#!/usr/bin/env python
"""
CGI interface to tidy and xsltproc
Share and Enjoy.
"""
import cgi, md5, os, sys, time, urlparse, urllib
from stat import *
from urlparse import urlparse, urlunparse
# Configuation variables
XSLT='/usr/bin/xsltproc'
TIDY='/usr/local/bin/tidy'
CURL='/usr/bin/curl'
cache_dir='/www/decafbad.com/data/tidyxslt'
cache_time=3*60*60 # 3 hours
def cacheFunc(func, **kwargs):
# Come up with a cache file name based on the hash of argument values
keys = kwargs.keys()
keys.sort()
m = md5.new()
for k in keys:
# Skip the cache argument
if k != 'cache':
m.update(str(kwargs[k]))
fn = os.path.join(cache_dir, m.hexdigest())
if kwargs['cache'] == '0' or \
not os.path.isfile(fn) or \
(time.time() - os.stat(fn)[ST_MTIME]) > cache_time:
# Caching is disabled, there's no cached output, or the cache
# has expired, so call the function.
out = func(**kwargs)
fout = open(fn, 'w')
fout.write(out)
fout.close()
else:
# The cache file exists and has not expired, so serve it up.
fin = open(fn, 'r')
out = fin.read()
fin.close()
return out
def tidyxslt(**kwargs):
docAddr = urlunparse(urlparse(kwargs['docAddr']))
xslAddr = urlunparse(urlparse(kwargs['xslAddr']))
out = "Content-Type: text/xml\n\n"
cmd = '%s --compressed -s %s | ' % (CURL, docAddr) + \
'%s --indent yes --doctype strict --output-encoding latin1 ' % (TIDY)+ \
' --force-output yes -asxml -q -f /dev/null | '+ \
'%s %s -' % (XSLT, xslAddr)
p = os.popen(cmd, 'r')
out = out + p.read()
p.close()
return out
def serveRequest():
fields = cgi.FieldStorage()
if not fields.has_key('doc'):
# Serve up a simple form interface
print "Content-Type: text/html"
print
print """
xsltproc service
Tidy and XSLT
"""
else:
print cacheFunc(tidyxslt,
docAddr=fields.getvalue('doc', ''),
xslAddr=fields.getvalue('xsl', ''),
cache=fields.getvalue('cache', '1'))
if __name__ == '__main__':
if os.environ.has_key('SCRIPT_NAME'): serveRequest()