Function Strs_init_from_pyarrow

python/stringzilla.c:6850–6993

The efficient `Strs_init` path initializing from PyArrow array capsules.

// The efficient `Strs_init` path initializing from PyArrow array capsules.
static int Strs_init_from_pyarrow(Strs *self, PyObject *sequence_obj, int view) {
    // Handle Arrow array
    PyObject *capsules = PyObject_CallMethod(sequence_obj, "__arrow_c_array__", NULL);
    if (!capsules || !PyTuple_Check(capsules) || PyTuple_Size(capsules) != 2) {
        Py_XDECREF(capsules);
        PyErr_SetString(PyExc_ValueError, "__arrow_c_array__ must return a tuple of 2 capsules");
        return -1;
    }

    PyObject *schema_capsule = PyTuple_GET_ITEM(capsules, 0);
    PyObject *array_capsule = PyTuple_GET_ITEM(capsules, 1);

    if (!PyCapsule_CheckExact(schema_capsule) || !PyCapsule_CheckExact(array_capsule)) {
        Py_DECREF(capsules);
        PyErr_SetString(PyExc_ValueError, "Expected PyCapsule objects from __arrow_c_array__");
        return -1;
    }

    struct ArrowSchema *schema = (struct ArrowSchema *)PyCapsule_GetPointer(schema_capsule, "arrow_schema");
    struct ArrowArray *array = (struct ArrowArray *)PyCapsule_GetPointer(array_capsule, "arrow_array");

    if (!schema || !array) {
        Py_DECREF(capsules);
        PyErr_SetString(PyExc_ValueError, "Failed to extract Arrow C structures");
        return -1;
    }

    // Validate string array layout
    if (!schema->format || (strcmp(schema->format, "u") != 0 && strcmp(schema->format, "U") != 0 &&
                            strcmp(schema->format, "z") != 0 && strcmp(schema->format, "Z") != 0)) {
        Py_DECREF(capsules);
        PyErr_SetString(PyExc_ValueError, "Arrow array must be string layout");
        return -1;
    }

    if (array->n_buffers != 3) {
        Py_DECREF(capsules);
        PyErr_SetString(PyExc_ValueError, "String Arrow array must have 3 buffers");
        return -1;
    }

    // Determine if 32-bit or 64-bit offsets
    int use_64bit = (strcmp(schema->format, "U") == 0 || strcmp(schema->format, "Z") == 0);
    void const **buffers = (void const **)array->buffers;
    sz_u8_t const *validity = (sz_u8_t const *)buffers[0]; // May be NULL
    sz_cptr_t data_buffer = (sz_cptr_t)buffers[2];
    sz_size_t length = array->length;

    // Zero-copy mode for Arrow arrays
    if (view) {
        if (use_64bit) {
            sz_i64_t const *offsets_64 = (sz_i64_t const *)buffers[1];
            self->layout = STRS_U64_TAPE_VIEW;
            self->data.u64_tape_view.count = length;
            self->data.u64_tape_view.parent = capsules;
            self->data.u64_tape_view.data = data_buffer;
            self->data.u64_tape_view.offsets = (sz_u64_t *)offsets_64;
            Py_INCREF(capsules);