Automatic file converting from pdf to jpg

pdf2jpg.py

# coding: utf-8

import os
import os.path
import fnmatch
import subprocess
import argparse

# Parse the arguments
parser = argparse.ArgumentParser(description='Tile the first num-pages horizontally into a JPEG image.')
parser.add_argument("-paper_dir", type=str, default='/Users/yosssshi/Documents/projects--code-conference/project--citeseeing/src/18_CiteSeeing/data/CVPR2016/pdf', help='Path to PDF')
args = parser.parse_args()

def execute(root_path):
    for dirpath, _, filenames in os.walk(root_path):
        for filename in filenames:
            if fnmatch.fnmatch(filename, u"*.pdf"):
                org_path = os.path.join(dirpath, filename)
                jpg_path = org_path.replace("pdf", "jpg")

                print("convert {0} to {1}".format(org_path, jpg_path))

                if os.path.exists(jpg_path[0:len(jpg_path)-4]):
                    print("dir exists.")
                else:
                    os.mkdir(jpg_path[0:len(jpg_path)-4])
                
                #image_path = os.path.join(jpg_path[0:len(jpg_path)-4],filename.replace("pdf", "jpg")) 
                image_path = os.path.join(jpg_path[0:len(jpg_path)-4],filename.replace(".pdf", "")) 

                print("org_path: ", org_path)
                print("image_path: ", image_path)

                #if subprocess.call(["convert", "-density", "300", "-trim", org_path, image_path]) != 0:
                #if subprocess.call(["convert", "-density", "500", org_path, image_path]) != 0: # Ghostscript
                if subprocess.call(["gs", "-dBATCH", "-dNOPAUSE", "-sDEVICE=jpeg", "-r500", "-sOutputFile={0}_%01d.jpg".format(image_path), org_path]) != 0:
                    print ("failed: {0}".format(org_path))

if __name__ == '__main__':
    execute(args.paper_dir)