import math
import pymysql
# Connect to the database

prob = 1

sentence = input("Please, enter a sentence : ")
words = sentence.split()

length = len(words)
print("--------------------------------")
print("Length of the sentece : ",length)
print("--------------------------------")
print("Unigram : ",words)
print("--------------------------------")

def retrieve_ngrams(txt, n):
  return [txt[i:i+n] for i in range(len(txt)-(n-1))]

teGrams = retrieve_ngrams(words, 2)

print("Bigram : ",teGrams)
print("--------------------------------")

def hesapla(bigram):
  connection = pymysql.connect(host='localhost',
                               user='root',
                               password='ruveyda123',
                               db='dbhw2',
                               charset='utf8mb4',
                               cursorclass=pymysql.cursors.DictCursor)
  try:
      ilk= "%"+bigram[0] + "%"
      ikinci = "%"+bigram[1] + "%"
      with connection.cursor() as cursor:
          sql1 = "Select count(*) from table_db where IG like %s";

          cursor.execute(sql1, (ilk))

          result = cursor.fetchone()
          sayi1 = result['count(*)']


         # print(result['count(*)'])
          #print("--------------------------------")

          sql3 = "Select count(*) from table_db as R " \
                 "INNER JOIN (select file_no, S_no, IX, IG from table_db where IG like %s) AS T " \
                 "where R.file_no = T.file_no AND R.S_no = T.S_no AND R.IX = T.IX + 1 and R.IG like %s;"

          cursor.execute(sql3,(ilk, ikinci))

          result2 = cursor.fetchone()
          sayi2 = result2['count(*)']

         # print(result2['count(*)'])
          #print("--------------------------------")

          if(sayi1==0&sayi2!=0):
              sayi1=sayi1+(1/11909) # add-one(1/corpusSize) smoothing
              pr = sayi2 / sayi1

          elif (sayi2==0&sayi1!=0):
              sayi2 = sayi2 + (1 / 11909)
              pr = sayi2 / sayi1
          elif (sayi2 != 0 & sayi2 != 0):
              pr = sayi2/sayi1
          else :
              pr = math.pow((1/11909), length)

      return pr

  finally:
      connection.close()


for bigram in teGrams:
   prob = prob*hesapla(bigram)
   print(bigram,prob)


print("Probability of the sentence is : ",prob)
print("--------------------------------")