test.py

from datetime import datetime
from tabulate import tabulate
from decimal import Decimal
from colorama import  Fore ,Style

from cassandra.cluster import Cluster
from cassandra.query import dict_factory
import distutils               
import time
from distutils import util


def main():





    #f = open("fdccontextids.txt", "w")
    #for i in range(1,10001):
        #f.write(str(i)+"\n")
    #f.close()
    
    cluster = Cluster(['rio.pdf.com'],port=9042)
    session = cluster.connect()    
    session.execute("USE fdc_treateddata_cass;")
    
    '''
    f = open("fdccontextids.txt", "w")
    rows = session.execute(' SELECT DISTINCT fdccontextid FROM treateddata LIMIT 10000;')
    for row in rows:
        f.write(str(row.fdccontextid)+"\n")
    
    
    '''

    f = open("fdccontextids.txt", "r")
    content = f.read().splitlines()

    query = 'select fdccontextid,indicatorid from fdc_treateddata_cass.treateddata where fdccontextid in ('

    for fdc in content:
            if fdc == content[-1]:
                query += fdc +');'
            else:
                query += fdc + ','
        
    time_1 = time.time()
    pStatement = session.prepare(query)
    rows = session.execute(pStatement)
    
    # time to execute the query : 0.010839223861694336 seconds : without print
    # time to execute the query : 250.32061100006104 seconds   : with all row 
    # time to execute the query : 41.78741955757141 seconds    :with jsut fdccontextid
 

    #time.sleep(2)
    for row in rows:
        print(row.fdccontextid)
    time_2 = time.time()
    time_interval = time_2 - time_1
    print(Fore.RED+"time to execute the query : "+ str(time_interval) +" seconds"+Style.RESET_ALL)



    #print(content)

    #print(f.readline())
    #for x in f:
        #print(x)
    '''
    query = 'select fdccontextid,indicatorid from fdc_treateddata_cass.treateddata where fdccontextid in ('

    for fdc in content:
        if fdc == content[-1]:
            query += fdc +');'
        else:
            query += fdc + ','
    print(type(content[0]))
    ps = session.prepare('select fdccontextid,indicatorid from fdc_treateddata_cass.treateddata where fdccontextid=?')
    time1 = time.time()
    for fdc in content:
        future = session.execute_async(ps.bind([int(fdc)]))
        try:
            rows = future.result()
            for row in rows:
                print(row.fdccontextid)
        except Exception:
            print("error")
    # '84.81679034233093 s' : with print 
    # '1.075 s' : without print
        
    print((time.time())- time1)
        
    '''  
        
        
        
    '''
        try:
            rows = future.result()
            for row in rows:
                print(row.fdccontextid)
        except Exception:
            print("error")
        '''
    '''
    #print(query)
    cluster = Cluster(['rio.pdf.com'],port=9042)
    session = cluster.connect()    
    session.execute("USE fdc_treateddata_cass;")
    pStatement = session.prepare(query)

    time_1 = time.time()

    rows = session.execute(pStatement)
    time_2 = time.time()
    time_interval = time_2 - time_1
    print(Fore.RED+"time to execute the query : "+ str(time_interval) +" seconds"+Style.RESET_ALL)

    print(tabulate(rows,tablefmt='pretty',headers='keys',showindex=True))
    '''
'''
    fdcs = ['589169','626069','638424','555165','579506','529509','553643','586462','403248','528165']
    cluster = Cluster(['rio.pdf.com'],port=9042)
    session = cluster.connect()    
    session.execute("USE fdc_treateddata_cass;")
    time_1 = time.time()
    for fdc in fdcs:
        rows = session.execute('select fdccontextid,indicatorid from fdc_treateddata_cass.treateddata where fdccontextid='+fdc+' ;')
        
    time_2 = time.time()
    time_interval = time_2 - time_1
    print(Fore.RED+"time to execute the query : "+ str(time_interval) +" seconds"+Style.RESET_ALL)
    print('---------------------------------')
    time_3 = time.time()
    rows = session.execute('select fdccontextid,indicatorid from fdc_treateddata_cass.treateddata where fdccontextid in (589169,626069,638424,555165,579506) ;')
    rows = session.execute('select fdccontextid,indicatorid from fdc_treateddata_cass.treateddata where fdccontextid in (529509,553643,586462,403248,528165) ;')
    time_4 = time.time()
    time_interval2 = time_4 - time_3
    print(Fore.RED+"time to execute the query : "+ str(time_interval2) +" seconds"+Style.RESET_ALL)



    

'''


    #print(tabulate(rows,tablefmt='pretty',headers='keys',showindex=True))



    
"""
    print(bool_value)
    types1 = {'1':'text' , '2':'int' , '3':'date','4':'boolean','5':'UUID'}
    types=types1.items()

    #print(types1)
    cluster = Cluster(['rio.pdf.com'],port=9042)
    session = cluster.connect()
    session.execute("USE test;")
    session.row_factory = dict_factory
    rows = session.execute("SELECT column_name, type FROM system_schema.columns  WHERE keyspace_name = 'test' AND table_name = 'person';")
    
    #print(tabulate(rows,tablefmt='pretty',headers='keys',showindex=True))
   
    fields = {}
    for row in rows:
        print (row)
        value = input("Insert the value of "+row['column_name']+"\n")
        if row['type'] == 'int':
            try:
                val = int(value)
                
            except ValueError:
                print("That's not an int!")

        elif row['type'] == 'date':
            try:
                datetime.datetime.strptime(value, '%Y-%m-%d')
            except ValueError:
                raise ValueError("Incorrect data format, should be YYYY-MM-DD")        
        
        fields[row['column_name']] = value
    print(fields)    



    '''
    YesNO = input("Do you want to view tables in keyspace[Y/N]?")
    if YesNO == 'Y':
        print("yes")
    else:
        pass  
  


    keyspace= 'test'
    name ='person'
    types1 = {'1':'text' , '2':'int' , '3':'date','4':'boolean','5':'UUID'}
    types=types1.items()
    fields = {}
    Cname = input("Column Name : ")

    while str(Cname) != 'e':
                    #and Cname != '':
        if Cname != '':
                    
            print(tabulate(types, tablefmt='grid',colalign=("center",),stralign="center"))
            type = input(Fore.CYAN+"Type number : "+Style.RESET_ALL)
            if (type == '1' or type == '2' or type == '3' or type == '4' or type == '5'):
                    fields[Cname] = types1[type]
                    
            else:
                print(Fore.RED+"Invalid Type"+Style.RESET_ALL)
                continue
        else : 
            print(Fore.RED+"Press 'e' to exit or try again ... "+Style.RESET_ALL)
        Cname = input("Column Name : ")
    pri_key = input(" what is the primary key? ")
    statement1='CREATE TABLE IF NOT EXISTS '+keyspace+'.'+name+' '+' '
    
    for key,value in fields.items():
        if key == pri_key:
            statement1 += key + ' '+ value + ' PRIMARY KEY'    
        else:
            statement1 += key + ' '+ value   

        if value == list(fields.values())[-1]:
            statement1 += ' ; '
        else:
            statement1 += ' , '

    print(statement1)

   a = 5
   b = 10
   count = 0
   while count < a or count < b:
      print(f"count: {count}, a: {a}, b: {b}")
      count += 1
   
   #type=[{"0","text"},{"1":"int"},{"2","bool"}]
 
   types1 = {'1':'text' , '2':'int' , '3':'date','4':'boolean'}
   types=types1.items()
   print(tabulate(types, tablefmt='grid',colalign=("center",),stralign="center"))
   fields = {}
   Cname = input("Column Name : ")
        
   while str(Cname) != 'e' :            #and Cname != '':
            if Cname != '':
                
                print(tabulate(types, tablefmt='grid',colalign=("center",),stralign="center"))
                type = input(Fore.CYAN+"Type number : "+Style.RESET_ALL)
                if (type == '1' or type == '2' or type == '3' or type == '4'):
                    fields[Cname] = types1[type]
                
                else:
                    print(Fore.RED+"Invalid Type"+Style.RESET_ALL)
                    continue
            else : 
                print(Fore.RED+"Press 'e' to exit or try again ... "+Style.RESET_ALL)
            Cname = input("Column Name : ")
            
   print(fields)         


'''


   # print("hellow")
    
    #a=[["name","id","gender"],['yousef','1','male'],['marah','2','female'],['adam','3','male']]
    #print(tabulate(a,headers='firstrow',tablefmt='pretty',showindex=True,))



   #value='marah'
   # print("you entered {value} >> ")
"""




if __name__ == "__main__": main()