Python: creating a dictionary that writes high scores to a file

Question

First: you don't have to code this for me, unless you're a super awesome nice guy. But since you're all great at programming and understand it so much better than me and all, it might just be easier (since it's probably not too many lines of code) than writing paragraph after paragraph trying to make me understand it.

So - I need to make a list of high scores that updates itself upon new entries. So here it goes:

First step - done

I have player-entered input, which has been taken as a data for a few calculations:

import time
import datetime

print "Current time:", time1.strftime("%d.%m.%Y, %H:%M")
time1 = datetime.datetime.now()
a = raw_input("Enter weight: ")    
b = raw_input("Enter height: ")
c = a/b

Second step - making high score list

Here, I would need some sort of a dictionary or a thing that would read the previous entries and check if the score (c) is (at least) better than the score of the last one in "high scores", and if it is, it would prompt you to enter your name.

After you entered your name, it would post your name, your a, b, c, and time in a high score list.

This is what I came up with, and it definitely doesn't work:

list = [("CPU", 200, 100, 2, time1)]
player = "CPU"
a = 200
b = 100
c = 2
time1 = "20.12.2012, 21:38"
list.append((player, a, b, c, time1))
list.sort()

import pickle
scores = open("scores", "w")
pickle.dump(list[-5:], scores)
scores.close()

scores = open("scores", "r")
oldscores = pickle.load(scores)
scores.close()
print oldscores()

I know I did something terribly stupid, but anyways, thanks for reading this and I hope you can help me out with this one. :-)

Answer 1

First, don't use list as a variable name. It shadows the built-in list object. Second, avoid using just plain date strings, since it is much easier to work with datetime objects, which support proper comparisons and easy conversions.

Here is a full example of your code, with individual functions to help divide up the steps. I am trying not to use any more advanced modules or functionality, since you are obviously just learning:

import os
import datetime
import cPickle

# just a constants we can use to define our score file location
SCORES_FILE = "scores.pickle"

def get_user_data():
    time1 = datetime.datetime.now()
    print "Current time:", time1.strftime("%d.%m.%Y, %H:%M")

    a = None
    while True:
        a = raw_input("Enter weight: ")    
        try:
            a = float(a)
        except:
            continue
        else:
            break

    b = None
    while True:
        b = raw_input("Enter height: ")    
        try:
            b = float(b)
        except:
            continue
        else:
            break

    c = a/b

    return ['', a, b, c, time1]

def read_high_scores():
    # initialize an empty score file if it does
    # not exist already, and return an empty list
    if not os.path.isfile(SCORES_FILE):
        write_high_scores([])
        return []

    with open(SCORES_FILE, 'r') as f:
        scores = cPickle.load(f)
    return scores

def write_high_scores(scores):
    with open(SCORES_FILE, 'w') as f:
        cPickle.dump(scores, f)

def update_scores(newScore, highScores):
    # reuse an anonymous function for looking
    # up the `c` (4th item) score from the object
    key = lambda item: item[3]

    # make a local copy of the scores
    highScores = highScores[:]

    lowest = None
    if highScores:
        lowest = min(highScores, key=key)

    # only add the new score if the high scores
    # are empty, or it beats the lowest one
    if lowest is None or (newScore[3] > lowest[3]):
        newScore[0] = raw_input("Enter name: ")
        highScores.append(newScore)

    # take only the highest 5 scores and return them
    highScores.sort(key=key, reverse=True)
    return highScores[:5]

def print_high_scores(scores):
    # loop over scores using enumerate to also
    # get an int counter for printing
    for i, score in enumerate(scores):
        name, a, b, c, time1 = score
        # #1    50.0    jdi    (20.12.2012, 15:02)
        print "#%d\t%s\t%s\t(%s)" % \
            (i+1, c, name, time1.strftime("%d.%m.%Y, %H:%M"))


def main():
    score = get_user_data()
    highScores = read_high_scores()

    highScores = update_scores(score, highScores)

    write_high_scores(highScores)
    print_high_scores(highScores)

if __name__ == "__main__":
    main()

What it does now is only add new scores if there were no high scores or it beats the lowest. You could modify it to always add a new score if there are less than 5 previous scores, instead of requiring it to beat the lowest one. And then just perform the lowest check after the size of highscores >= 5

Answer 2

You definitely don't want a dictionary here. The whole point of a dictionary is to be able to map keys to values, without any sorting. What you want is a sorted list. And you've already got that.

Well, as Tamás points out, you've actually got a list sorted by the player name, not the score. On top of that, you want to sort in downward order, not upward. You could use the decorate-sort-undecorate pattern, or a key function, or whatever, but you need to do something. Also, you've put it in a variable named list, which is a very bad idea, because that's already the name of the list type.

Anyway, you can find out whether to add something into a sorted list, and where to insert it if so, using the bisect module in the standard library. But it's probably simpler to just use something like SortedCollection or blist.

Here's an example:

highscores = SortedCollection(scores, key=lambda x: -x[3])

Now, when you finish the game:

highscores.insert_right((player, a, b, newscore, time1))
del highscores[-1]

That's it. If you were actually not in the top 10, you'll be added at #11, then removed. If you were in the top 10, you'll be added, and the old #10 will now be #11 and be removed.

If you don't want to prepopulate the list with 10 fake scores the way old arcade games used to, just change it to this:

highscores.insert_right((player, a, b, newscore, time1))
del highscores[10:]

Now, if there were already 10 scores, when you get added, #11 will get deleted, but if there were only 3, nothing gets deleted, and now there are 4.

Meanwhile, I'm not sure why you're writing the new scores out to a pickle file, and then reading the same thing back in. You probably want to do the reading before adding the highscore to the list, and then do the writing after adding it.

You also asked how to "beautify the list". Well, there are three sides to that.

First of all, in the code, (player, a, b, c, time1) isn't very meaningful. Giving the variables better names would help, of course, but ultimately you still come down to the fact that when accessing list, you have to do entry[3] to get the score or entry[4] to get the time.

There are at least three ways to solve this:

• Store a list (or SortedCollection) of dicts instead of tuples. The code gets a bit more verbose, but a lot more readable. You write {'player': player, 'height': a, 'weight': b, 'score': c, 'time': time1}, and then when accessing the list, you do entry['score'] instead of entry[3].
• Use a collection of namedtuples. Now you can actually just insert ScoreEntry(player, a, b, c, time1), or you can insert ScoreEntry(player=player, height=a, weight=b, score=c, time=time1), whichever is more readable in a given case, and they both work the same way. And you can access entry.score or as entry[3], again using whichever is more readable.
• Write an explicit class for score entries. This is pretty similar to the previous one, but there's more code to write, and you can't do indexed access anymore, but on the plus side you don't have to understand namedtuple.

Second, if you just print the entries, they look like a mess. The way to deal with that is string formatting. Instead of print scores, you do something like this:

print '\n'.join("{}: height {}, weight {}, score {} at {}".format(entry) for entry in highscores)

If you're using a class or namedtuple instead of just a tuple, you can even format by name instead of by position, making the code much more readable.

Finally, the highscore file itself is an unreadable mess, because pickle is not meant for human consumption. If you want it to be human-readable, you have to pick a format, and write the code to serialize that format. Fortunately, the CSV format is pretty human-readable, and most of the code is already written for you in the csv module. (You may want to look at the DictReader and DictWriter classes, especially if you want to write a header line. Again, there's the tradeoff of a bit more code for a lot more readability.)

Answer 3

Here are the things I notice.

These lines seem to be in the wrong order:

print "Current time:", time1.strftime("%d.%m.%Y, %H:%M")
time1 = datetime.datetime.now()

When the user enters the height and weight, they are going to be read in as strings, not integers, so you will get a TypeError on this line:

c = a/b

You could solve this by casting a and b to float like so:

a = float(raw_input("Enter weight: "))

But you'll probably need to wrap this in a try/catch block, in case the user puts in garbage, basically anything that can't be cast to a float. Put the whole thing in a while block until they get it right.

So, something like this:

b = None
while b == None:
    try:
        b = float(raw_input("Enter height: "))
    except:
        print "Weight should be entered using only digits, like '187'"

So, on to the second part, you shouldn't use list as a variable name, since it's a builtin, I'll use high_scores.

# Add one default entry to the list
high_scores = [("CPU", 200, 100, 2, "20.12.2012, 4:20")]

You say you want to check the player score against the high score, to see if it's best, but if that's the case, why a list? Why not just a single entry? Anyhow, that's confusing me, not sure if you really want a high score list, or just one high score.

So, let's just add the score, no matter what:

Assume you've gotten their name into the name variable.

high_score.append((name, a, b, c, time1))

Then apply the other answer from @Tamás

Answer 4

The first thing I noticed is that you did not tell list.sort() that the sorting should be based on the last element of each entry. By default, list.sort() will use Python's default sorting order, which will sort entries based on the first element of each entry (i.e. the name), then mode on to the second element, the third element and so on. So, you have to tell list.sort() which item to use for sorting:

from operator import itemgetter
[...]
list.sort(key=itemgetter(3))

This will sort entries based on the item with index 3 in each tuple, i.e. the fourth item.

Also, print oldscores() will definitely not work since oldscores is not a function, hence you cannot call it with the () operator. print oldscores is probably better.