"""
"""

# $Header: /cvsroot/dbagg2/lib/dbagg2/recommend.py,v 1.3 2004/05/12 02:16:44 deusx Exp $

import sys, os, os.path, time, traceback, shelve
from mx.DateTime import *
from dbagg2.config import *
from dbagg2.model import *
from mx.DateTime import *

from spambayes import classifier, mboxutils, hammie, Corpus, storage
from spambayes.hammie import Hammie

global log, oneHour
log = logging.getLogger("%s"%__name__)


def getHammie():
    if conf.get('data', 'driver') == 'mysql':
        host=conf.get('data_mysql', 'host')
        user=conf.get('data_mysql', 'user')
        passwd=conf.get('data_mysql', 'passwd')
        db=conf.get('data_mysql', 'db')

        dsn = "host=%s user=%s pass=%s dbname=%s" % \
              (host, user, passwd, db)

        log.info("Opening ham store in MySQL: %s" % (dsn))
        ham_store = storage.mySQLClassifier(dsn)
    else:
        log.error("Can't open a storage type for Hammie")
        return
        
    return Hammie(ham_store)
    
def train():
    classifier = getHammie()
    
    #seen_db    = os.path.join('data', 'bayes-seen')
    #seen       = shelve.open(seen_db)

    items = Item.select(
        (Item.q.created < (now() - oneDay)) &
        (Item.q.created > (now() - (oneDay*28)))
        )

    #for item in Item.select(): #items:
    for item in items:
        try:
            msg = toMsg(item)
            if item.visited:
                log.info("\tTraining pos #%s (%s)" % (item.id, item.title[:25]))
                classifier.train_ham(msg)
            else:
                log.info("\tTraining neg #%s (%s)" % (item.id, item.title[:25]))
                classifier.train_spam(msg)
            #classifier.store()
        except:
            (e_type, e_val, e_tb) = sys.exc_info()
            exc = ''.join(traceback.format_exception(e_type, e_val, e_tb))
            log.error("Unexpected exception:\n%s" % (exc))
            
    classifier.store()

def score():
    classifier = getHammie()
    
    items = Item.select(
        (Item.q.created > (now() - (oneHour*2))),
        orderBy='-created'
        )

    for item in items:
        try:
            msg = toMsg(item)
            (prob, clues) = classifier.score(msg, True)

            # Invert and round off the prob
            prob = round(1-prob, 3)
            
            item.rating = prob
            if prob < 0.7:
                log.info("\tScore #%s %s (%s)" % (item.id, prob, item.title))
                #log.info("\tScore #%s %s [%s] (%s)" % (item.id, prob, classifier.formatclues(clues), item.title))
        except:
            (e_type, e_val, e_tb) = sys.exc_info()
            exc = ''.join(traceback.format_exception(e_type, e_val, e_tb))
            log.error("Unexpected exception:\n%s" % (exc))
            

def toMsg(item):
    """
    Return a representation of the item which vaguely resembles
    an email message.
    """
    out = []
    out.append("X-Link: %s" % (item.link))
    out.append("X-Item-Id: %s" % (item.id))
    out.append("From: %s" % (item.source.title.encode('ascii', 'ignore')))
    out.append("Subject: %s" % (item.title.encode('ascii', 'ignore')))
    out.append("Date: %s" % (item.created))
    out.append("")
    out.append("%s" % item.description.encode('ascii', 'ignore'))
    return "\n".join(out)

if __name__ == '__main__':
    if len(sys.argv) < 2:
        train()
    elif sys.argv[1] == 'score':
        score()
    elif sys.argv[1] == 'train':
        train()
    else:
        train()