# Program to count word frequencies in a file
# Sort by frequency count, high to low
#
# Matt Bishop, MHI 289I, Fall 2024

#
# get the input file name
#
def getfname():
    try:
        fname = input("Enter the file name: ")
    except EOFError:
        # exit gracefully
        return None
    except Exception as msg:
        # report problem and exit gracefully
        print("Error reading file name:", msg)
        return None
    return fname


#
# open the file; be graceful on failure
#
def openfile(fname):
    try:
        inpf = open(fname)
    except Exception as msg:
        print("Could not open file", fname+":", msg)
        inpf = None
    return inpf

#
# this does the actual work
#
def main():
    #
    # get file name
    #
    fname = getfname()
    if fname == None:
        return
    #
    # open the file
    #
    fc = openfile(fname)
    if fc == None:
        return
    #
    # create empty dictionary
    #
    d = dict()

    # read the fileline by line, and get
    # the words
    # words are separated by whitespace
    for line in fc.readlines():
        # split into words
        words=line.split(' ')
        for tmpword in words:
            # now eliminate all non-alphanumerics
            word = ""
            for i in tmpword:
                if i.isalnum():
                    word = word + i
            # if we don't have anything, skip it
            # otherwise, enter it into the dictionary
            if word != "":
                d[word] = d.get(word, 0) + 1
    #
    # sort on the keys
    #
    worditems = list(d.items())
    #
    # print them out
    #
    for k in sorted(worditems, key=lambda x: x[1], reverse=True):
        print("%-30s%5d" % k)
    
#
# go!
#
main()