Function double_from_ucs4

numpy/core/src/multiarray/textreading/conversions.c:50–109

* In order to not pack a whole copy of a floating point parser, we copy the * result into ascii and call the Python one. Float parsing isn't super quick * so this is not terrible, but avoiding it would speed up things. * * Also note that parsing the first float of a complex will copy the whole * string to ascii rather than just the first part. * TODO: A tweak of the break might be a simple

 * @param result Output stored as double value.
 */
static inline int
double_from_ucs4(
        const Py_UCS4 *str, const Py_UCS4 *end,
        bool strip_whitespace, double *result, const Py_UCS4 **p_end)
{
    /* skip leading whitespace */
    if (strip_whitespace) {
        while (Py_UNICODE_ISSPACE(*str)) {
            str++;
        }
    }
    if (str == end) {
        return -1;  /* empty or only whitespace: not a floating point number */
    }

    /* We convert to ASCII for the Python parser, use stack if small: */
    char stack_buf[128];
    char *heap_buf = NULL;
    char *ascii = stack_buf;

    size_t str_len = end - str + 1;
    if (str_len > 128) {
        heap_buf = PyMem_MALLOC(str_len);
        if (heap_buf == NULL) {
            PyErr_NoMemory();
            return -1;
        }
        ascii = heap_buf;
    }
    char *c = ascii;
    for (; str < end; str++, c++) {
        if (NPY_UNLIKELY(*str >= 128)) {
            /* Character cannot be used, ignore for end calculation and stop */
            end = str;
            break;
        }
        *c = (char)(*str);
    }
    *c = '\0';

    char *end_parsed;
    *result = PyOS_string_to_double(ascii, &end_parsed, NULL);
    /* Rewind `end` to the first UCS4 character not parsed: */
    end = end - (c - end_parsed);

    PyMem_FREE(heap_buf);

    if (*result == -1. && PyErr_Occurred()) {
        return -1;
    }

    if (strip_whitespace) {
        /* and then skip any remaining whitespace: */
        while (Py_UNICODE_ISSPACE(*end)) {
            end++;
        }
    }
    *p_end = end;