automated generation of .csv file from sparse matrix

[jpiesing]

@FritzHeiden explained that the .csv file used to combine the test content and the HTML+JS templates is manually edited by him and that this has introduced errors in the past. He further explained that the sparse matrix is missing some information that would be needed to automate this process. For example paths to MPDs.

@nicholas-fr explained that his script to validate the encoded AVC test content has the expected properties reads the sparse matrix and works out the paths based on the agreed directory structure.

Next steps: @nicholas-fr to provide information about what / how his script works so @FritzHeiden and @louaybassbouss can review and see what would be needed for automating this step.

If something is missing, extra information can be added to the sparse matrix, potentially some kind of id or key that can be used as an index into database.json to find that information.

[yanj-github]

@FritzHeiden Is this to automate tests.csv file geration. But setting parameters for tests in test-config.json file, such as playout, is remain manual process?

[nicholas-fr]

We agreed a common folder/file structure:

<cmaf_brand>_sets/<sub_media_type (frame_rate_family|audio_codec)>/<stream_id>/<upload_date>/
            <stream_id>.zip
            stream.mpd
            <switching_set_number>/
                    init.mp4
                    <segment_start_time>.m4s

The root also contains the following for switching set tests after extraction from switching set streamid.zip (e.g. ss1.zip): Video: `_stream.mpd Audio:_stream.mpd`

The spreadsheet contains: cmaf_brand and stream_id for each test vector/stream. The following is fixed/predefined: 3 frame_rate_families, MPD and segment filenames. The path is contained in database.json for each test vector, including the upload_date e.g. cfhd_sets/12.5_25_50/t1/2022-10-17/.

So by using the sparse matrix and database.json, all info is available.