###########################################################

#   Computer Project #6
#
#   open_file()
#       prompt for input
#       check that input is correct (loop if not)
#       return fp
#
#   read_file(fp)
#       open csv reader
#       for each line in reader
#           if correct data exists
#               create tuple of specific data
#               append to list
#       return list
#
#   remove_duplicate_sites(list)
#       for each tuple in list
#       if site seen before
#           add it to main & seen list
#       else
#           ignore
#       return list
#
#   top_sites_per_country(list, country)
#       for each tuple in list
#           if country = [4]
#               add to new list
#       return list
#
#   top_sites_per_view(list)
#       sort list by views
#       remove_duplicate_sites(list)
#       sort list by views
#       return list
#
#   main()
#       open_file()
#       generate list by read_file(fp)
#       while true
#           prompt
#           ask user for input
#           if input is 1
#               ask for country
#               for sites in top_sites_per_country()
#                   print with formatting
#           elif input is 2
#               prompt for search
#                   for sites in full_list
#                       if search = full_list[1]
#                           print with formatting
#                   if no sites found
#                       print message
#           elif input is 3
#               for sites in top_sites_per_view()
#                   print with formatting
#           elif input is q
#               exit message
#               kill loop
#           else
#               prompt again

###########################################################


import csv  # Used for CSV Reader
from operator import itemgetter  # Used for sorting

PROMPT = '''
Choose
         (1) Top sites by country
         (2) Search by web site name
         (3) Top sites by views
         (q) Quit
         '''  # Main prompt


def open_file():
    """
    This function opens a file for reading
    Returns: file pointer
    """
    cont = True
    # loop to ensure we open a file correctly
    while cont:
        try_file = input("Input a filename: ")
        try:  # try/except to ensure file exists and opens correctly
            file_obj = open(try_file, "r", encoding="ISO-8859-1")
            cont = False
        except:
            print("Error: file not found.")
            cont = True
    return file_obj  # return the correct file


def read_file(fp):
    """
    This function reads file from open_file() and collects interesting data
    Returns: sorted list of tuples
    """
    website_list = []  # functional list
    reader = csv.reader(fp)  # open csv reader
    next(reader, None)  # skip header
    for row in reader:  # iterate for evert tuple
        try:
            row_min = (int(row[0]), row[1], int(row[14].replace(" ", "")),
                       int(row[5].replace(" ", "")),
                       row[30])  # interesting data
            website_list.append(row_min)  # append to main list
        except:
            pass  # ignore rows that fail that test - ex: int(N/A)
    return sorted(website_list, key=itemgetter(0, 4))  # double sorted list


def remove_duplicate_sites(L_of_L):
    """
    This function iterates through a list, and if a website has been seen before
    it gets ignored and disregarded from output (deduplication)
    Returns: sorted list of tuples
    """
    seen_sites, new_list = [], []  # create seen list and list to be sorted later
    for entry in L_of_L:
        site = entry[1].split('.')  # split website to actual name
        if site[1] in seen_sites:  # check if in list
            pass
        else:
            seen_sites.append(site[1])  # add to list
            new_list.append(entry)  # add to list
    return sorted(new_list, key=itemgetter(0, 1))  # return double sorted list


def top_sites_per_country(L_of_L, country):
    """
    This function itterates through the list and checks for top sites based on
    user input (country)
    Returns: list of 20 tuples
    """
    new_list = []
    for entry in L_of_L:
        if entry[4] == country:  # check if countries match
            new_list.append(entry)
    new_list = sorted(new_list, key=itemgetter(0))  # single sort
    return new_list[:20]  # return first 20


def top_sites_per_views(L_of_L):
    """
    This function sorts a list by top views, deduplicated, then sorts again
    Returns: list of 20 tuples
    """
    L_of_L.sort(reverse=True, key=itemgetter(3))  # sort by views
    L_of_L = remove_duplicate_sites(L_of_L)  # deduplicate
    L_of_L.sort(reverse=True, key=itemgetter(3))  # sort by views again
    return L_of_L[:20]  # return first 20


def main():
    """
    This function asks user for input, then loops through options ( see PROMPT )
    then runs functions based on the user input, while formatting the data in a
    way that is uniform.
    Returns: None
    """
    print("----- Web Data -----")
    file_pointer = open_file()  # open file
    full_list = read_file(file_pointer)  # save list before loop
    cont = True
    while cont:
        print(PROMPT)
        user_input = input("Choice: ")  # user input
        if user_input == "1":
            print("--------- Top 20 by Country -----------")
            country = input("Country: ")
            print("{:30s} {:>15s}{:>25s}".format("Website", "Traffic Rank",
                                                 "Average Daily Page Views"))
            for x in top_sites_per_country(full_list, country):
                print("{:30s} {:>15d}{:>30,d}".format(x[1], x[2], x[3]))
        elif user_input == "2":
            c = 0  # Add counter to check if none found
            search = input("Search: ").lower()
            print("{:^50s}".format("Websites Matching Query"))
            for x in full_list:
                if search in x[1]:
                    print("{:^10s}".format(x[1]))
                    c += 1
            if c == 0:
                print("None found")
        elif user_input == "3":
            print("--------- Top 20 by Page View -----------")
            print("{:30s} {:>15s}".format("Website", "Ave Daily Page Views"))
            for x in top_sites_per_views(full_list):
                print("{:30s} {:>15,d}".format(x[1], x[3]))
        elif user_input.lower() == "q":
            cont = False
        else:
            print("Incorrect input. Try again.")


if __name__ == "__main__":  # Initialize the function
    main()  # Start function