Function LevenshteinDistancesUTF8_init

python/stringzillas.c:656–700

}

static int LevenshteinDistancesUTF8_init(LevenshteinDistancesUTF8 *self, PyObject *args, PyObject *kwargs) {
    int match = 0, mismatch = 1, open = 1, extend = 1;
    PyObject *capabilities_tuple = NULL;
    sz_capability_t capabilities = default_hardware_capabilities;

    static char *kwlist[] = {"match", "mismatch", "open", "extend", "capabilities", NULL};
    if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|iiiiO", kwlist, &match, &mismatch, &open, &extend,
                                     &capabilities_tuple))
        return -1;

    // Validate range of values
    if (match < -128 || match > 127) {
        PyErr_SetString(PyExc_ValueError, "match cost must fit in 8-bit signed integer");
        return -1;
    }
    if (mismatch < -128 || mismatch > 127) {
        PyErr_SetString(PyExc_ValueError, "mismatch cost must fit in 8-bit signed integer");
        return -1;
    }
    if (open < -128 || open > 127) {
        PyErr_SetString(PyExc_ValueError, "open cost must fit in 8-bit signed integer");
        return -1;
    }
    if (extend < -128 || extend > 127) {
        PyErr_SetString(PyExc_ValueError, "extend cost must fit in 8-bit signed integer");
        return -1;
    }

    // Parse capabilities if provided
    if (capabilities_tuple) {
        if (parse_and_intersect_capabilities(capabilities_tuple, &capabilities) != 0) { return -1; }
    }

    char const *error_detail = NULL;
    sz_status_t status = szs_levenshtein_distances_utf8_init(match, mismatch, open, extend, NULL, capabilities,
                                                             &self->handle, &error_detail);

    if (status != sz_success_k) {
        set_stringzilla_error(status, error_detail, "UTF-8 Levenshtein distances initialization");
        return -1;
    }
    snprintf(self->description, sizeof(self->description), "%d,%d,%d,%d", match, mismatch, open, extend);
    self->capabilities = capabilities;
    return 0;
}

static PyObject *LevenshteinDistancesUTF8_repr(LevenshteinDistancesUTF8 *self) {
    return PyUnicode_FromFormat("LevenshteinDistancesUTF8(match,mismatch,open,extend=%s)", self->description);