pyCS-Grads-Analysis

def obtainFile():

    # MJK I'm not keen on arbitrary flags to control flow.
    # For me a big feature of functions is the ability to exit at any point when a return condition is found.
    # (I admit, not everyone agrees with this! But for me it makes code shorter and clearer).
    # Here I use break to exit the loop, then we can return

    while True:                     # Keep looping until a valid file is given
        inFile = input('Please enter the name of the data file: ')

        # MJK I think it is worth adding a check that the file entered is a .csv filw
        if not inFile.lower().endswith('.csv'):
            print("Must be a .csv file, try again ...")
            continue

        try:
            newFile = open(inFile, 'r')
            break

        except IOError:
            print("Invalid file name, try again ...")

    return newFile  # MJK Exit from function when valid file name is input


def getData():
    yearList = []
    totalList = []
    menList = []
    womenList = []
    inFile = obtainFile()               # MJK Get the file handle of the .csv file

    lineReader = inFile.readline()      # MJK Read and skip the header row

    # MJK Read data rows of the .csv file
    for lineReader in inFile:
        fields = lineReader.split(',')              # MJK read one row and split into fields
        if len(fields) < 4:
            continue                                # expecting at least 4 columns per row

        # get these 4 columns individually, as we are not sure how many columns actually exist in the .csv
        year, total, men, women = fields[0], fields[1], fields[2], fields[3]

        # Note: Probably a good idea to use .strip() on each column data
        # We could put these lines within a try: in case of any bad data
        yearList.append(int(year.strip()))
        totalList.append(int(total.strip()))
        menList.append(int(men.strip()))
        womenList.append(int(women.strip()))

    inFile.close()
    return yearList, totalList, menList, womenList              # Returning all data (same func for Q1 and Q2)


# MJK New utility function to get any integer from user input
def getInt(prompt):
    while True:                             # Loop until an integer is entered
        user_in = input(prompt)
        if user_in.isdigit():               # check that the input consists of digits only
            break
        print("Non-numeric input. Try again\n")

    return int(user_in)                     # Return integer value when an integer is input

# MJK Here is a function to get a year within the required range (range as defined by the .CSV file data itself)
# yearList is the full list of years extracted from the CSV file
# For Question 2, year must be between earliestYear+1, and latestYear, inclusive
# This is better than using a fixed range of 1972 to 2011 inclusive
def getYear(prompt, yearList, question):
    # Find the minimum year, and the maximum year, from the list of years in the CSV file
    minYear = min(yearList)
    if question == 2:                       # for Q2, minimum allowed year is earliest year +1
        minYear += 1
    maxYear = max(yearList)                 # Get the latest year from the CSV

    while True:
        y = getInt(prompt)                  # Get a valid integer from user input
        if minYear <= y <=  maxYear:        # Check user year within required range
            break

        print("Year out of range. Try again\n")

    return y                    # Exit when a valid year is given


# This function is for Question 1 only
# This returns 2 years, with the 2nd being greater than the first
# yearList is the complete list of years found in the CSV file
def getYears(yearList, question):
    while True:
        year1 = getYear("Enter earlier of two years: ", yearList, question)
        year2 = getYear("Enter later of two years: ", yearList, question)
        if year2 > year1:
            break
        print("2nd year not greater than first. Try again\n")

    return year1, year2


def computePercentChange(yearList, year1, womenList):

    # storeFileName = ("womenGradChangeFrom" + str(year1) + ".csv")
    # MJK use f-string formatting is nicer
    storeFileName = f"womenGradChangeFrom{year1}.csv"
    f = open(storeFileName, "w")

    chosenYearIndex = yearList.index(year1)  # position of chosen year in list

    for i in range(chosenYearIndex, len(yearList)):
        difference = (womenList[i] - womenList[i - 1]) / womenList[i - 1]  #

        # MJK NOTE: They have asked for a percentage difference, so multiple ratio by 100
        percentDiff = 100 * difference

        # f.write(str((yearList[i])) + ',' + str(percentDiff) + '\n')
        # MJK Again, f-string formatting is nice
        f.write(f"{yearList[i]},{percentDiff}\n")
    f.close()
    print("Your data has been written to the file " + storeFileName)


def main():
    # MJK: Use the same function for Q1 and Q2 for getting data
    yearList, totalList, menList, womenList = getData()

    year1 = getYear("Enter a year: ", yearList, 2)          # Question 2
    computePercentChange(yearList, year1, womenList)


main()