def run(self):
connection = DataBase().connection()
cursor = connection.cursor()
cursor.execute(DataBase().trunk_ticket_info)
connection.commit()
cursor.execute(DataBase().ssidcountquery)
count = cursor.fetchall()
count = count[0][0]
print "data==", count
limit = 1
start = 0
stop = 100
loopCount = 100
try:
limit = count / loopCount
except:
limit = 15
pass
print "no of threads==", limit + 1
for i in range(0, limit + 1):
t = threading.Thread(target=self.bms_ssid_count,
args=("thread_name_" + str(i), start, stop))
t.start()
start = start + 100
stop = 100
try:
connection.close()
print "Initial Connection Closed"
except:
print "Initial connection Not closed"
def run(self):
connection = DataBase().connection()
cursor = connection.cursor()
# cursor.execute("truncate moviesfyi_stage.bk_movies")
connection.commit()
cursor.execute(DataBase().showtimesCountQuery)
count = cursor.fetchall()
count = count[0][0]
print "data==", count
limit = 1
start = 1
stop = 300
loopCount = 300
try:
limit = count / loopCount
except:
limit = 15
pass
print "no of threads==", limit + 1
for i in range(0, limit + 1):
t = threading.Thread(target=self.bms_movie_count,
args=("thread_name_" + str(i), start, stop))
t.start()
start = stop + 1
stop = start + loopCount - 1
try:
connection.close()
print "Initial Connection Closed"
except:
print "Initial connection Not closed"
def run(self):
connection = DataBase().connection()
cursor = connection.cursor()
cursor.execute("truncate collections.bms_showtimes")
connection.commit()
print DataBase().showtimesCountQuery
cursor.execute(DataBase().showtimesCountQuery)
count = cursor.fetchall()
count = count[0][0]
limit = 1
start = 1
stop = 2000
loopCount = 2000
try:
limit = count / loopCount
except:
limit = 15
pass
for i in range(0, limit + 1):
t = threading.Thread(target=self.bms_showtime_count,
args=("thread_name_" + str(i), start, stop))
t.start()
start = stop + 1
stop = start + loopCount - 1
try:
connection.close()
print "Initial Connection Closed"
except:
print "Initial connection Not closed"
def ticketsinfo(self, starts, stops):
connection = DataBase().connection()
cursor = connection.cursor()
date = datetime.date.today()
date = str(date)
date = date.replace("-", "")
query = ""
# print date
seats_ssid = "SELECT theater_code ,showdate,showtime,sessionid FROM collections.bms_showtimes where showdate =CURDATE() AND STR_TO_DATE(showtime,'%l:%i %p') BETWEEN SUBTIME(ADDTIME(current_time(),'5:30:00.00'),5900) AND ADDTIME(ADDTIME(current_time(),'5:30:00.00'),5900) group by sessionid limit " + str(
starts) + " , " + str(stops)
print seats_ssid
cursor.execute(seats_ssid)
results = cursor.fetchall()
# print "length======================================================================================",len(results)
for j in results:
print j[0]
url = "https://in.bookmyshow.com/serv/getData?cmd=GETSHOWINFO&vid=" + j[
0] + "&ssid=" + str(j[3])
print url
r = requests.get(url)
data = r.text
val = ""
myjson = json.loads(data.replace("arrShowInfo=", "")[:-1])
# print "length==",len(myjson)
for mjson in range(len(myjson)):
try:
tid = myjson[mjson][0]
ssid = myjson[mjson][1]
tc_class = myjson[mjson][2]
t_class = myjson[mjson][3]
price = myjson[mjson][4]
screen_name = myjson[mjson][33]
m = myjson[mjson].index(date)
l = myjson[mjson][m:]
totalseats = l[2]
availableseats = l[3]
val = val + "('" + tid + "','" + screen_name + "'," + "'" + ssid + "'," + "'" + tc_class + "'," + "'" + t_class + "'," + "'" + price + "'," + "'" + totalseats + "'," + "'" + availableseats + "'),"
except:
"DNA"
val = val[:-1]
query = DataBase().seats_info + val
# print query
try:
cursor.execute(query)
connection.commit()
if cursor.lastrowid:
print('last insert id', cursor.lastrowid)
else:
print('last insert id not found')
connection.commit()
except:
print query
def getBmsShowTimes(self, starts, stops):
connection = DataBase().connection()
cursor = connection.cursor()
sql = "SELECT id,theater_url from collections.bms_theaters where theater_url is not null and id between " + str(
starts) + " and " + str(stops)
cursor.execute(sql)
results = cursor.fetchall()
date = datetime.date.today()
date = str(date)
date = date.replace("-", "")
movies_ins = ""
count = 0
for j in results:
testurl = j[1]
url = j[1]
da = "https://in.bookmyshow.com"
url = da + url
r = requests.get(url)
data = r.text
soup = BeautifulSoup(data)
query = ""
list = ""
for list in soup.find_all('a', href="javascript:;"):
try:
list = list['onclick']
if ("blnVistaCine" in list):
list = list
list = list.split("callSeatLayout(")
teatinfo = list[1].split(")")
teatinfo = teatinfo[0].split(",")
theater_code = teatinfo[0]
sessionid = teatinfo[1]
movie_code = teatinfo[2]
showtime = teatinfo[3]
date = date
query = query + "(" + movie_code + ",''," + theater_code + ",''," + "'y','" + date + "'," + showtime + "," + "'y'," + sessionid + "),"
except:
""
query = query[:-1]
insquery = DataBase().show_time_ssid + query
try:
cursor.execute(insquery)
connection.commit()
except:
""
connection.close()
def bms_showtime_count(self, thread_name, starts, stops):
connection = DataBase().connection()
cursor = connection.cursor()
BmsShowTimes().getBmsShowTimes(starts, stops)
'''
Created on 16-Jun-2017
@author: H-7
'''
from bs4 import BeautifulSoup
import re
import requests
import string
import json
import mysql.connector
from Database import DataBase
from mysql.connector import MySQLConnection, Error
connection = DataBase().connection()
cursor = connection.cursor()
query = 'SELECT distinct(city) FROM moviesfyi_stage.bk_theaters where theater_url is null;'
cursor.execute(query)
results = cursor.fetchall()
city = ""
k = 1
for j in results:
if (" " in j[0]):
city = j[0].replace(" ", "-")
print city
print "++++++++++++"
if ("(" in city):
city = city.replace("(", "")
city = city.replace(")", "")
else:
city = j[0].replace(" ", "-")
else:
def getmovie(self, starts, stops):
sql = "SELECT id,city,theater_url,state from moviesfyi_stage.bk_theaters_final where theater_url is not null and id between " + str(
starts) + " and " + str(stops)
try:
connection = DataBase().connection()
cursor = connection.cursor()
cursor.execute(sql)
results = cursor.fetchall()
for i in results:
city = i[1]
# city="bengaluru"
theater_url = i[2]
state = i[3]
url = "https://in.bookmyshow.com/" + theater_url
# print url
r = requests.get(url)
data = r.text
soup = BeautifulSoup(data, "lxml")
for j in soup.find_all("ul", id="showEvents"):
try:
movies_ins = ""
for k in j.find_all("div", class_="listing-info"):
# print k
# print "@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@"
lang = k.find("div", class_="eventInfo")
url = k.find("a")
languag = lang.text
mov_url = url["href"]
movie_url = mov_url
'''
BMS Movies Page
'''
qwe = requests.get("https://in.bookmyshow.com" +
movie_url)
asdfg = qwe.text
moviecook = BeautifulSoup(asdfg, "lxml")
relda = moviecook.find(
"span", class_="__release-date").text
relda = relda.replace(",", "")
relda = relda.replace(" ", "-")
asd = relda.split("-")
releasedate = datetime.datetime.strptime(
relda, '%d-%b-%Y').strftime('%Y-%m-%d')
gene = ""
for bgen in moviecook.find_all("span",
itemprop="genre"):
gene = gene + bgen.text + "|"
genere = "|" + gene
censo = moviecook.find("span", class_="__censor")
senso = censo.find("use")
# print senso
# print "----------------------------"
sensor = senso["xlink:href"]
movie_censor = sensor[sensor.rindex("-") + 1:]
movie_censor = movie_censor.upper()
mv_pstr = moviecook.find("div",
class_="poster wow")
mov_posr = mv_pstr.find("meta")
movie_poster = mov_posr["content"]
movie_banner = movie_poster.replace(
"poster", "banner")
# print movie_banner
movie_tari = moviecook.find(
"div", class_="banner-main synopsis-banner")
movie_trailer_code = movie_tari[
"data-trailer-code"]
# print movie_trailer_code
movie_trailer = "https://www.youtube.com/watch?v=" + movie_trailer_code
movie_description = moviecook.find(
"meta", itemprop="description")
movie_description = movie_description["content"]
movie_description = specialCharacterConverter.decode_unicode_references(
movie_description)
movie_description = movie_description.replace(
"'", "")
# print movie_description
'''
Back to Theatres
'''
dimes = ""
movie_code = movie_url[movie_url.rindex("/") + 1:]
mov_url = mov_url.replace("/movies/", "")
mov_url = mov_url.replace("/" + movie_code, "")
mov = url.text.strip()
deumlan = ""
movie_dim = "2D"
rm_lan = "- " + languag
if rm_lan in mov:
movi = mov.replace(rm_lan, "")
if "(" in movi:
deumlan = movi[movi.rindex("(") + 1:]
dimes = deumlan
if " " in dimes:
movie_dim = dimes.split(" ")
movie_dim = movie_dim[0]
else:
movie_dim = "2D"
movi = movi.split("(")
movi = movi[0]
deumlan = deumlan.replace(")", "").strip()
deumlan = deumlan.replace(" ", "-")
deumlan = deumlan.lower()
if deumlan in mov_url:
mov_url = mov_url.replace(
"-" + deumlan, "")
else:
mov_url = mov_url
else:
movi = movi
# print languag
# print mov_url
# print movi
# print movie_code
# print releasedate
# print movie_dim
movie_flag = city + movie_code
movies_ins = movies_ins + "('" + state + "','" + city + "','" + movi + "','" + genere + "','" + movie_censor + "','" + movie_banner + "', '', '" + movie_poster + "', '" + languag + "', '" + movie_dim + "', '" + releasedate + "', '" + movie_description + "', '" + movie_trailer + "', '', '','" + movie_code + "', '" + mov_url + "','" + movie_flag + "'),"
# print "--------------------"
movies_ins = movies_ins[:-1]
# print movies_ins
insmovieinfo = "INSERT ignore INTO moviesfyi_stage.bk_movies (state, city, title, genre, censor, desktop_banner, mobile_banner, poster, language, dimension, release_date, description, trailer_url, rating, votes, movie_code, movie_url, book_groupcode) values " + movies_ins
# print insmovieinfo
try:
cursor.execute(insmovieinfo)
connection.commit()
except:
print sys.exc_info()
# print testurl
print insmovieinfo
except:
""
except:
""