@@ -12,7 +12,9 @@ from urllib import urlretrieve from xmlrpclib import ServerProxy from time import sleep import feedparser +import simplejson as json import re +from pprint import pprint dl = open('%s/downloaded' % RSSNZB_DIR, 'r') downloaded = [] @@ -20,28 +22,41 @@ for line in dl.readlines(): downloaded.append(line[:-1]) dl.close() -def download_nzbs(url, hella): - feed = feedparser.parse(url) - for entry in feed['entries']: - if entry['report_progress'] == 'Report is complete' and not entry['report_id'] in downloaded: - print 'Downloading %s' % (entry.title) - hella.enqueuenewzbin(int(entry['report_id'])) - file('%s/downloaded' % RSSNZB_DIR, 'a').write('%s\n' % entry['report_id']) - sleep(1) - #nzb = file('%s/%s.nzb' % (DOWNLOAD_DIR, entry.title), 'w') - #query = { - # 'reportid': str(entry['report_id']), - #} - #nzb.write(directnzb(query)[1]) - #nzb.close() - +def download_nzbs(urls, hella): + entries = [] + for url in urls: + feed = feedparser.parse(url) + entries += feed['entries'] + downloaded = json.load(file('%s/downloaded' % RSSNZB_DIR, 'r')) + queue = {} + + for entry in entries: + if entry['report_progress'] != 'Report is complete': + continue + if entry['title'] in downloaded: + existing = downloaded[entry['title']] + if int(entry['report_size']) <= int(existing['report_size']): + continue + queue[entry['title']] = entry['report_id'] + downloaded[entry['title']] = entry + + for title in queue: + print 'Downloading', title + hella.enqueuenewzbin(queue[title]) + sleep(1) + + serial = {} + for key in downloaded: + serial[key] = {} + for k in downloaded[key]: + if k in ['report_id', 'report_size', 'title']: + serial[key][k] = downloaded[key][k] + json.dump(serial, file('%s/downloaded' % RSSNZB_DIR, 'w'), indent=4) def main(): hella = ServerProxy('http://hellanzb:massive@localhost:8760/') - for url in file('%s/savedsearches' % RSSNZB_DIR).readlines(): - url = url.rstrip('\r\n') - download_nzbs(url, hella) + download_nzbs([x.rstrip('\r\n') for x in file('%s/savedsearches' % RSSNZB_DIR, 'r').readlines()], hella) if __name__ == '__main__': main() ssnzb.py ebe38f58a6e4cabe7629a30dfe4e16c45198a25d Jeremy Grosser synack@neohippie.net http://github.com/synack/rssnzb/commit/9116abf6f9d39faf439ea617ac707110b86afa85 9116abf6f9d39faf439ea617ac707110b86afa85 2009-09-25T21:17:36-07:00 2009-09-25T21:17:36-07:00 Be clever when multiple copies of the same show appear in the feed. 375a812b417a91a0ca36de5bcd2eac0d3a291758 Jeremy Grosser synack@neohippie.net