Spaces:
Sleeping
Sleeping
| import gradio as gr #type: ignore | |
| import pymupdf #type: ignore | |
| from PIL import Image | |
| import os | |
| from functions import get_image_informations | |
| from excel_to_pdf import excel_to_pdf | |
| from dataSchema import * | |
| # import shutil | |
| def Noc_timeSheet_pdf_to_img(pdf_path,dpi: int = 300, quality: int = 95): | |
| pdf_document = pymupdf.open(pdf_path) | |
| output_path="output.jpg" | |
| # Get the first page of the PDF | |
| page = pdf_document.load_page(0) # 0 is the first page | |
| # Convert the page to a pixmap (image) | |
| pix = page.get_pixmap(dpi=dpi) | |
| # Convert the pixmap to a PIL Image and save as JPG | |
| image = Image.frombytes("RGB", [pix.width, pix.height], pix.samples) | |
| width, height = image.size | |
| start_y_total_table = int(height * 0.42) | |
| end_y_first_table = int(height * 0.30) | |
| croped1 = image.crop((0, 0, width, end_y_first_table)) | |
| croped2 = image.crop((0, start_y_total_table, width, height)) | |
| upper_width, upper_height = croped1.size | |
| lower_width, lower_height = croped2.size | |
| combined_image = Image.new('RGB', (upper_width, upper_height + lower_height)) | |
| # Paste the upper image (croped1) on top | |
| combined_image.paste(croped1, (0, 0)) | |
| # Paste the lower image (croped2) below the upper image | |
| combined_image.paste(croped2, (0, upper_height)) | |
| # Save the combined image | |
| combined_image.save(output_path, "JPEG", quality=quality) | |
| def Clauses_in_invoice(pdf_path: str) -> bool: | |
| """ | |
| Extract text from the last page of a PDF. | |
| """ | |
| try: | |
| pdf_document = pymupdf.open(pdf_path) | |
| total_pages = pdf_document.page_count | |
| if total_pages < 2: | |
| print("The PDF has fewer than 2 pages.") | |
| return False | |
| last_page = pdf_document.load_page(total_pages - 1) | |
| text = last_page.get_text() | |
| last_page = text.lower() | |
| if "clauses" in last_page: | |
| return True | |
| else: | |
| return False | |
| except Exception as e: | |
| print(f"error :{e}") | |
| return False | |
| finally: | |
| # Ensure the PDF document is closed | |
| if 'pdf_document' in locals(): | |
| pdf_document.close() | |
| def Clauses_in_invoice_2nd_version(pdf_path: str) -> bool: | |
| """ | |
| Extract text from the last page of a PDF. | |
| """ | |
| try: | |
| pdf_document = pymupdf.open(pdf_path) | |
| total_pages = pdf_document.page_count | |
| if total_pages < 2: | |
| print("The PDF has fewer than 2 pages.") | |
| return False | |
| second_to_last_page = pdf_document.load_page(total_pages - 2) | |
| text = second_to_last_page.get_text() | |
| if "clauses" in text.lower(): | |
| return True | |
| else: | |
| return False | |
| except Exception as e: | |
| print(f"error :{e}") | |
| return False | |
| finally: | |
| # Ensure the PDF document is closed | |
| if 'pdf_document' in locals(): | |
| pdf_document.close() | |
| def Noc_invoice_pdf_to_img(pdf_path: str, folder_path: str, dpi: int = 300, quality: int = 95): | |
| pdf_document = pymupdf.open(pdf_path) | |
| folder_path = folder_path.rstrip(os.sep) | |
| os.makedirs(folder_path, exist_ok=True) | |
| pdf_name = os.path.splitext(os.path.basename(pdf_path))[0] | |
| total_pages = pdf_document.page_count | |
| image_paths = [] | |
| for page_num in range(total_pages): | |
| page = pdf_document.load_page(page_num) | |
| pix = page.get_pixmap(dpi=dpi) | |
| image = Image.frombytes("RGB", [pix.width, pix.height], pix.samples) | |
| output_path = os.path.join(folder_path, f"{pdf_name}_page_{page_num + 1}.jpg") | |
| image.save(output_path, "JPEG", quality=quality) | |
| image_paths.append(output_path) | |
| pdf_document.close() | |
| return image_paths | |
| def delete_images(image_paths): | |
| # Iterate through the list of image paths | |
| for image_path in image_paths: | |
| try: | |
| # Check if the file exists before attempting to delete | |
| if os.path.exists(image_path): | |
| os.remove(image_path) | |
| print(f"Deleted: {image_path}") | |
| else: | |
| print(f"File not found: {image_path}") | |
| except Exception as e: | |
| print(f"Error deleting {image_path}: {e}") | |
| def noc_invoice_extraction(pdf_path: str, folder_path): | |
| image_paths = Noc_invoice_pdf_to_img(pdf_path, folder_path) | |
| data = {} | |
| result = get_image_informations(image_paths[0], invoice_first_page_prompt, Noc_PurchaseOrder_information_parser) | |
| data.update(result) | |
| result = get_image_informations(image_paths[1], invoice_item_page1_prompt, Noc_PurchaseOrder_item1_parser) | |
| data.update(result) | |
| if Clauses_in_invoice(pdf_path): | |
| for pic in range(len(image_paths) - 4): | |
| new_item = get_image_informations(image_paths[pic + 2], invoice_item_pages_prompt, Noc_PurchaseOrder_items_parser) | |
| for item in new_item["items"]: | |
| data["items"].append(item) | |
| result = get_image_informations(image_paths[-2], invoice_total_page_prompt, Noc_PurchaseOrder_total_parser) | |
| data.update(result) | |
| result = get_image_informations(image_paths[-1], invoice_clauses_page_prompt, Noc_PurchaseOrder_clauses_parser) | |
| data.update(result) | |
| delete_images(image_paths) | |
| return data | |
| else: | |
| for pic in range(len(image_paths) - 3): | |
| new_item = get_image_informations(image_paths[pic + 2], invoice_item_pages_prompt, Noc_PurchaseOrder_items_parser) | |
| for item in new_item["items"]: | |
| data["items"].append(item) | |
| result = get_image_informations(image_paths[-1], invoice_total_page_prompt, Noc_PurchaseOrder_total_parser) | |
| data.update(result) | |
| delete_images(image_paths) | |
| return data | |
| def pdf_to_img(pdf_path, dpi: float = 300, quality: float = 95): | |
| pdf_document = pymupdf.open(pdf_path) | |
| page = pdf_document.load_page(0) # Load the first page | |
| output_path = "output.jpg" | |
| # Convert the page to a pixmap (image) | |
| pix = page.get_pixmap(dpi=dpi) | |
| # Convert the pixmap to a PIL Image | |
| image = Image.frombytes("RGB", [pix.width, pix.height], pix.samples) | |
| image.save(output_path, "JPEG",quality=quality) | |
| def process_file(file, option): | |
| if file is None: | |
| return "Please upload a PDF or image file." | |
| try: | |
| save_dir = "uploaded_files" | |
| os.makedirs(save_dir, exist_ok=True) # Create the directory if it doesn't exist | |
| file_path = file.name | |
| file_extension = os.path.splitext(file_path)[1].lower() | |
| print(file_extension) | |
| if file_extension in ['.pdf']: | |
| # Process PDF files | |
| if option == "Noc_timesheet_residential_old": | |
| Noc_timeSheet_pdf_to_img(file_path) | |
| print("here 2") | |
| result = get_image_informations("output.jpg", Noc_Res_timesheet_prompt, Noc_Res_timeSheet_parser) | |
| return result | |
| elif option == "Noc_timesheet_rotational_old": | |
| Noc_timeSheet_pdf_to_img(file_path) | |
| result = get_image_informations("output.jpg", Noc_Rot_timesheet_prompt, Noc_Rot_timeSheet_parser) | |
| return result | |
| elif option == "Noc_PO": | |
| result = noc_invoice_extraction(file_path, save_dir) | |
| return result | |
| elif option =="Noc_timesheet_new": | |
| pdf_to_img(file_path) | |
| result = get_image_informations("output.jpg", Noc_timesheet_prompt, Noc_timesheet_parser_v1) | |
| return result | |
| elif option == "Noc_invoice": | |
| pdf_to_img(file_path) | |
| result = get_image_informations("output.jpg", Noc_invoice_prompt, Noc_invoice_parser_v1) | |
| return result | |
| elif file_extension in ['.jpg', '.jpeg', '.png']: | |
| # Process image files directly | |
| if option == "Noc_timesheet_residential_old": | |
| result = get_image_informations(file_path, Noc_Res_timesheet_prompt, Noc_Res_timeSheet_parser) | |
| return result | |
| elif option == "Noc_timesheet_rotational_old": | |
| result = get_image_informations(file_path, Noc_Rot_timesheet_prompt, Noc_Rot_timeSheet_parser) | |
| return result | |
| elif option == "Noc_PO": | |
| # For invoice images, we assume it's a single page | |
| result = get_image_informations(file_path, invoice_first_page_prompt, Noc_PurchaseOrder_information_parser) | |
| return result | |
| elif option =="Noc_timesheet_new": | |
| result = get_image_informations(file_path, Noc_timesheet_prompt, Noc_timesheet_parser_v1) | |
| return result | |
| elif option == "Noc_invoice": | |
| result = get_image_informations(file_path, Noc_invoice_prompt, Noc_invoice_parser_v1) | |
| return result | |
| elif file_extension in ['.xls','.xlsx']: | |
| if option == "Noc_timesheet_residential_old": | |
| Noc_timeSheet_pdf_to_img(file_path) | |
| print("here 2") | |
| result = get_image_informations("output.jpg", Noc_Res_timesheet_prompt, Noc_Res_timeSheet_parser) | |
| return result | |
| elif option == "Noc_timesheet_rotational_old": | |
| Noc_timeSheet_pdf_to_img(file_path) | |
| result = get_image_informations("output.jpg", Noc_Rot_timesheet_prompt, Noc_Rot_timeSheet_parser) | |
| return result | |
| elif option == "Noc_PO": | |
| result = noc_invoice_extraction(file_path, save_dir) | |
| return result | |
| elif option =="Noc_timesheet_new": | |
| pdf_to_img(file_path) | |
| result = get_image_informations("output.jpg", Noc_timesheet_prompt, Noc_timesheet_parser_v1) | |
| return result | |
| elif option == "Noc_invoice": | |
| pdf_to_img(file_path) | |
| result = get_image_informations("output.jpg", Noc_invoice_prompt, Noc_invoice_parser_v1) | |
| return result | |
| else: | |
| return "Unsupported file type. Please upload a PDF or image file." | |
| except Exception as e: | |
| return f"An error occurred: {e}" | |
| # Define the Gradio interface | |
| demo = gr.Interface( | |
| fn=process_file, | |
| inputs=[ | |
| gr.File(label="Upload PDF,Image or Excel"), # File upload input | |
| gr.Radio(["Noc_timesheet_new","Noc_invoice","Noc_timesheet_residential_old", "Noc_timesheet_rotational_old", "Noc_PO"], label="Choose an option") # Radio buttons for options | |
| ], | |
| outputs="text", # Text output | |
| title="Document Processor", | |
| description="Upload a PDF or image and choose an option to process the content." | |
| ) | |
| with gr.Blocks() as app: | |
| demo.render() | |
| gr.Markdown("### Examples") # Section title | |
| with gr.Row(): | |
| gr.Image("TS.png", label="NOC timesheet example") | |
| gr.Image("invoice.png", label="NOC invoice example") | |
| app.launch() |