Spaces:
Running
Running
| import fitz | |
| from PIL import Image | |
| from io import BytesIO | |
| def create_test_pdf(): | |
| doc = fitz.open() | |
| p = doc.new_page() | |
| img = Image.new('RGBA', (500, 500), color=(255, 0, 0, 128)) | |
| buf = BytesIO() | |
| img.save(buf, format='PNG') | |
| p.insert_image(fitz.Rect(0,0,100,100), stream=buf.getvalue()) | |
| doc.save("test_img.pdf") | |
| doc.close() | |
| def compress(): | |
| doc = fitz.open("test_img.pdf") | |
| img_info = doc[0].get_images()[0] | |
| xref = img_info[0] | |
| smask = img_info[1] | |
| base_image = doc.extract_image(xref) | |
| image_bytes = base_image["image"] | |
| pil_img = Image.open(BytesIO(image_bytes)) | |
| # Resize | |
| pil_img = pil_img.resize((250, 250), Image.Resampling.LANCZOS) | |
| if pil_img.mode == 'RGBA' and smask > 0: | |
| print("Handling SMask!") | |
| r, g, b, a = pil_img.split() | |
| rgb_img = Image.merge("RGB", (r, g, b)) | |
| rgb_io = BytesIO() | |
| rgb_img.save(rgb_io, format='JPEG', quality=30) | |
| a_io = BytesIO() | |
| a.save(a_io, format='JPEG', quality=30) # Grayscale JPEG for mask | |
| # Update main image | |
| doc.update_stream(xref, rgb_io.getvalue(), compress=0) | |
| doc.xref_set_key(xref, "Width", "250") | |
| doc.xref_set_key(xref, "Height", "250") | |
| doc.xref_set_key(xref, "Filter", "/DCTDecode") | |
| doc.xref_set_key(xref, "ColorSpace", "/DeviceRGB") | |
| doc.xref_set_key(xref, "DecodeParms", "null") | |
| # Update smask | |
| doc.update_stream(smask, a_io.getvalue(), compress=0) | |
| doc.xref_set_key(smask, "Width", "250") | |
| doc.xref_set_key(smask, "Height", "250") | |
| doc.xref_set_key(smask, "Filter", "/DCTDecode") | |
| doc.xref_set_key(smask, "ColorSpace", "/DeviceGray") | |
| doc.xref_set_key(smask, "DecodeParms", "null") | |
| doc.save("test_img_compressed.pdf", garbage=4, deflate=True) | |
| doc.close() | |
| create_test_pdf() | |
| compress() | |