import streamlit as st import os import img2pdf from pypdf import PdfWriter, PdfReader import io import subprocess import sys import re # Define the output directory within the app folder of the project # Assumes the script is in /path/to/project/app/app_pdf_compiler.py # The output will be in /path/to/project/app/ # Get the directory where the script is located SCRIPT_DIR = os.path.dirname(__file__) # Assume the 'app' folder is where the script is, and project root is one level up APP_FOLDER = SCRIPT_DIR OUTPUT_PDF_NAME = "compiled_document.pdf" OUTPUT_PDF_PATH = os.path.abspath(os.path.join(APP_FOLDER, OUTPUT_PDF_NAME)) # Ensure the app folder exists (it should, as the script is in it) # os.makedirs(APP_FOLDER, exist_ok=True) # Not strictly necessary if script is in APP_FOLDER def natural_sort_key(s): """ Natural sort key that handles numeric substrings so that 'file2' comes before 'file10'. """ return [int(text) if text.isdigit() else text.lower() for text in re.split(r'(\d+)', s)] def compile_pdfs_and_images(folder_path, output_path): """ Compiles all supported files (PDF, PNG, JPG, JPEG) from a given folder into a single PDF file, sorted by filename. Scales all pages to a consistent width while maintaining aspect ratio. """ allowed_extensions = ('.pdf', '.png', '.jpg', '.jpeg') files_to_process = [] try: for item in os.listdir(folder_path): if os.path.isfile(os.path.join(folder_path, item)): if item.lower().endswith(allowed_extensions): files_to_process.append(item) except FileNotFoundError: st.error(f"Error: Folder not found at '{folder_path}'. Please check the path.") return False, 0 except PermissionError: st.error(f"Error: Permission denied to access folder '{folder_path}'.") return False, 0 except Exception as e: st.error(f"An unexpected error occurred while listing files: {e}") return False, 0 if not files_to_process: st.warning("No compatible files (PDF, PNG, JPG, JPEG) found in the selected folder.") return False, 0 # Sort files by name in ascending natural order (handles numbers in filenames) files_to_process.sort(key=natural_sort_key) pdf_writer = PdfWriter() processed_file_count = 0 total_files = len(files_to_process) # Define a target width for all pages. 2000px at 96 DPI is ~1500 points. TARGET_WIDTH = 1500.0 st.info(f"Found {total_files} files to process. Starting compilation...") progress_bar = st.progress(0) status_text = st.empty() for i, filename in enumerate(files_to_process): file_path = os.path.join(folder_path, filename) status_text.text(f"Processing file {i+1}/{total_files}: {filename}") try: if filename.lower().endswith('.pdf'): reader = PdfReader(file_path) for page in reader.pages: original_width = page.mediabox.width if original_width > 0: scale_factor = TARGET_WIDTH / float(original_width) page.scale(sx=scale_factor, sy=scale_factor) pdf_writer.add_page(page) elif filename.lower().endswith(('.png', '.jpg', '.jpeg')): # Convert image to PDF bytes pdf_bytes = img2pdf.convert(file_path) # Read the converted PDF img_pdf_stream = io.BytesIO(pdf_bytes) img_pdf_reader = PdfReader(img_pdf_stream) if img_pdf_reader.pages: page = img_pdf_reader.pages[0] original_width = page.mediabox.width if original_width > 0: scale_factor = TARGET_WIDTH / float(original_width) page.scale(sx=scale_factor, sy=scale_factor) pdf_writer.add_page(page) else: st.warning(f"Could not convert image {filename} to a PDF page. Skipping.") continue # Skip this file processed_file_count += 1 except Exception as e: st.error(f"Could not process file {filename}: {e}") progress_bar.progress((i + 1) / total_files) status_text.text("Compilation finished.") if processed_file_count == 0 and total_files > 0: st.error("No files were successfully processed. Output PDF not created.") return False, 0 try: with open(output_path, "wb") as f_out: pdf_writer.write(f_out) return True, processed_file_count except Exception as e: st.error(f"Error writing compiled PDF: {e}") return False, processed_file_count st.set_page_config(layout="wide") st.title("📄 PDF and Image Compiler") # Initialize session state for folder path if it doesn't exist if 'folder_path' not in st.session_state: st.session_state.folder_path = "" st.markdown(""" Select a folder containing PDF, PNG, JPG, or JPEG files. This app will sort them by name (ascending) and compile them into a single PDF document. The output file will be named `compiled_document.pdf` and saved in the same folder as this script (the `app` folder). """) # Function to open folder dialog using helper script and update session state def select_folder(): try: # Determine the path to the helper script (assuming it's in the same 'app' directory) helper_script_path = os.path.join(os.path.dirname(__file__), "folder_picker_helper.py") # Run the helper script as a subprocess # Use sys.executable to ensure it runs with the same Python interpreter process = subprocess.Popen( [sys.executable, helper_script_path], stdout=subprocess.PIPE, stderr=subprocess.PIPE, text=True ) stdout, stderr = process.communicate(timeout=30) # Added timeout if process.returncode == 0 and stdout: folder_selected = stdout.strip() if folder_selected: st.session_state.folder_path = folder_selected else: # User might have cancelled the dialog, which is not an error # st.session_state.folder_path = "" # Or keep previous selection pass elif stderr: st.error(f"Error selecting folder: {stderr}") # else: # No stdout, no stderr, but also no success - could be cancellation # st.warning("Folder selection was cancelled or failed without error message.") except subprocess.TimeoutExpired: st.error("Folder selection timed out.") except FileNotFoundError: st.error(f"Error: The helper script 'folder_picker_helper.py' was not found in the app directory.") except Exception as e: st.error(f"An unexpected error occurred during folder selection: {e}") # Button to trigger folder selection if st.button("Browse and Select Folder", key="browse_folder"): select_folder() # Display the selected folder path if st.session_state.folder_path: st.info(f"Selected folder: `{st.session_state.folder_path}`") else: st.warning("No folder selected yet. Please select a folder to proceed.") # Compile button - disabled if no folder is selected if st.button("Compile to PDF", type="primary", disabled=not st.session_state.folder_path): if st.session_state.folder_path: # Redundant check due to 'disabled' but good for clarity if os.path.isdir(st.session_state.folder_path): st.markdown("---") success, count = compile_pdfs_and_images(st.session_state.folder_path, OUTPUT_PDF_PATH) if success: st.success(f"Successfully compiled {count} files into '{os.path.basename(OUTPUT_PDF_PATH)}' (located at: {OUTPUT_PDF_PATH}).") try: with open(OUTPUT_PDF_PATH, "rb") as pdf_file: st.download_button( label="Download Compiled PDF", data=pdf_file, file_name=OUTPUT_PDF_NAME, mime="application/pdf" ) except PermissionError: st.error( f"PermissionError: Could not open the compiled PDF at '{OUTPUT_PDF_PATH}' for download. " f"On macOS, this might be due to missing 'Full Disk Access' for your Terminal or Python. " f"Please check System Settings > Privacy & Security > Full Disk Access. " f"The file was created, but the app couldn't read it back for the download button." ) except Exception as e: st.error(f"An unexpected error occurred while preparing the PDF for download: {e}") elif count == 0 and not success: # Handles cases where no files were found or folder error pass # Messages are already shown by compile_pdfs_and_images else: st.error("PDF compilation failed. Check messages above.") else: st.error("The provided path is not a valid directory. Please enter a correct folder path.") else: st.warning("Please enter a folder path.") st.markdown("---") st.caption("Ensure the folder path is accessible by this application.")