#!/usr/bin/python

"""
This program analyzes the a text that is given to it, counting every
individual word. It is then able to report on them to the user. It can also
give basic statistics on the hash table, such as its size, load factor, etc.

It relies on a HashTable object, that you must write. The API is as given in
the assignment.
"""

__author__ =  "Adam A. Smith"
__version__ = "2023.09.28"

from hashtable import HashTable
import re

# load a file & make a HashTable from its words
def _make_hash_table(filename:str) -> HashTable:
    hash_table = HashTable()

    with open(filename) as file:
        for line in file:
            tokens = re.split(r"[^A-Za-z0-9'_]+", line.lower())

            for token in tokens:
                # remove leading/trailing apostrophes
                if token.startswith("'"): token = re.sub(r"^'+", "", token)
                if token.endswith("'"): token = re.sub(r"'+$", "", token)

                # skip if it was just apostrophes or otherwise empty
                if len(token) == 0: continue

                # add to table
                if (hash_table.contains_key(token)):
                    freq = hash_table.get(token) + 1
                    hash_table.put(token, freq)
                    
                else: hash_table.put(token, 1)

    return hash_table

# enter a loop to query the user for searches & commands
def _query_user(hash_table : HashTable):
    print("The text contains " +str(len(hash_table))+ " unique words.")
    print('Please enter a word to query, or "!help" for help, or "!exit" to exit.')
    while True:
        print("> ", end="")
        query = input()

        # do nothing on empty string
        if len(query) == 0: pass

        # quit on "!quit" or "!exit"
        elif query == "!quit" or query == "!exit": break
        
        # check for size
        elif query == "#": print("The text contains " +str(len(hash_table))+ " unique words.")

        # print stats on the backing table
        elif query == "!stats":
            print("Hash table statistics:")
            print("\tSize (n): " +str(len(hash_table)))
            print("\tTable size (m): " +str(hash_table.get_table_size()))
            print("\tLoad factor (α):", "{:1.3f}".format(hash_table.get_load_factor()))
            print("\tEmpty units: " +str(hash_table.count_empty_slots()))
            print("\tLongest run: " +str(hash_table.find_longest_run()))

        # help menu
        elif query == "!help": _print_help_menu()

        # unknown special command
        elif query.startswith("!"):
            print("I didn't recognize that. Try \"!help\"?")

        # deletion
        elif query.startswith("-"):
            word_to_delete = query[1:]
            if len(word_to_delete) == 0: print("What word do you want to delete? Please try again.")
            else:
                value = hash_table.delete(word_to_delete)
                if (value is None): print('"' +word_to_delete+ '" was not present.')
                elif value == 1: print('1 entry of "' +word_to_delete+ '" has been deleted.')
                else: print(str(value) + ' entries of "' +word_to_delete+ '" have been deleted.')

        # add a word or increase its frequency
        elif query.startswith("+"):
            new_word = query[1:]
            if len(new_word) == 0: print("What word do you want to add? Please try again.")
            elif hash_table.contains_key(new_word):
                freq = hash_table.get(new_word) + 1
                hash_table.put(new_word, freq)
                print('"' +new_word+ '" now appears ' +str(freq)+ '×.')
            else:
                hash_table.put(new_word, 1)
                print('"' +new_word+ '" now appears 1×.')

        # regular query
        else:
            freq = hash_table.get(query)
            if freq is None or freq == 0: print('"' +query+ '" is not in the text.')
            else: print('"' +query+ '" appears ' +str(freq)+ '× in the text.')

    print("Goodbye!")

# just prints a help menu
def _print_help_menu():
    print("\t<word>\t\tprints word frequency")
    print("\t-<word>\t\tdeletes word")
    print("\t+<word>\tadds word")
    print("\t#\t\tprints the table size")
    print("\t!help\t\tprints this help menu")
    print("\t!stats\t\tprints in-depth stats on the table")
    print("\t!exit\t\texits the program")

# pseudo main() function
if __name__ == "__main__":
    # test command-line args
    from sys import argv, exit, stderr

    # make sure we have a file
    if len(argv) != 2:
        print("Please enter a file name!", file=stderr)
        exit(0)

    # make the hash, and query on it
    try:
        hash_table = _make_hash_table(argv[1])
        _query_user(hash_table)
    except FileNotFoundError:
        print("Couldn't open file \"" +argv[1]+ "\".", file=stderr)
        exit(1)