lfs-get-tarballs

   1 #!/usr/bin/python
   2 #
   3 # Download all the LFS (development) tarballs from LFS page
   4
   5 from os import system
   6 from sys import exit
   7 import sgmllib, urllib
   8
   9 class LFSPackParser(sgmllib.SGMLParser):
  10         "LFS download page parser"
  11
  12         def parse(self, s):
  13                 "Parse the given string 's'."
  14                 self.feed(s)
  15                 self.close()
  16
  17         def __init__(self, verbose=0):
  18                 "Initialise an object, passing 'verbose' to the superclass."
  19                 sgmllib.SGMLParser.__init__(self, verbose)
  20                 self.hyperlinks = []
  21
  22         def start_a(self, attributes):
  23                 "Process a hyperlink and its 'attributes'."
  24
  25                 for name, value in attributes:
  26                         if name == "href":
  27                                 self.hyperlinks.append(value)
  28
  29         def get_hyperlinks(self):
  30                 "Return the list of hyperlinks."
  31                 return self.hyperlinks
  32
  33 def get_page_contents():
  34         page = urllib.urlopen("http://www.linuxfromscratch.org/lfs/view/development/chapter03/packages.html")
  35         contents = page.read()
  36         page.close()
  37         return contents
  38
  39 def get_package_list(page):
  40         parser = LFSPackParser()
  41         parser.parse(page)
  42         packages = []
  43         for name in parser.get_hyperlinks():
  44                 if name.find('.tar') != -1:
  45                         packages.append(name)
  46         return packages
  47
  48 def main():
  49         page = get_page_contents()
  50         for name in get_package_list(page):
  51                 system("wget " + name)
  52
  53 if __name__ == "__main__":
  54         main()
  55
  56 exit(0)