import time import re import json import io import os import boto3 from tqdm import tqdm from PIL import Image, ImageChops, ImageFile, ImageDraw ImageFile.LOAD_TRUNCATED_IMAGES = True from typing import List, Dict, Tuple import pandas as pd #from presidio_image_redactor.entities import ImageRecognizerResult from pdfminer.high_level import extract_pages from pdfminer.layout import LTTextContainer, LTChar, LTTextLine, LTTextLineHorizontal, LTAnno from pikepdf import Pdf, Dictionary, Name import pymupdf from pymupdf import Rect from fitz import Document, Page import gradio as gr from gradio import Progress from collections import defaultdict # For efficient grouping from presidio_analyzer import RecognizerResult from tools.aws_functions import RUN_AWS_FUNCTIONS from tools.custom_image_analyser_engine import CustomImageAnalyzerEngine, OCRResult, combine_ocr_results, CustomImageRecognizerResult 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, tesseract_ocr_option, text_ocr_option, textract_option, local_pii_detector, aws_pii_detector from tools.file_conversion import process_file, is_pdf, is_pdf_or_image from tools.aws_textract import analyse_page_with_textract, json_to_ocrresult from tools.presidio_analyzer_custom import recognizer_result_from_dict # Number of pages to loop through before breaking. Currently set very high, as functions are breaking on time metrics (e.g. every 105 seconds), rather than on number of pages redacted. page_break_value = get_or_create_env_var('page_break_value', '500') print(f'The value of page_break_value is {page_break_value}') max_time_value = get_or_create_env_var('max_time_value', '105') print(f'The value of max_time_value is {max_time_value}') def sum_numbers_before_seconds(string:str): """Extracts numbers that precede the word 'seconds' from a string and adds them up. Args: string: The input string. Returns: The sum of all numbers before 'seconds' in the string. """ # Extract numbers before 'seconds' using regular expression numbers = re.findall(r'(\d+\.\d+)?\s*seconds', string) # Extract the numbers from the matches numbers = [float(num.split()[0]) for num in numbers] # Sum up the extracted numbers sum_of_numbers = round(sum(numbers),1) return sum_of_numbers def choose_and_run_redactor(file_paths:List[str], prepared_pdf_file_paths:List[str], prepared_pdf_image_paths:List[str], language:str, chosen_redact_entities:List[str], chosen_redact_comprehend_entities:List[str], in_redact_method:str, in_allow_list:List[List[str]]=None, latest_file_completed:int=0, out_message:list=[], out_file_paths:list=[], log_files_output_paths:list=[], first_loop_state:bool=False, page_min:int=0, page_max:int=999, estimated_time_taken_state:float=0.0, handwrite_signature_checkbox:List[str]=["Redact all identified handwriting", "Redact all identified signatures"], all_request_metadata_str:str = "", annotations_all_pages:dict={}, all_line_level_ocr_results_df=[], all_decision_process_table=[], pymupdf_doc=[], current_loop_page:int=0, page_break_return:bool=False, pii_identification_method:str="Local", comprehend_query_number:int=0, progress=gr.Progress(track_tqdm=True)): ''' This function orchestrates the redaction process based on the specified method and parameters. It takes the following inputs: - file_paths (List[str]): A list of paths to the files to be redacted. - prepared_pdf_file_paths (List[str]): A list of paths to the PDF files prepared for redaction. - prepared_pdf_image_paths (List[str]): A list of paths to the PDF files converted to images for redaction. - language (str): The language of the text in the files. - chosen_redact_entities (List[str]): A list of entity types to redact from the files using the local model (spacy) with Microsoft Presidio. - chosen_redact_comprehend_entities (List[str]): A list of entity types to redact from files, chosen from the official list from AWS Comprehend service - in_redact_method (str): The method to use for redaction. - in_allow_list (List[List[str]], optional): A list of allowed terms for redaction. Defaults to None. - latest_file_completed (int, optional): The index of the last completed file. Defaults to 0. - out_message (list, optional): A list to store output messages. Defaults to an empty list. - out_file_paths (list, optional): A list to store paths to the output files. Defaults to an empty list. - log_files_output_paths (list, optional): A list to store paths to the log files. Defaults to an empty list. - first_loop_state (bool, optional): A flag indicating if this is the first iteration. Defaults to False. - page_min (int, optional): The minimum page number to start redaction from. Defaults to 0. - page_max (int, optional): The maximum page number to end redaction at. Defaults to 999. - estimated_time_taken_state (float, optional): The estimated time taken for the redaction process. Defaults to 0.0. - handwrite_signature_checkbox (List[str], optional): A list of options for redacting handwriting and signatures. Defaults to ["Redact all identified handwriting", "Redact all identified signatures"]. - all_request_metadata_str (str, optional): A string containing all request metadata. Defaults to an empty string. - annotations_all_pages (dict, optional): A dictionary containing all image annotations. Defaults to an empty dictionary. - all_line_level_ocr_results_df (optional): A DataFrame containing all line-level OCR results. Defaults to an empty DataFrame. - all_decision_process_table (optional): A DataFrame containing all decision process tables. Defaults to an empty DataFrame. - pymupdf_doc (optional): A list containing the PDF document object. Defaults to an empty list. - current_loop_page (int, optional): The current page being processed in the loop. Defaults to 0. - page_break_return (bool, optional): A flag indicating if the function should return after a page break. Defaults to False. - pii_identification_method (str, optional): The method to redact personal information. Either 'Local' (spacy model), or 'AWS Comprehend' (AWS Comprehend API). - comprehend_query_number (int, optional): A counter tracking the number of queries to AWS Comprehend. - progress (gr.Progress, optional): A progress tracker for the redaction process. Defaults to a Progress object with track_tqdm set to True. The function returns a redacted document along with processing logs. ''' combined_out_message = "" tic = time.perf_counter() all_request_metadata = all_request_metadata_str.split('\n') if all_request_metadata_str else [] # If this is the first time around, set variables to 0/blank if first_loop_state==True: #print("First_loop_state is True") latest_file_completed = 0 current_loop_page = 0 out_file_paths = [] estimate_total_processing_time = 0 estimated_time_taken_state = 0 # If not the first time around, and the current page loop has been set to a huge number (been through all pages), reset current page to 0 elif (first_loop_state == False) & (current_loop_page == 999): current_loop_page = 0 if not out_file_paths: out_file_paths = [] latest_file_completed = int(latest_file_completed) number_of_pages = len(prepared_pdf_image_paths) if isinstance(file_paths,str): number_of_files = 1 else: number_of_files = len(file_paths) # If we have already redacted the last file, return the input out_message and file list to the relevant components if latest_file_completed >= number_of_files: print("Completed last file") # Set to a very high number so as not to mix up with subsequent file processing by the user # latest_file_completed = 99 current_loop_page = 0 if isinstance(out_message, list): combined_out_message = '\n'.join(out_message) else: combined_out_message = out_message estimate_total_processing_time = sum_numbers_before_seconds(combined_out_message) print("Estimated total processing time:", str(estimate_total_processing_time)) return combined_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 # If we have reached the last page, return message if current_loop_page >= number_of_pages: print("current_loop_page:", current_loop_page, "is equal to or greater than number of pages in document:", number_of_pages) # Set to a very high number so as not to mix up with subsequent file processing by the user current_loop_page = 999 combined_out_message = out_message return combined_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 = False, label="Page break reached", visible=False), all_line_level_ocr_results_df, all_decision_process_table, comprehend_query_number # Create allow list # If string, assume file path if isinstance(in_allow_list, str): in_allow_list = pd.read_csv(in_allow_list) if not in_allow_list.empty: in_allow_list_flat = in_allow_list.iloc[:,0].tolist() print("In allow list:", in_allow_list_flat) else: in_allow_list_flat = [] # Try to connect to AWS services only if RUN_AWS_FUNCTIONS environmental variable is 1 if pii_identification_method == "AWS Comprehend": print("Trying to connect to AWS Comprehend service") if RUN_AWS_FUNCTIONS == "1": comprehend_client = boto3.client('comprehend') else: comprehend_client = "" out_message = "Cannot connect to AWS Comprehend service. Please choose another PII identification method." print(out_message) 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 else: comprehend_client = "" if in_redact_method == textract_option: print("Trying to connect to AWS Comprehend service") if RUN_AWS_FUNCTIONS == "1": textract_client = boto3.client('textract') else: textract_client = "" out_message = "Cannot connect to AWS Textract. Please choose another text extraction method." print(out_message) 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 else: textract_client = "" progress(0.5, desc="Redacting file") if isinstance(file_paths, str): file_paths_list = [os.path.abspath(file_paths)] file_paths_loop = file_paths_list elif isinstance(file_paths, dict): file_paths = file_paths["name"] file_paths_list = [os.path.abspath(file_paths)] file_paths_loop = file_paths_list else: file_paths_list = file_paths file_paths_loop = [file_paths_list[int(latest_file_completed)]] print("file_paths_list in choose_redactor function:", file_paths_list) for file in file_paths_loop: if isinstance(file, str): file_path = file else: file_path = file.name if file_path: file_path_without_ext = get_file_path_end(file_path) print("Redacting file:", file_path_without_ext) is_a_pdf = is_pdf(file_path) == True if is_a_pdf == False: # If user has not submitted a pdf, assume it's an image print("File is not a pdf, assuming that image analysis needs to be used.") in_redact_method = tesseract_ocr_option else: out_message = "No file selected" print(out_message) return combined_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 if in_redact_method == tesseract_ocr_option or in_redact_method == textract_option: #Analyse and redact image-based pdf or image if is_pdf_or_image(file_path) == False: out_message = "Please upload a PDF file or image file (JPG, PNG) for image analysis." 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 print("Redacting file " + file_path_without_ext + " as an image-based file") pymupdf_doc,all_decision_process_table,logging_file_paths,new_request_metadata, annotations_all_pages, current_loop_page, page_break_return, all_line_level_ocr_results_df, comprehend_query_number = redact_image_pdf(file_path, prepared_pdf_image_paths, language, chosen_redact_entities, chosen_redact_comprehend_entities, in_allow_list_flat, is_a_pdf, page_min, page_max, in_redact_method, handwrite_signature_checkbox, "", current_loop_page, page_break_return, prepared_pdf_image_paths, annotations_all_pages, all_line_level_ocr_results_df, all_decision_process_table, pymupdf_doc, pii_identification_method, comprehend_query_number, comprehend_client, textract_client) # Save Textract request metadata (if exists) if new_request_metadata: print("Request metadata:", new_request_metadata) all_request_metadata.append(new_request_metadata) elif in_redact_method == text_ocr_option: logging_file_paths = "" if is_pdf(file_path) == False: out_message = "Please upload a PDF file for text analysis. If you have an image, select 'Image analysis'." 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 # Analyse text-based pdf print('Redacting file as text-based PDF') pymupdf_doc, all_decision_process_table, all_line_level_ocr_results_df, annotations_all_pages, current_loop_page, page_break_return, comprehend_query_number = redact_text_pdf(file_path, prepared_pdf_image_paths,language, chosen_redact_entities, chosen_redact_comprehend_entities, in_allow_list_flat, page_min, page_max, text_ocr_option, current_loop_page, page_break_return, annotations_all_pages, all_line_level_ocr_results_df, all_decision_process_table, pymupdf_doc, pii_identification_method, comprehend_query_number, comprehend_client) else: out_message = "No redaction method selected" print(out_message) 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 # If at last page, save to file if current_loop_page >= number_of_pages: print("Current page loop:", current_loop_page, "is greater or equal to number of pages:", number_of_pages) latest_file_completed += 1 current_loop_page = 999 if latest_file_completed != len(file_paths_list): print("Completed file number:", str(latest_file_completed), "there are more files to do") # Save file if is_pdf(file_path) == False: 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" pymupdf_doc.save(out_image_file_path) out_file_paths.append(out_image_file_path) if logging_file_paths: log_files_output_paths.extend(logging_file_paths) logs_output_file_name = out_image_file_path + "_decision_process_output.csv" all_decision_process_table.to_csv(logs_output_file_name, index = None, encoding="utf-8") out_file_paths.append(logs_output_file_name) all_text_output_file_name = out_image_file_path + "_ocr_output.csv" all_line_level_ocr_results_df.to_csv(all_text_output_file_name, index = None, encoding="utf-8") out_file_paths.append(all_text_output_file_name) # Save the gradio_annotation_boxes to a JSON file try: out_annotation_file_path = out_image_file_path + '_redactions.json' with open(out_annotation_file_path, 'w') as f: json.dump(annotations_all_pages, f) out_file_paths.append(out_annotation_file_path) except: print("Could not save annotations to json file.") # Make a combined message for the file if isinstance(out_message, list): combined_out_message = '\n'.join(out_message) # Ensure out_message is a list of strings else: combined_out_message = out_message toc = time.perf_counter() time_taken = toc - tic estimated_time_taken_state = estimated_time_taken_state + time_taken out_time_message = f" Redacted in {estimated_time_taken_state:0.1f} seconds." combined_out_message = combined_out_message + " " + out_time_message # Ensure this is a single string estimate_total_processing_time = sum_numbers_before_seconds(combined_out_message) print("Estimated total processing time:", str(estimate_total_processing_time)) else: toc = time.perf_counter() time_taken = toc - tic estimated_time_taken_state = estimated_time_taken_state + time_taken # If textract requests made, write to logging file if all_request_metadata: all_request_metadata_str = '\n'.join(all_request_metadata).strip() all_request_metadata_file_path = output_folder + file_path_without_ext + "_textract_request_metadata.txt" with open(all_request_metadata_file_path, "w") as f: f.write(all_request_metadata_str) # Add the request metadata to the log outputs if not there already if all_request_metadata_file_path not in log_files_output_paths: log_files_output_paths.append(all_request_metadata_file_path) if combined_out_message: out_message = combined_out_message #print("\nout_message at choose_and_run_redactor end is:", out_message) # Ensure no duplicated output files log_files_output_paths = list(set(log_files_output_paths)) out_file_paths = list(set(out_file_paths)) 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_pymupdf(pymupdf_page, pikepdf_bbox): ''' 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 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 media_reference_y_diff = media_height - reference_box_height media_reference_x_diff = media_width - reference_box_width 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 # Unpack coordinates x1, y1, x2, y2 = rect_coordinates 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): ''' Convert annotations from pikepdf coordinates to image coordinates. ''' # Get the dimensions of the page in points with pymupdf rect_height = pymupdf_page.rect.height rect_width = pymupdf_page.rect.width # Get the dimensions of the image 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 # Extract the /Rect field rect_field = annot["/Rect"] # Convert the extracted /Rect field to a list of floats rect_coordinates = [float(coord) for coord in rect_field] # Convert the Y-coordinates (flip using the image height) x1, y1, x2, y2 = rect_coordinates x1_image = x1 * scale_width new_y1_image = image_page_height - (y2 * scale_height) # Flip Y0 (since it starts from bottom) x2_image = x2 * scale_width new_y2_image = image_page_height - (y1 * scale_height) # Flip Y1 return x1_image, new_y1_image, x2_image, new_y2_image def convert_image_coords_to_pymupdf(pymupdf_page, annot:CustomImageRecognizerResult, image:Image): ''' Converts an image with redaction coordinates from a CustomImageRecognizerResult to pymupdf 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 PIL image and pymupdf scale_width = rect_width / image_page_width scale_height = rect_height / image_page_height # Calculate scaled coordinates x1 = (annot.left * scale_width)# + page_x_adjust new_y1 = (annot.top * scale_height)# - page_y_adjust # Flip Y0 (since it starts from bottom) x2 = ((annot.left + annot.width) * scale_width)# + page_x_adjust # Calculate x1 new_y2 = ((annot.top + annot.height) * scale_height)# - page_y_adjust # Calculate y1 correctly 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, 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. ''' rect_height = pymupdf_page.rect.height rect_width = pymupdf_page.rect.width image_page_width, image_page_height = image.size # Calculate scaling factors between PIL image and pymupdf scale_width = rect_width / image_page_width scale_height = rect_height / image_page_height # Calculate scaled coordinates x1 = (annot["xmin"] * scale_width)# + page_x_adjust new_y1 = (annot["ymin"] * scale_height)# - page_y_adjust # Flip Y0 (since it starts from bottom) x2 = ((annot["xmax"]) * scale_width)# + page_x_adjust # Calculate x1 new_y2 = ((annot["ymax"]) * scale_height)# - page_y_adjust # Calculate y1 correctly return x1, new_y1, x2, new_y2 def move_page_info(file_path: str) -> str: # Split the string at '.png' base, extension = file_path.rsplit('.pdf', 1) # Extract the page info page_info = base.split('page ')[1].split(' of')[0] # Get the page number new_base = base.replace(f'page {page_info} of ', '') # Remove the page info from the original position # Construct the new file path new_file_path = f"{new_base}_page_{page_info}.png" return new_file_path def redact_page_with_pymupdf(page:Page, annotations_on_page, image = None): mediabox_height = page.mediabox[3] - page.mediabox[1] mediabox_width = page.mediabox[2] - page.mediabox[0] rect_height = page.rect.height rect_width = page.rect.width out_annotation_boxes = {} all_image_annotation_boxes = [] image_path = "" if isinstance(image, Image.Image): image_path = move_page_info(str(page)) image.save(image_path) elif isinstance(image, str): image_path = image image = Image.open(image_path) # Check if this is an object used in the Gradio Annotation component if isinstance (annotations_on_page, dict): annotations_on_page = annotations_on_page["boxes"] for annot in annotations_on_page: # Check if an Image recogniser result, or a Gradio annotation object if (isinstance(annot, CustomImageRecognizerResult)) | isinstance(annot, dict): img_annotation_box = {} # Should already be in correct format if img_annotator_box is an input if isinstance(annot, dict): img_annotation_box = annot x1, pymupdf_y1, x2, pymupdf_y2 = convert_gradio_annotation_coords_to_pymupdf(page, annot, image) # Else should be CustomImageRecognizerResult else: x1, pymupdf_y1, x2, pymupdf_y2 = convert_image_coords_to_pymupdf(page, annot, image) img_annotation_box["xmin"] = annot.left img_annotation_box["ymin"] = annot.top img_annotation_box["xmax"] = annot.left + annot.width img_annotation_box["ymax"] = annot.top + annot.height img_annotation_box["color"] = (0,0,0) try: img_annotation_box["label"] = annot.entity_type except: img_annotation_box["label"] = "Redaction" rect = Rect(x1, pymupdf_y1, x2, pymupdf_y2) # Create the PyMuPDF Rect # Else it should be a pikepdf annotation object else: 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: 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): img_annotation_box["label"] = str(annot["/T"]) else: img_annotation_box["label"] = "REDACTION" # Convert to a PyMuPDF Rect object #rect = Rect(rect_coordinates) all_image_annotation_boxes.append(img_annotation_box) # Calculate the middle y value and set height to 1 pixel middle_y = (pymupdf_y1 + pymupdf_y2) / 2 rect_single_pixel_height = Rect(x1, middle_y - 2, x2, middle_y + 2) # Small height in middle of word to remove text # Add the annotation to the middle of the character line, so that it doesn't delete text from adjacent lines page.add_redact_annot(rect_single_pixel_height) # Set up drawing a black box over the whole rect shape = page.new_shape() shape.draw_rect(rect) shape.finish(color=(0, 0, 0), fill=(0, 0, 0)) # Black fill for the rectangle shape.commit() out_annotation_boxes = { "image": image_path, #Image.open(image_path), #image_path, "boxes": all_image_annotation_boxes } page.apply_redactions(images=0, graphics=0) page.clean_contents() return page, out_annotation_boxes def bounding_boxes_overlap(box1, box2): """Check if two bounding boxes overlap.""" return (box1[0] < box2[2] and box2[0] < box1[2] and box1[1] < box2[3] and box2[1] < box1[3]) def merge_img_bboxes(bboxes, combined_results: Dict, signature_recogniser_results=[], handwriting_recogniser_results=[], handwrite_signature_checkbox: List[str]=["Redact all identified handwriting", "Redact all identified signatures"], horizontal_threshold:int=50, vertical_threshold:int=12): 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: #print("bbox:", bbox) bbox_box = (bbox.left, bbox.top, bbox.left + bbox.width, bbox.top + bbox.height) for line_text, line_info in combined_results.items(): line_box = line_info['bounding_box'] if bounding_boxes_overlap(bbox_box, line_box): if bbox.text in line_text: start_char = line_text.index(bbox.text) end_char = start_char + len(bbox.text) relevant_words = [] current_char = 0 for word in line_info['words']: word_end = current_char + len(word['text']) if current_char <= start_char < word_end or current_char < end_char <= word_end or (start_char <= current_char and word_end <= end_char): relevant_words.append(word) if word_end >= end_char: break current_char = word_end if not word['text'].endswith(' '): current_char += 1 # +1 for space if the word doesn't already end with a space if relevant_words: #print("Relevant words:", relevant_words) left = min(word['bounding_box'][0] for word in relevant_words) top = min(word['bounding_box'][1] for word in relevant_words) right = max(word['bounding_box'][2] for word in relevant_words) bottom = max(word['bounding_box'][3] for word in relevant_words) # Combine the text of all relevant words combined_text = " ".join(word['text'] for word in relevant_words) # Calculate new dimensions for the merged box reconstructed_bbox = CustomImageRecognizerResult( bbox.entity_type, bbox.start, bbox.end, bbox.score, left, top, right - left, # width bottom - top, # height combined_text ) reconstructed_bboxes.append(reconstructed_bbox) break else: # If the bbox text is not found in any line in combined_results, keep the original bbox reconstructed_bboxes.append(bbox) # Group reconstructed bboxes by approximate vertical proximity for box in reconstructed_bboxes: grouped_bboxes[round(box.top / vertical_threshold)].append(box) # Merge within each group for _, group in grouped_bboxes.items(): group.sort(key=lambda box: box.left) merged_box = group[0] for next_box in group[1:]: if next_box.left - (merged_box.left + merged_box.width) <= horizontal_threshold: # Calculate new dimensions for the merged box if merged_box.text == next_box.text: new_text = merged_box.text else: new_text = merged_box.text + " " + next_box.text if merged_box.text == next_box.text: new_text = merged_box.text new_entity_type = merged_box.entity_type # Keep the original entity type else: new_text = merged_box.text + " " + next_box.text new_entity_type = merged_box.entity_type + " - " + next_box.entity_type # Concatenate entity types new_left = min(merged_box.left, next_box.left) new_top = min(merged_box.top, next_box.top) new_width = max(merged_box.left + merged_box.width, next_box.left + next_box.width) - new_left new_height = max(merged_box.top + merged_box.height, next_box.top + next_box.height) - new_top merged_box = CustomImageRecognizerResult( new_entity_type, merged_box.start, merged_box.end, merged_box.score, new_left, new_top, new_width, new_height, new_text ) else: merged_bboxes.append(merged_box) merged_box = next_box merged_bboxes.append(merged_box) #print("bboxes:", bboxes) return merged_bboxes def redact_image_pdf(file_path:str, prepared_pdf_file_paths:List[str], language:str, chosen_redact_entities:List[str], chosen_redact_comprehend_entities:List[str], allow_list:List[str]=None, is_a_pdf:bool=True, page_min:int=0, page_max:int=999, analysis_type:str=tesseract_ocr_option, handwrite_signature_checkbox:List[str]=["Redact all identified handwriting", "Redact all identified signatures"], request_metadata:str="", current_loop_page:int=0, page_break_return:bool=False, images=[], annotations_all_pages:List=[], all_line_level_ocr_results_df = pd.DataFrame(), all_decision_process_table = pd.DataFrame(), pymupdf_doc = [], pii_identification_method:str="Local", comprehend_query_number:int=0, comprehend_client="", textract_client="", page_break_val:int=int(page_break_value), logging_file_paths:List=[], max_time:int=int(max_time_value), progress=Progress(track_tqdm=True)): ''' This function redacts sensitive information from a PDF document. It takes the following parameters: - file_path (str): The path to the PDF file to be redacted. - prepared_pdf_file_paths (List[str]): A list of paths to the PDF file pages converted to images. - language (str): The language of the text in the PDF. - chosen_redact_entities (List[str]): A list of entity types to redact from the PDF. - chosen_redact_comprehend_entities (List[str]): A list of entity types to redact from the list allowed by the AWS Comprehend service. - allow_list (List[str], optional): A list of entity types to allow in the PDF. Defaults to None. - is_a_pdf (bool, optional): Indicates if the input file is a PDF. Defaults to True. - page_min (int, optional): The minimum page number to start redaction from. Defaults to 0. - page_max (int, optional): The maximum page number to end redaction at. Defaults to 999. - analysis_type (str, optional): The type of analysis to perform on the PDF. Defaults to tesseract_ocr_option. - handwrite_signature_checkbox (List[str], optional): A list of options for redacting handwriting and signatures. Defaults to ["Redact all identified handwriting", "Redact all identified signatures"]. - request_metadata (str, optional): Metadata related to the redaction request. Defaults to an empty string. - page_break_return (bool, optional): Indicates if the function should return after a page break. Defaults to False. - images (list, optional): List of image objects for each PDF page. - annotations_all_pages (List, optional): List of annotations on all pages that is used by the gradio_image_annotation object. - all_line_level_ocr_results_df (pd.DataFrame(), optional): All line level OCR results for the document as a Pandas dataframe, - all_decision_process_table (pd.DataFrame(), optional): All redaction decisions for document as a Pandas dataframe. - pymupdf_doc (List, optional): The document as a PyMupdf object. - pii_identification_method (str, optional): The method to redact personal information. Either 'Local' (spacy model), or 'AWS Comprehend' (AWS Comprehend API). - comprehend_query_number (int, optional): A counter tracking the number of queries to AWS Comprehend. - comprehend_client (optional): A connection to the AWS Comprehend service via the boto3 package. - textract_client (optional): A connection to the AWS Textract service via the boto3 package. - page_break_val (int, optional): The value at which to trigger a page break. Defaults to 3. - logging_file_paths (List, optional): List of file paths used for saving redaction process logging results. - max_time (int, optional): The maximum amount of time (s) that the function should be running before it breaks. To avoid timeout errors with some APIs. - progress (Progress, optional): A progress tracker for the redaction process. Defaults to a Progress object with track_tqdm set to True. The function returns a fully or partially-redacted PDF document. ''' file_name = get_file_path_end(file_path) fill = (0, 0, 0) # Fill colour image_analyser = CustomImageAnalyzerEngine(nlp_analyser) comprehend_query_number_new = 0 if pii_identification_method == "AWS Comprehend" and comprehend_client == "": print("Connection to AWS Comprehend service unsuccessful.") return pymupdf_doc, all_decision_process_table, logging_file_paths, request_metadata, annotations_all_pages, current_loop_page, page_break_return, all_line_level_ocr_results_df, comprehend_query_number if analysis_type == textract_option and textract_client == "": print("Connection to AWS Textract service unsuccessful.") return pymupdf_doc, all_decision_process_table, logging_file_paths, request_metadata, annotations_all_pages, current_loop_page, page_break_return, all_line_level_ocr_results_df, comprehend_query_number tic = time.perf_counter() if not prepared_pdf_file_paths: out_message = "PDF does not exist as images. Converting pages to image" print(out_message) prepared_pdf_file_paths = process_file(file_path) number_of_pages = len(prepared_pdf_file_paths) print("Number of pages:", str(number_of_pages)) # Check that page_min and page_max are within expected ranges if page_max > number_of_pages or page_max == 0: page_max = number_of_pages if page_min <= 0: page_min = 0 else: page_min = page_min - 1 print("Page range:", str(page_min + 1), "to", str(page_max)) #print("Current_loop_page:", current_loop_page) if analysis_type == tesseract_ocr_option: ocr_results_file_path = output_folder + "ocr_results_" + file_name + "_pages_" + str(page_min + 1) + "_" + str(page_max) + ".csv" elif analysis_type == textract_option: ocr_results_file_path = output_folder + "ocr_results_" + file_name + "_pages_" + str(page_min + 1) + "_" + str(page_max) + "_textract.csv" if current_loop_page == 0: page_loop_start = 0 else: page_loop_start = current_loop_page progress_bar = tqdm(range(page_loop_start, number_of_pages), unit="pages remaining", desc="Redacting pages") for page_no in progress_bar: handwriting_or_signature_boxes = [] signature_recogniser_results = [] handwriting_recogniser_results = [] page_break_return = False reported_page_number = str(page_no + 1) print("Redacting page:", reported_page_number) # Assuming prepared_pdf_file_paths[page_no] is a PIL image object try: image = prepared_pdf_file_paths[page_no]#.copy() #print("image:", image) except Exception as e: print("Could not redact page:", reported_page_number, "due to:") print(e) continue image_annotations = {"image": image, "boxes": []} pymupdf_page = pymupdf_doc.load_page(page_no) if page_no >= page_min and page_no < page_max: #print("Image is in range of pages to redact") if isinstance(image, str): #print("image is a file path") image = Image.open(image) # Need image size to convert textract OCR outputs to the correct sizes page_width, page_height = image.size # Possibility to use different languages if language == 'en': ocr_lang = 'eng' else: ocr_lang = language # Step 1: Perform OCR. Either with Tesseract, or with AWS Textract if analysis_type == tesseract_ocr_option: word_level_ocr_results = image_analyser.perform_ocr(image) # Combine OCR results line_level_ocr_results, line_level_ocr_results_with_children = combine_ocr_results(word_level_ocr_results) # Import results from json and convert if analysis_type == textract_option: # Convert the image to bytes using an in-memory buffer image_buffer = io.BytesIO() image.save(image_buffer, format='PNG') # Save as PNG, or adjust format if needed pdf_page_as_bytes = image_buffer.getvalue() #json_file_path = output_folder + file_name + "_page_" + reported_page_number + "_textract.json" json_file_path = output_folder + file_name + "_textract.json" if not os.path.exists(json_file_path): text_blocks, new_request_metadata = analyse_page_with_textract(pdf_page_as_bytes, reported_page_number, textract_client) # Analyse page with Textract logging_file_paths.append(json_file_path) request_metadata = request_metadata + "\n" + new_request_metadata wrapped_text_blocks = {"pages":[text_blocks]} # Write the updated existing_data back to the JSON file with open(json_file_path, 'w') as json_file: json.dump(wrapped_text_blocks, json_file, indent=4) # indent=4 makes the JSON file pretty-printed else: # Open the file and load the JSON data print("Found existing Textract json results file.") with open(json_file_path, 'r') as json_file: existing_data = json.load(json_file) # Check if the current reported_page_number exists in the loaded JSON page_exists = any(page['page_no'] == reported_page_number for page in existing_data.get("pages", [])) if not page_exists: # If the page does not exist, analyze again print(f"Page number {reported_page_number} not found in existing data. Analyzing again.") text_blocks, new_request_metadata = analyse_page_with_textract(pdf_page_as_bytes, reported_page_number) # Analyse page with Textract # Check if "pages" key exists, if not, initialize it as an empty list if "pages" not in existing_data: existing_data["pages"] = [] # Append the new page data existing_data["pages"].append(text_blocks) # Write the updated existing_data back to the JSON file with open(json_file_path, 'w') as json_file: json.dump(existing_data, json_file, indent=4) # indent=4 makes the JSON file pretty-printed logging_file_paths.append(json_file_path) request_metadata = request_metadata + "\n" + new_request_metadata else: # If the page exists, retrieve the data text_blocks = next(page['data'] for page in existing_data["pages"] if page['page_no'] == reported_page_number) line_level_ocr_results, handwriting_or_signature_boxes, signature_recogniser_results, handwriting_recogniser_results, line_level_ocr_results_with_children = json_to_ocrresult(text_blocks, page_width, page_height, reported_page_number) # Step 2: Analyze text and identify PII if chosen_redact_entities: redaction_bboxes, comprehend_query_number_new = image_analyser.analyze_text( line_level_ocr_results, line_level_ocr_results_with_children, chosen_redact_comprehend_entities = chosen_redact_comprehend_entities, pii_identification_method = pii_identification_method, comprehend_client=comprehend_client, language=language, entities=chosen_redact_entities, allow_list=allow_list, score_threshold=score_threshold ) comprehend_query_number = comprehend_query_number + comprehend_query_number_new else: redaction_bboxes = [] if analysis_type == tesseract_ocr_option: interim_results_file_path = output_folder + "interim_analyser_bboxes_" + file_name + "_pages_" + str(page_min + 1) + "_" + str(page_max) + ".txt" elif analysis_type == textract_option: interim_results_file_path = output_folder + "interim_analyser_bboxes_" + file_name + "_pages_" + str(page_min + 1) + "_" + str(page_max) + "_textract.txt" # Save decision making process bboxes_str = str(redaction_bboxes) with open(interim_results_file_path, "w") as f: f.write(bboxes_str) # Merge close bounding boxes merged_redaction_bboxes = merge_img_bboxes(redaction_bboxes, line_level_ocr_results_with_children, signature_recogniser_results, handwriting_recogniser_results, handwrite_signature_checkbox) # 3. Draw the merged boxes if is_pdf(file_path) == False: draw = ImageDraw.Draw(image) all_image_annotations_boxes = [] for box in merged_redaction_bboxes: print("box:", box) x0 = box.left y0 = box.top x1 = x0 + box.width y1 = y0 + box.height try: label = box.entity_type except: label = "Redaction" # Directly append the dictionary with the required keys all_image_annotations_boxes.append({ "xmin": x0, "ymin": y0, "xmax": x1, "ymax": y1, "label": label, "color": (0, 0, 0) }) draw.rectangle([x0, y0, x1, y1], fill=fill) # Adjusted to use a list for rectangle image_annotations = {"image": file_path, "boxes": all_image_annotations_boxes} ## Apply annotations with pymupdf else: pymupdf_page, image_annotations = redact_page_with_pymupdf(pymupdf_page, merged_redaction_bboxes, image) # Convert decision process to table decision_process_table = pd.DataFrame([{ 'page': reported_page_number, 'entity_type': result.entity_type, 'start': result.start, 'end': result.end, 'score': result.score, 'left': result.left, 'top': result.top, 'width': result.width, 'height': result.height, 'text': result.text } for result in merged_redaction_bboxes]) all_decision_process_table = pd.concat([all_decision_process_table, decision_process_table]) # Convert to DataFrame and add to ongoing logging table line_level_ocr_results_df = pd.DataFrame([{ 'page': reported_page_number, 'text': result.text, 'left': result.left, 'top': result.top, 'width': result.width, 'height': result.height } for result in line_level_ocr_results]) all_line_level_ocr_results_df = pd.concat([all_line_level_ocr_results_df, line_level_ocr_results_df]) toc = time.perf_counter() time_taken = toc - tic #print("toc - tic:", time_taken) # Break if time taken is greater than max_time seconds if time_taken > max_time: print("Processing for", max_time, "seconds, breaking loop.") page_break_return = True progress.close(_tqdm=progress_bar) tqdm._instances.clear() if is_pdf(file_path) == False: images.append(image) pymupdf_doc = images annotations_all_pages.append(image_annotations) current_loop_page += 1 return pymupdf_doc, all_decision_process_table, logging_file_paths, request_metadata, annotations_all_pages, current_loop_page, page_break_return, all_line_level_ocr_results_df, comprehend_query_number if is_pdf(file_path) == False: images.append(image) pymupdf_doc = images annotations_all_pages.append(image_annotations) current_loop_page += 1 # Break if new page is a multiple of chosen page_break_val if current_loop_page % page_break_val == 0: page_break_return = True progress.close(_tqdm=progress_bar) tqdm._instances.clear() return pymupdf_doc, all_decision_process_table, logging_file_paths, request_metadata, annotations_all_pages, current_loop_page, page_break_return, all_line_level_ocr_results_df, comprehend_query_number return pymupdf_doc, all_decision_process_table, logging_file_paths, request_metadata, annotations_all_pages, current_loop_page, page_break_return, all_line_level_ocr_results_df, comprehend_query_number ### # PIKEPDF TEXT PDF REDACTION ### def get_text_container_characters(text_container:LTTextContainer): if isinstance(text_container, LTTextContainer): characters = [char for line in text_container if isinstance(line, LTTextLine) or isinstance(line, LTTextLineHorizontal) for char in line] return characters return [] def create_text_bounding_boxes_from_characters(char_objects:List[LTChar]) -> Tuple[List[OCRResult], List[LTChar]]: ''' Create an OCRResult object based on a list of pdfminer LTChar objects. ''' line_level_results_out = [] line_level_characters_out = [] #all_line_level_characters_out = [] character_objects_out = [] # New list to store character objects # Initialize variables full_text = "" added_text = "" overall_bbox = [float('inf'), float('inf'), float('-inf'), float('-inf')] # [x0, y0, x1, y1] word_bboxes = [] # Iterate through the character objects current_word = "" current_word_bbox = [float('inf'), float('inf'), float('-inf'), float('-inf')] # [x0, y0, x1, y1] for char in char_objects: character_objects_out.append(char) # Collect character objects if isinstance(char, LTAnno): # Handle space separately by finalizing the word full_text += char.get_text() # Adds space or newline if current_word: # Only finalize if there is a current word word_bboxes.append((current_word, current_word_bbox)) current_word = "" current_word_bbox = [float('inf'), float('inf'), float('-inf'), float('-inf')] # Reset for next word # Check for line break (assuming a new line is indicated by a specific character) if '\n' in char.get_text(): #print("char_anno:", char) # Finalize the current line if current_word: word_bboxes.append((current_word, current_word_bbox)) # Create an OCRResult for the current line line_level_results_out.append(OCRResult(full_text, round(overall_bbox[0], 2), round(overall_bbox[1], 2), round(overall_bbox[2] - overall_bbox[0], 2), round(overall_bbox[3] - overall_bbox[1], 2))) line_level_characters_out.append(character_objects_out) # Reset for the next line character_objects_out = [] full_text = "" overall_bbox = [float('inf'), float('inf'), float('-inf'), float('-inf')] current_word = "" current_word_bbox = [float('inf'), float('inf'), float('-inf'), float('-inf')] continue # Concatenate text for LTChar #full_text += char.get_text() #added_text = re.sub(r'[^\x00-\x7F]+', ' ', char.get_text()) added_text = char.get_text() if re.search(r'[^\x00-\x7F]', added_text): # Matches any non-ASCII character #added_text.encode('latin1', errors='replace').decode('utf-8') added_text = clean_unicode_text(added_text) full_text += added_text # Adds space or newline, removing # Update overall bounding box x0, y0, x1, y1 = char.bbox overall_bbox[0] = min(overall_bbox[0], x0) # x0 overall_bbox[1] = min(overall_bbox[1], y0) # y0 overall_bbox[2] = max(overall_bbox[2], x1) # x1 overall_bbox[3] = max(overall_bbox[3], y1) # y1 # Update current word #current_word += char.get_text() current_word += added_text # Update current word bounding box current_word_bbox[0] = min(current_word_bbox[0], x0) # x0 current_word_bbox[1] = min(current_word_bbox[1], y0) # y0 current_word_bbox[2] = max(current_word_bbox[2], x1) # x1 current_word_bbox[3] = max(current_word_bbox[3], y1) # y1 # Finalize the last word if any if current_word: word_bboxes.append((current_word, current_word_bbox)) if full_text: #print("full_text before:", full_text) if re.search(r'[^\x00-\x7F]', full_text): # Matches any non-ASCII character # Convert special characters to a human-readable format #full_text = full_text.encode('latin1', errors='replace').decode('utf-8') full_text = clean_unicode_text(full_text) #print("full_text:", full_text) line_level_results_out.append(OCRResult(full_text, round(overall_bbox[0],2), round(overall_bbox[1], 2), round(overall_bbox[2]-overall_bbox[0],2), round(overall_bbox[3]-overall_bbox[1],2))) #line_level_characters_out = character_objects_out return line_level_results_out, line_level_characters_out # Return both results and character objects def merge_text_bounding_boxes(analyser_results:CustomImageRecognizerResult, characters:List[LTChar], combine_pixel_dist:int=20, vertical_padding:int=0): ''' Merge identified bounding boxes containing PII that are very close to one another ''' analysed_bounding_boxes = [] if len(analyser_results) > 0 and len(characters) > 0: # Extract bounding box coordinates for sorting bounding_boxes = [] text_out = [] for result in analyser_results: char_boxes = [char.bbox for char in characters[result.start:result.end] if isinstance(char, LTChar)] char_text = [char._text for char in characters[result.start:result.end] if isinstance(char, LTChar)] if char_boxes: # Calculate the bounding box that encompasses all characters left = min(box[0] for box in char_boxes) bottom = min(box[1] for box in char_boxes) right = max(box[2] for box in char_boxes) top = max(box[3] for box in char_boxes) + vertical_padding bounding_boxes.append((bottom, left, result, [left, bottom, right, top], char_text)) # (y, x, result, bbox, text) char_text = "".join(char_text) # Sort the results by y-coordinate and then by x-coordinate bounding_boxes.sort() merged_bounding_boxes = [] current_box = None current_y = None current_result = None current_text = [] for y, x, result, char_box, text in bounding_boxes: #print(f"Considering result: {result}") #print(f"Character box: {char_box}") if current_y is None or current_box is None: current_box = char_box current_y = char_box[1] current_result = result current_text = list(text) #print(f"Starting new box: {current_box}") else: vertical_diff_bboxes = abs(char_box[1] - current_y) horizontal_diff_bboxes = abs(char_box[0] - current_box[2]) #print(f"Comparing boxes: current_box={current_box}, char_box={char_box}, current_text={current_text}, char_text={text}") #print(f"Vertical diff: {vertical_diff_bboxes}, Horizontal diff: {horizontal_diff_bboxes}") if ( vertical_diff_bboxes <= 5 and horizontal_diff_bboxes <= combine_pixel_dist ): #print("box is being extended") current_box[2] = char_box[2] # Extend the current box horizontally current_box[3] = max(current_box[3], char_box[3]) # Ensure the top is the highest current_result.end = max(current_result.end, result.end) # Extend the text range try: current_result.entity_type = current_result.entity_type + " - " + result.entity_type except Exception as e: print("Unable to combine result entity types:") print(e) # Add a space if current_text is not empty if current_text: current_text.append(" ") # Add space between texts current_text.extend(text) #print(f"Latest merged box: {current_box[-1]}") else: merged_bounding_boxes.append( {"text":"".join(current_text),"boundingBox": current_box, "result": current_result}) #print(f"Appending merged box: {current_box}") #print(f"Latest merged box: {merged_bounding_boxes[-1]}") # Reset current_box and current_y after appending current_box = char_box current_y = char_box[1] current_result = result current_text = list(text) #print(f"Starting new box: {current_box}") # After finishing with the current result, add the last box for this result if current_box: merged_bounding_boxes.append({"text":"".join(current_text), "boundingBox": current_box, "result": current_result}) #print(f"Appending final box for result: {current_box}") if not merged_bounding_boxes: analysed_bounding_boxes.extend( {"text":text, "boundingBox": char.bbox, "result": result} for result in analyser_results for char in characters[result.start:result.end] if isinstance(char, LTChar) ) else: analysed_bounding_boxes.extend(merged_bounding_boxes) #print("Analyzed bounding boxes:\n\n", analysed_bounding_boxes) return analysed_bounding_boxes def create_text_redaction_process_results(analyser_results, analysed_bounding_boxes, page_num): decision_process_table = pd.DataFrame() if len(analyser_results) > 0: # Create summary df of annotations to be made analysed_bounding_boxes_df_new = pd.DataFrame(analysed_bounding_boxes) analysed_bounding_boxes_df_text = analysed_bounding_boxes_df_new['result'].astype(str).str.split(",",expand=True).replace(".*: ", "", regex=True) analysed_bounding_boxes_df_text.columns = ["type", "start", "end", "score"] analysed_bounding_boxes_df_new = pd.concat([analysed_bounding_boxes_df_new, analysed_bounding_boxes_df_text], axis = 1) analysed_bounding_boxes_df_new['page'] = page_num + 1 decision_process_table = pd.concat([decision_process_table, analysed_bounding_boxes_df_new], axis = 0).drop('result', axis=1) #print('\n\ndecision_process_table:\n\n', decision_process_table) return decision_process_table def create_annotations_for_bounding_boxes(analysed_bounding_boxes): annotations_on_page = [] for analysed_bounding_box in analysed_bounding_boxes: bounding_box = analysed_bounding_box["boundingBox"] annotation = Dictionary( Type=Name.Annot, Subtype=Name.Square, #Name.Highlight, QuadPoints=[bounding_box[0], bounding_box[3], bounding_box[2], bounding_box[3], bounding_box[0], bounding_box[1], bounding_box[2], bounding_box[1]], Rect=[bounding_box[0], bounding_box[1], bounding_box[2], bounding_box[3]], C=[0, 0, 0], IC=[0, 0, 0], CA=1, # Transparency T=analysed_bounding_box["result"].entity_type, BS=Dictionary( W=0, # Border width: 1 point S=Name.S # Border style: solid ) ) annotations_on_page.append(annotation) return annotations_on_page def redact_text_pdf( filename: str, # Path to the PDF file to be redacted prepared_pdf_image_path: str, # Path to the prepared PDF image for redaction language: str, # Language of the PDF content chosen_redact_entities: List[str], # List of entities to be redacted chosen_redact_comprehend_entities: List[str], allow_list: List[str] = None, # Optional list of allowed entities page_min: int = 0, # Minimum page number to start redaction page_max: int = 999, # Maximum page number to end redaction analysis_type: str = text_ocr_option, # Type of analysis to perform current_loop_page: int = 0, # Current page being processed in the loop page_break_return: bool = False, # Flag to indicate if a page break should be returned annotations_all_pages: List = [], # List of annotations across all pages all_line_level_ocr_results_df: pd.DataFrame = pd.DataFrame(), # DataFrame for OCR results all_decision_process_table: pd.DataFrame = pd.DataFrame(), # DataFrame for decision process table pymupdf_doc: List = [], # List of PyMuPDF documents pii_identification_method: str = "Local", comprehend_query_number:int = 0, comprehend_client="", page_break_val: int = int(page_break_value), # Value for page break max_time: int = int(max_time_value), progress: Progress = Progress(track_tqdm=True) # Progress tracking object ): ''' Redact chosen entities from a PDF that is made up of multiple pages that are not images. Input Variables: - filename: Path to the PDF file to be redacted - prepared_pdf_image_path: Path to the prepared PDF image for redaction - language: Language of the PDF content - chosen_redact_entities: List of entities to be redacted - chosen_redact_comprehend_entities: List of entities to be redacted for AWS Comprehend - allow_list: Optional list of allowed entities - page_min: Minimum page number to start redaction - page_max: Maximum page number to end redaction - analysis_type: Type of analysis to perform - current_loop_page: Current page being processed in the loop - page_break_return: Flag to indicate if a page break should be returned - annotations_all_pages: List of annotations across all pages - all_line_level_ocr_results_df: DataFrame for OCR results - all_decision_process_table: DataFrame for decision process table - pymupdf_doc: List of PyMuPDF documents - pii_identification_method (str, optional): The method to redact personal information. Either 'Local' (spacy model), or 'AWS Comprehend' (AWS Comprehend API). - comprehend_query_number (int, optional): A counter tracking the number of queries to AWS Comprehend. - comprehend_client (optional): A connection to the AWS Comprehend service via the boto3 package. - page_break_val: Value for page break - max_time (int, optional): The maximum amount of time (s) that the function should be running before it breaks. To avoid timeout errors with some APIs. - progress: Progress tracking object ''' if pii_identification_method == "AWS Comprehend" and comprehend_client == "": print("Connection to AWS Comprehend service not found.") return pymupdf_doc, all_decision_process_table, all_line_level_ocr_results_df, annotations_all_pages, current_loop_page, page_break_return, comprehend_query_number tic = time.perf_counter() # Open with Pikepdf to get text lines pikepdf_pdf = Pdf.open(filename) number_of_pages = len(pikepdf_pdf.pages) # Check that page_min and page_max are within expected ranges if page_max > number_of_pages or page_max == 0: page_max = number_of_pages if page_min <= 0: page_min = 0 else: page_min = page_min - 1 print("Page range is",str(page_min + 1), "to", str(page_max)) print("Current_loop_page:", current_loop_page) if current_loop_page == 0: page_loop_start = 0 else: page_loop_start = current_loop_page progress_bar = tqdm(range(current_loop_page, number_of_pages), unit="pages remaining", desc="Redacting pages") #for page_no in range(0, number_of_pages): for page_no in progress_bar: reported_page_number = str(page_no + 1) print("Redacting page:", reported_page_number) # Assuming prepared_pdf_file_paths[page_no] is a PIL image object try: image = prepared_pdf_image_path[page_no]#.copy() #print("image:", image) except Exception as e: print("Could not redact page:", reported_page_number, "due to:") print(e) continue image_annotations = {"image": image, "boxes": []} pymupdf_page = pymupdf_doc.load_page(page_no) if page_min <= page_no < page_max: for page_layout in extract_pages(filename, page_numbers = [page_no], maxpages=1): page_analyser_results = [] page_analysed_bounding_boxes = [] characters = [] annotations_on_page = [] decision_process_table_on_page = pd.DataFrame() page_text_outputs = pd.DataFrame() if analysis_type == text_ocr_option: for n, text_container in enumerate(page_layout): text_container_analyser_results = [] text_container_analysed_bounding_boxes = [] characters = [] if isinstance(text_container, LTTextContainer) or isinstance(text_container, LTAnno): characters = get_text_container_characters(text_container) # Create dataframe for all the text on the page line_level_text_results_list, line_characters = create_text_bounding_boxes_from_characters(characters) # Create page_text_outputs (OCR format outputs) if line_level_text_results_list: # Convert to DataFrame and add to ongoing logging table line_level_text_results_df = pd.DataFrame([{ 'page': page_no + 1, 'text': result.text, 'left': result.left, 'top': result.top, 'width': result.width, 'height': result.height } for result in line_level_text_results_list]) page_text_outputs = pd.concat([page_text_outputs, line_level_text_results_df]) # Initialize batching variables current_batch = "" current_batch_mapping = [] # List of (start_pos, line_index, OCRResult) tuples all_text_line_results = [] # Store results for all lines # First pass: collect all lines into batches for i, text_line in enumerate(line_level_text_results_list): if chosen_redact_entities: if pii_identification_method == "Local": # Process immediately for local analysis text_line_analyser_result = nlp_analyser.analyze( text=text_line.text, language=language, entities=chosen_redact_entities, score_threshold=score_threshold, return_decision_process=True, allow_list=allow_list ) 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: current_batch += " | " start_pos = len(current_batch) current_batch += text_line.text current_batch_mapping.append((start_pos, i, text_line)) # Process batch if approaching 300 characters or last line if len(current_batch) >= 200 or i == len(line_level_text_results_list) - 1: print("length of text for Comprehend:", len(current_batch)) try: response = comprehend_client.detect_pii_entities( Text=current_batch, LanguageCode=language ) except Exception as e: print(e) time.sleep(3) response = comprehend_client.detect_pii_entities( Text=current_batch, LanguageCode=language ) comprehend_query_number += 1 # Process response and map back to original lines if response and "Entities" in response: for entity in response["Entities"]: entity_start = entity["BeginOffset"] entity_end = entity["EndOffset"] # Find which line this entity belongs to for batch_start, line_idx, original_line in current_batch_mapping: batch_end = batch_start + len(original_line.text) # Check if entity belongs to this line if batch_start <= entity_start < batch_end: # Adjust offsets relative to original line relative_start = entity_start - batch_start relative_end = min(entity_end - batch_start, len(original_line.text)) result_text = original_line.text[relative_start:relative_end] if result_text not in allow_list: if entity.get("Type") in chosen_redact_comprehend_entities: # Create adjusted entity adjusted_entity = entity.copy() adjusted_entity["BeginOffset"] = relative_start adjusted_entity["EndOffset"] = relative_end recogniser_entity = recognizer_result_from_dict(adjusted_entity) # Add to results for this line existing_results = next((results for idx, results in all_text_line_results if idx == line_idx), []) if not existing_results: all_text_line_results.append((line_idx, [recogniser_entity])) else: existing_results.append(recogniser_entity) # Reset batch current_batch = "" current_batch_mapping = [] # Second pass: process results for each line for i, text_line in enumerate(line_level_text_results_list): text_line_analyser_result = [] text_line_bounding_boxes = [] # Get results for this line line_results = next((results for idx, results in all_text_line_results if idx == i), []) if line_results: text_line_analyser_result = line_results #print("Analysed text container, now merging bounding boxes") # Merge bounding boxes if very close together text_line_bounding_boxes = merge_text_bounding_boxes(text_line_analyser_result, line_characters[i]) #print("merged bounding boxes") text_container_analyser_results.extend(text_line_analyser_result) text_container_analysed_bounding_boxes.extend(text_line_bounding_boxes) page_analyser_results.extend(text_container_analyser_results) page_analysed_bounding_boxes.extend(text_container_analysed_bounding_boxes) # Annotate redactions on page annotations_on_page = create_annotations_for_bounding_boxes(page_analysed_bounding_boxes) # Make pymupdf page redactions pymupdf_page, image_annotations = redact_page_with_pymupdf(pymupdf_page, annotations_on_page, image) #print("Did redact_page_with_pymupdf function") print("For page number:", page_no, "there are", len(image_annotations["boxes"]), "annotations") # Write logs # Create decision process table decision_process_table_on_page = create_text_redaction_process_results(page_analyser_results, page_analysed_bounding_boxes, current_loop_page) if not decision_process_table_on_page.empty: all_decision_process_table = pd.concat([all_decision_process_table, decision_process_table_on_page]) if not page_text_outputs.empty: page_text_outputs = page_text_outputs.sort_values(["top", "left"], ascending=[False, False]).reset_index(drop=True) all_line_level_ocr_results_df = pd.concat([all_line_level_ocr_results_df, page_text_outputs]) toc = time.perf_counter() time_taken = toc - tic #print("toc - tic:", time_taken) # Break if time taken is greater than max_time seconds if time_taken > max_time: print("Processing for", max_time, "seconds, breaking.") page_break_return = True progress.close(_tqdm=progress_bar) tqdm._instances.clear() annotations_all_pages.append(image_annotations) current_loop_page += 1 return pymupdf_doc, all_decision_process_table, all_line_level_ocr_results_df, annotations_all_pages, current_loop_page, page_break_return, comprehend_query_number annotations_all_pages.append(image_annotations) current_loop_page += 1 # Break if new page is a multiple of 10 if current_loop_page % page_break_val == 0: page_break_return = True progress.close(_tqdm=progress_bar) return pymupdf_doc, all_decision_process_table, all_line_level_ocr_results_df, annotations_all_pages, current_loop_page, page_break_return, comprehend_query_number return pymupdf_doc, all_decision_process_table, all_line_level_ocr_results_df, annotations_all_pages, current_loop_page, page_break_return, comprehend_query_number