Cython:将unicode字符串转换为wchar数组

时间:2013-10-29 05:16:06

标签: python unicode cython python-2.6

我正在使用Cython与外部C API接口,该API接受UCS2格式的unicode字符串(wchar数组)。 (我理解UCS2相对于UTF-16的局限性,但它是第三方API。)

  • Cython版本:0.15.1
  • Python版本:2.6(缩小unicode版本)
  • 操作系统:FreeBSD

Cython用户指南广泛涉及将unicode转换为字节字符串,但我无法弄清楚如何转换为16位数组。我意识到我首先需要编码为UTF-16(我现在假设不会出现超出BMP的代码点)。接下来我该怎么办?请帮忙。

提前致谢。

1 个答案:

答案 0 :(得分:2)

这在 Python 3 上很有可能,解决方案就是这样:

# cython: language_level=3

from libc.stddef cimport wchar_t

cdef extern from "Python.h":
    wchar_t* PyUnicode_AsWideCharString(object, Py_ssize_t *)

cdef extern from "wchar.h":
    int wprintf(const wchar_t *, ...)

my_string = u"Foobar\n"
cdef Py_ssize_t length
cdef wchar_t *my_wchars = PyUnicode_AsWideCharString(my_string, &length)

wprintf(my_wchars)
print("Length:", <long>length)
print("Null End:", my_wchars[7] == 0)

接下来是一个不太好的Python 2方法,但它可能处理未定义或损坏的行为,所以我不太容易相信它:

# cython: language_level=2

from cpython.ref cimport PyObject
from libc.stddef cimport wchar_t
from libc.stdio  cimport fflush, stdout
from libc.stdlib cimport malloc, free

cdef extern from "Python.h":
    ctypedef PyObject PyUnicodeObject
    Py_ssize_t PyUnicode_AsWideChar(PyUnicodeObject *o, wchar_t *w, Py_ssize_t size)

my_string = u"Foobar\n"
cdef Py_ssize_t length = len(my_string.encode("UTF-16")) // 2 # cheating
cdef wchar_t *my_wchars = <wchar_t *>malloc(length * sizeof(wchar_t))
cdef Py_ssize_t number_written = PyUnicode_AsWideChar(<PyUnicodeObject *>my_string, my_wchars, length)

# wprintf breaks things for some reason
print [my_wchars[i] for i in range(length)]
print "Length:", <long>length
print "Number Written:", <long>number_written
print "Null End:", my_wchars[7] == 0

free(my_wchars)