app.py

import gradio as gr
import PyPDF2
from reportlab.lib.pagesizes import letter
from reportlab.pdfgen import canvas
import random
import string
import io

def generate_random_gibberish(length=1000, line_length=100):
    """Generate random gibberish consisting of alphanumeric characters and symbols, split into lines of 100 chars."""
    chars = string.ascii_letters + string.digits + string.punctuation
    gibberish_chars = []
    
    i = 0
    while i < length:
        # Generate word length between 3-5 characters
        word_length = random.randint(3, 5)
        
        # Generate a "word" of random characters
        word = ''.join(random.choice(chars) for _ in range(word_length))
        gibberish_chars.append(word)
        
        # Add space after the word (except for the last word)
        if i + word_length < length:
            gibberish_chars.append(' ')
        
        i += word_length + 1  # +1 for the space
    
    # Join all characters and words
    gibberish = ''.join(gibberish_chars)
    
    # Split the gibberish into lines of specified length
    return [gibberish[i:i+line_length] for i in range(0, len(gibberish), line_length)]

def add_invisible_text_to_pdf(input_pdf_bytes, num_pages_to_modify=5):
    import tempfile
    import os
    
    # Create a file-like object from the uploaded PDF bytes
    input_pdf_path = io.BytesIO(input_pdf_bytes)
    
    reader = PyPDF2.PdfReader(input_pdf_path)
    writer = PyPDF2.PdfWriter()

    # Loop through the pages of the existing PDF
    for page_num in range(len(reader.pages)):
        page = reader.pages[page_num]

        # If this page is in the set of pages we want to modify
        if page_num < num_pages_to_modify:
            packet = io.BytesIO()

            # Create a new PDF to overlay on the existing one
            c = canvas.Canvas(packet, pagesize=letter)

            # Set invisible text color (white)
            c.setFillColorRGB(1, 1, 1, 0)  # White color (invisible on white background)
            c.setFont("Helvetica", 6)  # Use a smaller font for better coverage
            width, height = letter  # Get the page dimensions

            # Generate enough gibberish to cover the entire page
            gibberish_lines = generate_random_gibberish(length=10000, line_length=100)
            y_position = height - 10  # Start near the top of the page

            # Loop through the text and write it line by line
            for line in gibberish_lines:
                if y_position < 0:  # If we go past the bottom of the page, stop
                    break
                c.drawString(10, y_position, line)  # Draw the text starting from left to right
                y_position -= 8  # Move down for the next line

            c.save()

            # Move the "packet" to the start of the file and merge it with the original PDF
            packet.seek(0)
            overlay_pdf = PyPDF2.PdfReader(packet)
            overlay_page = overlay_pdf.pages[0]

            # Merge the overlay page with the original page
            page.merge_page(overlay_page)

        # Add the (modified or original) page to the writer
        writer.add_page(page)

    # Save the new PDF to a temporary file
    with tempfile.NamedTemporaryFile(delete=False, suffix='.pdf') as tmp_file:
        writer.write(tmp_file)
        tmp_file_path = tmp_file.name

    return tmp_file_path

# Define Gradio interface
def gradio_app(input_pdf, num_pages_to_modify):
    output_pdf = add_invisible_text_to_pdf(input_pdf, num_pages_to_modify)
    return output_pdf

# Create the Gradio interface
iface = gr.Interface(
    fn=gradio_app,
    inputs=[
        gr.File(label="Upload PDF", type="binary"),
        gr.Slider(minimum=1, maximum=20, value=5, label="Number of Pages to Modify")
    ],
    outputs=gr.File(label="Output PDF"),
    title="PDF Gibberish Inserter",
    # description="Upload a PDF and this app will add invisible gibberish text to the first few pages. You can choose how many pages to modify."
)

# Launch the app
iface.launch()