| #!/usr/bin/env python | |
| # -*- coding: utf-8 -*- | |
| import os | |
| from bs4 import BeautifulSoup | |
| import requests | |
| import urllib | |
| import argparse | |
| ap = argparse.ArgumentParser() | |
| ap.add_argument('url', type=str) | |
| ap.add_argument('dir', type=str, default="art") | |
| args = ap.parse_args() | |
| url = "http://gatherer.wizards.com/Pages/Search/Default.aspx?action=advanced&color=+[U]" | |
| def get_no_of_pages(url): | |
| r = requests.get(url) | |
| soup = BeautifulSoup(r.text, 'html.parser') | |
| for x in soup.find_all('div', {"class", "paging"}): | |
| for a in x.find_all('a'): | |
| lastpage = a["href"] | |
| lastpageid = lastpage.split('=')[1].split("&")[0] | |
| return int(lastpageid) | |
| def fetch_art_from_page(url, dirname): | |
| r = requests.get(url) | |
| soup = BeautifulSoup(r.text, 'html.parser') | |
| for r in soup.find_all('td', {"class", "leftCol"}): | |
| x = r.find('a') | |
| h = x["href"] | |
| # check if interesting | |
| if not "multiverseid" in h: | |
| continue | |
| # check if cardname is there | |
| try: | |
| title = x.find('img')["alt"].replace("/", "") | |
| except: | |
| continue | |
| cardid = h.split('=')[1] | |
| filename = dirname + '/' + title + '.jpg' | |
| arturl = 'http://gatherer.wizards.com/Handlers/Image.ashx?multiverseid=' + cardid + '&type=card' | |
| print("Downloading %s..." % title) | |
| urllib.urlretrieve(arturl, filename) | |
| def main(args=args): | |
| if not os.path.isdir(args.dir): | |
| return False | |
| page = 0 | |
| pages = get_no_of_pages(args.url) | |
| while page <= pages: | |
| print("Fetching page %s of %s" % (page, pages)) | |
| fetch_art_from_page(args.url + '&page=' + str(page), args.dir) | |
| page += 1 | |
| if __name__ == '__main__': | |
| main(args) |
pip3.9 install BeautifulSoup4
for example download all lands
mkdir lands
./archivist.py 'http://gatherer.wizards.com/Pages/Search/Default.aspx?type=+[Land]' lands/
or download all green cards from Urza's Saga
mkdir urzas-saga-green
./archivist.py http://gatherer.wizards.com/Pages/Search/Default.aspx?color=%7C[G]&set=[%22Urza's%20Saga%22] urzas-saga-green/