"""
gna2atom.py

Parses Google News Alerts received via e-mail and processes them.
Includes functionality to post to an Atom-enabled blog, and create an
Atom feed suitable for syndication.

copyright 2004, Nathan R. Yergler, Creative Commons
Licensed under the GNU GPL 2.

$Id: gna2atom.py,v 1.7 2004/11/08 02:48:11 nyergler Exp $
"""

import poplib
import urlparse
import email
import smtplib
import urllib2
import urllib

mail_server = 'mail'
username = 'xxx'
password = 'xxx'

gna_from = 'Google Alerts <googlealerts-noreply@google.com>'

class NewsAlert:
    def __init__(self, alerttext):
        self.title = self.summary = self.url = self.author = ''
        self.parseMessage(alerttext)

    def parseMessage(self, message):
        """Extract """

        lines = message.split('\n')
        self.title = lines[0]
        self.author = lines[1]

        # determine the line containing the article url
        if (len(lines[-3]) > 8) and lines[-3][:8] == '<http://':
            url_line = -3
        else:
            url_line = -1
            
        self.url = lines[url_line][1:-1]
        self.summary = "\n".join(lines[2:url_line - 1])

    def atom(self):
        """Return the wrapped entry as an Atom entry."""
        return """
        <entry xmlns="http://example.com/newformat#" >
          <title>%s</title>
          <summary>%s</summary>

          <author>
            <name>%s</name>
            <url>%s</url>
          </author>

          <issued>%s</issued>
          <created>%s</created>

          <link rel="alternate" type="text/html"
            href="%s"/>

          <content type="application/xhtml+xml" xml:lang="en-us">
           %s
          </content>

        </entry> 
        """ % (self.title, self.summary, self.author_name, self.author_url,
        self.issued, self.created, self.link, self.content)


###############################################################
#
# POP3 support functions
#

def connect(server, username, passwd):
    """Connect to the specified server using the username and password;
    return the server instance."""
    
    server = poplib.POP3(server)
    print server.user(username)
    print server.pass_(passwd)

    return server

def gnaMessages(server, delete=False):
    """Retrieve messages from the server; if the message is from the
    Google news service, yield the NewsAlert instance.  If [delete] is
    True, delete the message once control returns to this function.
    After all messages have been processed, disconnect from the server."""
    
    num_msgs = server.stat()[0]

    # for each message on the server
    for i in range(num_msgs):
        
        # retrieve the lines
        msg = email.message_from_string("\n".join(server.retr(i+1)[1]) )

        if msg['From'] == gna_from:

            # extract the message payload
            segments = msg.get_payload()

            # check if list or string
            try:
                segments.append('foo')
                del segments[-1]
            except:
                segments = [segments]

            for segment in segments:
                
                if segment.get_content_type() == 'text/plain':
                    # plain text segment; break on blank lines
                    msg_text = segment.get_payload()
                    msg_text = "\n".join(msg_text.split("\n")[:-12])
                    
                    for alert in msg_text.split('\n\n'):
                        aentry = NewsAlert(alert)
                        yield aentry

            if delete:
                # delete the message
                server.dele(i+1)

    # disconnect from the server
    server.quit()


###############################################################
#
# Google News Alert message handlers
#

def nullHandler(msg):
    """NewsAlert handler which performs no processing.  A handler should
    require a single parameter, an instance of NewsAlert.
    """

def printMessage(msg):
    print msg.title
    print msg.author
    print msg.summary
    print msg.url
    print

def SimpleBlogHandler(msg):
    POSTDRAFT_URL = 'http://apps.creativecommons.org:7080/blog/press/postdraft'
    SMTP_SERVER = 'localhost'
    SEND_TO = ('foo@bar.com',)
    
    post_data = {'title':msg.title,
                 'bodytext':"""%s<br><a href="%s">%s</a>""" % (
        msg.summary, msg.url, msg.url),
                 'username':"Google News Import",
                 
                 }

    result = urllib2.urlopen(POSTDRAFT_URL,
                             data=urllib.urlencode(post_data)).read()

    # send an email to the specified parties to
    # let them know a draft was created
    alert_msg = "From: gnaimport@yergler.net\r\nTo: %s\r\nSubject: Google News Alert draft blog entry created\r\n\r\nA news alert has been imported into the blog.  The URL is:\r\n\r\n%s\r\n\r\n" % (", ".join(SEND_TO), result)
    
    server = smtplib.SMTP(SMTP_SERVER)
    server.sendmail('gnaimport@yergler.net', SEND_TO, alert_msg)
    server.quit()
    
    print result
    
###############################################################
#
# main control functions
#

if __name__ == '__main__':
    # set up the message handlers
    # handlers = (nullHandler, printMessage, SimpleBlogHandler)
    handlers = (nullHandler, SimpleBlogHandler)

    # connect to the mail server
    popserve = connect(mail_server, username, password)

    # retrieve waiting messages from the mail server
    for msg in gnaMessages(popserve, True):
        for handler in handlers:
            handler(msg)