#------------------------------------------------------------------------------
#
#  RandomInput5.py
#  Creates a random input file for cse 101 pa5. This is identical to the input
#  file generator from pa1. 
#
#  An input file is created by selecting random words from file English.txt, 
#  and random characters from the delimiter string 
#
#               " \t\\\"\',<.>/?;:[{]}|`~!@#$%^&*()-_=+0123456789"
#
#  and creating random lines from these objects. Lines may be blank at random.
#  Words will be capitalized at random. The probabilities for these choices
#  and other parameters can be changed below.
#
#  To run this program on Linux do: 
#
#               python3 RandomInput1.py 
#
#------------------------------------------------------------------------------
from random import uniform, randint, choice, choices  

blankLineProbability  = 0.20  
delimProbability      = 0.30
capitalizeProbability = 0.25
minLineLength = 20 
maxLineLength = 1000
numLines = 200

wordfile = open('English.txt','r')  # file English.txt must exist

# get file name and open
fileName = input('Enter name of input file to be created: ')
file  = open(fileName,'w')

# define delimiter chars, leave out newline
D = list(" \t\\\"\',<.>/?;:[{]}|`~!@#$%^&*()-_=+0123456789")

# read lines of wordfile, create a list of words with repetition
F = wordfile.readlines()
wordfile.close()
W = [word.strip() for word in F]
W = choices(W, k=len(W)//50)  # a subset
W = choices(W, k=len(W)*20)   # a subset, with repetitions

# create random input file
for i in range(int(numLines)):

   # chose blank or number of chars in this line
   if uniform(0,1) < blankLineProbability:
      file.write('\n')
      continue
   else:  # non-blank line
      numchars = randint(minLineLength, maxLineLength)

   # create random line
   line = [choice(W)]    # pick a random word
   count = len(line[0])  # number of chars in this line
   while count < numchars:
      if uniform(0,1) < delimProbability:
         line.append(choice(D))
         count += 1
      else:
         word = choice(W)
         if uniform(0,1) < capitalizeProbability: 
            char_list = list(word)
            char_list[0] = char_list[0].upper()
            word = ''.join(char_list)
         item = ' '+word
         line.append(item)
         count += len(item)
   if count > numchars:
      item = line.pop()
      count -= len(item)

   # write line to file
   file.write(''.join(line)+'\n')

# close file
file.close()