sub: remove vobsub reader in favor of ffmpeg vobsub demuxer

ffmpeg recently added a demuxer that can read vobsubs (pairs of .sub and
.idx files). Get rid of the internal vobsub reader, and use the ffmpeg
demuxer instead.

Sneak in an unrelated manpage change (autosub default).

1 files changed, 27 insertions, 1 deletions

diff --git a/sub/spudec.c b/sub/spudec.c
index 22f121e92c..59c3058251 100644
--- a/sub/spudec.c
+++ b/sub/spudec.c
@@ -43,7 +43,6 @@
 #include "core/mp_msg.h"
 
 #include "spudec.h"
-#include "vobsub.h"
 #include "sub.h"
 #include "core/mp_common.h"
 #include "video/csputils.h"
@@ -659,6 +658,33 @@ void spudec_get_indexed(void *this, struct mp_osd_res *dim,
     }
 }
 
+static unsigned int vobsub_palette_to_yuv(unsigned int pal)
+{
+    int r, g, b, y, u, v;
+    // Palette in idx file is not rgb value, it was calculated by wrong formula.
+    // Here's reversed formula of the one used to generate palette in idx file.
+    r = pal >> 16 & 0xff;
+    g = pal >> 8 & 0xff;
+    b = pal & 0xff;
+    y = av_clip_uint8( 0.1494  * r + 0.6061 * g + 0.2445 * b);
+    u = av_clip_uint8( 0.6066  * r - 0.4322 * g - 0.1744 * b + 128);
+    v = av_clip_uint8(-0.08435 * r - 0.3422 * g + 0.4266 * b + 128);
+    y = y * 219 / 255 + 16;
+    return y << 16 | u << 8 | v;
+}
+
+static unsigned int vobsub_rgb_to_yuv(unsigned int rgb)
+{
+    int r, g, b, y, u, v;
+    r = rgb >> 16 & 0xff;
+    g = rgb >> 8 & 0xff;
+    b = rgb & 0xff;
+    y = ( 0.299   * r + 0.587   * g + 0.114   * b) * 219 / 255 + 16.5;
+    u = (-0.16874 * r - 0.33126 * g + 0.5     * b) * 224 / 255 + 128.5;
+    v = ( 0.5     * r - 0.41869 * g - 0.08131 * b) * 224 / 255 + 128.5;
+    return y << 16 | u << 8 | v;
+}
+
 static void spudec_parse_extradata(spudec_handle_t *this,
                                    uint8_t *extradata, int extradata_len)
 {