collections.Counter.most_common stops working correctly in 3 digit numbers

Question

I am working on a discord.py bot and when I am trying to sort the balance of each user for a "top balance" command, it stops sorting correctly if keys have 3 digit numbers.

# bot and discord are defined above here, btw
@bot.command(name='top')
async def top(context):
    scores.read("brewscores.ini")
    tops = scores['scores']
    print('hi\n', tops)
    print('sotred')
    from collections import Counter #todo : move this somewhere else
    c = Counter(tops)
    a = c.most_common(5) #//https://stackoverflow.com/a/40496562/9654083
    string = """"""
    await context.send("Loading balancers...")
    for item in a:
        print(item)
        g, s = item
        string += (f"{g}: {s}\n")
    em = discord.Embed(title="Top 5 Balancers", description=f'The top 5 contestants are!:{string}')
    await context.send(embed=em)

The scores:

[scores]
placeholder = 0
(username censored for privacy) = 35
(username censored for privacy) = 49
No other balancers! = 0
You can stop reading now... = 0
rats#3234 = 100

Output both me and the owner get in Discord when running the command:

(censored for privacy): 49
(censored for privacy): 35
rats#3234: 100
placeholder: 0
you can stop reading now...: 0

The one we expect:

rats#3234: 100
(censored for privacy): 49
(censored for privacy): 35
placeholder: 0
you can stop reading now...: 0

Notice how "rats#3234" was misplaced. We did not see this behaviour in 2- or 1-digit numbers like 99 or 3. Why does this occur?

Can you provide a snippet with the content of `tops` that work as you expect and one that does not? How are you splitting the score from the name? What are you counting? — Pietro, Mar 26 '21 at 16:48
Not exactly: do you want to sort the rows of the file based on the score? What is inside `scores` and `tops` after you read the `ini` file? Even more important to reproduce your code, what is `scores`? — Pietro, Mar 26 '21 at 16:55
`scores` is the ini file read. `tops` is the contents of the `scores` part of `scores`. I actually have a solution, I'll provide an answer just now. — TheTechRobo Stands for Ukraine, Mar 26 '21 at 16:58
Ok cool, btw I meant `scores.read("brewscores.ini")`: where do you define `scores`? What python object is that? — Pietro, Mar 26 '21 at 17:00
Ok that helped: I think that the problem is that when you read the `.ini` file, the values are interpreted as `str`. That is why the sort does not work, `100` comes before `99` if you order lexicographically and not as numbers. — Pietro, Mar 26 '21 at 17:07
Ohhh... so our answer is only correct when it's strings! Thanks, I'll update the answer. :D — TheTechRobo Stands for Ukraine, Mar 26 '21 at 17:07

score 1 · Accepted Answer · answered Mar 26 '21 at 17:15

There is no need for Counter:

scores.read("brewscores.ini")
print(scores)
tops = scores["scores"]
sortedtopkeys = sorted(tops, key=lambda k: int(tops[k]), reverse=True)
print(sortedtopkeys)

This way you sort the dict on the values, as shown in the answer you have in the comment, but interpreting the value as an int.

Cheers!

TheTechRobo Stands for Ukraine · Answer 2 · 2021-03-26T20:43:02.037

update

see https://stackoverflow.com/a/66821488/9654083 for a better answer if you don't care about the details of why this error occurs. imo this answer best supplements that answer, not replace.

So it turns out that when it's a string, collections.Counter.most_common looks at the first digit first to figure out which is greater, then the second, then the third. Not it all together.

So that means, this:

[scores]
placeholder = 0
(censored) = 35
(censored) = 49
rats#3234 = 9

will put rats first since it sees it as

[scores]
placeholder = 0
(censored) = 3
(censored) = 4
rats#3234 = 9

If there are two ones with the same first digit, it will continue by comparing the first and second digit:

[scores]
placeholder = 0
(censored for privacy) = 35
(censored) = 49

instead of just the first digit, and so on.

This means that contrary to what we thought, it is not just because it's 3 digits. It's because it has the largest first digit.

The other developer found this out because 9 being the value of rats would output

rats#3234: 9
(censored): 49
(censored): 35
placeholder: 0

This means that we should have to convert to int before sorting.

Give me one second while I test.

Try to convert the values to `int` – Pietro Mar 26 '21 at 17:07 — Pietro, Mar 26 '21 at 17:07
I've updated the answer. i"ll test in a few minutes. – TheTechRobo Stands for Ukraine Mar 26 '21 at 17:09 — TheTechRobo Stands for Ukraine, Mar 26 '21 at 17:09

collections.Counter.most_common stops working correctly in 3 digit numbers

2 Answers2

update