[FFmpeg-devel] [PATCH] ffprobe: implement generic reindent logic in the JSON writer

[Stefano Sabatini]

On date Sunday 2012-01-08 13:28:11 +0100, Clément Bœsch encoded:
> On Sun, Jan 08, 2012 at 11:26:03AM +0100, Stefano Sabatini wrote:
> > On date Sunday 2012-01-08 02:35:00 +0100, Clément Bœsch encoded:
> > > On Sun, Jan 08, 2012 at 02:27:20AM +0100, Stefano Sabatini wrote:
> > > > Clarify/generalize indent logic, possibly slower.
> > > > ---
> > > >  ffprobe.c |   54 ++++++++++++++++++++++++++++++++++++++----------------
> > > >  1 files changed, 38 insertions(+), 16 deletions(-)
> > [...]
> > > > +#define INDENT "    "
> > > > +
> > > > +#define JSON_INDENT() { int i; for (i = 0; i < json->indent_level; i++) printf(INDENT); }
> > > > +
> > > 
> > > I think I already suggested in a "very" old patch that this could be done
> > > with "%*c", json->indent_level * 4, ' '. It might be faster (and cleaner).
> > > 
> > > Are you against this?
> > 
> > No, I was not aware of the syntax and failed to find reference to it
> > the first time. Updated.
> 
> Before:
> 
> {
>   "streams": [{
>     "index": 0,
>     "codec_name": "mpeg1video",
>     ...
>   },{
>     "index": 1,
>     "codec_name": "mp2",
>     ...
>   }],
>   "format": {
>     "filename": "foobar.mpg",
>     "nb_streams": 2,
>     ...
>   }
> }
> 
> After:
> 

> {
>     "streams": [{
>             "index": 0,
>             "codec_name": "mpeg1video",
>             ...
>         },{
>             "index": 1,
>             "codec_name": "mp2",
>             ...
>         }],
>     "format": {
>         "filename": "foobar.mpg",
>         "nb_streams": 2,
>         ...
>     }
> }
> 
> 

> Is this the wanted behaviour?

Code is changing the indent level two times for the nested stream
section, so this is consistent with the implemented logic.

> I'm fine with the second one if it helps
> improving the code, but the change of behaviour should be notified in the
> commit message IMO.

Patch is supposed to be applied together with the next one, when the
output becomes:

{
    "streams": [
        {
            "index": 0,
            "codec_name": "mpeg1video",
            ...
        },
        {
            "index": 1,
            "codec_name": "mp2",
            ...
        }
    ],
    "format": {
        "filename": "foobar.mpg",
        "nb_streams": 2,
        ...
    }
}

which I find more readable (as it is more clear the relation between
the chapter and nested sections).
-- 
FFmpeg = Fabulous Fast Mysterious Plastic Ermetic Guru