#!/usr/bin/env python

import ConfigParser, os, string, sys, StringIO, time, threading

# Set up a module path that covers the global lib and all plugins
BASE_PATH = os.path.normpath(os.path.join(\
    os.path.abspath(sys.argv[0]), os.pardir, os.pardir))
sys.path.insert(0, os.path.join(BASE_PATH, 'lib'))

import logging, logging.config, sqlite, iso8601
import dbagg
import dbagg.storage.sqlite
from spambayes import classifier, mboxutils, hammie, Corpus

logging.config.fileConfig(os.path.join(BASE_PATH, 'conf', 'global.conf'))
main_log = logging.getLogger("")
main_log.info("Starting up.");

# Configure data sources
conn     = sqlite.connect("data/agg", autocommit=1)
sources  = dbagg.storage.sqlite.SQLiteSourceCollection(conn)
items    = dbagg.storage.sqlite.SQLiteItemCollection(conn)

# Configure the classifier
shown_db   = 'data/bayes-interesting.db'
usedb      = True
classifier = hammie.open(shown_db, usedb, 'c')

main_log.info("Searching for items...")

train_items = items.sql_iter( \
    """
    SELECT items.*
    FROM items
    WHERE items.created > "2003-08-12T00:00:00"
    """,
    (  ) );

main_log.info("Training...")

for item in train_items:
    trained = item.get_meta('bayes:trained')
    if trained == 'pos':
        classifier.train_spam(item.toMsg())
        main_log.info("\tTrained pos #%s (%s)" % (item.id, item.title))
    elif trained == 'neg':
        classifier.train_ham(item.toMsg())
        main_log.info("\tTrained neg #%s (%s)" % (item.id, item.title))
classifier.store()