#! /usr/bin/env python
"""
    Usage:
    python rdfparser.py {-or|-fm|-last|-debug| <url>}
    The rdf is saved into lastrdf.xml in the current dir
    Options:
    -or           for accessing to objectsroot.com
    -fm           for accessing to freshmeat
    -last         read the last parsed rdf
    -debug        for debug
    <url>    Use that url for parsing rdf
"""
import sys
import os
import string
import xml.dom.minidom
import urllib

print "Python RDF Script Rev $id$"

def getText(nodelist):
    """ Legge una stringa """
    rc = ""
    for node in nodelist:
        if node.nodeType == node.TEXT_NODE:
            rc = rc + node.data
    return rc


def handleRDF(rdf):
    """ A simple RDF Parser """
    #rss=rdf.getElementsByTagName("rss")[0]
    #print "==== Rss Version:" + rss.getAttribute("version")
    channel=rdf.getElementsByTagName("channel")[0]    
    title=channel.getElementsByTagName("title")[0]
    handleRDFTitle(title)
    news = rdf.getElementsByTagName("item")
    handleNews(news)
    print "==================="

def handleNews(l):
    for news in l:
       printNews(news)

def printNews(news):
    handleNewsTitle(news.getElementsByTagName("title")[0])
    
    handleLink(news.getElementsByTagName("link")[0])

def handleLink(link):
    print " "+getText(link.childNodes)

def handleNewsTitle(title):
    print "News Title:" + getText(title.childNodes)

def handleRDFTitle(title):
    print "Site Title:" + getText(title.childNodes)


def printIntro():
    print "RDF Parser"
    print "Giovanni Giorgi rdfparser@objectstoot.com"


## MAIN ##


printIntro()
rdfurl=""
opts={ "-fm":"http://freshmeat.net/backend/fm.rdf",
       "-or":"http://freezope2.nipltd.net/gio/jj/JDot/rdf",
       "-debug":"http://localhost:8080/freezopeSiteBackup/JDot/rdf"
}
if len(sys.argv)<=1:
    print "Usage:"
    print __doc__
    
else:
    if sys.argv[1] == "-last":
        rdfurl=''
        print "Reading last"
        f=open("lastrdf.xml","r")
        text=f.read()
        f.close()
    else:
        # Opzioni diverse da -last
        if opts.has_key(sys.argv[1]):        
            rdfurl=opts[sys.argv[1]]
        else:
            # Interpretato come URL
            rdfurl=sys.argv[1]                    
        print "Fetching RDF. from URL:"+rdfurl   
        f = urllib.urlopen(rdfurl)
        text=f.read()
        lastf=open("lastrdf.xml","w")
        lastf.write(text)
        lastf.close()                
    #For debug only:    
    ###print text
    print "Parsing..."
    dom = xml.dom.minidom.parseString( text )
    handleRDF(dom)