Perform slow task on lots of data in RecyclerView

Question

I am building a music player that gets the BPM ID3v2 tag from all mp3 files on a device and allows sorting and filtering by BPM. To get the tag, I am using the mp3agic library. It turns out that this tag fetching takes some time for each song, and performing the fetch when adding each song to my list of songs renders the player without function for a while. I am not sure if I should use an AsyncTask on the entire list when it is done building or an AsyncTask for each item to retrieve its BPM and insert it into the song or even using some other, faster method of getting the BPM value. Can anyone offer any guidance?

Here is my code to get the BPM:

public int getBpmFromId(long id) {
    int bpm = -1;
    Uri uri = Uri.withAppendedPath(MediaStore.Audio.Media.EXTERNAL_CONTENT_URI, Long.toString(id));
    try {
        Mp3File file = new Mp3File(getRealPathFromURI(getApplicationContext(), uri));
        if(file.hasId3v2Tag()) {
            ID3v2 id3v2Tag = file.getId3v2Tag();
            bpm = id3v2Tag.getBPM();
            Log.d("MP3AGIC", "Got BPM for track: " + id + ": " + bpm);

        }
    } catch (Exception e) {
        e.printStackTrace();
    }

    return bpm;
}

and here is the code I am using to build my list of songs:

public void getSongList() {
    ContentResolver musicResolver = getContentResolver();
    Uri musicUri = MediaStore.Audio.Media.EXTERNAL_CONTENT_URI;

    String selectionMimeType = MediaStore.Files.FileColumns.MIME_TYPE + "=?";
    String selection = MediaStore.Audio.Media.IS_MUSIC + " != 0 AND " + selectionMimeType;
    String sortOrder = null;
    String[] projection = null;
    String mimeType = MimeTypeMap.getSingleton().getMimeTypeFromExtension("mp3");
    String[] selectionArgsMp3 = new String[] { mimeType };


    Cursor musicCursor = musicResolver.query(musicUri, projection, selection, selectionArgsMp3, sortOrder);

    if(musicCursor != null && musicCursor.moveToFirst()) {
        int idColumn = musicCursor.getColumnIndex(MediaStore.Audio.Media._ID);
        int titleColumn = musicCursor.getColumnIndex(MediaStore.Audio.Media.TITLE);
        int artistColumn = musicCursor.getColumnIndex(MediaStore.Audio.Media.ARTIST);

        do {
            long id = musicCursor.getLong(idColumn);
            String title = musicCursor.getString(titleColumn);
            String artist = musicCursor.getString(artistColumn);

            //bpm processing
            int bpm = getBpmFromId(id);

            songList.add(new Song(id, title, artist));
        } while (musicCursor.moveToNext());
    }
    if(musicCursor != null)
        musicCursor.close();
}

EDIT: It seems that what might be taking so long is getting the file path from the Uri, which is done in this method:

    public String getRealPathFromURI(Context context, Uri contentUri) {
    Cursor cursor = null;
    try {
        String[] proj = { MediaStore.Images.Media.DATA };
        cursor = context.getContentResolver().query(contentUri,  proj, null, null, null);
        int column_index = cursor.getColumnIndexOrThrow(MediaStore.Images.Media.DATA);
        cursor.moveToFirst();
        return cursor.getString(column_index);
    } finally {
        if (cursor != null) {
            cursor.close();
        }
    }
}

I'm not sure how, or if I can, optimize this.

Answer 1

The decision to use one asynctask for the entire list or use the list as a queue for many asynctasks depends on your desire for concurrency.

• If you plan to parallelize the tag-getting, then use multiple asynctasks (this would be good if the operation involves a lot of disk IO rather than CPU time).
• Otherwise if you plan to execute them in sequence, then you should use only one asynctask. There is no need to create, then destroy tens/hundreds of background threads. This is inefficient.