from django.core.management.base import BaseCommand, CommandError
from catalog.models import User, News, Publisher, Topic, audioTime, Search
from django.db.models import Q
import operator
from django.core.paginator import Paginator

import os
#import simplejson as json
import json
from django.db.models import Q
import sys
sys.path.append("/home/mario/git/")
from Guns_N_Roses import M3GunsNRoses
import datetime



#from os import path
import codecs
import nltk
from nltk.stem import SnowballStemmer
from textblob.classifiers import (NaiveBayesClassifier)
import dill as pickle
#-------------------------------------------------------------------------------------------------
def file2list(fname):
	with open(fname) as f:
		lines = f.readlines()
	lines = [l.strip('\n') for l in lines if not l is '']
	return lines

#-------------------------------------------------------------------------------------------------
class Command(BaseCommand):
	help = 'Report database'


	def add_arguments(self, parser):
#		parser.add_argument('model', nargs=1, type=str) #(options['basedir'][0]
		parser.add_argument('words', nargs=1, type=str) 

	def handle(self, *args, **options):

		wordList = file2list(options['words'][0])
		print(list)
		#with open(options['word'][0]) as f:
    		#	cfg = json.load(f)

		pub=[3, 46, 39, 43, 45, 41, 40, 6, 7, 8, 27, 9, 10, 12, 13, 14, 15, 17, 31, 21, 24]
		publishers = Publisher.objects.all().filter(id__in=pub)
		
#		myQuery = Q(publisher__in=publishers) &  Q(date__gte=datetime.date(2014, 1, 1))  & Q(search_vector__in=wordList)
#		newsList = News.objects.all().filter(myQuery).order_by('id')
#		print(newsList.count())

		ids = set()
		for w in wordList:
			myQuery = Q(publisher__in=publishers) &  Q(date__gte=datetime.date(2014, 1, 1)) & Q(search_vector=w)
			newsList = News.objects.all().filter(myQuery)
			for i in newsList:
				ids.add(i.id)
	
		myQuery = Q(id__in=ids)
		newsList = News.objects.all().filter(myQuery)

		print("end")
		print(newsList.count())

		doclist = [{"text":n.text, "id":n.id, "date":n.date.strftime("%Y-%m-%d"), "url":n.url} for n in newsList]

		with open('dataNews.json', 'w', encoding='utf8') as outfile:
			s = json.dumps(doclist,  ensure_ascii=False, indent=2)
#			print (s)
			outfile.write(s)