Abstract

This PEP proposes to change the str() built-in function so that it can return unicode strings. This change would make it easier to write code that works with either string type and would also make some existing code handle unicode strings. The C function PyObject_Str() would remain unchanged and the function PyString_New() would be added instead.

Rationale

Python has had a Unicode string type for some time now but use of it is not yet widespread. There is a large amount of Python code that assumes that string data is represented as str instances. The long-term plan for Python is to phase out the str type and use unicode for all string data. Clearly, a smooth migration path must be provided.

We need to upgrade existing libraries, written for str instances, to be made capable of operating in an all-unicode string world. We can’t change to an all-unicode world until all essential libraries are made capable for it. Upgrading the libraries in one shot does not seem feasible. A more realistic strategy is to individually make the libraries capable of operating on unicode strings while preserving their current all-str environment behaviour.

First, we need to be able to write code that can accept unicode instances without attempting to coerce them to str instances. Let us label such code as Unicode-safe. Unicode-safe libraries can be used in an all-unicode world.

Second, we need to be able to write code that, when provided only str instances, will not create unicode results. Let us label such code as str-stable. Libraries that are str-stable can be used by libraries and applications that are not yet Unicode-safe.

Sometimes it is simple to write code that is both str-stable and Unicode-safe. For example, the following function just works:

def appendx(s):
    return s + 'x'

That’s not too surprising since the unicode type is designed to make the task easier. The principle is that when str and unicode instances meet, the result is a unicode instance. One notable difficulty arises when code requires a string representation of an object; an operation traditionally accomplished by using the str() built-in function.

Using the current str() function makes the code not Unicode-safe. Replacing a str() call with a unicode() call makes the code not str-stable. Changing str() so that it could return unicode instances would solve this problem. As a further benefit, some code that is currently not Unicode-safe because it uses str() would become Unicode-safe.

Specification

A Python implementation of the str() built-in follows:

def str(s):
    """Return a nice string representation of the object.  The
    return value is a str or unicode instance.
    """
    if type(s) is str or type(s) is unicode:
        return s
    r = s.__str__()
    if not isinstance(r, (str, unicode)):
        raise TypeError('__str__ returned non-string')
    return r

The following function would be added to the C API and would be the equivalent to the str() built-in (ideally it be called PyObject_Str, but changing that function could cause a massive number of compatibility problems):

PyObject *PyString_New(PyObject *);

A reference implementation is available on Sourceforge [1] as a patch.

Backwards Compatibility

Some code may require that str() returns a str instance. In the standard library, only one such case has been found so far. The function email.header_decode() requires a str instance and the email.Header.decode_header() function tries to ensure this by calling str() on its argument. The code was fixed by changing the line “header = str(header)” to:

if isinstance(header, unicode):
    header = header.encode('ascii')

Whether this is truly a bug is questionable since decode_header() really operates on byte strings, not character strings. Code that passes it a unicode instance could itself be considered buggy.

Alternative Solutions

A new built-in function could be added instead of changing str(). Doing so would introduce virtually no backwards compatibility problems. However, since the compatibility problems are expected to rare, changing str() seems preferable to adding a new built-in.

The basestring type could be changed to have the proposed behaviour, rather than changing str(). However, that would be confusing behaviour for an abstract base type.

References

[1]

https://bugs.python.org/issue1266570

Copyright

This document has been placed in the public domain.