Esempio n. 1
0
static int
encode_locale_ex(const wchar_t *text, char **str, size_t *error_pos,
                 const char **reason,
                 int raw_malloc, int current_locale, int surrogateescape)
{
    if (current_locale) {
#ifdef __ANDROID__
        return _Py_EncodeUTF8Ex(text, str, error_pos, reason,
                                raw_malloc, surrogateescape);
#else
        return encode_current_locale(text, str, error_pos, reason,
                                     raw_malloc, surrogateescape);
#endif
    }

#if defined(__APPLE__) || defined(__ANDROID__)
    return _Py_EncodeUTF8Ex(text, str, error_pos, reason,
                            raw_malloc, surrogateescape);
#else   /* __APPLE__ */
    if (Py_UTF8Mode == 1) {
        return _Py_EncodeUTF8Ex(text, str, error_pos, reason,
                                raw_malloc, surrogateescape);
    }

#ifdef USE_FORCE_ASCII
    if (force_ascii == -1) {
        force_ascii = check_force_ascii();
    }

    if (force_ascii) {
        return encode_ascii(text, str, error_pos, reason,
                            raw_malloc, surrogateescape);
    }
#endif

    return encode_current_locale(text, str, error_pos, reason,
                                 raw_malloc, surrogateescape);
#endif   /* __APPLE__ or __ANDROID__ */
}
Esempio n. 2
0
/* Decode a byte string from the locale encoding.

   Use the strict error handler if 'surrogateescape' is zero.  Use the
   surrogateescape error handler if 'surrogateescape' is non-zero: undecodable
   bytes are decoded as characters in range U+DC80..U+DCFF. If a byte sequence
   can be decoded as a surrogate character, escape the bytes using the
   surrogateescape error handler instead of decoding them.

   On success, return 0 and write the newly allocated wide character string into
   *wstr (use PyMem_RawFree() to free the memory). If wlen is not NULL, write
   the number of wide characters excluding the null character into *wlen.

   On memory allocation failure, return -1.

   On decoding error, return -2. If wlen is not NULL, write the start of
   invalid byte sequence in the input string into *wlen. If reason is not NULL,
   write the decoding error message into *reason.

   Use the Py_EncodeLocaleEx() function to encode the character string back to
   a byte string. */
int
_Py_DecodeLocaleEx(const char* arg, wchar_t **wstr, size_t *wlen,
                   const char **reason,
                   int current_locale, int surrogateescape)
{
    if (current_locale) {
#ifdef __ANDROID__
        return _Py_DecodeUTF8Ex(arg, strlen(arg), wstr, wlen, reason,
                                surrogateescape);
#else
        return decode_current_locale(arg, wstr, wlen, reason, surrogateescape);
#endif
    }

#if defined(__APPLE__) || defined(__ANDROID__)
    return _Py_DecodeUTF8Ex(arg, strlen(arg), wstr, wlen, reason,
                            surrogateescape);
#else
    if (Py_UTF8Mode == 1) {
        return _Py_DecodeUTF8Ex(arg, strlen(arg), wstr, wlen, reason,
                                surrogateescape);
    }

#ifdef USE_FORCE_ASCII
    if (force_ascii == -1) {
        force_ascii = check_force_ascii();
    }

    if (force_ascii) {
        /* force ASCII encoding to workaround mbstowcs() issue */
        return decode_ascii(arg, wstr, wlen, reason, surrogateescape);
    }
#endif

    return decode_current_locale(arg, wstr, wlen, reason, surrogateescape);
#endif   /* __APPLE__ or __ANDROID__ */
}
Esempio n. 3
0
/* Encode a (wide) character string to the locale encoding with the
   surrogateescape error handler (characters in range U+DC80..U+DCFF are
   converted to bytes 0x80..0xFF).

   This function is the reverse of _Py_char2wchar().

   Return a pointer to a newly allocated byte string (use PyMem_Free() to free
   the memory), or NULL on encoding or memory allocation error.

   If error_pos is not NULL: *error_pos is the index of the invalid character
   on encoding error, or (size_t)-1 otherwise. */
char*
_Py_wchar2char(const wchar_t *text, size_t *error_pos)
{
#ifdef __APPLE__
    Py_ssize_t len;
    PyObject *unicode, *bytes = NULL;
    char *cpath;

    unicode = PyUnicode_FromWideChar(text, wcslen(text));
    if (unicode == NULL)
        return NULL;

    bytes = _PyUnicode_AsUTF8String(unicode, "surrogateescape");
    Py_DECREF(unicode);
    if (bytes == NULL) {
        PyErr_Clear();
        if (error_pos != NULL)
            *error_pos = (size_t)-1;
        return NULL;
    }

    len = PyBytes_GET_SIZE(bytes);
    cpath = PyMem_Malloc(len+1);
    if (cpath == NULL) {
        PyErr_Clear();
        Py_DECREF(bytes);
        if (error_pos != NULL)
            *error_pos = (size_t)-1;
        return NULL;
    }
    memcpy(cpath, PyBytes_AsString(bytes), len + 1);
    Py_DECREF(bytes);
    return cpath;
#else   /* __APPLE__ */
    const size_t len = wcslen(text);
    char *result = NULL, *bytes = NULL;
    size_t i, size, converted;
    wchar_t c, buf[2];

#ifndef MS_WINDOWS
    if (force_ascii == -1)
        force_ascii = check_force_ascii();

    if (force_ascii)
        return encode_ascii_surrogateescape(text, error_pos);
#endif

    /* The function works in two steps:
       1. compute the length of the output buffer in bytes (size)
       2. outputs the bytes */
    size = 0;
    buf[1] = 0;
    while (1) {
        for (i=0; i < len; i++) {
            c = text[i];
            if (c >= 0xdc80 && c <= 0xdcff) {
                /* UTF-8b surrogate */
                if (bytes != NULL) {
                    *bytes++ = c - 0xdc00;
                    size--;
                }
                else
                    size++;
                continue;
            }
            else {
                buf[0] = c;
                if (bytes != NULL)
                    converted = wcstombs(bytes, buf, size);
                else
                    converted = wcstombs(NULL, buf, 0);
                if (converted == (size_t)-1) {
                    if (result != NULL)
                        PyMem_Free(result);
                    if (error_pos != NULL)
                        *error_pos = i;
                    return NULL;
                }
                if (bytes != NULL) {
                    bytes += converted;
                    size -= converted;
                }
                else
                    size += converted;
            }
        }
        if (result != NULL) {
            *bytes = '\0';
            break;
        }

        size += 1; /* nul byte at the end */
        result = PyMem_Malloc(size);
        if (result == NULL) {
            if (error_pos != NULL)
                *error_pos = (size_t)-1;
            return NULL;
        }
        bytes = result;
    }
    return result;
#endif   /* __APPLE__ */
}
Esempio n. 4
0
/* Decode a byte string from the locale encoding with the
   surrogateescape error handler (undecodable bytes are decoded as characters
   in range U+DC80..U+DCFF). If a byte sequence can be decoded as a surrogate
   character, escape the bytes using the surrogateescape error handler instead
   of decoding them.

   Use _Py_wchar2char() to encode the character string back to a byte string.

   Return a pointer to a newly allocated wide character string (use
   PyMem_RawFree() to free the memory) and write the number of written wide
   characters excluding the null character into *size if size is not NULL, or
   NULL on error (decoding or memory allocation error). If size is not NULL,
   *size is set to (size_t)-1 on memory error and (size_t)-2 on decoding
   error.

   Conversion errors should never happen, unless there is a bug in the C
   library. */
wchar_t*
_Py_char2wchar(const char* arg, size_t *size)
{
#ifdef __APPLE__
    wchar_t *wstr;
    wstr = _Py_DecodeUTF8_surrogateescape(arg, strlen(arg));
    if (size != NULL) {
        if (wstr != NULL)
            *size = wcslen(wstr);
        else
            *size = (size_t)-1;
    }
    return wstr;
#else
    wchar_t *res;
    size_t argsize;
    size_t count;
#ifdef HAVE_MBRTOWC
    unsigned char *in;
    wchar_t *out;
    mbstate_t mbs;
#endif

#ifndef MS_WINDOWS
    if (force_ascii == -1)
        force_ascii = check_force_ascii();

    if (force_ascii) {
        /* force ASCII encoding to workaround mbstowcs() issue */
        res = decode_ascii_surrogateescape(arg, size);
        if (res == NULL)
            goto oom;
        return res;
    }
#endif

#ifdef HAVE_BROKEN_MBSTOWCS
    /* Some platforms have a broken implementation of
     * mbstowcs which does not count the characters that
     * would result from conversion.  Use an upper bound.
     */
    argsize = strlen(arg);
#else
    argsize = mbstowcs(NULL, arg, 0);
#endif
    if (argsize != (size_t)-1) {
        if (argsize == PY_SSIZE_T_MAX)
            goto oom;
        argsize += 1;
        if (argsize > PY_SSIZE_T_MAX/sizeof(wchar_t))
            goto oom;
        res = (wchar_t *)PyMem_RawMalloc(argsize*sizeof(wchar_t));
        if (!res)
            goto oom;
        count = mbstowcs(res, arg, argsize);
        if (count != (size_t)-1) {
            wchar_t *tmp;
            /* Only use the result if it contains no
               surrogate characters. */
            for (tmp = res; *tmp != 0 &&
                         !Py_UNICODE_IS_SURROGATE(*tmp); tmp++)
                ;
            if (*tmp == 0) {
                if (size != NULL)
                    *size = count;
                return res;
            }
        }
        PyMem_RawFree(res);
    }
    /* Conversion failed. Fall back to escaping with surrogateescape. */
#ifdef HAVE_MBRTOWC
    /* Try conversion with mbrtwoc (C99), and escape non-decodable bytes. */

    /* Overallocate; as multi-byte characters are in the argument, the
       actual output could use less memory. */
    argsize = strlen(arg) + 1;
    if (argsize > PY_SSIZE_T_MAX/sizeof(wchar_t))
        goto oom;
    res = (wchar_t*)PyMem_RawMalloc(argsize*sizeof(wchar_t));
    if (!res)
        goto oom;
    in = (unsigned char*)arg;
    out = res;
    memset(&mbs, 0, sizeof mbs);
    while (argsize) {
        size_t converted = mbrtowc(out, (char*)in, argsize, &mbs);
        if (converted == 0)
            /* Reached end of string; null char stored. */
            break;
        if (converted == (size_t)-2) {
            /* Incomplete character. This should never happen,
               since we provide everything that we have -
               unless there is a bug in the C library, or I
               misunderstood how mbrtowc works. */
            PyMem_RawFree(res);
            if (size != NULL)
                *size = (size_t)-2;
            return NULL;
        }
        if (converted == (size_t)-1) {
            /* Conversion error. Escape as UTF-8b, and start over
               in the initial shift state. */
            *out++ = 0xdc00 + *in++;
            argsize--;
            memset(&mbs, 0, sizeof mbs);
            continue;
        }
        if (Py_UNICODE_IS_SURROGATE(*out)) {
            /* Surrogate character.  Escape the original
               byte sequence with surrogateescape. */
            argsize -= converted;
            while (converted--)
                *out++ = 0xdc00 + *in++;
            continue;
        }
        /* successfully converted some bytes */
        in += converted;
        argsize -= converted;
        out++;
    }
    if (size != NULL)
        *size = out - res;
#else   /* HAVE_MBRTOWC */
    /* Cannot use C locale for escaping; manually escape as if charset
       is ASCII (i.e. escape all bytes > 128. This will still roundtrip
       correctly in the locale's charset, which must be an ASCII superset. */
    res = decode_ascii_surrogateescape(arg, size);
    if (res == NULL)
        goto oom;
#endif   /* HAVE_MBRTOWC */
    return res;
oom:
    if (size != NULL)
        *size = (size_t)-1;
    return NULL;
#endif   /* __APPLE__ */
}