diff --git a/refresh.py b/refresh.py
new file mode 100644
index 0000000..b752da7
--- /dev/null
+++ b/refresh.py
@@ -0,0 +1,185 @@
+"""
+This script parses a token.xml file, collects the picURLs within, and replaces
+the URLs to Scryfall cards with up-to-date URLs by querying Scryfall's API.
+"""
+
+from xml.sax import saxutils, make_parser, handler
+from urllib.parse import urlsplit
+from urllib.request import Request, urlopen
+import itertools
+import json
+import sys
+import time
+import os
+import tempfile
+import pathlib
+import shutil
+
+SCRYFALL_MAX_LIST_SIZE = 75
+
+def cards_collection(identifiers):
+    """
+    Get information about a set of cards using the Scryfall API.
+
+    This simply returns a list of dictionaries representing the Card objects as
+    returned by the /card/collection Scryfall API.
+
+    If the list of identifiers is larger than Scryfall's API limit,
+    cards_collection automatically splits the list into smaller chunks and
+    makes multiple requests.
+    """
+
+    start_time = 0
+    n = 0
+    while n < len(identifiers):
+        chunk = identifiers[n:n + SCRYFALL_MAX_LIST_SIZE]
+        print("Requesting chunk {}-{}/{}...".format(n, n + len(chunk), len(identifiers)))
+        n += SCRYFALL_MAX_LIST_SIZE
+
+        payload = json.dumps({'identifiers': chunk}).encode('utf-8')
+        req = Request('https://api.scryfall.com/cards/collection', payload,
+                      headers={'Content-Type': 'application/json'})
+
+        # Rate limiting
+        cur_time = time.time()
+        delta_time = cur_time - start_time
+        if delta_time < 0.1:
+            time.sleep(0.1 - delta_time)
+        start_time = time.time()
+
+        with urlopen(req) as f:
+            list_obj = json.load(f)
+            assert not list_obj.get('has_more', False)
+            assert 'warnings' not in list_obj
+
+            yield from list_obj['data']
+
+def parse_picurl(picurl):
+    """
+    Parse a Scryfall picURL into its components.
+
+    The Scryfall picURL must be in one of those forms:
+
+        - https://c1.scryfall.com/file/scryfall-cards/<version>/<face>/*/*/<uuid>.jpg
+        - https://cards.scryfall.io/<version>/<face>/*/*/<uuid>.jpg
+
+    If it is, a dictionary with keys 'uuid', 'version' and 'face' is returned.
+    Otherwise, an empty dictionary is returned.
+    """
+
+    obj = {}
+    urlinfo = urlsplit(picurl)
+    if urlinfo.netloc == 'c1.scryfall.com':
+        parts = urlinfo.path.split('/')
+        obj['version'] = parts[3]
+        obj['face'] = parts[4]
+        obj['uuid'] = parts[-1].rsplit('.')[0]
+    elif urlinfo.netloc == 'cards.scryfall.io':
+        parts = urlinfo.path.split('/')
+        obj['version'] = parts[1]
+        obj['face'] = parts[2]
+        obj['uuid'] = parts[-1].rsplit('.')[0]
+    return obj
+
+class URLCollector(handler.ContentHandler):
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+
+        self.urls = []
+
+    def startElement(self, name, attrs):
+        if name == 'set' and 'picURL' in attrs:
+            obj = parse_picurl(attrs['picURL'])
+            if obj:
+                assert 'uuid' in obj
+                self.urls.append(obj)
+
+class URLRewriter(handler.LexicalHandler, saxutils.XMLGenerator):
+    def __init__(self, images, **kwargs):
+        super().__init__(**kwargs)
+
+        self._images = images
+        self.started = False
+
+    def startElement(self, name, attrs):
+        self.started = True
+
+        if name == 'set' and 'picURL' in attrs:
+            obj = parse_picurl(attrs['picURL'])
+            if 'uuid' in obj and obj['uuid'] in self._images:
+                new_url = self._images[obj['uuid']][obj['face']][obj['version']]
+                if parse_picurl(new_url) != obj:
+                    raise RuntimeError(
+                        "URL `{}` was rewritten to `{}` that no longer resolves to the same card (hint: update parse_picurl).".format(
+                            attrs['picURL'], new_url))
+                attrs = dict(attrs, picURL=new_url)
+
+        super().startElement(name, attrs)
+
+    def endDocument(self):
+        super().endDocument()
+        self._write('\n')
+
+    def comment(self, content):
+        self.ignorableWhitespace('<!--{}-->{}'.format(content, '' if self.started else '\n'))
+
+def collect_urls(fname):
+    parser = make_parser()
+    uc = URLCollector()
+    parser.setContentHandler(uc)
+    parser.parse(fname)
+    return uc.urls
+
+def rewrite_urls(fname, images, *, out=None):
+    parser = make_parser()
+    ur = URLRewriter(images, out=out, encoding='UTF-8')
+    parser.setContentHandler(ur)
+    parser.setProperty(handler.property_lexical_handler, ur)
+    parser.parse(fname)
+
+def main(fname, *, out=None):
+    urls = collect_urls(fname)
+    identifiers = {obj['uuid'] for obj in urls}
+    identifiers = [{'id': uuid} for uuid in identifiers]
+
+    images = {}
+    for card in cards_collection(identifiers):
+        assert card['id'] not in images
+
+        if 'image_uris' in card:
+            images[card['id']] = {'front': card['image_uris']}
+
+        else:
+            assert 'card_faces' in card and len(card['card_faces']) == 2
+            images[card['id']] = {
+                'front': card['card_faces'][0]['image_uris'],
+                'back': card['card_faces'][1]['image_uris'],
+            }
+
+    rewrite_urls(fname, images, out=out)
+
+if __name__ == '__main__':
+    import argparse
+
+    parser = argparse.ArgumentParser(
+        description='Helper script to refresh scryfall image URLs')
+    parser.add_argument('filename', nargs='?', default='tokens.xml')
+
+    output_group = parser.add_mutually_exclusive_group(required=True)
+    output_group.add_argument('--output', '-o')
+    output_group.add_argument('--inplace', '-i', action='store_true')
+
+    ns = parser.parse_args()
+
+    if ns.inplace:
+        outpath = pathlib.Path(ns.filename)
+    else:
+        outpath = pathlib.Path(ns.output)
+    fd, temppath = tempfile.mkstemp(dir=outpath.parent)
+
+    try:
+        main(ns.filename, out=os.fdopen(fd, mode='w+b'))
+        os.replace(temppath, outpath)
+    except:
+        os.unlink(temppath)
+        raise