lfs-get-tarballs

   1 #!/usr/bin/python
   2 #
   3 # Download all the LFS tarballs (or patches) from the provided download page
   4 #
   5 # Luiz Fernando N. Capitulino
   6 # <lcapitulino@gmail.com>
   7
   8 from os import system
   9 from urllib import urlopen
  10 from sys import argv, exit
  11 from HTMLParser import HTMLParser
  12
  13 class LFSPackParser(HTMLParser):
  14         def is_package(self, url):
  15                 if url.find('.tar.') != -1:
  16                         return True
  17                 if url.endswith('.patch'):
  18                         return True
  19                 return False
  20
  21         def handle_starttag(self, tag, attrs):
  22                 if tag == 'a' and self.is_package(attrs[1][1]):
  23                         self.hyperlinks.append(attrs[1][1])
  24
  25         def get_hyperlinks(self):
  26                 return self.hyperlinks
  27
  28         def __init__(self):
  29                 HTMLParser.__init__(self)
  30                 self.hyperlinks = []
  31
  32 def get_package_list(addr):
  33         page = urlopen(addr)
  34         parser = LFSPackParser()
  35         parser.feed(page.read())
  36         parser.close()
  37         page.close()
  38         return parser.get_hyperlinks()
  39
  40 def main():
  41         if len(argv) != 2 or argv[1] == '-h':
  42                 print 'usage: lfs-get-tarballs < LFS download page >'
  43                 exit(1)
  44
  45         failed = []
  46         pack_list = get_package_list(argv[1])
  47
  48         print '\nDownloading ' + str(len(pack_list)) + ' packages\n'
  49
  50         for name in pack_list:
  51                 ret = system('wget ' + name)
  52                 if ret != 0:
  53                         failed.append(name)
  54
  55         if len(failed) != 0:
  56                 print '\nWARNING: could not download the following packages\n'
  57                 for name in failed:
  58                         print '\t-> ' + name
  59                 print
  60
  61 if __name__ == '__main__':
  62         main()