1 from subprocess import *
   2 from Ft.Lib import Uri
   3 
   4 import amara
   5 
   6 cmdline = "java -jar  tagsoup-1.2.jar --encoding='utf-8'"
   7 URL = r"http://www.example.com"
   8 
   9 process = Popen(cmdline, stdin=PIPE, stdout=PIPE, universal_newlines=True, shell=True)
  10 unsouped, perr = process.communicate(input=Uri.UrlOpen(URL).read())
  11 try:
  12     doc = amara.parse(unsouped)
  13 except:
  14     #Don't use this blanket exception in non-example code :-)   Oh, and use L10N for error messages
  15     raise ValueError('Empty or unexpected output from the command line.  Command line: "%s"'%TAGSOUP_CMDLINE)
  16 print doc.html.head.title.xml()
  17 

   1 import urllib2
   2 import amara
   3 import cStringIO
   4 import tempfile
   5 import popen2
   6 
   7 #java tagsoup cmdline
   8 cmdline = "java -jar  tagsoup-1.2.jar %s"
   9 # url to parse
  10 url = r"http://www.hitimewine.net/istar.asp?a=6&id=161153!1247"
  11 
  12 # retrieving html file
  13 html_file = tempfile.mktemp()
  14 f = open (html_file, 'w')
  15 f.write(urllib2.urlopen(url).read())
  16 f.close()
  17 
  18 # doing tagsoup
  19 pout, pin, perr = popen2.popen3(cmdline % html_file)
  20 
  21 # parsing with amara
  22 doc = amara.parse(pout.read())
  23 print doc.xml(indent=True)
  24 

   1 #DO NOT USE: Naive approach that easily succumbs to buffer deadlock
   2 import urllib2
   3 import popen2
   4 import amara
   5 
   6 TAGSOUP_CMDLINE = "java -jar  tagsoup-1.2.jar --encoding='utf-8'"
   7 URL = r"http://www.example.com"
   8 
   9 pout, pin, perr = popen2.popen3(TAGSOUP_CMDLINE)
  10 pin.write(urllib2.urlopen(URL).read())
  11 pin.close()
  12 try:
  13     doc = amara.parse(pout)
  14 except:
  15     #Don't use this blanket exception in non-example code :-)   Oh, and use L10N for error messages
  16     raise ValueError('Empty or unexpected output from the command line.  Command line: "%s"'%TAGSOUP_CMDLINE)
  17 print doc.html.head.title.xml()
  18