anlpPython

#!/usr/bin/python

# Each stage of the process is numbered and printed it so you can see the conversion at each step
# The print method also has a number in the form of a string, so it shows up in the console as documented.
#i.e print( 'Number. ' + printedVariable )


#imports
import re
import sys
from random import random
from math import log
from collections import defaultdict

#0. This is the original string, it is printed in the main body of the code below
input_string = "ASd1234567890 . @$$$$£ rupaulisbae  %" # Changed the input text to test the number function

#Creates a dictionary for counting the trigrams
tri_counts=defaultdict(int) #counts of all trigrams in input

## Checks that there is an input file to be read
#if len(sys.argv) != 2:
#    print("Usage: ", sys.argv[0], "<training_file>")
#    sys.exit(1)

#infile = sys.argv[1] #get input argument: the training file
##

def preprocess_line(inputText): # Had to change input to a different name, as input is already pre-defined by python
	# 1. make it lower case
	lower_input = inputText.lower();
	print('1. ' + lower_input)

	# 2. & 3. replace 1-9 -> 0
	# Replace 1-9 -> 0
		# It's not the most elegant but it does work, and we could refine it once it's all working
		# Takes the previous input of the text and outputs a zero in place of the other number
		# Cycles through this until it has been completed
			# See print(numConversion1) and print(numConversionDone)
	numConversion1 = lower_input.replace('1', '0')
	print('2. ' + numConversion1)

	numConversion2 = numConversion1.replace('2', '0')
	numConversion3 = numConversion2.replace('3', '0')
	numConversion4 = numConversion3.replace('4', '0')
	numConversion5 = numConversion4.replace('5', '0')
	numConversion6 = numConversion5.replace('6', '0')
	numConversion7 = numConversion6.replace('7', '0')
	numConversion8 = numConversion7.replace('8', '0')
	numConversion9 = numConversion8.replace('9', '0')

	numConversionDone = numConversion9
	print('3. '+ numConversionDone) #    - here for demonstration of num conversion

	#4. Performs a regular expression, finds characters that match a-z, 0, [space] and [.]
	#   Saves each value as an item in a list (converted to a string below)
	letters_numbers = re.findall('([a-z0\.\ ])',numConversionDone)
	print('4. '+ str(letters_numbers))
	#   Converts the above list as a string
	letters_numbers = "".join(letters_numbers)

	outputText = letters_numbers
	# an outputText variable is created and returned as the output of our function
	return outputText


#________MAIN BODY OF THE CODE________________________________________________________________
print('0. ' + input_string)

# 5. A variable is created with the returned variable (outputText)
#    and stored in a new variable preProcessedText
#    This is then printed in  the console
pre_processed_text = preprocess_line(input_string)
print('5. ' + pre_processed_text)

#This bit of code gives an example of how you might extract trigram counts
#from a file, line by line. If you plan to use or modify this code,
#please ensure you understand what it is actually doing, especially at the
#beginning and end of each line. Depending on how you write the rest of
#your program, you may need to modify this code.

# with open(infile) as file:
#    list_lines = preprocess_file(file)  # --------preprocess the file, splitting the file into a list of new lines

##Sorts
test_string_tri_counts = input_string
print('6. ' + input_string)

for line in input_string:
	line = preprocess_line(input_string)  # doesn't do anything yet.
	print(line)
	for j in range(len(line) - (3)):
		trigram = line[j:j + 3]
		tri_counts[trigram] += 1
print(tri_counts)



print("Trigram counts in ", test_string_tri_counts, ", sorted alphabetically:")


# replaced_input = re.sub()