//
// Copyright (c) 2009-2013 Mikko Mononen memon@inside.org
// Port of _gl.h to _d3d11.h by Chris Maughan
//
// This software is provided 'as-is', without any express or implied
// warranty.  In no event will the authors be held liable for any damages
// arising from the use of this software.
// Permission is granted to anyone to use this software for any purpose,
// including commercial applications, and to alter it and redistribute it
// freely, subject to the following restrictions:
// 1. The origin of this software must not be misrepresented; you must not
//    claim that you wrote the original software. If you use this software
//    in a product, an acknowledgment in the product documentation would be
//    appreciated but is not required.
// 2. Altered source versions must be plainly marked as such, and must not be
//    misrepresented as being the original software.
// 3. This notice may not be removed or altered from any source distribution.
//
#ifndef NANOVG_D3D11_H
#define NANOVG_D3D11_H

// Hide nameless struct/union warning for D3D headers
#ifdef _MSC_VER
#pragma warning (disable : 4201)
#endif
#include <d3d11.h>
#ifdef _MSC_VER
#pragma warning (default : 4201)
#endif

#ifdef __cplusplus
extern "C" {
#endif

// Flag indicating if geoemtry based anti-aliasing is used (may not be needed when using MSAA).
#define NVG_ANTIALIAS 1

// Flag indicating if strokes should be drawn using stencil buffer. The rendering will be a little
// slower, but path overlaps (i.e. self-intersecting or sharp turns) will be drawn just once.
#define NVG_STENCIL_STROKES 2

struct NVGcontext* nvgCreateD3D11(ID3D11Device* pDevice, int edgeaa);
void nvgDeleteD3D11(struct NVGcontext* ctx);

// These are additional flags on top of NVGimageFlags.
enum NVGimageFlagsD3D11 {
	NVG_IMAGE_NODELETE			= 1<<16,	// Do not delete texture object.
};

// Not done yet.  Simple enough to do though...
#ifdef IMPLEMENTED_IMAGE_FUNCS
int nvd3dCreateImageFromHandle(struct NVGcontext* ctx, void* texture, int w, int h, int flags);
unsigned int nvd3dImageHandle(struct NVGcontext* ctx, int image);
void nvd3dImageFlags(struct NVGcontext* ctx, int image, int flags);
#endif

#ifdef __cplusplus
}
#endif

#ifdef NANOVG_D3D11_IMPLEMENTATION
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <math.h>
#include <assert.h>
#include "nanovg.h"
#include <d3d11.h>

#include "nvg_shader/D3D11VertexShader.h"
#include "nvg_shader/D3D11PixelShaderAA.h"
#include "nvg_shader/D3D11PixelShader.h"

// The cpp calling is much simpler.
// For 'c' calling of DX, we need to do pPtr->lpVtbl->Func(pPtr, ...)
// There's probably a better way... (but note we can't use the IInterace_Method() helpers because
// They won't work when compiled for cpp)
#ifdef __cplusplus
#define D3D_API(p, name, arg1) p->name()
#define D3D_API_1(p, name, arg1) p->name(arg1)
#define D3D_API_2(p, name, arg1, arg2) p->name(arg1, arg2)
#define D3D_API_3(p, name, arg1, arg2, arg3) p->name(arg1, arg2, arg3)
#define D3D_API_4(p, name, arg1, arg2, arg3, arg4) p->name(arg1, arg2, arg3, arg4)
#define D3D_API_5(p, name, arg1, arg2, arg3, arg4, arg5) p->name(arg1, arg2, arg3, arg4, arg5)
#define D3D_API_6(p, name, arg1, arg2, arg3, arg4, arg5, arg6) p->name(arg1, arg2, arg3, arg4, arg5, arg6)
#define D3D_API_7(p, name, arg1, arg2, arg3, arg4, arg5, arg6, arg7) p->name(arg1, arg2, arg3, arg4, arg5, arg6, arg7)
#define D3D_API_8(p, name, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8) p->name(arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8)
#define D3D_API_RELEASE(p) { if ( (p) ) { (p)->Release(); (p) = NULL; } }
#else
#define D3D_API(p, name) p->lpVtbl->name(p)
#define D3D_API_1(p, name, arg1) p->lpVtbl->name(p, arg1)
#define D3D_API_2(p, name, arg1, arg2) p->lpVtbl->name(p, arg1, arg2)
#define D3D_API_3(p, name, arg1, arg2, arg3) p->lpVtbl->name(p, arg1, arg2, arg3)
#define D3D_API_4(p, name, arg1, arg2, arg3, arg4) p->lpVtbl->name(p, arg1, arg2, arg3, arg4)
#define D3D_API_5(p, name, arg1, arg2, arg3, arg4, arg5) p->lpVtbl->name(p, arg1, arg2, arg3, arg4, arg5)
#define D3D_API_6(p, name, arg1, arg2, arg3, arg4, arg5, arg6) p->lpVtbl->name(p, arg1, arg2, arg3, arg4, arg5, arg6)
#define D3D_API_7(p, name, arg1, arg2, arg3, arg4, arg5, arg6, arg7) p->lpVtbl->name(p, arg1, arg2, arg3, arg4, arg5, arg6, arg7)
#define D3D_API_8(p, name, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8) p->lpVtbl->name(p, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8)
#define D3D_API_RELEASE(p) { if ( (p) ) { (p)->lpVtbl->Release((p)); (p) = NULL; } }
#endif

#pragma pack(push)
#pragma pack(16)
struct D3DNVGfragUniforms
{
	float scissorMat[16];
	float scissorExt[4];
	float scissorScale[4];
	float paintMat[16];
	float extent[4];
	float radius[4];
	float feather[4];
	struct NVGcolor innerCol;
	struct NVGcolor outerCol;
	float strokeMult[4];
	int texType;
	int type;
};
#pragma pack(pop)

struct VS_CONSTANTS
{
	float dummy[16];
	float viewSize[2];
};

enum D3DNVGshaderType {
	NSVG_SHADER_FILLGRAD,
	NSVG_SHADER_FILLIMG,
	NSVG_SHADER_SIMPLE,
	NSVG_SHADER_IMG
};

struct D3DNVGshader {
	ID3D11PixelShader* frag;
	ID3D11VertexShader* vert;
	struct VS_CONSTANTS vc;
};

struct D3DNVGtexture {
	int id;
	ID3D11Texture2D* tex;
	ID3D11Texture2D* stagingTex;
	ID3D11ShaderResourceView* resourceView;
	int width, height;
	int type;
	int flags;
};

enum D3DNVGcallType {
	D3DNVG_NONE = 0,
	D3DNVG_FILL,
	D3DNVG_CONVEXFILL,
	D3DNVG_STROKE,
	D3DNVG_TRIANGLES
};

struct D3DNVGcall {
	int type;
	int image;
	int pathOffset;
	int pathCount;
	int triangleOffset;
	int triangleCount;
	int uniformOffset;
};

struct D3DNVGpath {
	int fillOffset;
	int fillCount;
	int strokeOffset;
	int strokeCount;
};

struct D3DNVGBuffer {
	unsigned int MaxBufferEntries;
	unsigned int CurrentBufferEntry;
	ID3D11Buffer* pBuffer;
};

struct D3DNVGcontext {

	struct D3DNVGshader shader;
	struct D3DNVGtexture* textures;
	float view[2];
	int ntextures;
	int ctextures;
	int textureId;
	ID3D11SamplerState* pSamplerState[4];

	int fragSize;
	int flags;

	// Per frame buffers
	struct D3DNVGcall* calls;
	int ccalls;
	int ncalls;
	struct D3DNVGpath* paths;
	int cpaths;
	int npaths;
	struct NVGvertex* verts;
	int cverts;
	int nverts;
	unsigned char* uniforms;
	int cuniforms;
	int nuniforms;

	// D3D
	// Geometry
	struct D3DNVGBuffer VertexBuffer;
	ID3D11Buffer* pFanIndexBuffer;
	ID3D11InputLayout* pLayoutRenderTriangles;

	// State
	ID3D11Buffer* pVSConstants;
	ID3D11Buffer* pPSConstants;

	ID3D11Device* pDevice;
	ID3D11DeviceContext* pDeviceContext;

	ID3D11BlendState* pBSBlend;
	ID3D11BlendState* pBSNoWrite;

	ID3D11RasterizerState* pRSNoCull;
	ID3D11RasterizerState* pRSCull;

	ID3D11DepthStencilState* pDepthStencilDrawShapes;
	ID3D11DepthStencilState* pDepthStencilDrawAA;
	ID3D11DepthStencilState* pDepthStencilFill;
	ID3D11DepthStencilState* pDepthStencilDefault;
};

static int D3Dnvg__maxi(int a, int b) { return a > b ? a : b; }

static struct D3DNVGtexture* D3Dnvg__allocTexture(struct D3DNVGcontext* D3D)
{
	struct D3DNVGtexture* tex = NULL;
	int i;

	for (i = 0; i < D3D->ntextures; i++) {
		if (D3D->textures[i].id == 0) {
			tex = &D3D->textures[i];
			break;
		}
	}
	if (tex == NULL) {
		if (D3D->ntextures + 1 > D3D->ctextures) {
			struct D3DNVGtexture* textures;
			int ctextures = D3Dnvg__maxi(D3D->ntextures+1, 4) +  D3D->ctextures/2; // 1.5x Overallocate
			textures = (struct D3DNVGtexture*)realloc(D3D->textures, sizeof(struct D3DNVGtexture)*ctextures);
			if (textures == NULL) return NULL;
			D3D->textures = textures;
			D3D->ctextures = ctextures;
		}
		tex = &D3D->textures[D3D->ntextures++];
	}

	memset(tex, 0, sizeof(*tex));
	tex->id = ++D3D->textureId;

	return tex;
}

static struct D3DNVGtexture* D3Dnvg__findTexture(struct D3DNVGcontext* D3D, int id)
{
	int i;
	for (i = 0; i < D3D->ntextures; i++)
		if (D3D->textures[i].id == id)
			return &D3D->textures[i];
	return NULL;
}

static int D3Dnvg__deleteTexture(struct D3DNVGcontext* D3D, int id)
{
	int i;
	for (i = 0; i < D3D->ntextures; i++) {
		if (D3D->textures[i].id == id) {
			if (D3D->textures[i].tex != 0 && (D3D->textures[i].flags & NVG_IMAGE_NODELETE) == 0)
			{
				D3D_API_RELEASE(D3D->textures[i].tex);
				D3D_API_RELEASE(D3D->textures[i].resourceView);
				if (D3D->textures[i].stagingTex != NULL) {
					D3D_API_RELEASE(D3D->textures[i].stagingTex);
					D3D->textures[i].stagingTex = NULL;
				}
			}
			memset(&D3D->textures[i], 0, sizeof(D3D->textures[i]));
			return 1;
		}
	}
	return 0;
}

// BEGIN D3D specific

static void D3Dnvg_copyMatrix3to4(float* pDest, const float* pSource)
{
	unsigned int i;
	for (i = 0; i < 4; i++)
	{
		memcpy(&pDest[i * 4], &pSource[i * 3], sizeof(float) * 3);
	}
}
// END D3D specific

static int D3Dnvg__checkError(HRESULT hr, const char* str)
{
	if (!SUCCEEDED(hr))
	{
		printf("Error %08lx after %s\n", hr, str);
		return 1;
	}
	return 0;
}

static int D3Dnvg__createShader(struct D3DNVGcontext* D3D, struct D3DNVGshader* shader, const void* vshader, unsigned int vshader_size, const void* fshader, unsigned int fshader_size)
{
	ID3D11VertexShader* vert = NULL;
	ID3D11PixelShader* frag = NULL;
	HRESULT hr;

	memset(shader, 0, sizeof(*shader));

	// Shader byte code is created at compile time from the .hlsl files.
	// No need for error checks; shader errors can be fixed in the IDE.
	hr = D3D_API_4(D3D->pDevice, CreateVertexShader, vshader, vshader_size, NULL, &vert);
	if (hr != S_OK) return 0;

	hr = D3D_API_4(D3D->pDevice, CreatePixelShader, fshader, fshader_size, NULL, &frag);
	if (hr != S_OK) return 0;

	shader->vert = vert;
	shader->frag = frag;

	return SUCCEEDED(hr) ? 1 : -1;
}

static void D3Dnvg__deleteShader(struct D3DNVGshader* shader)
{
	D3D_API_RELEASE(shader->vert);
	D3D_API_RELEASE(shader->frag);
}

void D3Dnvg_buildFanIndices(struct D3DNVGcontext* D3D)
{
	UINT32 index0 = 0;
	UINT32 index1 = 1;
	UINT32 index2 = 2;
	UINT32 current = 0;
	D3D11_MAPPED_SUBRESOURCE resource;
	UINT32* pIndices = NULL;

	D3D_API_5(D3D->pDeviceContext, Map, (ID3D11Resource*)D3D->pFanIndexBuffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &resource);
	pIndices = (UINT32*)resource.pData;

	while (current < (D3D->VertexBuffer.MaxBufferEntries - 3))
	{
		pIndices[current++] = index0;
		pIndices[current++] = index1++;
		pIndices[current++] = index2++;
	}
	D3D_API_2(D3D->pDeviceContext, Unmap, (ID3D11Resource*)D3D->pFanIndexBuffer, 0);
}

struct NVGvertex* D3Dnvg_beginVertexBuffer(struct D3DNVGcontext* D3D, unsigned int maxCount, unsigned int* baseOffset)
{
	D3D11_MAPPED_SUBRESOURCE resource;
	if (maxCount >= D3D->VertexBuffer.MaxBufferEntries)
	{
		D3Dnvg__checkError(E_FAIL, "Vertex buffer too small!");
		return NULL;
	}
	if ((D3D->VertexBuffer.CurrentBufferEntry + maxCount) >= D3D->VertexBuffer.MaxBufferEntries)
	{
		*baseOffset = 0;
		D3D->VertexBuffer.CurrentBufferEntry = maxCount;
		D3D_API_5(D3D->pDeviceContext, Map, (ID3D11Resource*)D3D->VertexBuffer.pBuffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &resource);
	}
	else
	{
		*baseOffset = D3D->VertexBuffer.CurrentBufferEntry;
		D3D->VertexBuffer.CurrentBufferEntry = *baseOffset + maxCount;
		D3D_API_5(D3D->pDeviceContext, Map, (ID3D11Resource*)D3D->VertexBuffer.pBuffer, 0, D3D11_MAP_WRITE_NO_OVERWRITE, 0, &resource);
	}
	return ((struct NVGvertex*)resource.pData + *baseOffset);
}

void D3Dnvg_endVertexBuffer(struct D3DNVGcontext* D3D)
{
	D3D_API_2(D3D->pDeviceContext, Unmap, (ID3D11Resource*)D3D->VertexBuffer.pBuffer, 0);
}

static void D3Dnvg__copyVerts(struct NVGvertex* pDest, const struct NVGvertex* pSource, unsigned int num)
{
	unsigned int i;
	for (i = 0; i < num; i++)
	{
		pDest[i].x = pSource[i].x;
		pDest[i].y = pSource[i].y;
		pDest[i].u = pSource[i].u;
		pDest[i].v = pSource[i].v;
	}
}

static unsigned int D3Dnvg_updateVertexBuffer(ID3D11DeviceContext* pContext, struct D3DNVGBuffer* buffer, const struct NVGvertex* verts, unsigned int nverts)
{
	D3D11_MAPPED_SUBRESOURCE resource;
	unsigned int retEntry;

	if (nverts > buffer->MaxBufferEntries)
	{
		D3Dnvg__checkError(E_FAIL, "Vertex buffer too small!");
		return 0;
	}
	if ((buffer->CurrentBufferEntry + nverts) >= buffer->MaxBufferEntries)
	{
		buffer->CurrentBufferEntry = 0;
		D3D_API_5(pContext, Map, (ID3D11Resource*)buffer->pBuffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &resource);
	}
	else
	{
		D3D_API_5(pContext, Map, (ID3D11Resource*)buffer->pBuffer, 0, D3D11_MAP_WRITE_NO_OVERWRITE, 0, &resource);
	}

	D3Dnvg__copyVerts((((struct NVGvertex*)resource.pData) + buffer->CurrentBufferEntry), (const struct NVGvertex*)verts, nverts);

	D3D_API_2(pContext, Unmap, (ID3D11Resource*)buffer->pBuffer, 0);
	retEntry = buffer->CurrentBufferEntry;
	buffer->CurrentBufferEntry += nverts;
	return retEntry;
}

static void D3Dnvg_setBuffers(struct D3DNVGcontext* D3D, unsigned int dynamicOffset)
{
	ID3D11Buffer* pBuffers[1];
	unsigned int strides[1];
	unsigned int offsets[1];

	pBuffers[0] = D3D->VertexBuffer.pBuffer;
	strides[0] = sizeof(struct NVGvertex);
	offsets[0] = dynamicOffset * sizeof(struct NVGvertex);

	D3D_API_3(D3D->pDeviceContext, IASetIndexBuffer, D3D->pFanIndexBuffer, DXGI_FORMAT_R32_UINT, 0);
	D3D_API_5(D3D->pDeviceContext, IASetVertexBuffers, 0, 1, pBuffers, strides, offsets);
	D3D_API_1(D3D->pDeviceContext, IASetInputLayout, D3D->pLayoutRenderTriangles);
}

static int D3Dnvg__renderCreate(void* uptr)
{
	HRESULT hr;
	D3D11_BUFFER_DESC bufferDesc;
	D3D11_RASTERIZER_DESC rasterDesc;
	D3D11_BLEND_DESC blendDesc;
	D3D11_DEPTH_STENCIL_DESC depthStencilDesc;
	D3D11_SAMPLER_DESC sampDesc;

	struct D3DNVGcontext* D3D = (struct D3DNVGcontext*)uptr;

	const D3D11_DEPTH_STENCILOP_DESC frontOp = { D3D11_STENCIL_OP_KEEP, D3D11_STENCIL_OP_KEEP, D3D11_STENCIL_OP_INCR, D3D11_COMPARISON_ALWAYS };
	const D3D11_DEPTH_STENCILOP_DESC backOp = { D3D11_STENCIL_OP_KEEP, D3D11_STENCIL_OP_KEEP, D3D11_STENCIL_OP_DECR, D3D11_COMPARISON_ALWAYS };

	const D3D11_DEPTH_STENCILOP_DESC aaOp = { D3D11_STENCIL_OP_KEEP, D3D11_STENCIL_OP_KEEP, D3D11_STENCIL_OP_KEEP, D3D11_COMPARISON_EQUAL };
	const D3D11_DEPTH_STENCILOP_DESC fillOp = { D3D11_STENCIL_OP_ZERO, D3D11_STENCIL_OP_ZERO, D3D11_STENCIL_OP_ZERO, D3D11_COMPARISON_NOT_EQUAL };

	const D3D11_DEPTH_STENCILOP_DESC defaultOp = { D3D11_STENCIL_OP_KEEP, D3D11_STENCIL_OP_KEEP, D3D11_STENCIL_OP_KEEP, D3D11_COMPARISON_ALWAYS };

	D3D11_INPUT_ELEMENT_DESC LayoutRenderTriangles[] =
	{
		{ "POSITION", 0, DXGI_FORMAT_R32G32_FLOAT, 0, 0, D3D11_INPUT_PER_VERTEX_DATA, 0 },
		{ "TEXCOORD", 0, DXGI_FORMAT_R32G32_FLOAT, 0, 8, D3D11_INPUT_PER_VERTEX_DATA, 0 }
	};

	if (D3D->flags & NVG_ANTIALIAS) {
		if (D3Dnvg__createShader(D3D, &D3D->shader, g_D3D11VertexShader_Main, sizeof(g_D3D11VertexShader_Main), g_D3D11PixelShaderAA_Main, sizeof(g_D3D11PixelShaderAA_Main)) == 0)
			return 0;
	}
	else {
		if (D3Dnvg__createShader(D3D, &D3D->shader, g_D3D11VertexShader_Main, sizeof(g_D3D11VertexShader_Main), g_D3D11PixelShader_Main, sizeof(g_D3D11PixelShader_Main)) == 0)
			return 0;
	}

	// Todo: Need to find a good value for this, and
	// Use the dynamic buffer fill technnique to handle overflow
	D3D->VertexBuffer.MaxBufferEntries = 1000000;
	D3D->VertexBuffer.CurrentBufferEntry = 0;

	memset(&bufferDesc, 0, sizeof(bufferDesc));
	bufferDesc.Usage = D3D11_USAGE_DYNAMIC;
	bufferDesc.BindFlags = D3D11_BIND_VERTEX_BUFFER;
	bufferDesc.CPUAccessFlags = D3D11_CPU_ACCESS_WRITE;
	bufferDesc.MiscFlags = 0;

	// Create the vertex buffer.
	bufferDesc.ByteWidth = sizeof(struct NVGvertex)* D3D->VertexBuffer.MaxBufferEntries;
	hr = D3D_API_3(D3D->pDevice, CreateBuffer, &bufferDesc, NULL, &D3D->VertexBuffer.pBuffer);
	D3Dnvg__checkError(hr, "Create Vertex Buffer");

	bufferDesc.BindFlags = D3D11_BIND_INDEX_BUFFER;
	bufferDesc.ByteWidth = sizeof(UINT32)* D3D->VertexBuffer.MaxBufferEntries;
	hr = D3D_API_3(D3D->pDevice, CreateBuffer, &bufferDesc, NULL, &D3D->pFanIndexBuffer);
	D3Dnvg__checkError(hr, "Create Vertex Buffer Static");

	D3Dnvg_buildFanIndices(D3D);

	hr = D3D_API_5(D3D->pDevice, CreateInputLayout, LayoutRenderTriangles, 2, g_D3D11VertexShader_Main, sizeof(g_D3D11VertexShader_Main), &D3D->pLayoutRenderTriangles);
	D3Dnvg__checkError(hr, "Create Input Layout");

	bufferDesc.Usage = D3D11_USAGE_DYNAMIC;
	bufferDesc.BindFlags = D3D11_BIND_CONSTANT_BUFFER;
	bufferDesc.CPUAccessFlags = D3D11_CPU_ACCESS_WRITE;
	bufferDesc.MiscFlags = 0;
	bufferDesc.ByteWidth = sizeof(struct VS_CONSTANTS);
	if ((bufferDesc.ByteWidth % 16) != 0)
	{
		bufferDesc.ByteWidth += 16 - (bufferDesc.ByteWidth % 16);
	}

	hr = D3D_API_3(D3D->pDevice, CreateBuffer, &bufferDesc, NULL, &D3D->pVSConstants);
	D3Dnvg__checkError(hr, "Create Constant Buffer");

	bufferDesc.ByteWidth = sizeof(struct D3DNVGfragUniforms);
	if ((bufferDesc.ByteWidth % 16) != 0)
	{
		bufferDesc.ByteWidth += 16 - (bufferDesc.ByteWidth % 16);
	}
	D3D->fragSize = bufferDesc.ByteWidth;

	hr = D3D_API_3(D3D->pDevice, CreateBuffer, &bufferDesc, NULL, &D3D->pPSConstants);
	D3Dnvg__checkError(hr, "Create Constant Buffer");

	ZeroMemory(&rasterDesc, sizeof(rasterDesc));
	rasterDesc.FillMode = D3D11_FILL_SOLID;
	rasterDesc.CullMode = D3D11_CULL_NONE;
	rasterDesc.DepthClipEnable = TRUE;
	rasterDesc.FrontCounterClockwise = TRUE;
	hr = D3D_API_2(D3D->pDevice, CreateRasterizerState, &rasterDesc, &D3D->pRSNoCull);

	rasterDesc.CullMode = D3D11_CULL_BACK;
	hr = D3D_API_2(D3D->pDevice, CreateRasterizerState, &rasterDesc, &D3D->pRSCull);

	ZeroMemory(&blendDesc, sizeof(blendDesc));
	blendDesc.RenderTarget[0].BlendEnable = TRUE;
	blendDesc.RenderTarget[0].RenderTargetWriteMask = D3D11_COLOR_WRITE_ENABLE_ALL;
	blendDesc.RenderTarget[0].BlendOp = D3D11_BLEND_OP_ADD;
	blendDesc.RenderTarget[0].BlendOpAlpha = D3D11_BLEND_OP_ADD;
	blendDesc.RenderTarget[0].SrcBlend = D3D11_BLEND_ONE;
	blendDesc.RenderTarget[0].SrcBlendAlpha = D3D11_BLEND_ONE;
	blendDesc.RenderTarget[0].DestBlend = D3D11_BLEND_INV_SRC_ALPHA;
	blendDesc.RenderTarget[0].DestBlendAlpha = D3D11_BLEND_INV_SRC_ALPHA;
	blendDesc.IndependentBlendEnable = FALSE;
	hr = D3D_API_2(D3D->pDevice, CreateBlendState, &blendDesc, &D3D->pBSBlend);


	blendDesc.RenderTarget[0].BlendEnable = FALSE;
	blendDesc.RenderTarget[0].RenderTargetWriteMask = 0;
	hr = D3D_API_2(D3D->pDevice, CreateBlendState, &blendDesc, &D3D->pBSNoWrite);

	// Stencil A Draw shapes
	ZeroMemory(&depthStencilDesc, sizeof(depthStencilDesc));
	depthStencilDesc.DepthEnable = FALSE;
	depthStencilDesc.DepthFunc = D3D11_COMPARISON_LESS_EQUAL;
	depthStencilDesc.DepthWriteMask = D3D11_DEPTH_WRITE_MASK_ZERO;
	depthStencilDesc.StencilEnable = TRUE;
	depthStencilDesc.StencilReadMask = D3D11_DEFAULT_STENCIL_READ_MASK;
	depthStencilDesc.StencilWriteMask = D3D11_DEFAULT_STENCIL_WRITE_MASK;

	depthStencilDesc.FrontFace = frontOp;
	depthStencilDesc.BackFace = backOp;

	// No color write
	hr = D3D_API_2(D3D->pDevice, CreateDepthStencilState, &depthStencilDesc, &D3D->pDepthStencilDrawShapes);

	// Draw AA
	depthStencilDesc.FrontFace = aaOp;
	depthStencilDesc.BackFace = aaOp;

	hr = D3D_API_2(D3D->pDevice, CreateDepthStencilState, &depthStencilDesc, &D3D->pDepthStencilDrawAA);

	// Stencil Fill
	depthStencilDesc.FrontFace = fillOp;
	depthStencilDesc.BackFace = fillOp;
	hr = D3D_API_2(D3D->pDevice, CreateDepthStencilState, &depthStencilDesc, &D3D->pDepthStencilFill);

	depthStencilDesc.FrontFace = defaultOp;
	depthStencilDesc.BackFace = defaultOp;
	depthStencilDesc.StencilEnable = FALSE;
	hr = D3D_API_2(D3D->pDevice, CreateDepthStencilState, &depthStencilDesc, &D3D->pDepthStencilDefault);

	ZeroMemory(&sampDesc, sizeof(sampDesc));
	sampDesc.Filter = D3D11_FILTER_MIN_MAG_MIP_LINEAR;
	sampDesc.AddressW = D3D11_TEXTURE_ADDRESS_WRAP;
	sampDesc.ComparisonFunc = D3D11_COMPARISON_NEVER;
	sampDesc.MinLOD = 0;
	sampDesc.MaxLOD = D3D11_FLOAT32_MAX;
	sampDesc.MipLODBias = 0.0f;//-1.0f;

	sampDesc.AddressU = D3D11_TEXTURE_ADDRESS_CLAMP;
	sampDesc.AddressV = D3D11_TEXTURE_ADDRESS_CLAMP;
	D3D_API_2(D3D->pDevice, CreateSamplerState, &sampDesc, &D3D->pSamplerState[0]);

	sampDesc.AddressU = D3D11_TEXTURE_ADDRESS_WRAP;
	sampDesc.AddressV = D3D11_TEXTURE_ADDRESS_CLAMP;
	D3D_API_2(D3D->pDevice, CreateSamplerState, &sampDesc, &D3D->pSamplerState[1]);

	sampDesc.AddressU = D3D11_TEXTURE_ADDRESS_CLAMP;
	sampDesc.AddressV = D3D11_TEXTURE_ADDRESS_WRAP;
	D3D_API_2(D3D->pDevice, CreateSamplerState, &sampDesc, &D3D->pSamplerState[2]);

	sampDesc.AddressU = D3D11_TEXTURE_ADDRESS_WRAP;
	sampDesc.AddressV = D3D11_TEXTURE_ADDRESS_WRAP;
	D3D_API_2(D3D->pDevice, CreateSamplerState, &sampDesc, &D3D->pSamplerState[3]);

	return 1;
}

static int D3Dnvg__renderCreateTexture(void* uptr, int type, int w, int h, int imageFlags, const unsigned char* data)
{
	struct D3DNVGcontext* D3D = (struct D3DNVGcontext*)uptr;
	struct D3DNVGtexture* tex = D3Dnvg__allocTexture(D3D);
	D3D11_TEXTURE2D_DESC texDesc;
	int pixelWidthBytes;
	HRESULT hr;
	D3D11_SHADER_RESOURCE_VIEW_DESC viewDesc;

	if (tex == NULL)
	{
		return 0;
	}

	tex->width = w;
	tex->height = h;
	tex->type = type;
	tex->flags = imageFlags;

	memset(&texDesc, 0, sizeof(texDesc));
	texDesc.ArraySize = 1;
	texDesc.BindFlags = D3D11_BIND_SHADER_RESOURCE;
	texDesc.CPUAccessFlags = 0;
	texDesc.MipLevels = 1;
	if (type == NVG_TEXTURE_RGBA) {
		texDesc.Format = DXGI_FORMAT_R8G8B8A8_UNORM;
		pixelWidthBytes = 4;

		// Mip maps
		if (imageFlags & NVG_IMAGE_GENERATE_MIPMAPS)
		{
			texDesc.MipLevels = 0;
			texDesc.BindFlags |= D3D11_BIND_RENDER_TARGET;
			texDesc.MiscFlags = D3D11_RESOURCE_MISC_GENERATE_MIPS;
		}
	} else {
		texDesc.Format = DXGI_FORMAT_R8_UNORM;
		pixelWidthBytes = 1;
		texDesc.MipLevels = 1;
	}
	texDesc.Height = tex->height;
	texDesc.Width = tex->width;
	texDesc.SampleDesc.Count = 1;
	texDesc.SampleDesc.Quality = 0;
	if (imageFlags & NVG_IMAGE_STREAMING) {
		texDesc.Usage = D3D11_USAGE_DYNAMIC;
		texDesc.CPUAccessFlags = D3D11_CPU_ACCESS_WRITE;
	} else {
		texDesc.Usage = D3D11_USAGE_DEFAULT;
	}

	hr = D3D_API_3(D3D->pDevice, CreateTexture2D, &texDesc, NULL, &tex->tex);
	if (FAILED(hr))
	{
		return 0;
	}

	if (data != NULL)
	{
		D3D_API_6(D3D->pDeviceContext, UpdateSubresource, (ID3D11Resource*)tex->tex, 0, NULL, data, tex->width * pixelWidthBytes, (tex->width * tex->height) * pixelWidthBytes);
	}

	viewDesc.Format = texDesc.Format;
	viewDesc.ViewDimension = D3D11_SRV_DIMENSION_TEXTURE2D;
	viewDesc.Texture2D.MipLevels = texDesc.MipLevels;
	viewDesc.Texture2D.MostDetailedMip = 0;

	D3D_API_3(D3D->pDevice, CreateShaderResourceView, (ID3D11Resource*)tex->tex, &viewDesc, &tex->resourceView);

	if (data != NULL && texDesc.MipLevels == 0)
	{
		D3D_API_1(D3D->pDeviceContext, GenerateMips, tex->resourceView);
	}

	if (D3Dnvg__checkError(hr, "create tex"))
		return 0;

	return tex->id;
}

static int D3Dnvg__renderDeleteTexture(void* uptr, int image)
{
	struct D3DNVGcontext* D3D = (struct D3DNVGcontext*)uptr;
	return D3Dnvg__deleteTexture(D3D, image);
}

static int D3Dnvg__renderUpdateTexture(void* uptr, int image, int x, int y, int w, int h, const unsigned char* data)
{
	struct D3DNVGcontext* D3D = (struct D3DNVGcontext*)uptr;
	struct D3DNVGtexture* tex = D3Dnvg__findTexture(D3D, image);
	D3D11_BOX box;
	unsigned int pixelWidthBytes;
	unsigned char* pData;

	if (tex == NULL)
	{
		return 0;
	}

	box.left = x;
	box.right = (x + w);
	box.top = y;
	box.bottom = (y + h);
	box.front = 0;
	box.back = 1;

	if (tex->type == NVG_TEXTURE_RGBA)
	{
		pixelWidthBytes = 4;
	}
	else
	{
		pixelWidthBytes = 1;
	}

	if (tex->flags & NVG_IMAGE_COPY_SWAP) {
		ID3D11Texture2D *stagingTexture;
		D3D11_TEXTURE2D_DESC stagingTextureDesc;
		D3D11_MAPPED_SUBRESOURCE textureMemory;
		HRESULT hr;
		const unsigned char *src;
		unsigned char *dst;
		// 缓存临时 texture
		if (tex->stagingTex) {
			stagingTexture = tex->stagingTex;
		} else {
			D3D_API_1(tex->tex, GetDesc, &stagingTextureDesc);
			stagingTextureDesc.Width = w;
			stagingTextureDesc.Height = h;
			stagingTextureDesc.BindFlags = 0;
			stagingTextureDesc.MiscFlags = 0;
			stagingTextureDesc.CPUAccessFlags = D3D11_CPU_ACCESS_WRITE;
			stagingTextureDesc.Usage = D3D11_USAGE_STAGING;
			hr = D3D_API_3(D3D->pDevice, CreateTexture2D, &stagingTextureDesc, NULL, &stagingTexture);
			if (FAILED(hr)) {
				return 0;
			}
			tex->stagingTex = stagingTexture;
		}

		hr = D3D_API_5(D3D->pDeviceContext, Map, (ID3D11Resource*)stagingTexture, 0, D3D11_MAP_WRITE, 0, &textureMemory);
		if (FAILED(hr)) {
			return 0;
		}
		unsigned int length = w * pixelWidthBytes;
		unsigned int pitch = length;
		src = (const unsigned char *)data;
		dst = (unsigned char *)textureMemory.pData;
		if (length == textureMemory.RowPitch) {
			memcpy(dst, src, length*h);
		} else {
			if (length > textureMemory.RowPitch) {
				length = textureMemory.RowPitch;
			}
			int row;
			for (row = 0; row < h; ++row) {
				memcpy(dst, src, length);
				src += pitch;
				dst += textureMemory.RowPitch;
			}
		}
		D3D_API_2(D3D->pDeviceContext, Unmap, (ID3D11Resource*)stagingTexture, 0);
		D3D_API_8(D3D->pDeviceContext, CopySubresourceRegion, (ID3D11Resource *)tex->tex, 0, x, y, 0, (ID3D11Resource *)stagingTexture, 0, NULL);
	} else {
		pData = (unsigned char*)data + (y * (tex->width * pixelWidthBytes)) + (x * pixelWidthBytes);
		D3D_API_6(D3D->pDeviceContext, UpdateSubresource, (ID3D11Resource*)tex->tex, 0, &box, pData, tex->width, tex->width * tex->height);
	}
	return 1;
}

static int D3Dnvg__renderGetTextureSize(void* uptr, int image, int* w, int* h)
{
	struct D3DNVGcontext* D3D = (struct D3DNVGcontext*)uptr;
	struct D3DNVGtexture* tex = D3Dnvg__findTexture(D3D, image);
	if (tex == NULL)
	{
		return 0;
	}
	*w = tex->width;
	*h = tex->height;
	return 1;
}

static void D3Dnvg__xformToMat3x3(float* m3, float* t)
{
	m3[0] = t[0];
	m3[1] = t[1];
	m3[2] = 0.0f;
	m3[3] = t[2];
	m3[4] = t[3];
	m3[5] = 0.0f;
	m3[6] = t[4];
	m3[7] = t[5];
	m3[8] = 1.0f;
}

static struct NVGcolor D3Dnvg__premulColor(struct NVGcolor c)
{
	c.r *= c.a;
	c.g *= c.a;
	c.b *= c.a;
	return c;
}

static int D3Dnvg__convertPaint(struct D3DNVGcontext* D3D, struct D3DNVGfragUniforms* frag,
	struct NVGpaint* paint, struct NVGscissor* scissor,
	float width, float fringe, float strokeThr)
{
	struct D3DNVGtexture* tex = NULL;
	float invxform[6], paintMat[9], scissorMat[9];

	memset(frag, 0, sizeof(*frag));

	frag->innerCol = D3Dnvg__premulColor(paint->innerColor);
	frag->outerCol = D3Dnvg__premulColor(paint->outerColor);

	if (scissor->extent[0] < -0.5f || scissor->extent[1] < -0.5f)
	{
		memset(scissorMat, 0, sizeof(scissorMat));
		frag->scissorExt[0] = 1.0f;
		frag->scissorExt[1] = 1.0f;
		frag->scissorScale[0] = 1.0f;
		frag->scissorScale[1] = 1.0f;
	}
	else
	{
		nvgTransformInverse(invxform, scissor->xform);
		D3Dnvg__xformToMat3x3(scissorMat, invxform);
		frag->scissorExt[0] = scissor->extent[0];
		frag->scissorExt[1] = scissor->extent[1];
		frag->scissorScale[0] = sqrtf(scissor->xform[0] * scissor->xform[0] + scissor->xform[2] * scissor->xform[2]) / fringe;
		frag->scissorScale[1] = sqrtf(scissor->xform[1] * scissor->xform[1] + scissor->xform[3] * scissor->xform[3]) / fringe;
	}
	D3Dnvg_copyMatrix3to4(frag->scissorMat, scissorMat);


	frag->extent[0] = paint->extent[0];
	frag->extent[1] = paint->extent[1];

	frag->strokeMult[0] = (width*0.5f + fringe*0.5f) / fringe;
	frag->strokeMult[1] = strokeThr;

	if (paint->image != 0)
	{
		tex = D3Dnvg__findTexture(D3D, paint->image);
		if (tex == NULL)
		{
			return 0;
		}

		if ((tex->flags & NVG_IMAGE_FLIPY) != 0)
		{
			float m1[6], m2[6];
			nvgTransformTranslate(m1, 0.0f, frag->extent[1] * 0.5f);
			nvgTransformMultiply(m1, paint->xform);
			nvgTransformScale(m2, 1.0f, -1.0f);
			nvgTransformMultiply(m2, m1);
			nvgTransformTranslate(m1, 0.0f, -frag->extent[1] * 0.5f);
			nvgTransformMultiply(m1, m2);
			nvgTransformInverse(invxform, m1);
		}
		else
		{
			nvgTransformInverse(invxform, paint->xform);
		}
		frag->type = NSVG_SHADER_FILLIMG;

		#if NANOVG_GL_USE_UNIFORMBUFFER
		if (tex->type == NVG_TEXTURE_RGBA)
		{
			frag->texType = (tex->flags & NVG_IMAGE_PREMULTIPLIED) ? 0 : 1;
		}
		else
		{
			frag->texType = 2;
		}
		#else
		if (tex->type == NVG_TEXTURE_RGBA)
			frag->texType = (tex->flags & NVG_IMAGE_PREMULTIPLIED) ? 0.0f : 1.0f;
		else
			frag->texType = 2.0f;
		#endif
	}
	else
	{
		frag->type = NSVG_SHADER_FILLGRAD;
		frag->radius[0] = paint->radius;
		frag->feather[0] = paint->feather;
		nvgTransformInverse(invxform, paint->xform);
	}

	D3Dnvg__xformToMat3x3(paintMat, invxform);
	D3Dnvg_copyMatrix3to4(frag->paintMat, paintMat);

	//D3Dnvg_updateShaders(D3D);

	return 1;
}

static struct D3DNVGfragUniforms* nvg__fragUniformPtr(struct D3DNVGcontext* D3D, int i);

static void D3Dnvg__setUniforms(struct D3DNVGcontext* D3D, int uniformOffset, int image)
{
	struct D3DNVGfragUniforms* frag = nvg__fragUniformPtr(D3D, uniformOffset);

	D3D11_MAPPED_SUBRESOURCE MappedResource;

	// Pixel shader constants
	D3D_API_5(D3D->pDeviceContext, Map, (ID3D11Resource*)D3D->pPSConstants, 0, D3D11_MAP_WRITE_DISCARD, 0, &MappedResource);
	memcpy(MappedResource.pData, frag, sizeof(struct D3DNVGfragUniforms));
	D3D_API_2(D3D->pDeviceContext, Unmap, (ID3D11Resource*)D3D->pPSConstants, 0);

	if (image != 0)
	{
		struct D3DNVGtexture* tex = D3Dnvg__findTexture(D3D, image);
		if (tex != NULL)
		{
			D3D_API_3(D3D->pDeviceContext, PSSetShaderResources,0, 1, &tex->resourceView);
		}
		else
		{
		   // D3D_API_3(D3D->pDeviceContext, PSSetShaderResources,0, 1, NULL);
		}
	}
	else
	{
	   // D3D_API_3(D3D->pDeviceContext, PSSetShaderResources,0, 1, NULL);
	}

}

static void D3Dnvg__renderViewport(void* uptr, float width, float height, float devicePixelRatio)
{
	NVG_NOTUSED(devicePixelRatio);

	struct D3DNVGcontext* D3D = (struct D3DNVGcontext*)uptr;
	D3D11_MAPPED_SUBRESOURCE mappedResource;

	//D3D->alphaMode = alphaBlend;
	D3D->shader.vc.viewSize[0] = width;
	D3D->shader.vc.viewSize[1] = height;

	// Vertex parameters only change when viewport size changes
	D3D_API_5(D3D->pDeviceContext, Map, (ID3D11Resource*)D3D->pVSConstants, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource);
	memcpy(mappedResource.pData, &D3D->shader.vc, sizeof(struct VS_CONSTANTS));
	D3D_API_2(D3D->pDeviceContext, Unmap, (ID3D11Resource*)D3D->pVSConstants, 0);
}

static void D3Dnvg__fill(struct D3DNVGcontext* D3D, struct D3DNVGcall* call)
{
	struct D3DNVGpath* paths = &D3D->paths[call->pathOffset];
	int i, npaths = call->pathCount;

	// Draw shapes
	D3D_API_2(D3D->pDeviceContext, OMSetDepthStencilState, D3D->pDepthStencilDrawShapes, 0);
	D3D_API_3(D3D->pDeviceContext, OMSetBlendState, D3D->pBSNoWrite, NULL, 0xFFFFFFFF);
	D3D_API_1(D3D->pDeviceContext, RSSetState, D3D->pRSNoCull);

	// set bindpoint for solid loc
	D3Dnvg__setUniforms(D3D, call->uniformOffset, 0);

	D3D_API_1(D3D->pDeviceContext, IASetPrimitiveTopology, D3D11_PRIMITIVE_TOPOLOGY_TRIANGLELIST);
	for (i = 0; i < npaths; i++)
	{
		unsigned int numIndices = ((paths[i].fillCount - 2) * 3);
		assert(numIndices < D3D->VertexBuffer.MaxBufferEntries);
		if (numIndices < D3D->VertexBuffer.MaxBufferEntries)
		{
			D3D_API_3(D3D->pDeviceContext, DrawIndexed, numIndices, 0, paths[i].fillOffset);
		}
	}

	// Draw anti-aliased pixels
	D3D_API_1(D3D->pDeviceContext, RSSetState, D3D->pRSCull);
	D3D_API_1(D3D->pDeviceContext, IASetPrimitiveTopology, D3D11_PRIMITIVE_TOPOLOGY_TRIANGLESTRIP);
	D3D_API_3(D3D->pDeviceContext, OMSetBlendState, D3D->pBSBlend, NULL, 0xFFFFFFFF);

	D3Dnvg__setUniforms(D3D, call->uniformOffset + D3D->fragSize, call->image);

	if (D3D->flags & NVG_ANTIALIAS)
	{
		D3D_API_2(D3D->pDeviceContext, OMSetDepthStencilState, D3D->pDepthStencilDrawAA, 0);
		// Draw fringes
		for (i = 0; i < npaths; i++)
		{
			D3D_API_2(D3D->pDeviceContext, Draw, paths[i].strokeCount, paths[i].strokeOffset);
		}
	}

	// Draw fill
	D3D_API_1(D3D->pDeviceContext, RSSetState, D3D->pRSNoCull);

	D3D_API_2(D3D->pDeviceContext, OMSetDepthStencilState, D3D->pDepthStencilFill, 0);
	D3D_API_2(D3D->pDeviceContext, Draw, call->triangleCount, call->triangleOffset);

	D3D_API_2(D3D->pDeviceContext, OMSetDepthStencilState, D3D->pDepthStencilDefault, 0);

}

static void D3Dnvg__convexFill(struct D3DNVGcontext* D3D, struct D3DNVGcall* call)
{
	struct D3DNVGpath* paths = &D3D->paths[call->pathOffset];
	int i, npaths = call->pathCount;

	D3Dnvg__setUniforms(D3D, call->uniformOffset, call->image);

	D3D_API_1(D3D->pDeviceContext, IASetPrimitiveTopology, D3D11_PRIMITIVE_TOPOLOGY_TRIANGLELIST);
	for (i = 0; i < npaths; i++)
	{
		// Draws a fan using indices to fake it up, since there isn't a fan primitive in D3D11.
		if (paths[i].fillCount > 2)
		{
			unsigned int numIndices = ((paths[i].fillCount - 2) * 3);
			assert(numIndices < D3D->VertexBuffer.MaxBufferEntries);
			if (numIndices < D3D->VertexBuffer.MaxBufferEntries)
			{
				D3D_API_3(D3D->pDeviceContext, DrawIndexed, numIndices, 0, paths[i].fillOffset);
			}
		}
	}

	// Draw fringes
	D3D_API_1(D3D->pDeviceContext, IASetPrimitiveTopology, D3D11_PRIMITIVE_TOPOLOGY_TRIANGLESTRIP);
	for (i = 0; i < npaths; i++)
	{
		if (paths[i].strokeCount > 0) {
			D3D_API_2(D3D->pDeviceContext, Draw, paths[i].strokeCount, paths[i].strokeOffset);
		}
	}
}

static void D3Dnvg__stroke(struct D3DNVGcontext* D3D, struct D3DNVGcall* call)
{
	struct D3DNVGpath* paths = &D3D->paths[call->pathOffset];
	int npaths = call->pathCount, i;

	D3D_API_1(D3D->pDeviceContext, IASetPrimitiveTopology, D3D11_PRIMITIVE_TOPOLOGY_TRIANGLESTRIP);

	if (D3D->flags & NVG_STENCIL_STROKES)
	{
		// Fill the stroke base without overlap
		D3D_API_2(D3D->pDeviceContext, OMSetDepthStencilState, D3D->pDepthStencilDefault, 0);

		D3Dnvg__setUniforms(D3D, call->uniformOffset + D3D->fragSize, call->image);
		for (i = 0; i < npaths; i++)
		{
			D3D_API_2(D3D->pDeviceContext, Draw, paths[i].strokeCount, paths[i].strokeOffset);
		}

		// Draw anti-aliased pixels.
		D3Dnvg__setUniforms(D3D, call->uniformOffset, call->image);
		D3D_API_2(D3D->pDeviceContext, OMSetDepthStencilState, D3D->pDepthStencilDrawAA, 0);
		for (i = 0; i < npaths; i++)
		{
			D3D_API_2(D3D->pDeviceContext, Draw, paths[i].strokeCount, paths[i].strokeOffset);
		}

		// Clear stencil buffer.
		D3D_API_2(D3D->pDeviceContext, OMSetDepthStencilState, D3D->pDepthStencilFill, 0);
		for (i = 0; i < npaths; i++)
		{
			D3D_API_2(D3D->pDeviceContext, Draw, paths[i].strokeCount, paths[i].strokeOffset);
		}

		D3D_API_2(D3D->pDeviceContext, OMSetDepthStencilState, D3D->pDepthStencilDefault, 0);
	}
	else
	{
		D3Dnvg__setUniforms(D3D, call->uniformOffset, call->image);
		// Draw Strokes
		for (i = 0; i < npaths; i++)
		{
			D3D_API_2(D3D->pDeviceContext, Draw, paths[i].strokeCount, paths[i].strokeOffset);
		}
	}
}

static void D3Dnvg__triangles(struct D3DNVGcontext* D3D, struct D3DNVGcall* call)
{
	D3D_API_1(D3D->pDeviceContext, IASetPrimitiveTopology, D3D11_PRIMITIVE_TOPOLOGY_TRIANGLELIST);

	D3Dnvg__setUniforms(D3D, call->uniformOffset, call->image);

	D3D_API_2(D3D->pDeviceContext, Draw, call->triangleCount, call->triangleOffset);
}

static void D3Dnvg__renderCancel(void* uptr)
{
	struct D3DNVGcontext* D3D = (struct D3DNVGcontext*)uptr;
	D3D->nverts = 0;
	D3D->npaths = 0;
	D3D->ncalls = 0;
	D3D->nuniforms = 0;
}

static void D3Dnvg__renderFlush(void* uptr)
{
	struct D3DNVGcontext* D3D = (struct D3DNVGcontext*)uptr;
	int i;

	if (D3D->ncalls > 0)
	{
		unsigned int buffer0Offset = D3Dnvg_updateVertexBuffer(D3D->pDeviceContext, &D3D->VertexBuffer, D3D->verts, D3D->nverts);
		D3Dnvg_setBuffers(D3D, buffer0Offset);

		// Ensure valid state
		D3D_API_3(D3D->pDeviceContext, PSSetConstantBuffers, 0, 1, &D3D->pPSConstants);
		D3D_API_3(D3D->pDeviceContext, VSSetConstantBuffers, 0, 1, &D3D->pVSConstants);

		D3D_API_3(D3D->pDeviceContext, PSSetShader, D3D->shader.frag, NULL, 0);
		D3D_API_3(D3D->pDeviceContext, VSSetShader, D3D->shader.vert, NULL, 0);

		 // Draw shapes
		D3D_API_2(D3D->pDeviceContext, OMSetDepthStencilState, D3D->pDepthStencilDefault, 0);
		D3D_API_3(D3D->pDeviceContext, OMSetBlendState, D3D->pBSBlend, NULL, 0xFFFFFFFF);

		D3D_API_1(D3D->pDeviceContext, RSSetState, D3D->pRSCull);

		for (i = 0; i < D3D->ncalls; i++) {
			struct D3DNVGcall* call = &D3D->calls[i];

			if (call->image != 0)
			{
				struct D3DNVGtexture* tex = D3Dnvg__findTexture(D3D, call->image);
				if (tex != NULL)
				{
					D3D_API_3(D3D->pDeviceContext, PSSetSamplers, 0, 1, &D3D->pSamplerState[(tex->flags & NVG_IMAGE_REPEATX ? 1 : 0) + (tex->flags & NVG_IMAGE_REPEATY ? 2 : 0)]);
				}
			}

			if (call->type == D3DNVG_FILL)
				D3Dnvg__fill(D3D, call);
			else if (call->type == D3DNVG_CONVEXFILL)
				D3Dnvg__convexFill(D3D, call);
			else if (call->type == D3DNVG_STROKE)
				D3Dnvg__stroke(D3D, call);
			else if (call->type == D3DNVG_TRIANGLES)
				D3Dnvg__triangles(D3D, call);
		}
	}

	// Reset calls
	D3D->nverts = 0;
	D3D->npaths = 0;
	D3D->ncalls = 0;
	D3D->nuniforms = 0;
}

static int D3Dnvg__maxVertCount(const struct NVGpath* paths, int npaths)
{
	int i, count = 0;
	for (i = 0; i < npaths; i++) {
		count += paths[i].nfill;
		count += paths[i].nstroke;
	}
	return count;
}

static struct D3DNVGcall* D3Dnvg__allocCall(struct D3DNVGcontext* D3D)
{
	struct D3DNVGcall* ret = NULL;
	if (D3D->ncalls+1 > D3D->ccalls) {
		struct D3DNVGcall* calls;
		int ccalls = D3Dnvg__maxi(D3D->ncalls+1, 128) + D3D->ccalls/2; // 1.5x Overallocate
		calls = (struct D3DNVGcall*)realloc(D3D->calls, sizeof(struct D3DNVGcall) * ccalls);
		if (calls == NULL) return NULL;
		D3D->calls = calls;
		D3D->ccalls = ccalls;
	}
	ret = &D3D->calls[D3D->ncalls++];
	memset(ret, 0, sizeof(struct D3DNVGcall));
	return ret;
}

static int D3Dnvg__allocPaths(struct D3DNVGcontext* D3D, int n)
{
	int ret = 0;
	if (D3D->npaths+n > D3D->cpaths) {
		struct D3DNVGpath* paths;
		int cpaths = D3Dnvg__maxi(D3D->npaths + n, 128) + D3D->cpaths/2; // 1.5x Overallocate
		paths = (struct D3DNVGpath*)realloc(D3D->paths, sizeof(struct D3DNVGpath) * cpaths);
		if (paths == NULL) return -1;
		D3D->paths = paths;
		D3D->cpaths = cpaths;
	}
	ret = D3D->npaths;
	D3D->npaths += n;
	return ret;
}

static int D3Dnvg__allocVerts(struct D3DNVGcontext* D3D, int n)
{
	int ret = 0;
	if (D3D->nverts+n > D3D->cverts) {
		struct NVGvertex* verts;
		int cverts = D3Dnvg__maxi(D3D->nverts + n, 4096) + D3D->cverts/2; // 1.5x Overallocate
		verts = (struct NVGvertex*)realloc(D3D->verts, sizeof(struct NVGvertex) * cverts);
		if (verts == NULL) return -1;
		D3D->verts = verts;
		D3D->cverts = cverts;
	}
	ret = D3D->nverts;
	D3D->nverts += n;
	return ret;
}

static int D3Dnvg__allocFragUniforms(struct D3DNVGcontext* D3D, int n)
{
	int ret = 0, structSize = D3D->fragSize;
	if (D3D->nuniforms+n > D3D->cuniforms) {
		unsigned char* uniforms;
		int cuniforms = D3Dnvg__maxi(D3D->nuniforms+n, 128) + D3D->cuniforms/2; // 1.5x Overallocate
		uniforms = (unsigned char*)realloc(D3D->uniforms, structSize * cuniforms);
		if (uniforms == NULL) return -1;
		D3D->uniforms = uniforms;
		D3D->cuniforms = cuniforms;
	}
	ret = D3D->nuniforms * structSize;
	D3D->nuniforms += n;
	return ret;
}

static struct D3DNVGfragUniforms* nvg__fragUniformPtr(struct D3DNVGcontext* D3D, int i)
{
	return (struct D3DNVGfragUniforms*)&D3D->uniforms[i];
}

static void D3Dnvg__vset(struct NVGvertex* vtx, float x, float y, float u, float v)
{
	vtx->x = x;
	vtx->y = y;
	vtx->u = u;
	vtx->v = v;
}

static void D3Dnvg__renderFill(void* uptr, struct NVGpaint* paint, NVGcompositeOperationState compositeOperation, struct NVGscissor* scissor, float fringe,
							  const float* bounds, const struct NVGpath* paths, int npaths)
{
	struct D3DNVGcontext* D3D = (struct D3DNVGcontext*)uptr;
	struct D3DNVGcall* call = D3Dnvg__allocCall(D3D);
	struct NVGvertex* quad;
	struct D3DNVGfragUniforms* frag;
	int i, maxverts, offset;

	if (call == NULL) return;

	call->type = D3DNVG_FILL;
	call->triangleCount = 4;
	call->pathOffset = D3Dnvg__allocPaths(D3D, npaths);
	if (call->pathOffset == -1) goto error;
	call->pathCount = npaths;
	call->image = paint->image;
	NVG_NOTUSED(compositeOperation);

	if (npaths == 1 && paths[0].convex)
	{
		call->type = D3DNVG_CONVEXFILL;
		call->triangleCount = 0;	// Bounding box fill quad not needed for convex fill
	}

	// Allocate vertices for all the paths.
	maxverts = D3Dnvg__maxVertCount(paths, npaths) + call->triangleCount;
	offset = D3Dnvg__allocVerts(D3D, maxverts);
	if (offset == -1) goto error;

	for (i = 0; i < npaths; i++) {
		struct D3DNVGpath* copy = &D3D->paths[call->pathOffset + i];
		const struct NVGpath* path = &paths[i];
		memset(copy, 0, sizeof(struct D3DNVGpath));
		if (path->nfill > 0) {
			copy->fillOffset = offset;
			copy->fillCount = path->nfill;
			memcpy(&D3D->verts[offset], path->fill, sizeof(struct NVGvertex) * path->nfill);
			offset += path->nfill;
		}
		if (path->nstroke > 0) {
			copy->strokeOffset = offset;
			copy->strokeCount = path->nstroke;
			memcpy(&D3D->verts[offset], path->stroke, sizeof(struct NVGvertex) * path->nstroke);
			offset += path->nstroke;
		}
	}

	if (call->type == D3DNVG_FILL) {
		// Quad
		call->triangleOffset = offset;
		quad = &D3D->verts[call->triangleOffset];
		D3Dnvg__vset(&quad[0], bounds[2], bounds[3], 0.5f, 1.0f);
		D3Dnvg__vset(&quad[1], bounds[2], bounds[1], 0.5f, 1.0f);
		D3Dnvg__vset(&quad[2], bounds[0], bounds[3], 0.5f, 1.0f);
		D3Dnvg__vset(&quad[3], bounds[0], bounds[1], 0.5f, 1.0f);

		call->uniformOffset = D3Dnvg__allocFragUniforms(D3D, 2);
		if (call->uniformOffset == -1) goto error;
		// Simple shader for stencil
		frag = nvg__fragUniformPtr(D3D, call->uniformOffset);
		memset(frag, 0, sizeof(*frag));
		frag->strokeMult[1] = -1.0f;
		frag->type = NSVG_SHADER_SIMPLE;
		// Fill shader
		D3Dnvg__convertPaint(D3D, nvg__fragUniformPtr(D3D, call->uniformOffset + D3D->fragSize), paint, scissor, fringe, fringe, -1.0f);
	} else {
		call->uniformOffset = D3Dnvg__allocFragUniforms(D3D, 1);
		if (call->uniformOffset == -1) goto error;
		// Fill shader
		D3Dnvg__convertPaint(D3D, nvg__fragUniformPtr(D3D, call->uniformOffset), paint, scissor, fringe, fringe, -1.0f);
	}

	return;

error:
	// We get here if call alloc was ok, but something else is not.
	// Roll back the last call to prevent drawing it.
	if (D3D->ncalls > 0) D3D->ncalls--;
}

static void D3Dnvg__renderStroke(void* uptr, struct NVGpaint* paint, NVGcompositeOperationState compositeOperation, struct NVGscissor* scissor, float fringe,
								float strokeWidth, const struct NVGpath* paths, int npaths)
{
	struct D3DNVGcontext* D3D = (struct D3DNVGcontext*)uptr;
	struct D3DNVGcall* call = D3Dnvg__allocCall(D3D);
	int i, maxverts, offset;

	if (call == NULL) return;

	call->type = D3DNVG_STROKE;
	call->pathOffset = D3Dnvg__allocPaths(D3D, npaths);
	if (call->pathOffset == -1) goto error;
	call->pathCount = npaths;
	call->image = paint->image;
	NVG_NOTUSED(compositeOperation);

	// Allocate vertices for all the paths.
	maxverts = D3Dnvg__maxVertCount(paths, npaths);
	offset = D3Dnvg__allocVerts(D3D, maxverts);
	if (offset == -1) goto error;

	for (i = 0; i < npaths; i++) {
		struct D3DNVGpath* copy = &D3D->paths[call->pathOffset + i];
		const struct NVGpath* path = &paths[i];
		memset(copy, 0, sizeof(struct D3DNVGpath));
		if (path->nstroke) {
			copy->strokeOffset = offset;
			copy->strokeCount = path->nstroke;
			memcpy(&D3D->verts[offset], path->stroke, sizeof(struct NVGvertex) * path->nstroke);
			offset += path->nstroke;
		}
	}

	if (D3D->flags & NVG_STENCIL_STROKES) {
		// Fill shader
		call->uniformOffset = D3Dnvg__allocFragUniforms(D3D, 2);
		if (call->uniformOffset == -1) goto error;

		D3Dnvg__convertPaint(D3D, nvg__fragUniformPtr(D3D, call->uniformOffset), paint, scissor, strokeWidth, fringe, -1.0f);
		D3Dnvg__convertPaint(D3D, nvg__fragUniformPtr(D3D, call->uniformOffset + D3D->fragSize), paint, scissor, strokeWidth, fringe, 1.0f - 0.5f/255.0f);

	} else {
		// Fill shader
		call->uniformOffset = D3Dnvg__allocFragUniforms(D3D, 1);
		if (call->uniformOffset == -1) goto error;
		D3Dnvg__convertPaint(D3D, nvg__fragUniformPtr(D3D, call->uniformOffset), paint, scissor, strokeWidth, fringe, -1.0f);
	}

	return;

error:
	// We get here if call alloc was ok, but something else is not.
	// Roll back the last call to prevent drawing it.
	if (D3D->ncalls > 0) D3D->ncalls--;
}

static void D3Dnvg__renderTriangles(void* uptr, struct NVGpaint* paint, NVGcompositeOperationState compositeOperation, struct NVGscissor* scissor,
								   const struct NVGvertex* verts, int nverts, float fringe)
{
	struct D3DNVGcontext* D3D = (struct D3DNVGcontext*)uptr;
	struct D3DNVGcall* call = D3Dnvg__allocCall(D3D);
	struct D3DNVGfragUniforms* frag;

	if (call == NULL) return;

	call->type = D3DNVG_TRIANGLES;
	call->image = paint->image;
	NVG_NOTUSED(compositeOperation);

	// Allocate vertices for all the paths.
	call->triangleOffset = D3Dnvg__allocVerts(D3D, nverts);
	if (call->triangleOffset == -1) goto error;
	call->triangleCount = nverts;

	memcpy(&D3D->verts[call->triangleOffset], verts, sizeof(struct NVGvertex) * nverts);

	// Fill shader
	call->uniformOffset = D3Dnvg__allocFragUniforms(D3D, 1);
	if (call->uniformOffset == -1) goto error;
	frag = nvg__fragUniformPtr(D3D, call->uniformOffset);
	D3Dnvg__convertPaint(D3D, frag, paint, scissor, 1.0f, fringe, -1.0f);
	frag->type = NSVG_SHADER_IMG;

	return;

error:
	// We get here if call alloc was ok, but something else is not.
	// Roll back the last call to prevent drawing it.
	if (D3D->ncalls > 0) D3D->ncalls--;
}

static void D3Dnvg__renderDelete(void* uptr)
{
	struct D3DNVGcontext* D3D = (struct D3DNVGcontext*)uptr;
	int i;
	if (D3D == NULL)
	{
		return;
	}

	D3Dnvg__deleteShader(&D3D->shader);

	for (i = 0; i < D3D->ntextures; i++)
	{
		if (D3D->textures[i].tex != 0 && (D3D->textures[i].flags & NVG_IMAGE_NODELETE) == 0)
		{
			D3D_API_RELEASE(D3D->textures[i].tex);
			D3D_API_RELEASE(D3D->textures[i].resourceView);
		}
	}
	for (i = 0; i < 4; i++)
	{
		D3D_API_RELEASE(D3D->pSamplerState[i]);
	}

	D3D_API_RELEASE(D3D->VertexBuffer.pBuffer);

	D3D_API_RELEASE(D3D->pVSConstants);
	D3D_API_RELEASE(D3D->pPSConstants);

	D3D_API_RELEASE(D3D->pFanIndexBuffer);
	D3D_API_RELEASE(D3D->pLayoutRenderTriangles);

	D3D_API_RELEASE(D3D->pBSBlend);
	D3D_API_RELEASE(D3D->pBSNoWrite);
	D3D_API_RELEASE(D3D->pRSCull);
	D3D_API_RELEASE(D3D->pRSNoCull);
	D3D_API_RELEASE(D3D->pDepthStencilDrawShapes);
	D3D_API_RELEASE(D3D->pDepthStencilDrawAA);
	D3D_API_RELEASE(D3D->pDepthStencilFill);
	D3D_API_RELEASE(D3D->pDepthStencilDefault);

	// We took a reference to this
	// Don't delete the device though.
	D3D_API_RELEASE(D3D->pDeviceContext);

	free(D3D->textures);

	free(D3D->paths);
	free(D3D->verts);
	free(D3D->uniforms);
	free(D3D->calls);

	free(D3D);
}


struct NVGcontext* nvgCreateD3D11(ID3D11Device* pDevice, int flags)
{
	struct NVGparams params;
	struct NVGcontext* ctx = NULL;
	struct D3DNVGcontext* D3D = (struct D3DNVGcontext*)malloc(sizeof(struct D3DNVGcontext));
	if (D3D == NULL)
	{
		goto error;
	}
	memset(D3D, 0, sizeof(struct D3DNVGcontext));
	D3D->pDevice = pDevice;
	D3D_API_1(pDevice, GetImmediateContext, &D3D->pDeviceContext);

	memset(&params, 0, sizeof(params));
	params.renderCreate = D3Dnvg__renderCreate;
	params.renderCreateTexture = D3Dnvg__renderCreateTexture;
	params.renderDeleteTexture = D3Dnvg__renderDeleteTexture;
	params.renderUpdateTexture = D3Dnvg__renderUpdateTexture;
	params.renderGetTextureSize = D3Dnvg__renderGetTextureSize;
	params.renderViewport = D3Dnvg__renderViewport;
	params.renderCancel = D3Dnvg__renderCancel;
	params.renderFlush = D3Dnvg__renderFlush;
	params.renderFill = D3Dnvg__renderFill;
	params.renderStroke = D3Dnvg__renderStroke;
	params.renderTriangles = D3Dnvg__renderTriangles;
	params.renderDelete = D3Dnvg__renderDelete;
	params.userPtr = D3D;
	params.edgeAntiAlias = flags & NVG_ANTIALIAS ? 1 : 0;

	D3D->flags = flags;

	ctx = nvgCreateInternal(&params);
	if (ctx == NULL) goto error;

	return ctx;

error:
	// 'D3D' is freed by nvgDeleteInternal.
	if (ctx != NULL) nvgDeleteInternal(ctx);
	return NULL;
}

void nvgDeleteD3D11(struct NVGcontext* ctx)
{
	nvgDeleteInternal(ctx);
}

#ifdef IMPLEMENTED_IMAGE_FUNCS
int nvd3dCreateImageFromHandle(struct NVGcontext* ctx, void* textureId, int w, int h, int flags)
{

	/*struct D3DNVGcontext* gl = (struct D3DNVGcontext*)nvgInternalParams(ctx)->userPtr;
	struct D3DNVGtexture* tex = D3Dnvg__allocTexture(gl);

	if (tex == NULL) return 0;

	tex->type = NVG_TEXTURE_RGBA;
	tex->tex = textureId;
	tex->flags = flags;
	tex->width = w;
	tex->height = h;

	return tex->id;
*/
	return 0;
}

unsigned int nvd3dImageHandle(struct NVGcontext* ctx, int image)
{
	/*
	struct D3DNVGcontext* gl = (struct D3DNVGcontext*)nvgInternalParams(ctx)->userPtr;
	struct D3DNVGtexture* tex = D3Dnvg__findTexture(gl, image);
	return tex->tex;*/
	return 0;
}

void nvd3dImageFlags(struct NVGcontext* ctx, int image, int flags)
{
	/*
	struct D3DNVGcontext* gl = (struct D3DNVGcontext*)nvgInternalParams(ctx)->userPtr;
	struct D3DNVGtexture* tex = D3Dnvg__findTexture(gl, image);
	tex->flags = flags;
	*/
}
#endif

#endif //NANOVG_D3D11_IMPLEMENTATION
#endif //NANOVG_D3D11_H