e621.py

#!/usr/bin/env python

# usage: e621.py artist1 artist2 etc. to download images of each artist in succession
#    e621.py --update to download new images for existing folders

import urllib, json, sys, os, os.path
from threading import Thread
from Queue import Queue

def e621(artist):
    # you can edit these three variables to configure the script
    # os.getenv("USERPROFILE") automagically determines where your home folder is on Windows
    dir = os.getenv("USERPROFILE") + "/Furry/"
    # number of CPU threads to use
    threads = 4
    # tags to use every time the script is invoked, max 5,
    # if you don't want to use any just leave default as []
    default = ["-vore", "-gore"]

    if artist:
        dir += artist
    else:
        print "Error: no artist specified."
        sys.exit()

    if len(default) > 5:
        print "Error: too many default option tags. Limit is 5."
        sys.exit()

    try:
        verify = json.loads(urllib.urlopen("http://e621.net/tag/index.json?name=" + artist + "&type=1").read())
    except IOError:
        print "Error: site URL not valid."
        sys.exit()
    else:
        if verify:
            print "Downloading images for artist " + artist
        else:
            print "Error: first tag is not an artist."
            sys.exit()

    if not os.path.exists(dir):
        os.makedirs(dir)

    images = []
    for page in xrange(1, 5000):
        url = json.loads(urllib.urlopen("http://e621.net/post/index.json?tags=" + artist + "+" + "+".join(default) + "&page=" + str(page)).read())
        if url:
            for i in url:
                if not images.count(i[u'file_url']):
                    images.append(i[u'file_url'])
        else:
            if page == 1:
                print "Error: no images found for the desired artist."
                sys.exit()
            break

    images.reverse()
    queue = Queue(threads)

    def download(imagelist):
        queue.get()
        for image in imagelist:
            location = dir + "/" + artist + "_" + str(images.index(image) + 1) + "." + image.split(".")[-1]
            if not os.path.isfile(location):
                urllib.urlretrieve("http://e621.net" + image, location)
        queue.task_done()

    for i in [images[i:i+len(images)/threads] for i in xrange(0, len(images), len(images)/threads)]:
        queue.put(Thread(target=download, args=(i,)).start())

    queue.join()
    print str(len(os.listdir(dir))) + " images saved for artist " + artist

if __name__ == "__main__":
    try:
        if sys.argv[1] == "--update":
            for tag in os.listdir(os.getenv("USERPROFILE") + "/Furry/"):
                e621(tag)
        else:
            for tag in sys.argv[1:]:
                e621(tag)
    except IndexError:
        print "Error: no tags specified."
    except WindowsError:
        print "Error: nothing to update."