cpython/Objects/stringlib/codecs.h

/* stringlib: codec implementations */

#if !STRINGLIB_IS_UNICODE
# error "codecs.h is specific to Unicode"
#endif

#include "pycore_bitutils.h"      // _Py_bswap32()

/* Mask to quickly check whether a C 'size_t' contains a
   non-ASCII, UTF8-encoded char. */
#if (SIZEOF_SIZE_T == 8)
#define ASCII_CHAR_MASK …
#elif (SIZEOF_SIZE_T == 4)
#define ASCII_CHAR_MASK …
#else
# error C 'size_t' size should be either 4 or 8!
#endif

/* 10xxxxxx */
#define IS_CONTINUATION_BYTE(ch) …

Py_LOCAL_INLINE(Py_UCS4)
STRINGLIB(utf8_decode)(const char **inptr, const char *end,
                       STRINGLIB_CHAR *dest,
                       Py_ssize_t *outpos)
{ … }

#undef ASCII_CHAR_MASK


/* UTF-8 encoder specialized for a Unicode kind to avoid the slow
   PyUnicode_READ() macro. Delete some parts of the code depending on the kind:
   UCS-1 strings don't need to handle surrogates for example. */
Py_LOCAL_INLINE(char *)
STRINGLIB(utf8_encoder)(_PyBytesWriter *writer,
                        PyObject *unicode,
                        const STRINGLIB_CHAR *data,
                        Py_ssize_t size,
                        _Py_error_handler error_handler,
                        const char *errors)
{ … }

/* The pattern for constructing UCS2-repeated masks. */
#if SIZEOF_LONG == 8
#define UCS2_REPEAT_MASK …
#elif SIZEOF_LONG == 4
#define UCS2_REPEAT_MASK …
#else
# error C 'long' size should be either 4 or 8!
#endif

/* The mask for fast checking. */
#if STRINGLIB_SIZEOF_CHAR == 1
/* The mask for fast checking of whether a C 'long' contains a
   non-ASCII or non-Latin1 UTF16-encoded characters. */
#define FAST_CHAR_MASK …
#else
/* The mask for fast checking of whether a C 'long' may contain
   UTF16-encoded surrogate characters. This is an efficient heuristic,
   assuming that non-surrogate characters with a code point >= 0x8000 are
   rare in most input.
*/
#define FAST_CHAR_MASK …
#endif
/* The mask for fast byte-swapping. */
#define STRIPPED_MASK …
/* Swap bytes. */
#define SWAB …

Py_LOCAL_INLINE(Py_UCS4)
STRINGLIB(utf16_decode)(const unsigned char **inptr, const unsigned char *e,
                        STRINGLIB_CHAR *dest, Py_ssize_t *outpos,
                        int native_ordering)
{ … }
#undef UCS2_REPEAT_MASK
#undef FAST_CHAR_MASK
#undef STRIPPED_MASK
#undef SWAB


#if STRINGLIB_MAX_CHAR >= 0x80
Py_LOCAL_INLINE(Py_ssize_t)
STRINGLIB(utf16_encode)(const STRINGLIB_CHAR *in,
                        Py_ssize_t len,
                        unsigned short **outptr,
                        int native_ordering)
{
    unsigned short *out = *outptr;
    const STRINGLIB_CHAR *end = in + len;
#if STRINGLIB_SIZEOF_CHAR == 1
    if (native_ordering) {
        const STRINGLIB_CHAR *unrolled_end = in + _Py_SIZE_ROUND_DOWN(len, 4);
        while (in < unrolled_end) {
            out[0] = in[0];
            out[1] = in[1];
            out[2] = in[2];
            out[3] = in[3];
            in += 4; out += 4;
        }
        while (in < end) {
            *out++ = *in++;
        }
    } else {
#define SWAB2 …
        const STRINGLIB_CHAR *unrolled_end = in + _Py_SIZE_ROUND_DOWN(len, 4);
        while (in < unrolled_end) {
            out[0] = SWAB2(in[0]);
            out[1] = SWAB2(in[1]);
            out[2] = SWAB2(in[2]);
            out[3] = SWAB2(in[3]);
            in += 4; out += 4;
        }
        while (in < end) {
            Py_UCS4 ch = *in++;
            *out++ = SWAB2((Py_UCS2)ch);
        }
#undef SWAB2
    }
    *outptr = out;
    return len;
#else
    if (native_ordering) {
#if STRINGLIB_MAX_CHAR < 0x10000
        const STRINGLIB_CHAR *unrolled_end = in + _Py_SIZE_ROUND_DOWN(len, 4);
        while (in < unrolled_end) {
            /* check if any character is a surrogate character */
            if (((in[0] ^ 0xd800) &
                 (in[1] ^ 0xd800) &
                 (in[2] ^ 0xd800) &
                 (in[3] ^ 0xd800) & 0xf800) == 0)
                break;
            out[0] = in[0];
            out[1] = in[1];
            out[2] = in[2];
            out[3] = in[3];
            in += 4; out += 4;
        }
#endif
        while (in < end) {
            Py_UCS4 ch;
            ch = *in++;
            if (ch < 0xd800)
                *out++ = ch;
            else if (ch < 0xe000)
                /* reject surrogate characters (U+D800-U+DFFF) */
                goto fail;
#if STRINGLIB_MAX_CHAR >= 0x10000
            else if (ch >= 0x10000) {
                out[0] = Py_UNICODE_HIGH_SURROGATE(ch);
                out[1] = Py_UNICODE_LOW_SURROGATE(ch);
                out += 2;
            }
#endif
            else
                *out++ = ch;
        }
    } else {
#define SWAB2 …
#if STRINGLIB_MAX_CHAR < 0x10000
        const STRINGLIB_CHAR *unrolled_end = in + _Py_SIZE_ROUND_DOWN(len, 4);
        while (in < unrolled_end) {
            /* check if any character is a surrogate character */
            if (((in[0] ^ 0xd800) &
                 (in[1] ^ 0xd800) &
                 (in[2] ^ 0xd800) &
                 (in[3] ^ 0xd800) & 0xf800) == 0)
                break;
            out[0] = SWAB2(in[0]);
            out[1] = SWAB2(in[1]);
            out[2] = SWAB2(in[2]);
            out[3] = SWAB2(in[3]);
            in += 4; out += 4;
        }
#endif
        while (in < end) {
            Py_UCS4 ch = *in++;
            if (ch < 0xd800)
                *out++ = SWAB2((Py_UCS2)ch);
            else if (ch < 0xe000)
                /* reject surrogate characters (U+D800-U+DFFF) */
                goto fail;
#if STRINGLIB_MAX_CHAR >= 0x10000
            else if (ch >= 0x10000) {
                Py_UCS2 ch1 = Py_UNICODE_HIGH_SURROGATE(ch);
                Py_UCS2 ch2 = Py_UNICODE_LOW_SURROGATE(ch);
                out[0] = SWAB2(ch1);
                out[1] = SWAB2(ch2);
                out += 2;
            }
#endif
            else
                *out++ = SWAB2((Py_UCS2)ch);
        }
#undef SWAB2
    }
    *outptr = out;
    return len;
  fail:
    *outptr = out;
    return len - (end - in + 1);
#endif
}

static inline uint32_t
STRINGLIB(SWAB4)(STRINGLIB_CHAR ch)
{
    uint32_t word = ch;
#if STRINGLIB_SIZEOF_CHAR == 1
    /* high bytes are zero */
    return (word << 24);
#elif STRINGLIB_SIZEOF_CHAR == 2
    /* high bytes are zero */
    return ((word & 0x00FFu) << 24) | ((word & 0xFF00u) << 8);
#else
    return _Py_bswap32(word);
#endif
}

Py_LOCAL_INLINE(Py_ssize_t)
STRINGLIB(utf32_encode)(const STRINGLIB_CHAR *in,
                        Py_ssize_t len,
                        uint32_t **outptr,
                        int native_ordering)
{
    uint32_t *out = *outptr;
    const STRINGLIB_CHAR *end = in + len;
    if (native_ordering) {
        const STRINGLIB_CHAR *unrolled_end = in + _Py_SIZE_ROUND_DOWN(len, 4);
        while (in < unrolled_end) {
#if STRINGLIB_SIZEOF_CHAR > 1
            /* check if any character is a surrogate character */
            if (((in[0] ^ 0xd800) &
                 (in[1] ^ 0xd800) &
                 (in[2] ^ 0xd800) &
                 (in[3] ^ 0xd800) & 0xf800) == 0)
                break;
#endif
            out[0] = in[0];
            out[1] = in[1];
            out[2] = in[2];
            out[3] = in[3];
            in += 4; out += 4;
        }
        while (in < end) {
            Py_UCS4 ch;
            ch = *in++;
#if STRINGLIB_SIZEOF_CHAR > 1
            if (Py_UNICODE_IS_SURROGATE(ch)) {
                /* reject surrogate characters (U+D800-U+DFFF) */
                goto fail;
            }
#endif
            *out++ = ch;
        }
    } else {
        const STRINGLIB_CHAR *unrolled_end = in + _Py_SIZE_ROUND_DOWN(len, 4);
        while (in < unrolled_end) {
#if STRINGLIB_SIZEOF_CHAR > 1
            /* check if any character is a surrogate character */
            if (((in[0] ^ 0xd800) &
                 (in[1] ^ 0xd800) &
                 (in[2] ^ 0xd800) &
                 (in[3] ^ 0xd800) & 0xf800) == 0)
                break;
#endif
            out[0] = STRINGLIB(SWAB4)(in[0]);
            out[1] = STRINGLIB(SWAB4)(in[1]);
            out[2] = STRINGLIB(SWAB4)(in[2]);
            out[3] = STRINGLIB(SWAB4)(in[3]);
            in += 4; out += 4;
        }
        while (in < end) {
            Py_UCS4 ch = *in++;
#if STRINGLIB_SIZEOF_CHAR > 1
            if (Py_UNICODE_IS_SURROGATE(ch)) {
                /* reject surrogate characters (U+D800-U+DFFF) */
                goto fail;
            }
#endif
            *out++ = STRINGLIB(SWAB4)(ch);
        }
    }
    *outptr = out;
    return len;
#if STRINGLIB_SIZEOF_CHAR > 1
  fail:
    *outptr = out;
    return len - (end - in + 1);
#endif
}

#endif