#!/usr/bin/env python

# Original thanks to David Fraser <davidf@sjsoft.com>
# Caolan McNamara <caolanm@redhat.com>

import urllib2, cookielib, cgi
import os, sys

from HTMLParser import HTMLParser
from httplib import HTTPConnection

class cws:
    def __init__(self, modules, master):
        self.modules = modules
        self.master = master

class EISScraper(HTMLParser):
    def __init__(self):
        HTMLParser.__init__(self)
        self.state = -1;
	self.modules = {}
	self.master = ''
    def handle_starttag(self, tag, attrs):
	if tag == 'h3' and self.master == '':
            self.state = 0
	elif tag == 'h4':
            self.state = 1
	elif tag == 'tbody' and self.state == 2:
            self.state = 3
	elif tag == 'tr' and self.state == 3:
            self.state = 4
	elif tag == 'td' and self.state == 4:
            self.state = 5
    def handle_data(self, data):
        if self.state == 0 and self.master == '':
            self.master = data.split('\n')[1].split('/')[0].strip()
        elif self.state == 1 and data.find('Modules') != -1:
            self.state = 2
        elif self.state == 5:
            self.modules[data.strip()] = True
            self.state = 3
    def handle_endtag(self, tag):
	if tag == 'h4' and self.state == 1:
            self.state = -1 
        elif tag == 'tr' and self.state == 4:
            self.state = 3
        elif tag == 'td' and self.state == 5:
            self.state = 3
	elif tag == 'table' and self.state == 3:
            self.state = -1

class EIS:
    def __init__(self, cookiefile="eis.lwp"):
        self.cookiefile = cookiefile
        self.cookiejar = cookielib.LWPCookieJar()
        if os.path.isfile(self.cookiefile):
          self.cookiejar.load(self.cookiefile)
        opener = urllib2.build_opener(urllib2.HTTPCookieProcessor(self.cookiejar))
        urllib2.install_opener(opener)
        self.login()
        self.cache = {}

    def login(self):
        urllib2.urlopen("http://eis.services.openoffice.org/EIS2/GuestLogon").read()
        self.cookiejar.save(self.cookiefile)

    def cacheurl(self, url):
        if url in self.cache:
            return self.cache[url]
        else:
            try:
                contents = urllib2.urlopen(url).read()
            except urllib2.HTTPError, e:
                if e.code == 401:
                    self.login()
                    contents = urllib2.urlopen(url).read()
                else:
                    raise
            self.cache[url] = contents
            return contents
    def findcws(self, cws,):
       thiscwsid = None
       milestoneresults = self.cacheurl("http://eis.services.openoffice.org/EIS2/cws.SearchCWS?DATE_NULL_Integrated_After=&DATE_NULL_DueDateBefore=&INT_NULL_Priority=&Name=" + cws + "&SRC_Step=Search&INT_NULL_IsHelpRelevant=&RSV_NoWait=true&DATE_NULL_DueDateAfter=&TaskId=&DATE_NULL_Integrated_Before=&INT_NULL_IsUIRelevant=")
       for line in milestoneresults.replace("\r", "").split("\n"):
            # cws.ShowCWS?Path=SRC680%2Fm54%2Fdba15&Id=1431
            startmark, endmark = "'cws.ShowCWS?", "'"
            if startmark in line:
                cwsargs = line[line.find(startmark) + len(startmark):]
                cwsargs = cwsargs[:cwsargs.find(endmark)]
                cwsargs = cgi.parse_qs(cwsargs)
                thiscwsid = int(cwsargs["Id"][0])

       return thiscwsid

    def getcwsid(self, cwsname):
        somecwsid = self.findcws(cwsname)
        if somecwsid != None:
            return somecwsid
        raise ValueError("no id found for cws %s" % cwsname)

    def getcwsurl(self, cwsname):
        cwsid = self.getcwsid(cwsname)
        return self.cacheurl("http://tools.services.openoffice.org/EIS2/cws.ShowCWS?Id=%d&Section=All" % cwsid)

    def getcwsdetails(self, cwsname):
        cwsid = self.getcwsid(cwsname)

	conn = HTTPConnection("tools.services.openoffice.org")
        conn.request("GET", "/EIS2/cws.ShowCWS?Id=%d&Section=All" % cwsid)
        response = conn.getresponse()

 	foo = EISScraper()
	foo.feed(response.read())
        conn.close()
	foo.close()

	return cws(foo.modules.keys(), foo.master)

if len(sys.argv) < 2:
    print 'Usage: oooextractcws childworkspace [cvsserver]'
    print 'e.g. oooextractcws jaxpapi ' + \
        ':pserver:cmc@localhost:2401/shared/data/helm/cvs/repository'
else:
    if len(sys.argv) > 2:
        cvsserver = sys.argv[2]
    else:
        cvsserver = ':pserver:anoncvs@anoncvs.services.openoffice.org:2401/cvs'

    childws = sys.argv[1]

    eis = EIS()
    info = eis.getcwsdetails(childws)
    masterws = info.master

    lowertag='cws' + '_' + masterws.lower() + '_' + childws
    uppertag=lowertag.upper() + '_ANCHOR'

    for module in info.modules:
        os.system('cvs -q -d' + cvsserver + ' rdiff -kk -u -r' + uppertag + \
          ' -r' + lowertag + ' ' + module)