Last active
July 10, 2023 14:04
-
-
Save lebedov/cefbec588c2b2bd0251ba505bd8bf933 to your computer and use it in GitHub Desktop.
How to call pdfbox's API with JPype.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| #!/usr/bin/env python3 | |
| """ | |
| How to call pdfbox's API with JPype. | |
| """ | |
| import pathlib | |
| import pkg_resources | |
| import re | |
| import urllib.request | |
| import appdirs | |
| import jpype | |
| import jpype.imports | |
| import numpy as np | |
| # Replace with path to pdfbox jar file: | |
| a = appdirs.AppDirs('python-pdfbox') | |
| cache_dir = pathlib.Path(a.user_cache_dir) | |
| file_list = list(cache_dir.glob('pdfbox-app-*.jar')) | |
| def f(s): | |
| v = re.search('pdfbox-app-([\w\.\-]+)\.jar', s.name).group(1) | |
| return pkg_resources.parse_version(v) | |
| jpype.addClassPath(sorted(file_list, key=f)[-1]) | |
| if not jpype.isJVMStarted(): | |
| jpype.startJVM(jpype.getDefaultJVMPath(), '-Djava.awt.headless=true', convertStrings=False) | |
| from java.awt.image import BufferedImage | |
| from java.io import File | |
| from org.apache.pdfbox.pdmodel import PDDocument | |
| from org.apache.pdfbox.rendering import ImageType, PDFRenderer | |
| def extract_images(in_file, pages=None, dpi=72): | |
| """ | |
| Extract pages of PDF file as images. | |
| Parameters | |
| ---------- | |
| in_file : str | |
| Path to input PDF file. | |
| pages : iterable | |
| Numbers of pages to extract (0-indexed). If None, return all pages. | |
| dpi : int | |
| Resolution at which to render output images. | |
| Returns | |
| ------- | |
| output : list of numpy.ndarray | |
| PDF pages rendered into RGB numpy arrays. | |
| """ | |
| doc = PDDocument.load(File(in_file)) | |
| pdf_renderer = PDFRenderer(doc) | |
| output = [] | |
| if pages == None: | |
| pages = range(doc.getNumberOfPages()) | |
| for i in pages: | |
| im = pdf_renderer.renderImageWithDPI(i, dpi) | |
| h = im.getHeight() | |
| w = im.getWidth() | |
| # Retrieve data as numpy array of RGB values packed into int32: | |
| data = im.getRGB(0, 0, w, h, None, 0, w)[:] | |
| # Separate RGB channels, return as array of bytes: | |
| output.append(np.frombuffer(memoryview(data), np.uint8).reshape(h, w, 4)[..., :3]) | |
| return output | |
| if __name__ == '__main__': | |
| import os | |
| import tempfile | |
| import urllib | |
| # Download sample multipage PDF: | |
| data = urllib.request.urlopen('https://researchtorevenue.files.wordpress.com/2015/04/1r41ai10801601_fong.pdf').read() | |
| fd, name = tempfile.mkstemp() | |
| f = open(name, 'wb') | |
| f.write(data) | |
| f.close() | |
| result = extract_images(name) | |
| os.unlink(name) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
I know this is just an example, but for anyone copying this code, note that using a generator (yield expression) would be better in terms of memory management than stacking all bitmaps in a list.