The EIA-608 decoder expects fonts to be rendered with a monospace font,
so add the ability for decoders to make use of the "tt" tag, and tweak
the EIA-608 decoder to use that tag.

This patch supercedes folds in patch 9 from the previous series as well
as incorporates the changes Rafaël Carré solicited from the ML regarding
the fonts to use for non-Linux platforms.
Modified-by: Rafaël Carré <funman@videolan.org>
 - Simplify HandleTT()
 - Fix memleak in Create()
Signed-off-by: Rafaël Carré <funman@videolan.org>

This can happen if the font name is NULL
Signed-off-by: Rafaël Carré <funman@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

The left and right arrow keys have not any real advantage / functionality for
the playlist view or sidebar but are used by default for several hotkeys.
So lets allow these as VLC hotkeys (also for audio files).

Signed-off-by: Rafaël Carré <funman@videolan.org>

I added the shortcut 'm' in nvlc for toggle mute. The volume status is
replaced my 'Mute' when enabled.

I also remapped the shortcut of metadata box from 'm' to 'M', I think
it follows the principle of least surprise.
Signed-off-by: Rafaël Carré <funman@videolan.org>

When an output picure is requested from a Samsung (exynos?) device
it is not guarateed to be aligned to 128 bit (4 pixels).
Signed-off-by: Rafaël Carré <funman@videolan.org>

We want to center the subpicture region, but the text itself needs
to be rendered left justified in order to get consistent placement
of captions (e.g. indentation).

Introduce a new alignment type which centers the subpicture, but
not the text it contains within (and enable it for the EIA-608 decoder).
Signed-off-by: Rafaël Carré <funman@videolan.org>

In order to get consistent EIA-608 caption positioning, we need to be able
to size the fonts based on the size of the video.  Add support for
specifying the font height as a percentage of the video.
Signed-off-by: Rafaël Carré <funman@videolan.org>

We need to add quotes to the color attribute for the font tag, or else
the XML parser will choke and the HTML will not be rendered.
Signed-off-by: Rafaël Carré <funman@videolan.org>

Make the buffer larger, since there are more HTML tags in the resulting output
stream.
Signed-off-by: Rafaël Carré <funman@videolan.org>

If the character in the EIA-608 stream is one of the reserved characters,
escape it accordingly so that the HTML renders properly.
Signed-off-by: Rafaël Carré <funman@videolan.org>

If there are any whitespace characters before the start of text in a row,
convert it to non-breaking whitespace. We do this because the RenderHtml
function will strip any cases of more than one contigious whitespace. The
effect of such is that text that is not left justified will be forced to
the left.
Signed-off-by: Rafaël Carré <funman@videolan.org>

Because the subpicture field contains all the lines, we need to change the
layout.
Signed-off-by: Rafaël Carré <funman@videolan.org>

We need to render all the rows, or else the placement of popup captions will
not be correct.

The previous code worked fine when rendering text horizontally aligned at the
bottom of the screen, but won't work for a real layout.
Signed-off-by: Rafaël Carré <funman@videolan.org>

We need to use the HTML renderer if we want colors/styles to work
Signed-off-by: Rafaël Carré <funman@videolan.org>

According to Qt, props must have a read method

And more :)

hack adapted from the qt interface.

close #7876

avcodec contains codecs that don't like variable size frames, so feed
avcodec with frame_size chunks and and handle leftovers ourself. This
functionality was dropped when flushing was introduced and seemed to
work without just fine, I was proven wrong.