Function sz_py_replace_fragmented_allocator

python/stringzilla.c:747–840

}

static sz_bool_t sz_py_replace_fragmented_allocator(Strs *strs, sz_memory_allocator_t *old_allocator,
                                                    sz_memory_allocator_t *allocator) {
    struct fragmented_t *fragmented = &strs->data.fragmented;
    sz_assert_(fragmented->spans && "Expected spans to be allocated");

    // Calculate total size needed for consolidated tape
    sz_size_t total_bytes = 0;
    for (sz_size_t i = 0; i < fragmented->count; i++) total_bytes += fragmented->spans[i].length;

    // Choose 32-bit or 64-bit tape based on size
    sz_bool_t use_64bit = total_bytes >= UINT32_MAX;

    // Skip allocation if there's no data to allocate (empty strings case)
    if (total_bytes == 0) {
        // Convert to empty tape layout
        old_allocator->free(fragmented->spans, fragmented->count * sizeof(sz_string_view_t), old_allocator->handle);
        Py_XDECREF(fragmented->parent);

        strs->layout = STRS_U32_TAPE;
        strs->data.u32_tape.count = fragmented->count;
        strs->data.u32_tape.data = NULL;
        strs->data.u32_tape.offsets = NULL;
        strs->data.u32_tape.allocator = *allocator;
        return sz_true_k;
    }

    // Allocate consolidated data buffer and offsets array
    sz_ptr_t new_data = (sz_ptr_t)allocator->allocate(total_bytes, allocator->handle);
    if (!new_data) return sz_false_k;

    if (use_64bit) {
        sz_u64_t *new_offsets =
            (sz_u64_t *)allocator->allocate((fragmented->count + 1) * sizeof(sz_u64_t), allocator->handle);
        if (!new_offsets) {
            allocator->free(new_data, total_bytes, allocator->handle);
            return sz_false_k;
        }

        // Copy fragmented data into consolidated buffer
        sz_size_t current_offset = 0;
        new_offsets[0] = 0;
        for (sz_size_t i = 0; i < fragmented->count; i++) {
            sz_size_t len = fragmented->spans[i].length;
            if (len > 0) { memcpy(new_data + current_offset, fragmented->spans[i].start, len); }
            current_offset += len;
            new_offsets[i + 1] = current_offset;
        }

        // Free old fragmented data and convert to 64-bit tape
        old_allocator->free(fragmented->spans, fragmented->count * sizeof(sz_string_view_t), old_allocator->handle);
        Py_XDECREF(fragmented->parent);

        strs->layout = STRS_U64_TAPE;
        strs->data.u64_tape.count = fragmented->count;
        strs->data.u64_tape.data = new_data;
        strs->data.u64_tape.offsets = new_offsets;
        strs->data.u64_tape.allocator = *allocator;
    }