Android源代码分析(三) MediaScanner源码分析(下)
MediaScanner位置在frameworks\base\media\下，包括jni和java文件，
在Android的SDK里面是看不到这个类的，因为被google隐藏了。通过Android的源码我们可以看到MediaScanner的类注解多了一个@hide的标注。所以对于一般应用开发者，此文意义不是很大，大家可以绕道。在前两篇文章中，最后我们都了解了Android的媒体文件的扫描是在MediaScannerService中调用MediaScanner的scanDirectories或者scanSingleFile完成最终的扫描的。那么MediaScanner是如何工作的呢？
先来看看google对该类的注释：
1: /* In summary:
2: * Java MediaScannerService calls
3: * Java MediaScanner scanDirectories, which calls
4: * Java MediaScanner processDirectory (native method), which calls
5: * native MediaScanner processDirectory, which calls
6: * native MyMediaScannerClient scanFile, which calls
7: * Java MyMediaScannerClient scanFile, which calls
8: * Java MediaScannerClient doScanFile, which calls
9: * Java MediaScanner processFile (native method), which calls
10: * native MediaScanner processFile, which calls
11: * native parseMP3, parseMP4, parseMidi, parseOgg or parseWMA, which calls
12: * native MyMediaScanner handleStringTag, which calls
13: * Java MyMediaScanner handleStringTag.
14: * Once MediaScanner processFile returns, an entry is inserted in to the database.
15: *
16: * {@hide}
17: */
下面为调用时序图，如下：
先看看java实现。
这个类巨复杂，而且和MediaProvider交互频繁。在分析的时候要时刻回到MediaProvider去
看看。
1. 初始化
public class MediaScanner
{
static {
//libmedia_jni.so的加载是在MediaScanner类中完成的
//这么重要的so为何放在如此不起眼的地方加载？？？
System.loadLibrary("media_jni");
native_init();
}
public MediaScanner(Context c) {
native_setup();//调用jni层的初始化，暂时不用看了，无非就是一些
//初始化工作，待会在再进去看看
……..
}
刚才MSS中是调用scanDirectories函数，我们看看这个。
2. scanDirectories
public void scanDirectories(String[] directories, String volumeName) {
try {
long start = System.currentTimeMillis();
initialize(volumeName);//初始化
prescan(null);//扫描前的预处理
long prescan = System.currentTimeMillis();
for (int i = 0; i < directories.length; i++) {
//扫描文件夹，这里有一个很重要的参数 mClient
// processDirectory是一个native函数
processDirectory(directories[i], MediaFile.sFileExtensions, mClient);
}
long scan = System.currentTimeMillis();
postscan(directories);//扫描后处理
long end = System.currentTimeMillis();
…..打印时间，异常处理…没了…
下面简单讲讲initialize ，preScan和postScan都干嘛了。
private void initialize(String volumeName) {
//打开MediaProvider，获得它的一个实例
mMediaProvider = mContext.getContentResolver().acquireProvider("media");
//得到一些uri
mAudioUri = Audio.Media.getContentUri(volumeName);
mVideoUri = Video.Media.getContentUri(volumeName);
mImagesUri = Images.Media.getContentUri(volumeName);
mThumbsUri = Images.Thumbnails.getContentUri(volumeName);
//外部存储的话，可以支持播放列表之类的东西，搞了一些个缓存池之类的
//如mGenreCache等
if (!volumeName.equals("internal")) {
// we only support playlists on external media
mProcessPlaylists = true;
mGenreCache = new HashMap<String, Uri>();
…
preScan，这个函数很复杂：
大概就是创建一个FileCache，用来缓存扫描文件的一些信息，例如last_modified等。这个
FileCache是从MediaProvider中已有信息构建出来的，也就是历史信息。后面根据扫描得到的新
信息来对应更新历史信息。
postScan,这个函数做一些清除工作，例如以前有video生成了一些缩略图，现在video文件
被干掉了，则对应的缩略图也要被干掉。
另外还有一个mClient，这个是从MediaScannerClient派生下来的一个东西，里边保存了一
个文件的一些信息。后续再分析。
刚才说到，具体扫描工作是在processDirectory函数中完成的。这个是一个native函数。
在frameworks\base\media\jni\android_media_MediaScanner.cpp中。
3，mp->processDirectory(pathStr, extensionsStr, myClient, ExceptionCheck, env);
1: status_t MediaScanner::processDirectory(const char *path, const char* extensions,
2: MediaScannerClient& client, ExceptionCheck exceptionCheck, void* exceptionEnv)
3: {//这方法不知道干吗的，估计跟线程有关
4: InitializeForThread();
5:
6: int pathLength = strlen(path);
7: if (pathLength >= PATH_MAX) {
8: return PVMFFailure;
9: }
10: char* pathBuffer = (char *)malloc(PATH_MAX + 1);
11: if (!pathBuffer) {
12: return PVMFFailure;
13: }
14:
15: int pathRemaining = PATH_MAX - pathLength;
16: strcpy(pathBuffer, path);
17: if (pathBuffer[pathLength - 1] != '/') {
18: pathBuffer[pathLength] = '/';
19: pathBuffer[pathLength + 1] = 0;
20: --pathRemaining;
21: }
22:
23: client.setLocale(mLocale);
24: //有是一个关键点
25: status_t result = doProcessDirectory(pathBuffer, pathRemaining, extensions, client, exceptionCheck, exceptionEnv);
26: //释放内存
27: free(pathBuffer);
28: return result;
29: }
4，doProcessDirectory
1: status_t MediaScanner::doProcessDirectory(char *path, int pathRemaining, const char* extensions,
2: MediaScannerClient& client, ExceptionCheck exceptionCheck, void* exceptionEnv)
3: {
4: ……
5: ……
6: if (type == DT_REG || type == DT_DIR) {
7: int nameLength = strlen(name);
8: bool isDirectory = (type == DT_DIR);
9:
10: if (nameLength > pathRemaining || (isDirectory && nameLength + 1 > pathRemaining)) {
11: // path too long!
12: continue;
13: }
14:
15: strcpy(fileSpot, name);
16: if (isDirectory) {
17: // ignore directories with a name that starts with '.'
18: // for example, the Mac ".Trashes" directory
19: if (name[0] == '.') continue;
20:
21: strcat(fileSpot, "/");
22: //文件夹，递归调用
23: int err = doProcessDirectory(path, pathRemaining - nameLength - 1, extensions, client, exceptionCheck, exceptionEnv);
24: if (err) {
25: // pass exceptions up - ignore other errors
26: if (exceptionCheck && exceptionCheck(exceptionEnv)) goto failure;
27: LOGE("Error processing '%s' - skipping\n", path);
28: continue;
29: }
30: } else if (fileMatchesExtension(path, extensions)) {
31: //文件，扩展名符合
32: struct stat statbuf;
33: stat(path, &statbuf);
34: if (statbuf.st_size > 0) {
35: //调用client的scanFile方法
36: client.scanFile(path, statbuf.st_mtime, statbuf.st_size);
37: }
38: if (exceptionCheck && exceptionCheck(exceptionEnv)) goto failure;
39: }
40: }
41: ……
42: ……
1: // returns true if it succeeded, false if an exception occured in the Java code
2: virtual bool scanFile(const char* path, long long lastModified, long long fileSize)
3: {
4: jstring pathStr;
5: if ((pathStr = mEnv->NewStringUTF(path)) == NULL) return false;
6: //有点反射的感觉，调用java里面mClient中的scanFile方法
7: mEnv->CallVoidMethod(mClient, mScanFileMethodID, pathStr, lastModified, fileSize);
8:
9: mEnv->DeleteLocalRef(pathStr);
10: return (!mEnv->ExceptionCheck());
11: }
6，mEnv->CallVoidMethod(mClient, mScanFileMethodID, pathStr, lastModified, fileSize); 让我们回到Java
在android.media.MediaScanner.MyMediaScannerClient中的scanFile方法是直接调用doScanFile的，来看看doScanFile
1:
2: public Uri doScanFile(String path, String mimeType, long lastModified, long fileSize,
3: boolean scanAlways) {
4: Uri result = null;
5: // long t1 = System.currentTimeMillis();
6: try {
7: FileCacheEntry entry = beginFile(path, mimeType, lastModified, fileSize);
8: // rescan for metadata if file was modified since last scan
9: if (entry != null && (entry.mLastModifiedChanged || scanAlways)) {
10: String lowpath = path.toLowerCase();
11: boolean ringtones = (lowpath.indexOf(RINGTONES_DIR) > 0);
12: boolean notifications = (lowpath.indexOf(NOTIFICATIONS_DIR) > 0);
13: boolean alarms = (lowpath.indexOf(ALARMS_DIR) > 0);
14: boolean podcasts = (lowpath.indexOf(PODCAST_DIR) > 0);
15: boolean music = (lowpath.indexOf(MUSIC_DIR) > 0)
16: || (!ringtones && !notifications && !alarms && !podcasts);
17:
18: if (isMetadataSupported(mFileType)) {
19: // 调用jni方法
20: processFile(path, mimeType, this);
21: } else if (MediaFile.isImageFileType(mFileType)) {
22: // we used to compute the width and height but it's not
23: // worth it
24: }
25:
26: result = endFile(entry, ringtones, notifications, alarms, music, podcasts);
27: }
28: } catch (RemoteException e) {
29: Log.e(TAG, "RemoteException in MediaScanner.scanFile()", e);
30: }
31: // long t2 = System.currentTimeMillis();
32: // Log.v(TAG, "scanFile: " + path + " took " + (t2-t1));
33: return result;
34: }
补充：result = endFile(entry, ringtones, notifications, alarms, music, podcasts);就是在这里将媒体数据信息存放到数据库的
7，接着是native的 processFile
1: static void
2: android_media_MediaScanner_processFile(JNIEnv *env, jobject thiz, jstring path, jstring mimeType, jobject client)
3: {
4: MediaScanner *mp = (MediaScanner *)env->GetIntField(thiz, fields.context);
5:
6: if (path == NULL) {
7: jniThrowException(env, "java/lang/IllegalArgumentException", NULL);
8: return;
9: }
10:
11: const char *pathStr = env->GetStringUTFChars(path, NULL);
12: if (pathStr == NULL) { // Out of memory
13: jniThrowException(env, "java/lang/RuntimeException", "Out of memory");
14: return;
15: }
16: const char *mimeTypeStr = (mimeType ? env->GetStringUTFChars(mimeType, NULL) : NULL);
17: if (mimeType && mimeTypeStr == NULL) { // Out of memory
18: env->ReleaseStringUTFChars(path, pathStr);
19: jniThrowException(env, "java/lang/RuntimeException", "Out of memory");
20: return;
21: }
22:
23: MyMediaScannerClient myClient(env, client);
24: //调用MediaScanner的processFile
25: mp->processFile(pathStr, mimeTypeStr, myClient);
26: env->ReleaseStringUTFChars(path, pathStr);
27: if (mimeType) {
28: env->ReleaseStringUTFChars(mimeType, mimeTypeStr);
29: }
30: }
8，mp->processFile(pathStr, mimeTypeStr, myClient);
在此方法中根据不同的文件扩展名调用更加底层的解析方法，我想主要是ID3信息解析
1: status_t MediaScanner::processFile(const char *path, const char* mimeType, MediaScannerClient& client)
2: {
3: status_t result;
4: InitializeForThread();
5: //初始化client
6: client.setLocale(mLocale);
7: client.beginFile();
8:
9: //LOGD("processFile %s mimeType: %s\n", path, mimeType);
10: const char* extension = strrchr(path, '.');
11: //根据扩展名调用不同的解析方法
12: if (extension && strcasecmp(extension, ".mp3") == 0) {
13: result = parseMP3(path, client);
14: } else if (extension &&
15: (strcasecmp(extension, ".mp4") == 0 || strcasecmp(extension, ".m4a") == 0 ||
16: strcasecmp(extension, ".3gp") == 0 || strcasecmp(extension, ".3gpp") == 0 ||
17: strcasecmp(extension, ".3g2") == 0 || strcasecmp(extension, ".3gpp2") == 0)) {
18: result = parseMP4(path, client);
19: } else if (extension && strcasecmp(extension, ".ogg") == 0) {
20: result = parseOgg(path, client);
21: } else if (extension &&
22: ( strcasecmp(extension, ".mid") == 0 || strcasecmp(extension, ".smf") == 0
23: || strcasecmp(extension, ".imy") == 0)) {
24: result = parseMidi(path, client);
25: } else if (extension &&
26: (strcasecmp(extension, ".wma") == 0 || strcasecmp(extension, ".aac") == 0)) {
27: //TODO: parseWMA needs to be renamed to reflect what it is really doing,
28: //ie. using OpenCORE frame metadata utility(FMU) to retrieve metadata.
29: result = parseWMA(path, client);
30: } else {
31: result = PVMFFailure;
32: }
33: //调用client
34: client.endFile();
35:
36: return result;
37: }
9，client.endFile()
1: void MediaScannerClient::endFile()
2: {
3: if (mLocaleEncoding != kEncodingNone) {
4: int size = mNames->size();
5: uint32_t encoding = kEncodingAll;
6:
7: // compute a bit mask containing all possible encodings
8: for (int i = 0; i < mNames->size(); i++)
9: encoding &= possibleEncodings(mValues->getEntry(i));
10:
11: // if the locale encoding matches, then assume we have a native encoding.
12: if (encoding & mLocaleEncoding)
13: convertValues(mLocaleEncoding);
14:
15: // finally, push all name/value pairs to the client
16: for (int i = 0; i < mNames->size(); i++) {
17: //在handleStringTag中是通过类反射的方法调用java中的handleStringTag
18: if (!handleStringTag(mNames->getEntry(i), mValues->getEntry(i)))
19: break;
20: }
21: }
22: // else addStringTag() has done all the work so we have nothing to do
23:
24: delete mNames;
25: delete mValues;
26: mNames = NULL;
27: mValues = NULL;
28: }
10，java中的handleStringTag，这个方法主要处理那些在底层解析后的数据返回到java层
1: public void handleStringTag(String name, String value) {
2: if (name.equalsIgnoreCase("title") || name.startsWith("title;")) {
3: // Don't trim() here, to preserve the special \001 character
4: // used to force sorting. The media provider will trim() before
5: // inserting the title in to the database.
6: mTitle = value;
7: } else if (name.equalsIgnoreCase("artist") || name.startsWith("artist;")) {
8: mArtist = value.trim();
9: } else if (name.equalsIgnoreCase("albumartist") || name.startsWith("albumartist;")) {
10: mAlbumArtist = value.trim();
11: } else if (name.equalsIgnoreCase("album") || name.startsWith("album;")) {
12: mAlbum = value.trim();
13: } else if (name.equalsIgnoreCase("composer") || name.startsWith("composer;")) {
14: mComposer = value.trim();
15: } else if (name.equalsIgnoreCase("genre") || name.startsWith("genre;")) {
16: // handle numeric genres, which PV sometimes encodes like "(20)"
17: if (value.length() > 0) {
18: int genreCode = -1;
19: char ch = value.charAt(0);
20: if (ch == '(') {
21: genreCode = parseSubstring(value, 1, -1);
22: } else if (ch >= '0' && ch <= '9') {
23: genreCode = parseSubstring(value, 0, -1);
24: }
25: if (genreCode >= 0 && genreCode < ID3_GENRES.length) {
26: value = ID3_GENRES[genreCode];
27: }
28: }
29: mGenre = value;
30: } else if (name.equalsIgnoreCase("year") || name.startsWith("year;")) {
31: mYear = parseSubstring(value, 0, 0);
32: } else if (name.equalsIgnoreCase("tracknumber") || name.startsWith("tracknumber;")) {
33: // track number might be of the form "2/12"
34: // we just read the number before the slash
35: int num = parseSubstring(value, 0, 0);
36: mTrack = (mTrack / 1000) * 1000 + num;
37: } else if (name.equalsIgnoreCase("discnumber") ||
38: name.equals("set") || name.startsWith("set;")) {
39: // set number might be of the form "1/3"
40: // we just read the number before the slash
41: int num = parseSubstring(value, 0, 0);
42: mTrack = (num * 1000) + (mTrack % 1000);
43: } else if (name.equalsIgnoreCase("duration")) {
44: mDuration = parseSubstring(value, 0, 0);
45: } else if (name.equalsIgnoreCase("writer") || name.startsWith("writer;")) {
46: mWriter = value.trim();
47: }
48: }