Badminton-Scoreboard/Libraries/libil2cpp/include/utils/dynamic_array.h
2023-10-08 10:24:48 +08:00

379 lines
11 KiB
C++

#pragma once
#include "Memory.h"
#include <algorithm> // std::max
#include <memory> // std::uninitialized_fill
// dynamic_array - simplified version of std::vector<T>
//
// features:
// . always uses memcpy for copying elements. Your data structures must be simple and can't have internal pointers / rely on copy constructor.
// . EASTL like push_back(void) implementation
// Existing std STL implementations implement insertion operations by copying from an element.
// For example, resize(size() + 1) creates a throw-away temporary object.
// There is no way in existing std STL implementations to add an element to a container without implicitly or
// explicitly providing one to copy from (aside from some existing POD optimizations).
// For expensive-to-construct objects this creates a potentially serious performance problem.
// . grows X2 on reallocation
// . small code footprint
// . clear actually deallocates memory
// . resize does NOT initialize members!
//
// Changelog:
// Added pop_back()
// Added assign()
// Added clear() - frees the data, use resize(0) to clear w/o freeing
// zero allocation for empty array
//
namespace il2cpp
{
namespace utils
{
template<typename T>
struct AlignOfType
{
enum
{
align = ALIGN_OF(T)
};
};
template<typename T, size_t ALIGN = AlignOfType<T>::align>
struct dynamic_array
{
public:
enum
{
align = ALIGN
};
typedef T *iterator;
typedef const T *const_iterator;
typedef T value_type;
typedef size_t size_type;
typedef size_t difference_type;
typedef T &reference;
typedef const T &const_reference;
public:
dynamic_array() : m_data(NULL), m_size(0), m_capacity(0)
{
}
explicit dynamic_array(size_t size)
: m_size(size), m_capacity(size)
{
m_data = allocate(size);
}
dynamic_array(size_t size, T const &init_value)
: m_size(size), m_capacity(size)
{
m_data = allocate(size);
std::uninitialized_fill(m_data, m_data + size, init_value);
}
~dynamic_array()
{
if (owns_data())
m_data = deallocate(m_data);
}
dynamic_array(const dynamic_array &other) : m_size(0), m_capacity(0)
{
m_data = NULL;
assign(other.begin(), other.end());
}
dynamic_array &operator=(const dynamic_array &other)
{
// should not allocate memory unless we have to
if (&other != this)
assign(other.begin(), other.end());
return *this;
}
void clear()
{
if (owns_data())
m_data = deallocate(m_data);
m_size = 0;
m_capacity = 0;
}
void assign(const_iterator begin, const_iterator end)
{
Assert(begin <= end);
resize_uninitialized(end - begin);
memcpy(m_data, begin, m_size * sizeof(T));
}
iterator erase(iterator input_begin, iterator input_end)
{
Assert(input_begin <= input_end);
Assert(input_begin >= begin());
Assert(input_end <= end());
size_t leftOverSize = end() - input_end;
memmove(input_begin, input_end, leftOverSize * sizeof(T));
m_size -= input_end - input_begin;
return input_begin;
}
iterator erase(iterator it)
{
return erase(it, it + 1);
}
iterator erase_swap_back(iterator it)
{
m_size--;
memcpy(it, end(), sizeof(T));
return it;
}
iterator insert(iterator insert_before, const_iterator input_begin, const_iterator input_end)
{
Assert(input_begin <= input_end);
Assert(insert_before >= begin());
Assert(insert_before <= end());
// resize (make sure that insertBefore does not get invalid in the meantime because of a reallocation)
size_t insert_before_index = insert_before - begin();
size_t elements_to_be_moved = size() - insert_before_index;
resize_uninitialized((input_end - input_begin) + size(), true);
insert_before = begin() + insert_before_index;
size_t insertsize = input_end - input_begin;
// move to the end of where the inserted data will be
memmove(insert_before + insertsize, insert_before, elements_to_be_moved * sizeof(T));
// inject input data in the hole we just created
memcpy(insert_before, input_begin, insertsize * sizeof(T));
return insert_before;
}
iterator insert(iterator insertBefore, const T &t) { return insert(insertBefore, &t, &t + 1); }
void swap(dynamic_array &other) throw ()
{
std::swap(m_data, other.m_data);
std::swap(m_size, other.m_size);
std::swap(m_capacity, other.m_capacity);
}
// Returns the memory to the object.
// This does not call the constructor for the newly added element.
// You are expected to initialize all member variables of the returned data.
T &push_back()
{
if (++m_size > capacity())
reserve(std::max<size_t>(capacity() * 2, 1));
return back();
}
// push_back but it also calls the constructor for the newly added element.
T &push_back_construct()
{
if (++m_size > capacity())
reserve(std::max<size_t>(capacity() * 2, 1));
// construct
T *ptr = &back();
new(ptr) T;
return *ptr;
}
// push_back but assigns /t/ to the newly added element.
void push_back(const T &t)
{
push_back() = t;
}
void pop_back()
{
Assert(m_size >= 1);
m_size--;
}
void resize_uninitialized(size_t size, bool double_on_resize = false)
{
m_size = size;
if (m_size <= capacity())
return;
if (double_on_resize && size < capacity() * 2)
size = capacity() * 2;
reserve(size);
}
void resize_initialized(size_t size, const T &t = T(), bool double_on_resize = false)
{
if (size > capacity())
{
size_t requested_size = size;
if (double_on_resize && size < capacity() * 2)
requested_size = capacity() * 2;
reserve(requested_size);
}
if (size > m_size)
std::uninitialized_fill(m_data + m_size, m_data + size, t);
m_size = size;
}
void reserve(size_t inCapacity)
{
if (capacity() >= inCapacity)
return;
if (owns_data())
{
Assert((inCapacity & k_reference_bit) == 0 && "Dynamic array capacity overflow");
m_capacity = inCapacity;
m_data = reallocate(m_data, inCapacity);
}
else
{
T *newData = allocate(inCapacity);
memcpy(newData, m_data, m_size * sizeof(T));
// Invalidate old non-owned data, since using the data from two places is most likely a really really bad idea.
#if IL2CPP_DEBUG
memset(m_data, 0xCD, capacity() * sizeof(T));
#endif
m_capacity = inCapacity; // and clear reference bit
m_data = newData;
}
}
void assign_external(T *begin, T *end)
{
if (owns_data())
m_data = deallocate(m_data);
m_size = m_capacity = reinterpret_cast<value_type *>(end) - reinterpret_cast<value_type *>(begin);
Assert(m_size < k_reference_bit);
m_capacity |= k_reference_bit;
m_data = begin;
}
void set_owns_data(bool ownsData)
{
if (ownsData)
m_capacity &= ~k_reference_bit;
else
m_capacity |= k_reference_bit;
}
void shrink_to_fit()
{
if (owns_data())
{
m_capacity = m_size;
m_data = reallocate(m_data, m_size);
}
}
const T &back() const
{
Assert(m_size != 0);
return m_data[m_size - 1];
}
const T &front() const
{
Assert(m_size != 0);
return m_data[0];
}
T &back()
{
Assert(m_size != 0);
return m_data[m_size - 1];
}
T &front()
{
Assert(m_size != 0);
return m_data[0];
}
T *data() { return m_data; }
T const *data() const { return m_data; }
bool empty() const { return m_size == 0; }
size_t size() const { return m_size; }
size_t capacity() const { return m_capacity & ~k_reference_bit; }
T const &operator[](size_t index) const
{
Assert(index < m_size);
return m_data[index];
}
T &operator[](size_t index)
{
Assert(index < m_size);
return m_data[index];
}
T const *begin() const { return m_data; }
T *begin() { return m_data; }
T const *end() const { return m_data + m_size; }
T *end() { return m_data + m_size; }
bool owns_data() { return (m_capacity & k_reference_bit) == 0; }
bool equals(const dynamic_array &other) const
{
if (m_size != other.m_size)
return false;
for (int i = 0; i < m_size; i++)
{
if (!(m_data[i] == other.m_data[i]))
return false;
}
return true;
}
private:
static const size_t k_reference_bit = (size_t)1 << (sizeof(size_t) * 8 - 1);
T *allocate(size_t size)
{
return static_cast<T *>(IL2CPP_MALLOC_ALIGNED(size * sizeof(T), align));
}
T *deallocate(T *data)
{
Assert(owns_data());
IL2CPP_FREE_ALIGNED(data);
return NULL;
}
T *reallocate(T *data, size_t size)
{
Assert(owns_data());
return static_cast<T *>(IL2CPP_REALLOC_ALIGNED(data, size * sizeof(T), align));
}
T *m_data;
size_t m_size;
size_t m_capacity;
};
} //namespace il2cpp
} //namespace utils