前言
我們前面兩篇分析了CarAudioZone相關的聲音以及音頻焦點，基本控制流就差不多了，今天繼續看下關于CarAudioZone相關的資料流。
正文
資料流這塊與CarAudioZone的關系是什麼呢，資料流底層是一個bus的概念，那麼什麼是bus，是谷歌專為car弄得一套devices（這裡的device概念是framework層的），即AUDIO_DEVICE_OUT_BUS。它和普通的device有什麼差別呢，我們知道device的加載時通過audiopolicy來初始化的，在AudioPolicyService啟動後會通過AudioPolicyManager來load audio_policy_configuration.xml，而在audio_policy_configuration.xml裡定義了framework層所有的device。普通的device的type一定是不同的比如AUDIO_DEVICE_OUT_EARPIECE或者AUDIO_DEVICE_OUT_WIRED_HEADSET等，而使用bus的方式所有的device的type必須都是AUDIO_DEVICE_OUT_BUS，而且他們的address一定不為空，并且daddress的命名規則必須是busx_xxx或者bus00x_xxx，其中bus後面跟着的x必須是int，也就是這樣bus1_xxx或者bus001_xxx.那麼為什麼這麼命名呢，就進入了今天的正題。
CarAudioService的啟動過程這裡就不多說了，啟動後加載了init，在init中通過AudioManager的getDevices(AudioManager.GET_DEVICES_OUTPUTS)擷取了所有的用于輸出的device，然後把這些device中type是BUS的過濾出來，是以說想使用bus的這套邏輯首先type必須是AUDIO_DEVICE_OUT_BUS。拿到這些type是AUDIO_DEVICE_OUT_BUS的device後重新建構了CarAudioDeviceInfo。然後便進入了setupDynamicRouting，從這個方法名稱也可以看出設定動态路由，在setupDynamicRouting中我們之前的兩篇分析了Volume和AudioFocus部分。但其中夾雜了兩行代碼
// Setup dynamic routing rules by usage
final CarAudioDynamicRouting dynamicRouting = new CarAudioDynamicRouting(mCarAudioZones);
dynamicRouting.setupAudioDynamicRouting(builder);
我們進入CarAudioDynamicRouting看看
CarAudioDynamicRouting(CarAudioZone[] carAudioZones) {
mCarAudioZones = carAudioZones;
}
構造方法不說了，繼續看下dynamicRouting.setupAudioDynamicRouting(builder)
void setupAudioDynamicRouting(AudioPolicy.Builder builder) {
for (CarAudioZone zone : mCarAudioZones) {
for (CarVolumeGroup group : zone.getVolumeGroups()) {
setupAudioDynamicRoutingForGroup(group, builder);
}
}
}
兩個for循環，外層是mCarAudioZones即我們傳入的mCarAudioZones的循環，我們繼續看内層循環，我們還記得在分析Android10.0CarAudioZone（一）的時候說過CarVolumeGroup，每個CarAudioZone中包含多個CarVolumeGroup，這裡拿出每一個CarVolumeGroup，和builder（audiopolicy的bulider）傳遞給setupAudioDynamicRoutingForGroup（個人吐槽下覺得這個寫的還有優化的空間）
private void setupAudioDynamicRoutingForGroup(CarVolumeGroup group,
AudioPolicy.Builder builder) {
// Note that one can not register audio mix for same bus more than once.
for (int busNumber : group.getBusNumbers()) {
boolean hasContext = false;
CarAudioDeviceInfo info = group.getCarAudioDeviceInfoForBus(busNumber);
AudioFormat mixFormat = new AudioFormat.Builder()
.setSampleRate(info.getSampleRate())
.setEncoding(info.getEncodingFormat())
.setChannelMask(info.getChannelCount())
.build();
AudioMixingRule.Builder mixingRuleBuilder = new AudioMixingRule.Builder();
for (int contextNumber : group.getContextsForBus(busNumber)) {
hasContext = true;
int[] usages = getUsagesForContext(contextNumber);
for (int usage : usages) {
mixingRuleBuilder.addRule(
new AudioAttributes.Builder().setUsage(usage).build(),
AudioMixingRule.RULE_MATCH_ATTRIBUTE_USAGE);
}
Log.d(CarLog.TAG_AUDIO, "Bus number: " + busNumber
+ " contextNumber: " + contextNumber
+ " sampleRate: " + info.getSampleRate()
+ " channels: " + info.getChannelCount()
+ " usages: " + Arrays.toString(usages));
}
if (hasContext) {
// It's a valid case that an audio output bus is defined in
// audio_policy_configuration and no context is assigned to it.
// In such case, do not build a policy mix with zero rules.
AudioMix audioMix = new AudioMix.Builder(mixingRuleBuilder.build())
.setFormat(mixFormat)
.setDevice(info.getAudioDeviceInfo())
.setRouteFlags(AudioMix.ROUTE_FLAG_RENDER)
.build();
builder.addMix(audioMix);
}
}
}
又是各種for循環，我們一點一點分析，先看最外層的for，其中group.getBusNumbers()是什麼呢？
int[] getBusNumbers() {
final int[] busNumbers = new int[mBusToCarAudioDeviceInfo.size()];
for (int i = 0; i < busNumbers.length; i++) {
busNumbers[i] = mBusToCarAudioDeviceInfo.keyAt(i);
}
return busNumbers;
}
mBusToCarAudioDeviceInfo就是我們在CarVolumeGroup的bind的時候建構的
mContextToBus.put(contextNumber, busNumber);
mBusToCarAudioDeviceInfo.put(busNumber, info);
這裡不說了，想看的可參照Android10.0CarAudioZone（一），拿到busNumber後就可以得到CarAudioDeviceInfo，拿到info後又建構了一個AudioFormat。AudioFormat就不說了，接下來就是建立AudioMixingRule，然後便進入contextNumbers的一個循環，而contextNumbers是從mContextToBus根據bus取出的contextNUmbers
int[] getContextsForBus(int busNumber) {
List<Integer> contextNumbers = new ArrayList<>();
for (int i = 0; i < mContextToBus.size(); i++) {
int value = mContextToBus.valueAt(i);
if (value == busNumber) {
contextNumbers.add(mContextToBus.keyAt(i));
}
}
//list轉成了int數組
return contextNumbers.stream().mapToInt(i -> i).toArray();
}
拿到了每個device下的contextNumbers數組後，在根據每個contextNumber取了一個usage的數組，即getUsagesForContext(contextNumber)
private int[] getUsagesForContext(int contextNumber) {
final List<Integer> usages = new ArrayList<>();
for (int i = 0; i < CarAudioDynamicRouting.USAGE_TO_CONTEXT.size(); i++) {
if (CarAudioDynamicRouting.USAGE_TO_CONTEXT.valueAt(i) == contextNumber) {
usages.add(CarAudioDynamicRouting.USAGE_TO_CONTEXT.keyAt(i));
}
}
return usages.stream().mapToInt(i -> i).toArray();
}
我們看下USAGE_TO_CONTEXT中usage和contextNumber的對應關系如下：
static {
USAGE_TO_CONTEXT.put(AudioAttributes.USAGE_UNKNOWN, ContextNumber.MUSIC);
USAGE_TO_CONTEXT.put(AudioAttributes.USAGE_MEDIA, ContextNumber.MUSIC);
USAGE_TO_CONTEXT.put(AudioAttributes.USAGE_VOICE_COMMUNICATION, ContextNumber.CALL);
USAGE_TO_CONTEXT.put(AudioAttributes.USAGE_VOICE_COMMUNICATION_SIGNALLING,
ContextNumber.CALL);
USAGE_TO_CONTEXT.put(AudioAttributes.USAGE_ALARM, ContextNumber.ALARM);
USAGE_TO_CONTEXT.put(AudioAttributes.USAGE_NOTIFICATION, ContextNumber.NOTIFICATION);
USAGE_TO_CONTEXT.put(AudioAttributes.USAGE_NOTIFICATION_RINGTONE, ContextNumber.CALL_RING);
USAGE_TO_CONTEXT.put(AudioAttributes.USAGE_NOTIFICATION_COMMUNICATION_REQUEST,
ContextNumber.NOTIFICATION);
USAGE_TO_CONTEXT.put(AudioAttributes.USAGE_NOTIFICATION_COMMUNICATION_INSTANT,
ContextNumber.NOTIFICATION);
USAGE_TO_CONTEXT.put(AudioAttributes.USAGE_NOTIFICATION_COMMUNICATION_DELAYED,
ContextNumber.NOTIFICATION);
USAGE_TO_CONTEXT.put(AudioAttributes.USAGE_NOTIFICATION_EVENT, ContextNumber.NOTIFICATION);
USAGE_TO_CONTEXT.put(AudioAttributes.USAGE_ASSISTANCE_ACCESSIBILITY,
ContextNumber.VOICE_COMMAND);
USAGE_TO_CONTEXT.put(AudioAttributes.USAGE_ASSISTANCE_NAVIGATION_GUIDANCE,
ContextNumber.NAVIGATION);
USAGE_TO_CONTEXT.put(AudioAttributes.USAGE_ASSISTANCE_SONIFICATION,
ContextNumber.SYSTEM_SOUND);
USAGE_TO_CONTEXT.put(AudioAttributes.USAGE_GAME, ContextNumber.MUSIC);
USAGE_TO_CONTEXT.put(AudioAttributes.USAGE_VIRTUAL_SOURCE, ContextNumber.INVALID);
USAGE_TO_CONTEXT.put(AudioAttributes.USAGE_ASSISTANT, ContextNumber.VOICE_COMMAND);
}
拿到usage後便 mixingRuleBuilder.addRule
public Builder addRule(AudioAttributes attrToMatch, int rule)
throws IllegalArgumentException {
if (!isValidAttributesSystemApiRule(rule)) {
throw new IllegalArgumentException("Illegal rule value " + rule);
}
return checkAddRuleObjInternal(rule, attrToMatch);
}
先看下這個isValidAttributesSystemApiRule，我們傳入的是 AudioMixingRule.RULE_MATCH_ATTRIBUTE_USAGE
private static boolean isValidAttributesSystemApiRule(int rule) {
// API rules only expose the RULE_MATCH_* rules
switch (rule) {
case RULE_MATCH_ATTRIBUTE_USAGE:
case RULE_MATCH_ATTRIBUTE_CAPTURE_PRESET:
return true;
default:
return false;
}
}
，即return checkAddRuleObjInternal。我們繼續看下checkAddRuleObjInternal
private Builder checkAddRuleObjInternal(int rule, Object property)
throws IllegalArgumentException {
if (property == null) {
throw new IllegalArgumentException("Illegal null argument for mixing rule");
}
if (!isValidRule(rule)) {
throw new IllegalArgumentException("Illegal rule value " + rule);
}
final int match_rule = rule & ~RULE_EXCLUSION_MASK;
if (isAudioAttributeRule(match_rule)) {
if (!(property instanceof AudioAttributes)) {
throw new IllegalArgumentException("Invalid AudioAttributes argument");
}
return addRuleInternal((AudioAttributes) property, null, rule);
} else {
// implies integer match rule
if (!(property instanceof Integer)) {
throw new IllegalArgumentException("Invalid Integer argument");
}
return addRuleInternal(null, (Integer) property, rule);
}
}
又是一個isValidRule的判斷，這裡傳回時true，
private static boolean isValidRule(int rule) {
final int match_rule = rule & ~RULE_EXCLUSION_MASK;
switch (match_rule) {
case RULE_MATCH_ATTRIBUTE_USAGE:
case RULE_MATCH_ATTRIBUTE_CAPTURE_PRESET:
case RULE_MATCH_UID:
return true;
default:
return false;
}
}
繼續又做了一個isAudioAttributeRule判斷
private static boolean isAudioAttributeRule(int match_rule) {
switch(match_rule) {
case RULE_MATCH_ATTRIBUTE_USAGE:
case RULE_MATCH_ATTRIBUTE_CAPTURE_PRESET:
return true;
default:
return false;
}
}
還是return true，進入 addRuleInternal((AudioAttributes) property, null, rule);這個方法
private Builder addRuleInternal(AudioAttributes attrToMatch, Integer intProp, int rule)
throws IllegalArgumentException {
// as rules are added to the Builder, we verify they are consistent with the type
// of mix being built. When adding the first rule, the mix type is MIX_TYPE_INVALID.
//mTargetMixType 預設是MIX_TYPE_INVALID
if (mTargetMixType == AudioMix.MIX_TYPE_INVALID) {
//又判斷這個，此處為true
if (isPlayerRule(rule)) {
mTargetMixType = AudioMix.MIX_TYPE_PLAYERS;
} else {
mTargetMixType = AudioMix.MIX_TYPE_RECORDERS;
}
} else if (((mTargetMixType == AudioMix.MIX_TYPE_PLAYERS) && !isPlayerRule(rule))
|| ((mTargetMixType == AudioMix.MIX_TYPE_RECORDERS) && isPlayerRule(rule)))
{
throw new IllegalArgumentException("Incompatible rule for mix");
}
synchronized (mCriteria) {
Iterator<AudioMixMatchCriterion> crIterator = mCriteria.iterator();
final int match_rule = rule & ~RULE_EXCLUSION_MASK;
//第一調用crIterator 的size是0，是以不走while
while (crIterator.hasNext()) {
final AudioMixMatchCriterion criterion = crIterator.next();
if ((criterion.mRule & ~RULE_EXCLUSION_MASK) != match_rule) {
continue; // The two rules are not of the same type
}
switch (match_rule) {
case RULE_MATCH_ATTRIBUTE_USAGE:
// "usage"-based rule
if (criterion.mAttr.getUsage() == attrToMatch.getUsage()) {
if (criterion.mRule == rule) {
// rule already exists, we're done
return this;
} else {
// criterion already exists with a another rule,
// it is incompatible
throw new IllegalArgumentException("Contradictory rule exists"
+ " for " + attrToMatch);
}
}
break;
case RULE_MATCH_ATTRIBUTE_CAPTURE_PRESET:
// "capture preset"-base rule
if (criterion.mAttr.getCapturePreset() == attrToMatch.getCapturePreset()) {
if (criterion.mRule == rule) {
// rule already exists, we're done
return this;
} else {
// criterion already exists with a another rule,
// it is incompatible
throw new IllegalArgumentException("Contradictory rule exists"
+ " for " + attrToMatch);
}
}
break;
case RULE_MATCH_UID:
// "usage"-based rule
if (criterion.mIntProp == intProp.intValue()) {
if (criterion.mRule == rule) {
// rule already exists, we're done
return this;
} else {
// criterion already exists with a another rule,
// it is incompatible
throw new IllegalArgumentException("Contradictory rule exists"
+ " for UID " + intProp);
}
}
break;
}
}
// rule didn't exist, add it
switch (match_rule) {
case RULE_MATCH_ATTRIBUTE_USAGE:
case RULE_MATCH_ATTRIBUTE_CAPTURE_PRESET:
//将我們傳入的usage組成的audioAttribute和rule存入mCriteria中。
mCriteria.add(new AudioMixMatchCriterion(attrToMatch, rule));
break;
case RULE_MATCH_UID:
mCriteria.add(new AudioMixMatchCriterion(intProp, rule));
break;
default:
throw new IllegalStateException("Unreachable code in addRuleInternal()");
}
}
return this;
}
到此 addRule的過程就結束了，我們簡單總結一下**，首先通過傳入的mCarAudioZones周遊其中的每個CarAudioZone，每個CarAudioZone又包含一個CarVolumeGroup集合，周遊CarVolumeGroup裡的每個group，每個group又包含一個devices的集合（CarAudioDeviceInfo的map，這個map的key是busNumber），每個device又包含一個context的集合（contextNumber和busNumber組成的map），說的有點亂，但我們一定要把這些關系捋清楚。這樣層層循環後我們拿到了contextNumber後通過map拿到usage數組，歸根揭底就是把Audio Attribute的usage數組和context以及CarAudioDeviceInfo都關聯上**。最終的關聯我們又回到CarAudioDynamicRouting的 setupAudioDynamicRoutingForGroup中
if (hasContext) {
// It's a valid case that an audio output bus is defined in
// audio_policy_configuration and no context is assigned to it.
// In such case, do not build a policy mix with zero rules.
AudioMix audioMix = new AudioMix.Builder(mixingRuleBuilder.build())
.setFormat(mixFormat)
.setDevice(info.getAudioDeviceInfo())
.setRouteFlags(AudioMix.ROUTE_FLAG_RENDER)
.build();
builder.addMix(audioMix);
}
AudioMix 包含了有usage數組的mixingRuleBuilder和AudioDeviceInfo，這樣device與usage數組便對應到了一起，我們知道原生的Android是通過stream在AudioPolicyManager的Engine中選擇的device，而在Car的這套邏輯中通過usage和device在上層（java）就配置好了。這裡的builder就是audioPolicy，addMix是簡單的将audioMix傳遞到了audiopolicy中
public Builder addMix(@NonNull AudioMix mix) throws IllegalArgumentException {
if (mix == null) {
throw new IllegalArgumentException("Illegal null AudioMix argument");
}
mMixes.add(mix);
return this;
}
mMixes即
ArrayList<AudioMix> mMixes
我們繼續看下，就又回到了CarAudioService的setupDynamicRouting中
開始了audiopolicy的bulid
public AudioPolicy build() {
if (mStatusListener != null) {
// the AudioPolicy status listener includes updates on each mix activity state
for (AudioMix mix : mMixes) {
mix.mCallbackFlags |= AudioMix.CALLBACK_FLAG_NOTIFY_ACTIVITY;
}
}
if (mIsFocusPolicy && mFocusListener == null) {
throw new IllegalStateException("Cannot be a focus policy without "
+ "an AudioPolicyFocusListener");
}
return new AudioPolicy(new AudioPolicyConfig(mMixes), mContext, mLooper,
mFocusListener, mStatusListener, mIsFocusPolicy, mIsTestFocusPolicy,
mVolCb, mProjection);
}
}
建立了一個AudioPolicyConfig然後用來建構AudioPolicy，這裡也是簡單看下AudioPolicyConfig的構造函數吧
AudioPolicyConfig(ArrayList<AudioMix> mixes) {
mMixes = mixes;
}
簡單過了，再看下AudioPolicy的構造函數
private AudioPolicy(AudioPolicyConfig config, Context context, Looper looper,
AudioPolicyFocusListener fl, AudioPolicyStatusListener sl,
boolean isFocusPolicy, boolean isTestFocusPolicy,
AudioPolicyVolumeCallback vc, @Nullable MediaProjection projection) {
mConfig = config;
mStatus = POLICY_STATUS_UNREGISTERED;
mContext = context;
if (looper == null) {
looper = Looper.getMainLooper();
}
if (looper != null) {
mEventHandler = new EventHandler(this, looper);
} else {
mEventHandler = null;
Log.e(TAG, "No event handler due to looper without a thread");
}
mFocusListener = fl;
mStatusListener = sl;
mIsFocusPolicy = isFocusPolicy;
mIsTestFocusPolicy = isTestFocusPolicy;
mVolCb = vc;
mProjection = projection;
}
也是一些指派，具體看下，**其中mConfig是我們剛才new的，mFocusListener 是mFocusHandler即我們之前new的 CarZonesAudioFocus，mStatusListener這裡為null，mIsFocusPolicy我們之前設定為了true，mIsTestFocusPolicy是false，mVolCb 即mAudioPolicyVolumeCallback，mProjection這裡也為null。**到這裡我們AudioPolicy也build結束了。剩下來的就是
将AudioPolicy注冊下去了，我們下篇在講
總結
在CarAudioService的初始化過程中，通過car_audio_configuration.xml做了volume 、audiofocus以及device的多分區的配置，如何讓我們的應用對應到不同的分區上，主要是通過uid處理的（這塊邏輯後面說）。今天我們主要分析的是devic的動态路由，通過周遊mCarAudioZones中的每個CarAudioZone，然後再周遊每個CarAudioZone中的CarVolumeGroup集合，周遊CarVolumeGroup裡的每個group，再周遊每個group中的devices的集合（CarAudioDeviceInfo的map，這個map的key是busNumber），最後周遊device中的一個context的集合（contextNumber和busNumber組成的map），把所有conrextNumber對應的usage全部拿出來放到usage數組中。最後把這個usage數組以及他對應的外層的deviceInfo一起存入AudioMix中，再被add到AudioPolicy中，這塊歸根揭底就是把Audio Attribute的usage數組和CarAudioDeviceInfo關聯上。這樣我們最終通過mAudioManager.registerAudioPolicy(mAudioPolicy)注冊下去。
如有問題，歡迎大家随時溝通~
![Java String.format方法的簡單使用[圖]](data:image/gif;base64,R0lGODlhAQABAIAAAP///wAAACwAAAAAAQABAAACAkQBADs=)