Function Strs_shuffled

python/stringzilla.c:6090–6203

* @brief Shuffles the parts of a `Strs` object. * * This accepts a `Strs` object and potentially produces a new `Strs` object of a different layout: * - `STRS_U32_TAPE_VIEW` becomes `STRS_FRAGMENTED`, and keeps a link to the old as a parent. * - `STRS_U64_TAPE_VIEW` becomes `STRS_FRAGMENTED`, and keeps a link to the old as a parent. * - `STRS_U32_TAPE` becomes `STRS_FRAGMENTED`, and keep

 *  - `STRS_FRAGMENTED` returns a copy of itself, with the parts shuffled.
 */
static PyObject *Strs_shuffled(Strs *self, PyObject *const *args, Py_ssize_t positional_args_count,
                               PyObject *args_names_tuple) {

    // Check for positional arguments
    PyObject *seed_obj = positional_args_count == 1 ? args[0] : NULL;
    if (positional_args_count > 1) {
        PyErr_SetString(PyExc_TypeError, "shuffle() takes at most 1 positional argument");
        return NULL;
    }

    // Check for keyword arguments
    if (args_names_tuple) {
        Py_ssize_t args_names_count = PyTuple_GET_SIZE(args_names_tuple);
        for (Py_ssize_t i = 0; i < args_names_count; ++i) {
            PyObject *key = PyTuple_GET_ITEM(args_names_tuple, i);
            PyObject *value = args[positional_args_count + i];
            if (PyUnicode_CompareWithASCIIString(key, "seed") == 0 && !seed_obj) { seed_obj = value; }
            else if (PyErr_Format(PyExc_TypeError, "Got an unexpected keyword argument '%U'", key)) { return NULL; }
        }
    }

    // Fisher-Yates Shuffle Algorithm
    unsigned int seed = (unsigned int)time(NULL);
    if (seed_obj) {
        if (!PyLong_Check(seed_obj)) {
            PyErr_SetString(PyExc_TypeError, "The seed must be an integer");
            return NULL;
        }
        seed = PyLong_AsUnsignedLong(seed_obj);
    }

    // Determine the amount of memory needed
    sz_size_t substrings_count = 0;
    get_string_at_offset_t substring_getter = NULL;
    PyObject *parent_to_increment = NULL;
    sz_memory_allocator_t allocator;

    switch (self->layout) {
    case STRS_U32_TAPE:
        substring_getter = str_at_offset_u32_tape;
        substrings_count = self->data.u32_tape.count;
        parent_to_increment = (PyObject *)self;
        allocator = self->data.u32_tape.allocator;
        break;
    case STRS_U32_TAPE_VIEW:
        substring_getter = str_at_offset_u32_tape_view;
        substrings_count = self->data.u32_tape_view.count;
        parent_to_increment = self->data.u32_tape_view.parent;
        sz_memory_allocator_init_default(&allocator);
        break;
    case STRS_U64_TAPE:
        substring_getter = str_at_offset_u64_tape;
        substrings_count = self->data.u64_tape.count;
        parent_to_increment = (PyObject *)self;
        allocator = self->data.u64_tape.allocator;
        break;
    case STRS_U64_TAPE_VIEW:
        substring_getter = str_at_offset_u64_tape_view;