Cocos2d-x学习（二十二）：cocos2d-x中CCScale9Sprite的另一种实现
cocos2d 2.0之后加入了一种九宫格的实现，主要作用是用来拉伸图片，这样的好处在于保留图片四个角不变形的同时，对图片中间部分进行拉伸，来满足一些控件的自适应（PS: 比如包括按钮，对话框，最直观的形象就是ios里的短信气泡了），这就要求图片资源的中间部分是纯色或者是简单的渐变了！
1.cocos2d中九宫格CCScale9Sprite的实现
（1）原理
cocos2d的实现非常巧妙，是通过1个CCSpriteBatchNode和9个CCSprite来实现的，原理很简单，通过将原纹理资源切割成9部分（PS: 这也是叫九宫格的原因），根据想要的尺寸，完成以下的三个步骤：
a. 保持4个角部分不变形
b. 单向拉伸4条边（即在4个角两两之间的边，比如上边，只做横向拉伸）
c. 双向拉伸中间部分（即九宫格的中间部分，横向，纵向同时拉伸，PS：拉伸比例不一定相同）
（PS: 更多原理可参考 http://yannickloriot.com/2011/12/create-buttons-in-cocos2d-by-using-cccontrolbutton/）
（2）实现
CCSpriteBatchNode的资源为整个的纹理，9个CCSprite对应于纹理的9个部分（根据纹理不同，9部分所占比例会有所不同），根据想要的尺寸，将9部分拼装在一起！
（3）优缺点
优点：思路简单清晰；使用CCSpriteBatchNode，只需要一次绘制，效率较高
缺点：内存占用大，需要1个CCSpriteBatchNode和9个CCSprite对象；不支持CCSpriteBatchNode（如果控件很多，我们都需要对每个控件单独绘制一次，会影响效率）
2.cocos2d-x中CCSprite的绘制
在介绍我的九宫格实现之前，先简单介绍一下CCSprite的绘制原理
（1）顶点数据
每一个CCSprite都保持了一个关于顶点数据的结构体
// vertex coords, texture coords and color info
ccV3F_C4B_T2F_Quad m_sQuad;
这个Quad字眼的意思是一个矩形，参照ccV3F_C4B_T2F_Quad的定义，可以得知，是包含4个顶点数据的结构体（根据注释可知4个顶点分别为：左上，左下，右上，右下）
//! 4 ccVertex3FTex2FColor4B
typedef struct _ccV3F_C4B_T2F_Quad
{
//! top left
ccV3F_C4B_T2F tl;
//! bottom left
ccV3F_C4B_T2F bl;
//! top right
ccV3F_C4B_T2F tr;
//! bottom right
ccV3F_C4B_T2F br;
} ccV3F_C4B_T2F_Quad;
而ccV3F_C4B_T2F又是一个关于顶点信息的结构体，包括坐标（x, y, z），颜色（r, g, b, a），纹理坐标（x, y）
（PS：2D游戏中，坐标的z都为0，这里的z并不是Z-Order，Z-Order是指渲染的先后属性，z是代表3D的z轴坐标）
//! a Point with a vertex point, a tex coord point and a color 4B
typedef struct _ccV3F_C4B_T2F
{
//! vertices (3F)
ccVertex3F vertices; // 12 bytes
// char __padding__[4];
//! colors (4B)
ccColor4B colors; // 4 bytes
// char __padding2__[4];
// tex coords (2F)
ccTex2F texCoords; // 8 byts
} ccV3F_C4B_T2F;
（2）绘制
在初始化精灵之后，就将纹理的四个顶点信息保存在m_sQuad中了，接下来要做的，就是根据m_sQuad的信息来绘制
由于OpenGL是状态机的设计，所以要先将顶点信息保存，再根据顶点的关系进行绘制，主要的绘制代码如下：
#define kQuadSize sizeof(m_sQuad.bl)
int size = sizeof(m_sQuad.bl);
if (m_pobTexture)
{
glBindTexture(GL_TEXTURE_2D, m_pobTexture->getName());
}
else
{
glBindTexture(GL_TEXTURE_2D, 0);
}
long offset = (long)&m_sQuad;
// vertex
int diff = offsetof(ccV3F_C4B_T2F, vertices);
glVertexPointer(3, GL_FLOAT, kQuadSize, (void*)(offset + diff));
// color
diff = offsetof( ccV3F_C4B_T2F, colors);
glColorPointer(4, GL_UNSIGNED_BYTE, kQuadSize, (void*)(offset + diff));
// tex coords
diff = offsetof( ccV3F_C4B_T2F, texCoords);
glTexCoordPointer(2, GL_FLOAT, kQuadSize, (void*)(offset + diff));
glDrawArrays(GL_TRIANGLE_STRIP, 0, 4);
（PS: offsetof()函数是得到结构体中某一数据的地址偏移量）
根据注释可知，先将顶点的坐标数据保存，再将顶点的颜色数据保存，最后将顶点的纹理映射坐标保存
（吐槽一下：#define kQuadSize sizeof(m_sQuad.bl) 这个宏的名字把我迷惑了，我不知道为什么会有Quad字眼，我觉得应该是kVertexSize）
3. CCScaleNineSprite的实现
（吐槽一下：我没找到更好的关于九宫格的名字，于是偷懒将9换成了Nine。。。）
我的九宫格的实现和CCScale9Sprite略有不同，只是优化了其内存的问题，我将1个CCSpriteBatchNode和9个CCSprite用1个CCSprite来实现了，通过纹理映射做拉伸！
（PS：我目前也没有解决支持CCSpriteBatchNode，因为CCSpriteBatchNode的子节点要求是CCSprite类型，而我的CCScaleNineSprite并不是继承于CCSprite，而是于CCSprite是兄弟关系，因为其顶点的数据不同，所以我认为不是继承关系，当然可以考虑把CCSprite的顶点数据修改，使其不再被限制于固定4个顶点）
我偷懒将CCSprite.h和CCSprite.cpp拷贝了一份，注释掉了一些不常用的方法，以及对CCSpriteBatchNode的支持。。。
将m_sQuad替换为ccV3F_C4B_T2F mScaleNineVertices[16]（九宫格需要16个顶点，请根据上面的图计算，包括顶点和切割线的交点）
额外增加了1个设置九宫格比例的方法（重载了3份），通过比例计算出mScaleNineVertices的数据
public:
void CalculateScaleNineVertices(unsigned int widthFromLeft, unsigned int widthFromRight,
unsigned int heightFromTop, unsigned int heightFromBottom);
void CalculateScaleNineVertices(unsigned int widthFromLeft, unsigned int heightFromTop);
void CalculateScaleNineVertices(unsigned int offsetFromEdge);
void CCScaleNineSprite::CalculateScaleNineVertices(unsigned int widthFromLeft, unsigned int widthFromRight,
unsigned int heightFromTop, unsigned int heightFromBottom)
{
float textureOriginX = m_obRectInPixels.origin.x;
float textureOriginY = m_obRectInPixels.origin.y;
float textureWidth = m_obRectInPixels.size.width;
float textureHeight = m_obRectInPixels.size.height;
CCAssert((widthFromLeft < textureWidth) && (widthFromRight < textureWidth) &&
(heightFromTop < textureHeight) && (heightFromBottom < textureHeight), "The SIZE of Corner is too BIG!");
float contentWidth = m_tContentSizeInPixels.width;
float contentHeight = m_tContentSizeInPixels.height;
unsigned int textureAtlasWidth = getTexture()->getPixelsWide();
unsigned int textureAtlasHeight = getTexture()->getPixelsHigh();
ccV3F_C4B_T2F vertice;
// First Line
vertice.vertices.x = 0;
vertice.vertices.y = contentHeight;
vertice.vertices.z = 0;
vertice.colors.a = 255;
vertice.colors.r = 255;
vertice.colors.g = 255;
vertice.colors.b = 255;
vertice.texCoords.u = textureOriginX / textureAtlasWidth;
vertice.texCoords.v = textureOriginY / textureAtlasHeight;
mScaleNineVertices[0] = vertice;
vertice.vertices.x = (float) widthFromLeft;
vertice.texCoords.u = (float) (textureOriginX + widthFromLeft) / textureAtlasWidth;
mScaleNineVertices[1] = vertice;
vertice.vertices.x = (float) (contentWidth - widthFromRight);
vertice.texCoords.u = (float) (textureOriginX + textureWidth - widthFromRight) / textureAtlasWidth;
mScaleNineVertices[2] = vertice;
vertice.vertices.x = (float) contentWidth;
vertice.texCoords.u = (float) (textureOriginX + textureWidth) / textureAtlasWidth;
mScaleNineVertices[3] = vertice;
// Second Line
vertice.vertices.x = 0;
vertice.vertices.y = (float) (contentHeight - heightFromTop);
vertice.texCoords.u = textureOriginX / textureAtlasWidth;
vertice.texCoords.v = (float) (textureOriginY + heightFromTop) / textureAtlasHeight;
mScaleNineVertices[4] = vertice;
vertice.vertices.x = (float) widthFromLeft;
vertice.texCoords.u = (float) (textureOriginX + widthFromLeft) / textureAtlasWidth;
mScaleNineVertices[5] = vertice;
vertice.vertices.x = (float) (contentWidth - widthFromRight);
vertice.texCoords.u = (float) (textureOriginX + textureWidth - widthFromRight) / textureAtlasWidth;
mScaleNineVertices[6] = vertice;
vertice.vertices.x = (float) contentWidth;
vertice.texCoords.u = (float) (textureOriginX + textureWidth) / textureAtlasWidth;
mScaleNineVertices[7] = vertice;
// Third Line
vertice.vertices.x = 0;
vertice.vertices.y = (float) heightFromBottom;
vertice.texCoords.u = textureOriginX / textureAtlasWidth;
vertice.texCoords.v = (float) (textureOriginY + textureHeight - heightFromBottom) / textureAtlasHeight;
mScaleNineVertices[8] = vertice;
vertice.vertices.x = (float) widthFromLeft;
vertice.texCoords.u = (float) (textureOriginX + widthFromLeft) / textureAtlasWidth;
mScaleNineVertices[9] = vertice;
vertice.vertices.x = (float) (contentWidth - widthFromRight);
vertice.texCoords.u = (float) (textureOriginX + textureWidth - widthFromRight) / textureAtlasWidth;
mScaleNineVertices[10] = vertice;
vertice.vertices.x = (float) contentWidth;
vertice.texCoords.u = (float) (textureOriginX + textureWidth) / textureAtlasWidth;
mScaleNineVertices[11] = vertice;
// Fourth Line
vertice.vertices.x = 0;
vertice.vertices.y = 0;
vertice.texCoords.u = textureOriginX / textureAtlasWidth;
vertice.texCoords.v = (float) (textureOriginY + textureHeight) / textureAtlasHeight;
mScaleNineVertices[12] = vertice;
vertice.vertices.x = (float) widthFromLeft;
vertice.texCoords.u = (float) (textureOriginX + widthFromLeft) / textureAtlasWidth;
mScaleNineVertices[13] = vertice;
vertice.vertices.x = (float) (contentWidth - widthFromRight);
vertice.texCoords.u = (float) (textureOriginX + textureWidth - widthFromRight) / textureAtlasWidth;
mScaleNineVertices[14] = vertice;
vertice.vertices.x = (float) contentWidth;
vertice.texCoords.u = (float) (textureOriginX + textureWidth) / textureAtlasWidth;
mScaleNineVertices[15] = vertice;
}
计算好顶点数据之后，简单修改一下draw()函数就可以了（将之前的m_sQuad替换为mScaleNineVertices）
#define kVertexSize sizeof(ccV3F_C4B_T2F)
if (m_pobTexture)
{
glBindTexture(GL_TEXTURE_2D, m_pobTexture->getName());
}
else
{
glBindTexture(GL_TEXTURE_2D, 0);
}
long offset = (long) mScaleNineVertices;
// vertex
int diff = offsetof(ccV3F_C4B_T2F, vertices);
glVertexPointer(3, GL_FLOAT, kVertexSize, (void*)(offset + diff));
// color
diff = offsetof( ccV3F_C4B_T2F, colors);
glColorPointer(4, GL_UNSIGNED_BYTE, kVertexSize, (void*)(offset + diff));
// tex coords
diff = offsetof( ccV3F_C4B_T2F, texCoords);
glTexCoordPointer(2, GL_FLOAT, kVertexSize, (void*)(offset + diff));
glDrawElements(GL_TRIANGLES, 54, GL_UNSIGNED_SHORT, mVerticesIndex);
4.Demo
和CCSprite的使用差不太多，只是需要设置一下ContentSize（即展示的尺寸），并且需要设置九宫格切割的比例（以像素为单位，美术比较好理解！）
// Add a Scale Nine Sprite
CCTexture2D* texture = CCTextureCache::sharedTextureCache()->addImage("GreenButton.png");
CCScaleNineSprite* scaleNineSprite = CCScaleNineSprite::scaleNineSpriteWithTexture(texture);
scaleNineSprite->setContentSize(CCSizeMake(200, 100));
scaleNineSprite->CalculateScaleNineVertices(10);
scaleNineSprite->setPosition(CCPointMake(size.width / 2, size.height / 2));
this->addChild(scaleNineSprite);
效果如下：
原资源：