December 2017 - MPlayer-dev-eng - lists.mplayerhq.hu

[PATCH] stream_ffmpeg: add -lavfstreamopts option.
by Reimar Döffinger 01 Nov '20

01 Nov '20

Allows specifying options to ffmpeg-based streams. --- DOCS/man/en/mplayer.1 | 4 ++++ cfg-common.h | 1 + stream/stream.h | 1 + stream/stream_ffmpeg.c | 19 ++++++++++++++++++- 4 files changed, 24 insertions(+), 1 deletion(-) diff --git a/DOCS/man/en/mplayer.1 b/DOCS/man/en/mplayer.1 index 2a143cb..d8e9b3b 100644 --- a/DOCS/man/en/mplayer.1 +++ b/DOCS/man/en/mplayer.1 @@ -1806,6 +1806,10 @@ Skip the proxy for IPv6 addresses. It will still be used for IPv4 connections. . .TP +.B \-lavfstreamopts <options> +Specify extra options for libavformat based streams. +. +.TP .B \-loadidx <index file> The file from which to read the video index data saved by \-saveidx. This index will be used for seeking, overriding any index data diff --git a/cfg-common.h b/cfg-common.h index 3a422cf..ca4e9c9 100644 --- a/cfg-common.h +++ b/cfg-common.h @@ -545,6 +545,7 @@ const m_option_t common_opts[] = { #ifdef CONFIG_FFMPEG {"lavdopts", lavc_decode_opts_conf, CONF_TYPE_SUBCONFIG, 0, 0, 0, NULL}, {"lavfdopts", lavfdopts_conf, CONF_TYPE_SUBCONFIG, CONF_GLOBAL, 0, 0, NULL}, + {"lavfstreamopts", &lavfstreamopts, CONF_TYPE_STRING, CONF_GLOBAL, 0, 0, NULL}, #endif #ifdef CONFIG_XVID4 {"xvidopts", xvid_dec_opts, CONF_TYPE_SUBCONFIG, 0, 0, 0, NULL}, diff --git a/stream/stream.h b/stream/stream.h index 615f579..388c353 100644 --- a/stream/stream.h +++ b/stream/stream.h @@ -392,6 +392,7 @@ extern char *dvd_device; extern const m_option_t dvbin_opts_conf[]; extern char *rtsp_destination; +extern char *lavfstreamopts; typedef struct { int id; // 0 - 31 mpeg; 128 - 159 ac3; 160 - 191 pcm diff --git a/stream/stream_ffmpeg.c b/stream/stream_ffmpeg.c index 57d45e4..1883af3 100644 --- a/stream/stream_ffmpeg.c +++ b/stream/stream_ffmpeg.c @@ -27,6 +27,8 @@ #include "av_helpers.h" #include "libmpdemux/demuxer.h" +char *lavfstreamopts; + static int fill_buffer(stream_t *s, char *buffer, int max_len) { int r = avio_read(s->priv, buffer, max_len); @@ -90,6 +92,7 @@ static int open_f(stream_t *stream, int mode, void *opts, int *file_format) { int flags = 0; const char *filename; + AVDictionary *avopts = NULL; AVIOContext *ctx = NULL; int res = STREAM_ERROR; int64_t size; @@ -125,9 +128,23 @@ static int open_f(stream_t *stream, int mode, void *opts, int *file_format) dummy = !strncmp(filename, "rtsp:", 5) || !strncmp(filename, "dummy:", 6); mp_msg(MSGT_OPEN, MSGL_V, "[ffmpeg] Opening %s\n", filename); - if (!dummy && avio_open(&ctx, filename, flags) < 0) + if (lavfstreamopts && av_dict_parse_string(&avopts, lavfstreamopts, "=", ",", 0) < 0) { + mp_msg(MSGT_HEADER,MSGL_ERR, "Your options /%s/ look like gibberish to me pal\n", lavfstreamopts); + goto out; + } + + if (!dummy && avio_open2(&ctx, filename, flags, NULL, &avopts) < 0) goto out; + if (!dummy && av_dict_count(avopts)) { + AVDictionaryEntry *e = NULL; + while ((e = av_dict_get(avopts, "", e, AV_DICT_IGNORE_SUFFIX))) { + mp_msg(MSGT_HEADER,MSGL_ERR,"Unknown option %s\n", e->key); + } + goto out; + } + av_dict_free(&avopts); + stream->priv = ctx; size = dummy ? 0 : avio_size(ctx); if (size >= 0) -- 2.7.0

1 1

[PATCH] aalib subtitle fix
by za3k＠za3k.com 15 Apr '19

15 Apr '19

For some time now, aalib rendering has had broken subtitles. -vo aa is designed to use human-readable text at the size of one character per character. In actuality, it's somehow switched to trying to render a standard font using dithering since its introduction. This patch fixes it. Explanation: vo_font doesn't seem to get used anywhere, because OSD rendering is actually done in a completely separate way from other video output, but I kept it in just in case. As a side note, someone may want to investigate unicode support for subtitles in -vo aa. *** /var/src/MPlayer-1.3.0/libvo/vo_aa.c 2016-02-12 12:23:39.000000000 -0800 --- vo_aa.c 2017-05-07 02:07:11.900548663 -0700 *************** static int osdx, osdy; *** 92,97 **** --- 92,98 ---- static int osd_text_length = 0; int aaconfigmode=1; font_desc_t* vo_font_save = NULL; + font_desc_t* sub_font_save = NULL; static struct SwsContext *sws=NULL; /* configuration */ *************** config(uint32_t width, uint32_t height, *** 249,256 **** /* now init our own 'font' */ if(!vo_font_save) vo_font_save = vo_font; ! if(vo_font == vo_font_save) { ! vo_font=malloc(sizeof(font_desc_t));//if(!desc) return NULL; memset(vo_font,0,sizeof(font_desc_t)); vo_font->pic_a[0]=malloc(sizeof(raw_file)); memset(vo_font->pic_a[0],0,sizeof(raw_file)); --- 250,258 ---- /* now init our own 'font' */ if(!vo_font_save) vo_font_save = vo_font; ! if(!sub_font_save) sub_font_save = vo_font; ! if(vo_font == vo_font_save || sub_font == sub_font_save) { ! vo_font=sub_font=malloc(sizeof(font_desc_t));//if(!desc) return NUL L; memset(vo_font,0,sizeof(font_desc_t)); vo_font->pic_a[0]=malloc(sizeof(raw_file)); memset(vo_font->pic_a[0],0,sizeof(raw_file));

3 4

mplayer 1.4.0 release
by Compn 15 Apr '18

15 Apr '18

i see development has dropped a bit... are we going to get a 1.4.0 release this year? -compn

3 3

[PATCH] MNG detection fails with libjpeg-turbo
by The Wanderer 16 Jan '18

16 Jan '18

With current SVN (and for as long as I can remember checking), the configure check for the availability of libmng fails. Today I finally got around to looking into why. The test performs a compile check against libmng.h, which includes jpeglib.h. With libjpeg-turbo (the source of jpeglib.h in current Debian), this test fails because FILE is undefined. This is typedef is normally expected to come from stdio.h. https://github.com/libjpeg-turbo/libjpeg-turbo/issues/17 indicates that this is intentional, for backwards compatibility with libjpeg, and that "it is actually a design requirement of the libjpeg API that you include stdio.h before including jpeglib.h.". This would seem to indicate that this has been broken for a very long time, and quite possibly no one noticed (which would imply that no one bothered trying to compile with MNG support). The last MNG-related commits seem to have been in 2012; support was originally added in 2011. The compilation check is performed by the return_statement_check function, which only permits passing one header-file argument. There is an existing statement_check_broken function, specifically for use with headers which don't include some of their own dependencies; however, it hardcodes a 'return 0', whereas return_statement_check permits passing in an expression to be used in the return statement. Rather than try to add an ' 0' to all callers of statement_check_broken, I've duplicated it into return_statement_check_broken, which adds the return-statement argument from return_statement_check. With that, detection of libmng succeeds. With an include of stdio.h added to vo_mng.c, compilation also succeeds. The attached patch makes both changes. -- The Wanderer The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man. -- George Bernard Shaw

2 11

Adapt config.h's generation with some ffmpeg's commits
by Etienne Buira 08 Jan '18

08 Jan '18

Adapt config.h's generation with some ffmpeg's commits ffmpeg's related commits: e0d56f097f42bcdbe6c3b2f57df62a4da63f2094 - CONFIG_LINUX_PERF 5de85c1029881970dc69e529dc48e277f3b5336c - HAVE_LOCAL_ALIGNED c708b5403346255ea5adc776645616cc7c61f078 - HAVE_MACH_MACH_TIME_H & HAVE_MACH_ABSOLUTE_TIME 8f86e6623811f7713d5e72c13797e20fffb3df62 - HAVE_SIMD_ALIGN_64 b7a3d1f249c9da9f21d10ea9137e304961da02ba - HAVE_AVX512 1f4c62e84e586c52282a58ff878022a79feac4d1 - HAVE_DIRENT_H 465becbc4f1fd6dc7973fb69a2d46e8c8580e719 - HAVE_ACCESS d65b9114f35c1afe2a7061f0a1ec957d33ba02b5 - HAVE_LSTAT 6e073dd127f09f203bd21d40a1eb423afd183e7a - HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC 534a2231f69638c3fdd998853ba4158ed79322a8 - HAVE_FCNTL Regards

4 8

communicating to mplayer via the dbus mediaplayer2 interface
by Enrico Guiraud 18 Dec '17

18 Dec '17

Hello all, I use mplayer daily as audio player -- thank you for your work! I also use other applications that would like to communicate to my audio player via dbus, specifically using the MediaPlayer2 <https://specifications.freedesktop.org/mpris-spec/latest/Media_Player.html> interface specification. So I'm about to set off to write a small python wrapper for mplayer that makes it compatible with this interface. My question is whether this is something that has already been done by someone somewhere or whether there are intrinsic limitations or fundamental reasons why I should not do this. Thanks in advance for any feedback you might provide! Enrico

2 1