Ocr_check / app.py
cdcvd's picture
Upload app.py
8956ec9 verified
raw
history blame
3.65 kB
import os
import sys
from PIL import Image, ImageOps, ImageChops
import io
import fitz # PyMuPDF
from docx import Document
from rembg import remove
import gradio as gr
def create_output_folder(file_path):
base_name = os.path.splitext(os.path.basename(file_path))[0]
output_folder = os.path.join(os.path.dirname(file_path), base_name)
if not os.path.exists(output_folder):
os.makedirs(output_folder)
return output_folder
def convert_image_to_jpeg(input_path, output_folder):
image = Image.open(input_path)
image = image.convert('RGB')
output_path = os.path.join(output_folder, f"{os.path.splitext(os.path.basename(input_path))[0]}.jpg")
image.save(output_path, 'JPEG')
return output_path
def trim_whitespace(image):
gray_image = ImageOps.grayscale(image)
inverted_image = ImageChops.invert(gray_image)
bbox = inverted_image.getbbox()
trimmed_image = image.crop(bbox)
return trimmed_image
def convert_pdf_to_images(pdf_path, output_folder, zoom=2):
pdf_document = fitz.open(pdf_path)
name_with_extension = os.path.basename(pdf_path)
name = os.path.splitext(name_with_extension)[0]
output_paths = []
for page_num in range(len(pdf_document)):
page = pdf_document.load_page(page_num)
matrix = fitz.Matrix(zoom, zoom)
pix = page.get_pixmap(matrix=matrix)
image = Image.frombytes("RGB", [pix.width, pix.height], pix.samples)
trimmed_image = trim_whitespace(image)
output_path = os.path.join(output_folder, f"{name}_page_{page_num + 1}.jpg")
trimmed_image.save(output_path, 'JPEG')
output_paths.append(output_path)
return output_paths
def convert_docx_to_jpeg(docx_path, output_folder):
document = Document(docx_path)
output_paths = []
for i, image_shape in enumerate(document.inline_shapes, start=1):
image_stream = image_shape.image.blob
image = Image.open(io.BytesIO(image_stream))
output_path = os.path.join(output_folder, f"{os.path.splitext(os.path.basename(docx_path))[0]}_page_{i}.jpg")
image.save(output_path, 'JPEG')
output_paths.append(output_path)
return output_paths
def process_file(input_file):
file_extension = os.path.splitext(input_file.name)[1].lower()
output_folder = create_output_folder(input_file.name)
if file_extension in ['.png', '.jpeg', '.jpg', '.bmp', '.gif']:
output_path = convert_image_to_jpeg(input_file.name, output_folder)
return remove_background(output_path)
elif file_extension == '.pdf':
image_paths = convert_pdf_to_images(input_file.name, output_folder)
return [remove_background(path) for path in image_paths]
elif file_extension in ['.docx', '.doc']:
image_paths = convert_docx_to_jpeg(input_file.name, output_folder)
return [remove_background(path) for path in image_paths]
else:
return "File format not supported."
def remove_background(image_path):
input_image = Image.open(image_path)
output_image = remove(input_image)
output_path = image_path.replace('.jpg', '_no_bg.png')
output_image.save(output_path, 'PNG')
return output_path
def gradio_interface(input_file):
return process_file(input_file)
iface = gr.Interface(
fn=gradio_interface,
inputs=gr.inputs.File(label="Upload Word, PDF, or Image"),
outputs=gr.outputs.Image(type="file", label="Processed Image(s)"),
title="Document to Image Converter with Background Removal"
)
if __name__ == "__main__":
iface.launch()