Function Strs_init_from_iterable

python/stringzilla.c:7276–7472

The inefficient `Strs_init` path initializing from a Pythonic iterable of strings.

// The inefficient `Strs_init` path initializing from a Pythonic iterable of strings.
static int Strs_init_from_iterable(Strs *self, PyObject *sequence_obj, int view) {
    // Get an iterator from the object
    PyObject *iterator = PyObject_GetIter(sequence_obj);
    if (!iterator) {
        PyErr_SetString(PyExc_TypeError, "Object is not iterable");
        return -1;
    }

    if (view) {
        // View mode is not supported for iterators because we can't safely keep references
        // to all the individual string objects without significant overhead
        Py_DECREF(iterator);
        PyErr_SetString(PyExc_ValueError, "View mode (view=True) is not supported for iterators. "
                                          "Use view=False to create a copy, or convert to a list/tuple first.");
        return -1;
    }

    // Initialize allocator for memory management
    sz_memory_allocator_t allocator;
    sz_memory_allocator_init_default(&allocator);

    // Incrementally allocate a new tape to fit all of the items
    sz_size_t data_capacity = 4096;
    sz_size_t offsets_capacity = 16;
    sz_size_t count = 0;
    sz_size_t total_bytes = 0;
    int use_64bit = 0; // Start with 32-bit

    sz_ptr_t data_buffer = (sz_ptr_t)allocator.allocate(data_capacity, allocator.handle);
    void *offsets = allocator.allocate(offsets_capacity * sizeof(sz_u32_t), allocator.handle); // Start with 32-bit

    if (!data_buffer || !offsets) {
        if (data_buffer) allocator.free(data_buffer, data_capacity, allocator.handle);
        if (offsets) allocator.free(offsets, offsets_capacity * sizeof(sz_u32_t), allocator.handle);
        Py_DECREF(iterator);
        PyErr_NoMemory();
        return -1;
    }

    // Set initial offset to 0 (Apache Arrow format: N+1 offsets for N strings)
    if (use_64bit) { ((sz_u64_t *)offsets)[0] = 0; }
    else { ((sz_u32_t *)offsets)[0] = 0; }

    // Iterate through all items
    PyObject *item;
    while ((item = PyIter_Next(iterator))) {
        sz_cptr_t item_start;
        sz_size_t item_length;
        if (!sz_py_export_string_like(item, &item_start, &item_length)) {
            Py_DECREF(item);
            allocator.free(data_buffer, data_capacity, allocator.handle);
            allocator.free(offsets, offsets_capacity * (use_64bit ? sizeof(sz_u64_t) : sizeof(sz_u32_t)),
                           allocator.handle);
            Py_DECREF(iterator);
            PyErr_Format(PyExc_TypeError, "Item %zd is not a string-like object", count);
            return -1;
        }