#gproxy.py, written by Rusty Japikse, at Concepts NREC (3/06).
#This code is available as open source under the LGPL license
#
#About:
#       gproxy.py is a python based (duh!) cgi script to provide a secure proxy to a Google Mini
#       server. This script handles securely proxing a connection to different sub-collections
#	on a Google Mini server. Attempts by end users to set the sub-collection variable in
#	get requests (&restrict=secret_sub_collection) is stripped by regular expressions and
#	a default (SUB_COLLECTION Configuration Setting) sub-collection is specified.
#
#       As this script is open source there are no guaranties, if it breaks things, you get to
#       keep the pieces! That being said, this crawler works quite well in feeding our Google Mini.
#
#Usage:
#       1. Set the configuration variables
#	2. Place this script in a server directory as the default file to be served. Configure security for this
#	   directory (e.g. Windows Security Settings for IIS or .htaccess for Apache).
#       3. (Test then...) Deploy! 
#
#Other Useful Tips:
#       For feeding your Google Mini with a list of good URLs, please also look for a script
#	called CrawlFS.py. This script should be available in the python cookbook at activestate
#	and at http://www.japikse.com. Hope this helps someone to avoid reinventing the wheel!

################### Configuration Settings ##########################################
GMINI_URL='http://gmini/' #URL to your Google Mini, include the trailing slash
IMG_URL='<img border="0" src="acmecorp_fileserver_search.png">' #url of your personalized image 
FOOTER="""<div style="margin-top: 10px; width: 100%; font-size: 80%; text-align: center;">
	<a href="http://intranet.acmecorp.com/">Intranet</a> | 
	<a href="http://intranet.acmecorp.com/library/">Library Search</a> | 
	<a href="http://intranet.acmecorp.com/faces/">Facebook</a> |
	<a href="http://intranet.acmecorp.com/accounting/">Timesheets</a> | 
	<a href="http://www.acmecorp.com/">ACME Corp (External)</a>
	<p style="margin-top: 0; padding-top: 3px;"><small>Powered by Google</small></p></div>\n""" #custom footer
INCLUDE_OLD_TITLE=True #if true, resuse the old title e.g. <title>TITLE - Old Title</title>
TITLE='ACME Corportation Fileserver Search'
FAVICON='favicon.ico' #path to your favicon.ico file
SUB_COLLECTION='mark_cust' #the sub-collection to search from
REL_URL='./'  #Might be useful if you aren't serving this as the default page
#####################################################################################

import re, cgi, urllib
def rewrite_html(h):
	#rewrite the html send to the user to point toward this script, also provides for customization of the page 
	#functional modifications to the page
	r=re.compile('/search') #change the search target
	h=r.sub(REL_URL,h)
	r=re.compile('/basics.html') #rewrite where the basics.html link points and turn it into a get method
	h=r.sub('%s?basics=basics.html' %REL_URL, h)
	#cosmetic modifications to the page
	r=re.compile(r'<[^<]*src{1}[^>]*>', re.IGNORECASE | re.DOTALL)  #image urls
	h=r.sub(IMG_URL, h)
	r=re.compile(r'<center><br><br><p><font face="arial,sans-serif" size="-1" color="#2f2f2f">\s*(Powered by Google)+.*</center>|<div[^>]+class=footer[^>]*>.*</div>', re.DOTALL) #strip the footer from all pages
	h=r.sub(FOOTER,h)
	r=re.compile(r'<title>(.*?)</title>', re.DOTALL) #title and favicon swap
	title=''
	if INCLUDE_OLD_TITLE:
		m=r.search(h)
		title='- '+m.group()[7:-8]
	title='<title>%s %s</title>\n\t<link rel="icon" href="%s" type="image/x-icon" />\n\t<link rel="shortcut icon" href="%s" type="image/x-icon" />' %(TITLE,title,FAVICON,FAVICON)
	return r.sub(title,h)
	return h
def gmini_url():
	#based upon the incoming url, creates the correct query url to send to the gmini server 
	#import cgitb; cgitb.enable() #for debug purposes - comment out for production
	secure_url=clean_url()
	get_d=cgi.FormContentDict()
	url=GMINI_URL
	if get_d.has_key('basics'): #fetch the basics page
		url=GMINI_URL+'basics.html'
	elif get_d: #a search is being run
		url=GMINI_URL+'search?'+secure_url
	return url
	
def clean_url():
	#strip the sub-collection request from the query string before passing it
	#onto the gmini server - this should stop attempts by smart / hacker users 
	#to query other sub-collections
	url=''
	if cgi.os.environ.has_key('QUERY_STRING'):
		url='&'+cgi.os.environ['QUERY_STRING']
		r=re.compile(r'&restrict=?[^&]*|&basics=basics.html', re.IGNORECASE) #remove sub collection requests as well as requests for the basics.html page
		url=r.sub('',url)
	url+='&restrict=%s' %SUB_COLLECTION
	return url.strip('&')
	
if __name__=='__main__':
	print "Content-type: text/html" 
	print
	html=urllib.urlopen(gmini_url()).read()
	print rewrite_html(html)