Android对不同DPI的dimen选择优先级问题

Android对不同DPI的dimen选择优先级问题

问题描述

笔者最近发现这么一个问题，如果仅对values-ldpi和values-xxhdpi下的dimen进行设定，那么hdpi下的手机进行将会如何选择呢？

问题解决

笔者通过查找native层的源代码ResourceType.cpp中的isBetterThan()方法得到了确切的答案。先看看源码对这个屏幕密度适配做了怎样的处理。

bool ResTable_config::isBetterThan(const ResTable_config& o,
        const ResTable_config* requested) const {
        ..........

       if (screenType || o.screenType) {
            if (density != o.density) {
                // Use the system default density (DENSITY_MEDIUM, 160dpi) if none specified.
                const int thisDensity = density ? density : int(ResTable_config::DENSITY_MEDIUM);
                const int otherDensity = o.density ? o.density : int(ResTable_config::DENSITY_MEDIUM);

                // We always prefer DENSITY_ANY over scaling a density bucket.
                if (thisDensity == ResTable_config::DENSITY_ANY) {
                    return true;
                } else if (otherDensity == ResTable_config::DENSITY_ANY) {
                    return false;
                }

                int requestedDensity = requested->density;
                if (requested->density == 0 ||
                        requested->density == ResTable_config::DENSITY_ANY) {
                    requestedDensity = ResTable_config::DENSITY_MEDIUM;
                }

                // DENSITY_ANY is now dealt with. We should look to
                // pick a density bucket and potentially scale it.
                // Any density is potentially useful
                // because the system will scale it. Scaling down
                // is generally better than scaling up.
                int h = thisDensity;
                int l = otherDensity;
                bool bImBigger = true;
                if (l > h) {
                    int t = h;
                    h = l;
                    l = t;
                    bImBigger = false;
                }

                if (requestedDensity >= h) {
                    // requested value higher than both l and h, give h
                    return bImBigger;
                }
                if (l >= requestedDensity) {
                    // requested value lower than both l and h, give l
                    return !bImBigger;
                }
                // saying that scaling down is 2x better than up
                if (((2 * l) - requestedDensity) * h > requestedDensity * requestedDensity) {
                    return !bImBigger;
                } else {
                    return bImBigger;
                }
            }
    .......
}

总体来说走的就近原则，选择离当前密度最近的适配值。系统会便遍历整个values下对应的dimen值，选择一个可包含当前屏幕密度的最小区间，如果实在找不到这样的区间，如ldpi和hdpi进行了适配，但是手机是xxhdpi，那么此时就会选择离当前手机屏幕密度最近的端点进行适配，也就是(requestedDensity >= h)或者(l >= requestedDensity—)的情况。如果这样的区间存在，那么就执行 (((2 * l) - requestedDensity) * h > requestedDensity * requestedDensity) 的算法，这里requestedDensity就指的是需要适配手机屏幕密度。这里的l和h为其具体值（单位dpi，如240dpi）与系统默认密度160dpi的比值，也就是系统对dp转化为px的缩放比。

举个例子，就以开篇的那个问题为例，此时l = 0.75, h = 3.0,requestedDensity = 1.5, 那么可以计算得出((2 * l) - requestedDensity) * h < requestedDensity * requestedDensity)，返回true，也就是返回了h，也就是xxhdpi的dimen值

从哪里回去此篇代码呢？这个CPP是从GITHUB中直接clone过来的，如果读者需要，只要执行

$git clone https://github.com/android/platform_frameworks_base.git

即可此文件。