[vlc-devel] DCP subtiltes

[Nicolas Bertrand]

Hi all, so far I didn't work on DCP stuff ...

So, I decide to take a closer look on DCP subtitle.

A long time ago we proposed a patch,
https://patches.videolan.org/patch/4324/ , but its probably better to
have a dedicated codec module for DCP subtitle.

DCP subtitle is XML based and defined in SMPTE docs (SMPTE-428-10-2008)

As its XML based, I looked in ttml subs modules (demux and codec).

So, if I understand correctly, its needed to parse XML in demux for
timestamps and also in codec for font styling stuff. I joined a sample.

In ttml demux the XML parsing is done, and XML tags '<p/>' are recreated
(ie new generation of XML tags) to be after parsed by the codec.
Do you think is there a way to just send the entire node to the stream
by doing a memcopy?
For instance at each XML_NODE_STARTELEM of <Subtitle> tag get the size
and memory location of the entire node ( and subnodes) and send it as
elementary stream to codec ? This can avoid to parsing twice all XML
nodes as it is done in ttml module?
But I didn't find any solution to get the node size and location during
XML parsing.
Any Ideas?

Or regenerating XML tags is the only solution?

Or is this not the proper way to parse XML based subtitles?

Thanks in advance

Nicolas (Buxiness in IRC)






-------------- next part --------------
A non-text attachment was scrubbed...
Name: FA_G_REV.xml
Type: text/xml
Size: 9000 bytes
Desc: not available
URL: <http://mailman.videolan.org/pipermail/vlc-devel/attachments/20160906/11cfd4d8/attachment.xml>