* To be able to read data from several source, and encode them in the
same output stream.
* This will be done in the next commits.

The buffer could be different.

* _sourceBuffers to _decodedData.
* _frameBuffers to _convertedData.

To have a buffer of filtered data.

* In this case, '_inputStreams.size() == 1' is always true!
* Anyway, we should check the '_currentInputStream' attribute because it
is used in the method.

* No pointers are needed here.
* Updated FilterGraph parameters to launch a process.

…d data

Need to define 'Frame::operator='.

Confusing when we will have several input streams.

The VideoFrameDesc could be found from the given frame when decoding.

The AudioFrameDesc could be found from the given frame when decoding.

* Add provate method 'addDecoder'.
* Will be used to add several decoders.

* Need pointers here to be able to use polymorphism.
* Updated FilterGraph parameters to launch a process.
* Almost a revert of commit ecd3ddd

The _currentDecoder could be one of the decoder in the _inputDecoders
list.

* Renamed private method 'getProfileFromFile' to 'getProfileFromInput'.
* Removed private method 'addTranscodeStream'.

…tream

refFrame method allocates new data that will be freed by calling clear
method or the destructor of the referenced frame.

Avoid switching if the current case is a generator.

After calling 'avcodec_decode_audio4', sometimes the channel layout of
the frame is set to 0. I could be a problem in the next steps of the
process.

2 inputs is the default value for this filter.

…he profile as a string

Added vector<InputStreamDesc> in the SWIG interface.

* Need to have a description of the given frame buffer when decoding,
because the buffer could be invalid (clear at the end of decoding of an
other stream). This attribute could be seen as a backup.
* Almost a revert of commits 935ed64 and 924bf5b.

…into dev_transcoderUseComplexFilterToMapAudio

Conflicts:
	test/pyTest/testSetFrame.py

The profile is found from the inputs.

This code was never reach, since the previous intructions already check
if the frame is a valid audio frame.

Better to check a more complicated case, when the encoding profile is
found from the inputs.

If channel_layout attribute is not indicated, the method returns the
default channel layout which corresponds to the given number of
channels.

To testAddAStreamFromAFileWhichDoesNotExist

…of inputs

The width and height should be added as string.

* AV_PIX_FMT_NONE = -1
* YUV420P = 0

…puts

Added a python test to check this behavior.

…into dev_transcoderUseComplexFilterToMapAudio