How to find out if Python is compiled with UCS-2 or UCS-4?

Question

Just what the title says.

$ ./configure --help | grep -i ucs
  --enable-unicode[=ucs[24]]

Searching the official documentation, I found this:

sys.maxunicode: An integer giving the largest supported code point for a Unicode character. The value of this depends on the configuration option that specifies whether Unicode characters are stored as UCS-2 or UCS-4.

What is not clear here is - which value(s) correspond to UCS-2 and UCS-4.

The code is expected to work on Python 2.6+.

Om

Answer 1

When built with --enable-unicode=ucs4:

>>> import sys
>>> print sys.maxunicode
1114111

When built with --enable-unicode=ucs2:

>>> import sys
>>> print sys.maxunicode
65535

Fai

This is not universally correct anymore for Python 3. See docs.python.org/3.4/c-api/unicode.html: Since the implementation of PEP 393 in Python 3.3, Unicode objects internally use a variety of representations. python.org/dev/peps/pep-0393 – Neology

@Jan-PhilipGehrcke: deficient_unicode_build = (sys.maxunicode < 0x10ffff) works on any Python version (even if the flexible internal representation is used where sys.maxunicode == 0x10ffff). The flexible representations allows to get correct results like ucs4 did on previous versions while using less memory than ucs4 in some cases. – Muscle

Answer 2

It's 0xFFFF (or 65535) for UCS-2, and 0x10FFFF (or 1114111) for UCS-4:

Py_UNICODE
PyUnicode_GetMax(void)
{
#ifdef Py_UNICODE_WIDE
    return 0x10FFFF;
#else
    /* This is actually an illegal character, so it should
       not be passed to unichr. */
    return 0xFFFF;
#endif
}

The maximum character in UCS-4 mode is defined by the maxmimum value representable in UTF-16.

Reni

Note that this function is not used to implement sys.maxunicode after python 3.3 (ie, all maintained version of python as of this comment) - it only concerns the size of the now-deprecated Py_UNICODE typedef. maxunicode originates from SET_SYS_FROM_STRING("maxunicode", PyLong_FromLong(0x10FFFF));. – Cardialgia

Answer 3

I had this same issue once. I documented it for myself on my wiki at

http://arcoleo.org/dsawiki/Wiki.jsp?page=Python%20UTF%20-%20UCS2%20or%20UCS4

I wrote -

import sys
sys.maxunicode > 65536 and 'UCS4' or 'UCS2'

Decoration

For anyone wondering what this does: it is an old (< Python 2.5) way of doing 'UCS4' if sys.maxunicode > 65536 else 'UCS2'. – Retaliation

Answer 4

sysconfig will tell the unicode size from the configuration variables of python.

The buildflags can be queried like this.

Python 2.7:

import sysconfig
sysconfig.get_config_var('Py_UNICODE_SIZE')

Python 2.6:

import distutils
distutils.sysconfig.get_config_var('Py_UNICODE_SIZE')

Harpsichord

Answer 5

I had the same issue and found a semi-official piece of code that does exactly that and may be interesting for people with the same issue: https://bitbucket.org/pypa/wheel/src/cf4e2d98ecb1f168c50a6de496959b4a10c6b122/wheel/pep425tags.py?at=default&fileviewer=file-view-default#pep425tags.py-83:89.

It comes from the wheel project which needs to check if the python is compiled with ucs-2 or ucs-4 because it will change the name of the binary file generated.

Yetty

Answer 6

Another way is to create an Unicode array and look at the itemsize:

import array
bytes_per_char = array.array('u').itemsize

Quote from the array docs:

The 'u' typecode corresponds to Python’s unicode character. On narrow Unicode builds this is 2-bytes, on wide builds this is 4-bytes.

Note that the distinction between narrow and wide Unicode builds is dropped from Python 3.3 onward, see PEP393. The 'u' typecode for array is deprecated since 3.3 and scheduled for removal in Python 4.0.

Hyperplane

Answer 7

65535 is UCS-2:

Thus code point U+0000 is encoded as the number 0, and U+FFFF is encoded as 65535 (which is FFFF16 in hexadecimal).

Archiepiscopate