update_annotator, inputs=[all_image_annotations_state, page_min, annotator_zoom_number], outputs=[annotator, annotate_current_page, annotate_current_page_bottom]).\
                     then(fn=reveal_feedback_buttons, outputs=[pdf_feedback_radio, pdf_further_details_text, pdf_submit_feedback_btn, pdf_feedback_title])
         # latest_file_completed_text.change(fn = prepare_image_or_pdf, inputs=[in_doc_files, in_redaction_method, in_allow_list, latest_file_completed_text, output_summary, second_loop_state, annotate_max_pages, current_loop_page_number], outputs=[output_summary, prepared_pdf_state, images_pdf_state, annotate_max_pages, annotate_max_pages_bottom, pdf_doc_state]).\
     # then(fn = choose_and_run_redactor, inputs=[in_doc_files, prepared_pdf_state, images_pdf_state, in_redact_language, in_redact_entities, in_redaction_method, in_allow_list_state, latest_file_completed_text, output_summary, output_file_list_state, log_files_output_list_state, second_loop_state, page_min, page_max, estimated_time_taken_number, handwrite_signature_checkbox, textract_metadata_textbox, all_image_annotations_state, all_line_level_ocr_results_df_state, all_decision_process_table_state, pdf_doc_state, current_loop_page_number, page_break_return],
@@ -284,12 +299,20 @@ with app:
     # Page controls at top
     annotate_current_page.submit(
         modify_existing_page_redactions, inputs = [annotator, annotate_current_page, annotate_previous_page, all_image_annotations_state], outputs = [all_image_annotations_state, annotate_previous_page, annotate_current_page_bottom]).\
-        then(update_annotator, inputs=[all_image_annotations_state, annotate_current_page], outputs = [annotator, annotate_current_page, annotate_current_page_bottom])
+        then(update_annotator, inputs=[all_image_annotations_state, annotate_current_page, annotator_zoom_number], outputs = [annotator, annotate_current_page, annotate_current_page_bottom])
     
     annotation_last_page_button.click(fn=decrease_page, inputs=[annotate_current_page], outputs=[annotate_current_page, annotate_current_page_bottom]).\
-        then(update_annotator, inputs=[all_image_annotations_state, annotate_current_page], outputs = [annotator, annotate_current_page, annotate_current_page_bottom])
+        then(update_annotator, inputs=[all_image_annotations_state, annotate_current_page, annotator_zoom_number], outputs = [annotator, annotate_current_page, annotate_current_page_bottom])
     annotation_next_page_button.click(fn=increase_page, inputs=[annotate_current_page, all_image_annotations_state], outputs=[annotate_current_page, annotate_current_page_bottom]).\
-        then(update_annotator, inputs=[all_image_annotations_state, annotate_current_page], outputs = [annotator, annotate_current_page, annotate_current_page_bottom])
+        then(update_annotator, inputs=[all_image_annotations_state, annotate_current_page, annotator_zoom_number], outputs = [annotator, annotate_current_page, annotate_current_page_bottom])
+    
+    # Zoom in and out on annotator
+    annotate_zoom_in.click(modify_existing_page_redactions, inputs = [annotator, annotate_current_page, annotate_previous_page, all_image_annotations_state], outputs = [all_image_annotations_state, annotate_previous_page, annotate_current_page_bottom]).\
+        then(update_zoom, inputs=[annotator_zoom_number, zoom_true_bool], outputs=[annotator_zoom_number]).\
+        then(update_annotator, inputs=[all_image_annotations_state, annotate_current_page, annotator_zoom_number], outputs = [annotator, annotate_current_page, annotate_current_page_bottom])
+    annotate_zoom_out.click(modify_existing_page_redactions, inputs = [annotator, annotate_current_page, annotate_previous_page, all_image_annotations_state], outputs = [all_image_annotations_state, annotate_previous_page, annotate_current_page_bottom]).\
+        then(update_zoom, inputs=[annotator_zoom_number, zoom_false_bool], outputs=[annotator_zoom_number]).\
+        then(update_annotator, inputs=[all_image_annotations_state, annotate_current_page, annotator_zoom_number], outputs = [annotator, annotate_current_page, annotate_current_page_bottom])
 
     #annotation_button_get.click(get_boxes_json, annotator, json_boxes)
     annotation_button_apply.click(apply_redactions, inputs=[annotator, in_doc_files, pdf_doc_state, all_image_annotations_state, annotate_current_page], outputs=[pdf_doc_state, all_image_annotations_state, output_review_files], scroll_to_output=True)
@@ -297,12 +320,12 @@ with app:
     # Page controls at bottom
     annotate_current_page_bottom.submit(
         modify_existing_page_redactions, inputs = [annotator, annotate_current_page_bottom, annotate_previous_page, all_image_annotations_state], outputs = [all_image_annotations_state, annotate_previous_page, annotate_current_page]).\
-        then(update_annotator, inputs=[all_image_annotations_state, annotate_current_page], outputs = [annotator, annotate_current_page, annotate_current_page_bottom])
+        then(update_annotator, inputs=[all_image_annotations_state, annotate_current_page, annotator_zoom_number], outputs = [annotator, annotate_current_page, annotate_current_page_bottom])
 
     annotation_last_page_button_bottom.click(fn=decrease_page, inputs=[annotate_current_page], outputs=[annotate_current_page, annotate_current_page_bottom]).\
-        then(update_annotator, inputs=[all_image_annotations_state, annotate_current_page], outputs = [annotator, annotate_current_page, annotate_current_page_bottom])
+        then(update_annotator, inputs=[all_image_annotations_state, annotate_current_page, annotator_zoom_number], outputs = [annotator, annotate_current_page, annotate_current_page_bottom])
     annotation_next_page_button_bottom.click(fn=increase_page, inputs=[annotate_current_page, all_image_annotations_state], outputs=[annotate_current_page, annotate_current_page_bottom]).\
-        then(update_annotator, inputs=[all_image_annotations_state, annotate_current_page], outputs = [annotator, annotate_current_page, annotate_current_page_bottom])
+        then(update_annotator, inputs=[all_image_annotations_state, annotate_current_page, annotator_zoom_number], outputs = [annotator, annotate_current_page, annotate_current_page_bottom])
 
     ###
     # TABULAR DATA REDACTION

tools/aws_functions.py

@@ -10,7 +10,7 @@ PandasDataFrame = Type[pd.DataFrame]
 # Get AWS credentials
 bucket_name=""
 
-RUN_AWS_FUNCTIONS = get_or_create_env_var("RUN_AWS_FUNCTIONS", "0")
+RUN_AWS_FUNCTIONS = get_or_create_env_var("RUN_AWS_FUNCTIONS", "1")
 print(f'The value of RUN_AWS_FUNCTIONS is {RUN_AWS_FUNCTIONS}')
 
 AWS_REGION = get_or_create_env_var('AWS_REGION', 'eu-west-2')

tools/custom_image_analyser_engine.py

@@ -13,6 +13,7 @@ from copy import deepcopy
 from tools.helper_functions import clean_unicode_text
 from tools.aws_functions import comprehend_client
 from tools.presidio_analyzer_custom import recognizer_result_from_dict
+from tools.load_spacy_model_custom_recognisers import custom_entities
 #import string  # Import string to get a list of common punctuation characters
 
 @dataclass
@@ -491,6 +492,14 @@ class CustomImageAnalyzerEngine:
                 analyzer_results_by_line[i] = analyzer_result
 
             elif pii_identification_method == "AWS Comprehend":
+
+                # If using AWS Comprehend, Spacy model is only used to identify the custom entities created. Comprehend can't pick up Titles, Streetnames, and UKPostcodes specifically
+                text_analyzer_kwargs["entities"] = [entity for entity in chosen_redact_comprehend_entities if entity in custom_entities]
+
+                spacy_analyzer_result = self.analyzer_engine.analyze(
+                text=line_level_ocr_result.text, **text_analyzer_kwargs)
+                analyzer_results_by_line[i].extend(spacy_analyzer_result)
+
                 if len(line_level_ocr_result.text) >= 3:
                     # Add line to current batch with a separator
                     if current_batch:
@@ -509,6 +518,7 @@ class CustomImageAnalyzerEngine:
                                 Text=current_batch,
                                 LanguageCode=text_analyzer_kwargs["language"]
                             )
+
                         except Exception as e:
                             print(e)
                             time.sleep(3)

tools/file_conversion.py

@@ -11,6 +11,8 @@ import pymupdf
 from gradio import Progress
 from typing import List, Optional
 
+image_dpi = 300.0
+
 def is_pdf_or_image(filename):
     """
     Check if a file name is a PDF or an image file.
@@ -42,7 +44,7 @@ def is_pdf(filename):
 # %%
 ## Convert pdf to image if necessary
 
-def convert_pdf_to_images(pdf_path:str, page_min:int = 0, progress=Progress(track_tqdm=True)):
+def convert_pdf_to_images(pdf_path:str, page_min:int = 0, image_dpi:float = image_dpi, progress=Progress(track_tqdm=True)):
 
     # Get the number of pages in the PDF
     page_count = pdfinfo_from_path(pdf_path)['Pages']
@@ -70,7 +72,7 @@ def convert_pdf_to_images(pdf_path:str, page_min:int = 0, progress=Progress(trac
 
 
         else:
-            image_l = convert_from_path(pdf_path, first_page=page_num+1, last_page=page_num+1, dpi=300, use_cropbox=True, use_pdftocairo=False)
+            image_l = convert_from_path(pdf_path, first_page=page_num+1, last_page=page_num+1, dpi=image_dpi, use_cropbox=True, use_pdftocairo=False)
 
             image = image_l[0]
 
@@ -334,7 +336,7 @@ def prepare_image_or_pdf(
     
     return out_message_out, converted_file_paths, image_file_paths, number_of_pages, number_of_pages, pymupdf_doc
 
-def convert_text_pdf_to_img_pdf(in_file_path:str, out_text_file_path:List[str]):
+def convert_text_pdf_to_img_pdf(in_file_path:str, out_text_file_path:List[str], image_dpi:float=image_dpi):
     file_path_without_ext = get_file_path_end(in_file_path)
 
     out_file_paths = out_text_file_path
@@ -344,7 +346,7 @@ def convert_text_pdf_to_img_pdf(in_file_path:str, out_text_file_path:List[str]):
     
     pdf_text_image_paths = process_file(out_text_file_path[0])
     out_text_image_file_path = output_folder + file_path_without_ext + "_text_redacted_as_img.pdf"
-    pdf_text_image_paths[0].save(out_text_image_file_path, "PDF" ,resolution=300.0, save_all=True, append_images=pdf_text_image_paths[1:])
+    pdf_text_image_paths[0].save(out_text_image_file_path, "PDF" ,resolution=image_dpi, save_all=True, append_images=pdf_text_image_paths[1:])
 
     # out_file_paths.append(out_text_image_file_path)
 

tools/file_redaction.py

@@ -27,8 +27,8 @@ from collections import defaultdict  # For efficient grouping
 from presidio_analyzer import RecognizerResult
 
 from tools.custom_image_analyser_engine import CustomImageAnalyzerEngine, OCRResult, combine_ocr_results, CustomImageRecognizerResult
-from tools.file_conversion import process_file
-from tools.load_spacy_model_custom_recognisers import nlp_analyser, score_threshold
+from tools.file_conversion import process_file, image_dpi
+from tools.load_spacy_model_custom_recognisers import nlp_analyser, score_threshold, custom_entities
 from tools.helper_functions import get_file_path_end, output_folder, clean_unicode_text, get_or_create_env_var
 from tools.file_conversion import process_file, is_pdf, is_pdf_or_image
 # from tools.data_anonymise import generate_decision_process_output
@@ -314,8 +314,8 @@ def choose_and_run_redactor(file_paths:List[str],
 
             # Save file
             if is_pdf(file_path) == False:
-                out_image_file_path = output_folder + file_path_without_ext + "_redacted_as_img.pdf"
-                pymupdf_doc[0].save(out_image_file_path, "PDF" ,resolution=100.0, save_all=True, append_images=pymupdf_doc[1:])
+                out_image_file_path = output_folder + file_path_without_ext + "_redacted_as_pdf.pdf"
+                pymupdf_doc[0].save(out_image_file_path, "PDF" ,resolution=image_dpi, save_all=False)#, append_images=pymupdf_doc[:1])
             
             else:
                 out_image_file_path = output_folder + file_path_without_ext + "_redacted.pdf"
@@ -413,35 +413,40 @@ def choose_and_run_redactor(file_paths:List[str],
 
     return out_message, out_file_paths, out_file_paths, gr.Number(value=latest_file_completed, label="Number of documents redacted", interactive=False, visible=False), log_files_output_paths, log_files_output_paths, estimated_time_taken_state, all_request_metadata_str, pymupdf_doc, annotations_all_pages, gr.Number(value=current_loop_page, precision=0, interactive=False, label = "Last redacted page in document", visible=False), gr.Checkbox(value = True, label="Page break reached", visible=False), all_line_level_ocr_results_df, all_decision_process_table, comprehend_query_number
 
-def convert_pikepdf_coords_to_pymudf(pymupdf_page, annot):
+def convert_pikepdf_coords_to_pymupdf(pymupdf_page, pikepdf_bbox):
     '''
-    Convert annotations from pikepdf to pymupdf format
+    Convert annotations from pikepdf to pymupdf format, handling the mediabox larger than rect.
     '''
+    # Use cropbox if available, otherwise use mediabox
+    reference_box = pymupdf_page.rect
+    mediabox = pymupdf_page.mediabox
 
-    mediabox_height = pymupdf_page.mediabox[3] - pymupdf_page.mediabox[1]
-    mediabox_width = pymupdf_page.mediabox[2] - pymupdf_page.mediabox[0]
-    rect_height = pymupdf_page.rect.height
-    rect_width = pymupdf_page.rect.width  
+    reference_box_height = reference_box.height
+    reference_box_width = reference_box.width
+    
+    # Convert PyMuPDF coordinates back to PDF coordinates (bottom-left origin)
+    media_height = mediabox.height
+    media_width = mediabox.width
 
-    # Adjust coordinates based on scaling factors
-    page_x_adjust = (rect_width - mediabox_width) / 2  # Center adjustment
-    page_y_adjust = (rect_height - mediabox_height) / 2  # Center adjustment
+    media_reference_y_diff = media_height - reference_box_height
+    media_reference_x_diff = media_width - reference_box_width
 
-    #print("In the pikepdf conversion function")
-    # Extract the /Rect field
-    rect_field = annot["/Rect"]
-
-    # Convert the extracted /Rect field to a list of floats (since pikepdf uses Decimal objects)
-    rect_coordinates = [float(coord) for coord in rect_field]
+    y_diff_ratio = media_reference_y_diff / reference_box_height
+    x_diff_ratio = media_reference_x_diff / reference_box_width
+    
+    # Extract the annotation rectangle field
+    rect_field = pikepdf_bbox["/Rect"]
+    rect_coordinates = [float(coord) for coord in rect_field]  # Convert to floats
 
-    # Convert the Y-coordinates (flip using the page height)
+    # Unpack coordinates
     x1, y1, x2, y2 = rect_coordinates
-    x1 = x1 + page_x_adjust
-    new_y1 = (rect_height - y2) - page_y_adjust
-    x2 = x2 + page_x_adjust
-    new_y2 = (rect_height - y1) - page_y_adjust
-
-    return x1, new_y1, x2, new_y2
+    
+    new_x1 = x1 - (media_reference_x_diff * x_diff_ratio)
+    new_y1 = media_height - y2 - (media_reference_y_diff * y_diff_ratio)
+    new_x2 = x2 - (media_reference_x_diff * x_diff_ratio)
+    new_y2 = media_height - y1 - (media_reference_y_diff * y_diff_ratio)
+    
+    return new_x1, new_y1, new_x2, new_y2
 
 def convert_pikepdf_to_image_coords(pymupdf_page, annot, image:Image):
     '''
@@ -496,6 +501,64 @@ def convert_image_coords_to_pymupdf(pymupdf_page, annot:CustomImageRecognizerRes
 
     return x1, new_y1, x2, new_y2
 
+# def convert_pymupdf_to_image_coords(pymupdf_page, x1, y1, x2, y2, image: Image):
+#     '''
+#     Converts coordinates from pymupdf format to image coordinates.
+#     '''
+
+#     rect_height = pymupdf_page.rect.height
+#     rect_width = pymupdf_page.rect.width 
+
+#     image_page_width, image_page_height = image.size
+
+#     # Calculate scaling factors between pymupdf and PIL image
+#     scale_width = image_page_width / rect_width
+#     scale_height = image_page_height / rect_height
+
+#     x1_image = x1 * scale_width
+#     y1_image = ((rect_height - y2) * scale_height)
+#     x2_image = x2 * scale_width
+#     y2_image = ((rect_height - y1) * scale_height)
+
+#     return x1_image, y1_image, x2_image, y2_image
+
+def convert_pymupdf_to_image_coords(pymupdf_page, x1, y1, x2, y2, image: Image):
+    '''
+    Converts coordinates from pymupdf format to image coordinates,
+    accounting for mediabox dimensions.
+    '''
+    
+    rect_height = pymupdf_page.rect.height
+    rect_width = pymupdf_page.rect.width
+    
+    # Get mediabox dimensions
+    mediabox = pymupdf_page.mediabox
+    mediabox_width = mediabox.width
+    mediabox_height = mediabox.height
+    
+    image_page_width, image_page_height = image.size
+
+    # Calculate scaling factors using mediabox dimensions
+    scale_width = image_page_width / mediabox_width
+    scale_height = image_page_height / mediabox_height
+
+    print("scale_width:", scale_width)
+    print("scale_height:", scale_height)
+
+    rect_to_mediabox_x_scale = mediabox_width / rect_width
+    rect_to_mediabox_y_scale = mediabox_height / rect_height 
+
+    print("rect_to_mediabox_x_scale:", rect_to_mediabox_x_scale)
+    print("rect_to_mediabox_y_scale:", rect_to_mediabox_y_scale)
+
+    # Adjust coordinates based on scaling factors
+    x1_image = (x1 * scale_width) * rect_to_mediabox_x_scale
+    y1_image = (y1 * scale_height) * rect_to_mediabox_y_scale
+    x2_image = (x2 * scale_width) * rect_to_mediabox_x_scale
+    y2_image = (y2 * scale_height) * rect_to_mediabox_y_scale
+
+    return x1_image, y1_image, x2_image, y2_image
+
 def convert_gradio_annotation_coords_to_pymupdf(pymupdf_page:Page, annot:dict, image:Image):
     '''
     Converts an image with redaction coordinates from a gradio annotation component to pymupdf coordinates.
@@ -587,25 +650,25 @@ def redact_page_with_pymupdf(page:Page, annotations_on_page, image = None):
 
         # Else it should be a pikepdf annotation object
         else:           
-            x1, pymupdf_y1, x2, pymupdf_y2 = convert_pikepdf_coords_to_pymudf(page, annot)
+            x1, pymupdf_y1, x2, pymupdf_y2 = convert_pikepdf_coords_to_pymupdf(page, annot)
 
             rect = Rect(x1, pymupdf_y1, x2, pymupdf_y2)
 
             img_annotation_box = {}
 
             if image:
-                image_x1, image_y1, image_x2, image_y2 = convert_pikepdf_to_image_coords(page, annot, image)
-                
-                img_annotation_box["xmin"] = image_x1
-                img_annotation_box["ymin"] = image_y1
-                img_annotation_box["xmax"] = image_x2
-                img_annotation_box["ymax"] = image_y2
-                img_annotation_box["color"] = (0,0,0)
+                img_width, img_height = image.size
+
+                x1, image_y1, x2, image_y2 = convert_pymupdf_to_image_coords(page, x1, pymupdf_y1, x2, pymupdf_y2, image)
+
+                img_annotation_box["xmin"] = x1  #* (img_width / rect_width) # Use adjusted x1
+                img_annotation_box["ymin"] = image_y1  #* (img_width / rect_width) # Use adjusted y1
+                img_annotation_box["xmax"] = x2# * (img_height / rect_height) # Use adjusted x2
+                img_annotation_box["ymax"] = image_y2 #* (img_height / rect_height) # Use adjusted y2
+                img_annotation_box["color"] = (0, 0, 0)
 
                 if isinstance(annot, Dictionary):
-                    #print("Trying to get label out of annotation", annot["/T"])
                     img_annotation_box["label"] = str(annot["/T"])
-                    #print("Label is:", img_annotation_box["label"])
                 else:
                     img_annotation_box["label"] = "REDACTION"
 
@@ -646,6 +709,18 @@ def merge_img_bboxes(bboxes, combined_results: Dict, signature_recogniser_result
     merged_bboxes = []
     grouped_bboxes = defaultdict(list)
 
+    
+        # Process signature and handwriting results
+    if signature_recogniser_results or handwriting_recogniser_results:
+        if "Redact all identified handwriting" in handwrite_signature_checkbox:
+            #print("Handwriting boxes exist at merge:", handwriting_recogniser_results)
+            merged_bboxes.extend(handwriting_recogniser_results)
+
+        if "Redact all identified signatures" in handwrite_signature_checkbox:
+            #print("Signature boxes exist at merge:", signature_recogniser_results)
+            merged_bboxes.extend(signature_recogniser_results)
+
+
     # Reconstruct bounding boxes for substrings of interest
     reconstructed_bboxes = []
     for bbox in bboxes:
@@ -735,16 +810,6 @@ def merge_img_bboxes(bboxes, combined_results: Dict, signature_recogniser_result
 
         merged_bboxes.append(merged_box)
 
-        # Process signature and handwriting results
-    if signature_recogniser_results or handwriting_recogniser_results:
-        if "Redact all identified handwriting" in handwrite_signature_checkbox:
-            #print("Handwriting boxes exist at merge:", handwriting_recogniser_results)
-            merged_bboxes.extend(handwriting_recogniser_results)
-
-        if "Redact all identified signatures" in handwrite_signature_checkbox:
-            #print("Signature boxes exist at merge:", signature_recogniser_results)
-            merged_bboxes.extend(signature_recogniser_results)
-
     #print("bboxes:", bboxes)
 
     return merged_bboxes
@@ -1483,6 +1548,21 @@ def redact_text_pdf(
                                     all_text_line_results.append((i, text_line_analyser_result))
                                 
                                 elif pii_identification_method == "AWS Comprehend":
+
+                                    # First use the local Spacy model to pick up custom entities that AWS Comprehend can't search for.
+                                    custom_redact_entities = [entity for entity in chosen_redact_comprehend_entities if entity in custom_entities]
+
+                                    text_line_analyser_result = nlp_analyser.analyze(
+                                        text=text_line.text,
+                                        language=language,
+                                        entities=custom_redact_entities,
+                                        score_threshold=score_threshold,
+                                        return_decision_process=True,
+                                        allow_list=allow_list
+                                    )
+                                    all_text_line_results.append((i, text_line_analyser_result))
+
+
                                     if len(text_line.text) >= 3:
                                         # Add separator between lines
                                         if current_batch:

tools/helper_functions.py

@@ -3,9 +3,20 @@ import re
 import gradio as gr
 import pandas as pd
 import unicodedata
+from gradio_image_annotation import image_annotator
 
 def reset_state_vars():
-    return [], [], pd.DataFrame(), pd.DataFrame(), 0, ""
+    return [], [], pd.DataFrame(), pd.DataFrame(), 0, "", image_annotator(
+            label="Modify redaction boxes",
+            label_list=["Redaction"],
+            label_colors=[(0, 0, 0)],
+            show_label=False,
+            sources=None,#["upload"],
+            show_clear_button=False,
+            show_share_button=False,
+            show_remove_button=False,
+            interactive=False
+        )
 
 def get_or_create_env_var(var_name, default_value):
     # Get the environment variable if it exists

tools/load_spacy_model_custom_recognisers.py

@@ -10,6 +10,7 @@ import re
 # %%
 model_name = "en_core_web_lg" #"en_core_web_trf"
 score_threshold = 0.001
+custom_entities = ["TITLES", "UKPOSTCODE", "STREETNAME"]
 
 # %% [markdown]
 # #### Custom recognisers

tools/redaction_review.py

@@ -37,9 +37,22 @@ def increase_page(number:int, image_annotator_object:AnnotatedImageData):
     else:
         return max_pages, max_pages
 
-def update_annotator(image_annotator_object:AnnotatedImageData, page_num:int):
+def update_zoom(current_zoom_level:int, decrease:bool=True):
+    if decrease == False:
+        if current_zoom_level >= 50:
+            current_zoom_level -= 10
+    else:    
+        if current_zoom_level < 100:
+            current_zoom_level += 10
+        
+    return current_zoom_level
+
+
+def update_annotator(image_annotator_object:AnnotatedImageData, page_num:int, zoom:int=100):
     # print("\nImage annotator object:", image_annotator_object)
 
+    zoom_str = str(zoom) + '%'
+
     if not image_annotator_object:
         return image_annotator(
         label="Modify redaction boxes",
@@ -76,8 +89,8 @@ def update_annotator(image_annotator_object:AnnotatedImageData, page_num:int):
         #label_list=["Redaction"],
         #label_colors=[(0, 0, 0)],
         show_label=False,
-        height='100%',
-        width='100%',
+        height=zoom_str,
+        width=zoom_str,
         box_min_size=1,
         box_selected_thickness=2,
         handle_size=4,