ascii() was always materializing a new plain str, which discarded str subclass types even when __repr__ had already returned an ASCII-only string. This diverged from CPython, which returns the original repr object unchanged in that case.

• Behavior change

  • Update builtins.ascii() and PyObject::ascii() to short-circuit when repr(obj) is already ASCII.
  • Return the original repr object directly in that branch, preserving str subclass identity.
  • Keep the existing escape path for non-ASCII repr output, which still produces a new escaped plain str.

• Call site updates

  • Adjust internal %a / format-conversion consumers to work with a returned PyStr object instead of an AsciiString.
  • Preserve existing formatting behavior while allowing subclass-preserving results to flow through unchanged where appropriate.

• Regression coverage

  • Unskip the existing CPython-derived test_ascii case in Lib/test/test_str.py.
  • Add focused snippet coverage for both branches:
    • ASCII repr from a str subclass preserves the subclass
    • non-ASCII repr still escapes to a plain str

class MyStr(str):
    pass

class Foo:
    def __repr__(self):
        return MyStr("hello")

class Bar:
    def __repr__(self):
        return MyStr("héllo")

type(ascii(Foo()))  # MyStr
type(ascii(Bar()))  # str

---

This section details on the original issue you should resolve

<issue_title>ascii() discards str subclass type when repr result is already ASCII</issue_title>
<issue_description>## Description

ascii() always creates a new plain str from the repr result, even when the repr is already ASCII. CPython's PyObject_ASCII returns the repr object directly when it's already ASCII, preserving str subclass types. This is a CPython compatibility issue.

Root Cause

builtin_ascii() unconditionally converts the repr result to a new AsciiString via to_ascii(), which then becomes a plain str through ToPyObject. The same pattern exists in PyObject::ascii().

CPython reference: object.c#L848-L849

// Key branch: return repr directly when it's already ASCII
if (PyUnicode_IS_ASCII(repr))
    return repr;

Fix

Add an ASCII check before to_ascii() — if the repr is already ASCII, return the repr object directly instead of creating a new string.

Reproduction

class MyStr(str):
    pass

class Foo:
    def __repr__(self):
        return MyStr('hello')

result = ascii(Foo())
print(type(result))

Output

RustPython:

<class 'str'>

CPython:

<class '__main__.MyStr'>

Environment

• RustPython d248a04 (Python 3.14.0)
• CPython 3.14.3
• OS: Debian 12</issue_description>

Comments on the Issue (you are @copilot in this section)

• Fixes ascii() discards str subclass type when repr result is already ASCII #7451

📱 Kick off Copilot coding agent tasks wherever you are with GitHub Mobile, available on iOS and Android.

Summary by CodeRabbit

• Bug Fixes

  • Refined the ascii() builtin function to properly handle cases where __repr__ returns a string subclass, preserving the return type when the representation is already in ASCII format.

• Tests

  • Added test coverage for ascii() behavior with custom string subclasses.