Better support for metadata in mediafiles (chapters, title, subtitle)

[Frans]

Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'data/e/e2bj8RDjDoaSaZuoVDOczlDR':
  Metadata:
    major_brand     : M4VH
    minor_version   : 1
    compatible_brands: M4VHM4A mp42isom
    artist          : Medische Fysiologie
    album           : Medische Fysiologie
    title           : MF - College 4 en 5 - Inleiding zenuwstelsel
    composer        : Medische Fysiologie
    show            : Medische Fysiologie
  Duration: 01:04:37.96, start: 0.000000, bitrate: 286 kb/s
    Chapter #0.0: start 0.000000, end 50.000000
    Metadata:
      title           : Inledingx��
    Chapter #0.1: start 50.000000, end 367.320000
    Metadata:
      title           : Fysiologie�
    Chapter #0.2: start 367.320000, end 441.960000
    Metadata:
      title           : Onderwerpen
    Chapter #0.3: start 441.960000, end 760.120000
    Metadata:
      title           : CNS Cellenn
    Chapter #0.4: start 760.120000, end 1044.080000
    Metadata:
      title           : Zenuwcelenn
    Chapter #0.5: start 1044.080000, end 1467.000000
    Metadata:
      title           : Transport van informatie
    Chapter #0.6: start 1467.000000, end 1686.920000
    Metadata:
      title           : Klassificatie zenuwvezels
    Chapter #0.7: start 1686.920000, end 1865.320000
    Metadata:
      title           : Ladingsverdeling
    Chapter #0.8: start 1865.320000, end 1993.040000
    Metadata:
      title           : Distributie mobiele ionen
    Chapter #0.9: start 1993.040000, end 2363.440000
    Metadata:
      title           : Membraamevenwicht
    Chapter #0.10: start 2363.440000, end 2598.080000
    Metadata:
      title           : Nernstvergelijking
    Chapter #0.11: start 2598.080000, end 2861.160000
    Metadata:
      title           : Kanalenergelijking
    Chapter #0.12: start 2861.160000, end 2892.160000
    Metadata:
      title           : Potentialenlijking
    Chapter #0.13: start 2892.160000, end 3165.880000
    Metadata:
      title           : Variabele Potentialenonen
    Chapter #0.14: start 3165.880000, end 3315.440000
    Metadata:
      title           : Drukgevoelige receptorenn
    Chapter #0.15: start 3315.440000, end 3503.560000
    Metadata:
      title           : Actiepotentiaal
    Chapter #0.16: start 3503.560000, end 3542.240000
    Metadata:
      title           : Flowcurventiaal
    Chapter #0.17: start 3542.240000, end 3590.360000
    Metadata:
      title           : Verschillen tussen Na- en K-kanalen
    Chapter #0.18: start 3590.360000, end 3842.360000
    Metadata:
      title           : Myelinex��
    Chapter #0.19: start 3842.360000, end 3877.960000
    Metadata:
      title           : Samenvatting
    Stream #0.0(eng): Video: h264, yuv420p, 720x540, 155 kb/s, 25 fps, 25 tbr, 2500 tbn, 5k tbc
    Stream #0.1(eng): Audio: aac, 44100 Hz, stereo, s16, 126 kb/s
    Stream #0.2(eng): Subtitle: tx3g / 0x67337874
At least one output file must be specified

[Frans]

This is an example of a h264 mediafile with use of chapters, metadata and subtitles.
Would be interesting to use this info more than we do now.

[Frans]

Media metadata should be placed in db

[MC-arjen]

We can parse the output and:

- put in a mediafile_metadata field (to show and to (free-text-)search in )
- put in the asset metadata with options:

• append to existing metadata
• overwrite existing metadata
• only add where there is no metadata yet.

Question: what to do with non matchable metadata (for example: major_brand).
Question: what to do when we add a second original file to an asset (suggestion: ignore)

make this option an ega-setting, default: ignore (version 2 behavior)

[Michiel.Schok]

Agree to this being an optional EGA setting. I can imagine that for MediaMosa it will be on by default, for VP-Core it will be off. Plz turn it on for SURFmedia ;-)

Me thinks that by default:
* When the first original file is automatically analysed after upload, metadata should be appended to existing metadata (there most likely is none)
* When the restcall [post] /mediafile/{id}/analyse is called on an original file in an asset (be it the first, or any other one) the metadata could be replaced (default behaviour) or could possibly be appended to existing metadata.

These give sensible defaults for usecases involving just one original on an asset (those metadata are extracted by default, and replaced by a newer 'reanalyse' job which is probably better). In usecases with multiple originals, they are created intentionally, and can easily perform an /mediafile/{id}/analyse?metadata_append=true after upload.

I'm not sure about the 'non matchable metadata'. Have to see more examples of output to create an opinion.

[peter]

Didn't make it to definitive set of must have issues for MM 3.0.

[Frans]

Related to:
http://mediamosa.org/trac/ticket/532