The pprint module employs a key-value sorting when representing dict obj.

[Stranger-Jie]

Feature or enhancement

Proposal:

The pprint module in Python currently provides a sort_dicts parameter (defaulting to True) to control whether the keys of dict are sorted during output. This sorting method is based solely on the keys of the dict, disregarding the importance of the values. However, in certain scenarios, the keys of a dictionary are merely labels, and users are more concerned with the values associated with these labels. These values might often be class objects that have overridden comparison methods. The current implementation does not meet the need for sorting by values or custom sorting based on values. The use of the value sorting feature for dict in pprint not only retains the efficiency and aesthetics of pprint's output but also avoids potential errors that might be introduced when users manually implement sorting.

from pprint import pprint


class FinalExamResults:
    def __init__(self, score: float, name: str = "") -> None:
        self.score = score  # example

    def __lt__(self, other):
        return self.score < other.score

    def __repr__(self):
        return f"{self.score}"


classmates = {
    "Alice": FinalExamResults(4.0),
    "Bob": FinalExamResults(3.5),
    "Charlie": FinalExamResults(4.4),
    "David": FinalExamResults(3.2),
}


pprint(classmates)
# {'Alice': 4.0, 'Bob': 3.5, 'Charlie': 4.4, 'David': 3.2}

pprint(classmates, sort_dicts_by_key=False)
# {'Charlie': 4.4, 'Alice': 4.0, 'Bob': 3.5, 'David': 3.2}

Has this already been discussed elsewhere?

This is a minor feature, which does not need previous discussion elsewhere

Links to previous discussion of this feature:

pprint does not keep the order of dict keys #115985
Bug: pprint module print dict object in wrong key sequence #98232

Linked PRs

• gh-137448: Add sort by key or value to dict obj #137446

[skirpichev]

Why just not use sorted() on input or provide a key parameter with function?

[Stranger-Jie]

Thank you, I have considered this problem. When using sorted before pprint, users often forget to sort_dicts=False, which leads to a secondary sorting. If pprint sorts the values, it could enhance the usability of pprint, especially when the values are significant.

[terryjreedy]

pprint.pprint is now an alias for pprint.pp with sort_dicts = True instead of False, so use pp when sorting not wanted. In any case, sorted returns a list.

from pprint import pp

class FinalExamResults:
    def __init__(self, score: float, name: str = "") -> None:
        self.score = score  # example

    def __lt__(self, other):
        return self.score < other.score

    def __repr__(self):
        return f"{self.score}"

classmates = {
    "Alice": FinalExamResults(4.0),
    "Bob": FinalExamResults(3.5),
    "Charlie": FinalExamResults(4.4),
    "David": FinalExamResults(3.2),
}

pp(dict(sorted(classmates.items(), key=lambda item: item[1])))
# {'David': 3.2, 'Bob': 3.5, 'Alice': 4.0, 'Charlie': 4.4}

The sorted call can include reverse=True and one can feed the output list of tuples into pp directly or a custom transformer other than dict.

pairlist = sorted(classmates.items(), key=lambda it: repr(it[1]), reverse=True)
pp([f"{score} - {name}" for name, score in pairlist])
# ['4.4 - Charlie', '4.0 - Alice', '3.5 - Bob', '3.2 - David']