#!/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

def item_to_msg(item):
    return "%s ^^^ %s" % (item.title, item.description)

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
h_shown  = hammie.open(shown_db, usedb, 'c')

#
#since  = '2003-07-02T20:37:59.45-05:00'
since  = '2003-08-12T00:00:00-05:00'
offset = 0
max    = 80000

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

shown_items = items.sql_iter( \
    """
    SELECT items.*
    FROM items
    WHERE
        (items.created > %s)
    ORDER BY
        items.created DESC
    LIMIT %s OFFSET %s
    """,
    ( since, max, offset ) );
#    """, () )

main_log.info("scoring items")

for item in shown_items:
    msg = item_to_msg(item)
    (prob, clues) = h_shown.score(msg, True)
    prob = round(prob, 3)
    item.set_meta('bayes:score', prob)

    if prob > 0.6:
        main_log.info("\tScore #%s %s (%s)" % (item.id, prob, item.title))
        #main_log.info("\t\t%s" % (clues))
conn.commit()