import logging
import os
import sys
import time
from optparse import OptionParser

import psycopg2
from psycopg2.extras import DictCursor

sys.path.append(os.getcwd())

import fastText as ft
import pandas as pd
from pymongo import MongoClient
import yaml
from inference.inference import HierarchicalModel
from inference.models import ProductMeta
from dataset_tools.push_labels_into_file import load_label_into_idx

FORMAT = '%(asctime)-15s %(message)s'
logging.basicConfig(format=FORMAT, level=logging.INFO)
LOGGER = logging.getLogger(__name__)

"""
    Inference model
    Fetch all textual information of product includes title, description, product tags, product type, options of product
"""
if __name__ == '__main__':

    parser = OptionParser()
    parser.add_option("-c", "--config_path", dest="config_path",
                      help="Config path")
    parser.add_option("-f", "--file_path", dest="file_path",
                      help="file path for inference")

    (options, args) = parser.parse_args()
    config_path = options.config_path
    file_path = options.file_path

    FILL_EMPTY = " "

    config = yaml.load(open(config_path))

    hierarchical_model = HierarchicalModel(model_lv1_path=config['model_lv1_path'],
                                           model_lv2_path=config['model_lv2_path'],
                                           model_load_fn=ft.load_model,
                                           label_lv1_path=config['label_lv1_path'],
                                           label_lv2_path=config['label_lv2_path'],
                                           prefix=config['prefix'])

    POSTGRESQL_HOST = config['POSTGRESQL']['HOST']
    POSTGRESQL_PORT = config['POSTGRESQL']['PORT']
    POSTGRESQL_DBNAME = config['POSTGRESQL']['DBNAME']
    POSTGRESQL_USERNAME = config['POSTGRESQL']['USERNAME']
    POSTGRESQL_PASSWORD = config['POSTGRESQL']['PASSWORD']

    # Connect to postgresql

    # - *dbname*: the database name
    # - *database*: the database name (only as keyword argument)
    # - *user*: user name used to authenticate
    # - *password*: password used to authenticate
    # - *host*: database host address (defaults to UNIX socket if not provided)
    # - *port*: connection port number (defaults to 5432 if not provided)


    conn = psycopg2.connect(database=POSTGRESQL_DBNAME,
                            user=POSTGRESQL_USERNAME,
                            password=POSTGRESQL_PASSWORD,
                            port=POSTGRESQL_PORT)


    LOGGER.info("Connect to postgresql")
    cur = conn.cursor(cursor_factory=DictCursor)


    LOGGER.info("Loading model")
    hierarchical_model.load()

    labels_mapper = load_label_into_idx(config['label_lv2_path'])

    # Production mongodb
    client_production = MongoClient(host=config['MONGODB_PRODUCTION']['HOST'],
                                    port=config['MONGODB_PRODUCTION']['PORT'])
    db_product = client_production['beeketing']['Product']

    labels_dat = []
    LOGGER.info("Begin to process")

    items = cur.execute("SELECT shop_id, product_id from reporting.top_10_products")
    for idx, item in enumerate(items.fetch_all()):

        shop_id = int(float(item['shop_id']))
        product_id = item['product_id']

        stime = time.time()
        try:
            product = db_product.find_one({"_id": product_id})
            product_meta = ProductMeta(title=product.get("title", FILL_EMPTY),
                                       description=product.get("shortDescription", FILL_EMPTY),
                                       tags=product.get("tags", FILL_EMPTY),
                                       options=product.get("options", FILL_EMPTY))

            label, prob = hierarchical_model.predict(product_meta, threshold=0.6, k_top_2=1)[0]
            # Write to file
            labels_dat.append([shop_id, product_id, labels_mapper.get(label), prob])
            etime = time.time()
            LOGGER.info("Finish to inference: %s" % (etime - stime))
        except Exception as e:
            LOGGER.error("Error %s on shop %s and %s", e, shop_id, product_id)
            with open("data/shop_inference_error.txt", "a") as f:
                f.write("%s|%s\n" % (shop_id, product_id))

    dat = pd.DataFrame(labels_dat, columns=['shop_id', 'product_id', 'label', 'prob'])
    dat.to_csv("data/beeketing_top_10.csv", sep="|", index=None)

    LOGGER.info("Finished")