Function SmithWaterman_init

python/stringzillas.c:1238–1298

}

static int SmithWaterman_init(SmithWaterman *self, PyObject *args, PyObject *kwargs) {
    PyObject *substitution_matrix_obj = NULL;
    sz_error_cost_t open = -1, extend = -1;
    PyObject *capabilities_tuple = NULL;
    sz_capability_t capabilities = default_hardware_capabilities;

    // Parse arguments: substitution_matrix, open, extend, capabilities
    static char *kwlist[] = {"substitution_matrix", "open", "extend", "capabilities", NULL};
    if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|iiO", kwlist, &substitution_matrix_obj, &open, &extend,
                                     &capabilities_tuple))
        return -1;

    // Validate substitution matrix (should be a 256x256 numpy array)
    if (!numpy_available || !PyArray_Check(substitution_matrix_obj)) {
        PyErr_SetString(PyExc_TypeError, "substitution_matrix must be a NumPy array");
        return -1;
    }

    PyArrayObject *subs_array = (PyArrayObject *)substitution_matrix_obj;
    if (PyArray_NDIM(subs_array) != 2 || PyArray_DIM(subs_array, 0) != 256 || PyArray_DIM(subs_array, 1) != 256) {
        PyErr_SetString(PyExc_ValueError, "substitution_matrix must be a 256x256 array");
        return -1;
    }

    if (PyArray_TYPE(subs_array) != NPY_INT8) {
        PyErr_SetString(PyExc_TypeError, "substitution_matrix must have int8 dtype");
        return -1;
    }

    // Check that array is C-contiguous for safe memory access
    if (!PyArray_IS_C_CONTIGUOUS(subs_array)) {
        PyErr_SetString(PyExc_ValueError,
                        "substitution_matrix must be a C-contiguous array. Use np.ascontiguousarray() to convert.");
        return -1;
    }

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

    // Initialize the engine
    sz_error_cost_t *subs_data = (sz_error_cost_t *)PyArray_DATA(subs_array);
    char const *error_detail = NULL;
    sz_status_t status =
        szs_smith_waterman_scores_init(subs_data, open, extend, NULL, capabilities, &self->handle, &error_detail);

    if (status != sz_success_k) {
        set_stringzilla_error(status, error_detail, "SmithWaterman initialization");
        return -1;
    }

    // Create a simple checksum of the substitution matrix for the description
    sz_u32_t subs_checksum = 0;
    for (int i = 0; i < 256; i += 16)                      // Sample every 16th element
        subs_checksum += (sz_u32_t)subs_data[i * 256 + i]; // Diagonal elements

    snprintf(self->description, sizeof(self->description), "%X,%d,%d", subs_checksum & 0xFFFF, open, extend);