Merge branch 'cairocanvas' of git.ardour.org:ardour/ardour into cairocanvas
[ardour.git] / gtk2_ardour / transcode_ffmpeg.cc
index a38d9c45f094321d89898e5e43181ed61a452a01..9cbc40564cb4c30070ad9a1700e16d3da1b6042a 100644 (file)
@@ -17,8 +17,6 @@
     Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
 
 */
-#ifdef WITH_VIDEOTIMELINE
-
 #include <stdio.h>
 #include <string.h>
 #include <sstream>
@@ -42,27 +40,43 @@ TranscodeFfmpeg::TranscodeFfmpeg (std::string f)
        ffmpeg_exe = "";
        ffprobe_exe = "";
        m_duration = 0;
+       m_avoffset = m_lead_in = m_lead_out = 0;
+       m_width = m_height = 0;
+       m_aspect = m_fps = 0;
 #if 1 /* tentative debug mode */
        debug_enable = false;
 #endif
 
        std::string ff_file_path;
-       if (find_file_in_search_path (PBD::SearchPath(Glib::getenv("PATH")), X_("ffmpeg"), ff_file_path)) { ffmpeg_exe = ff_file_path; }
-       else if (find_file_in_search_path (PBD::SearchPath(Glib::getenv("PATH")), X_("avconv"), ff_file_path)) { ffmpeg_exe = ff_file_path; }
-       else if (find_file_in_search_path (PBD::SearchPath(std::string("/usr/local/bin/")), X_("ffmpeg_harvid"), ff_file_path)) { ffmpeg_exe = ff_file_path; }
+       if (find_file_in_search_path (PBD::SearchPath(Glib::getenv("PATH")), X_("ffmpeg_harvid"), ff_file_path)) { ffmpeg_exe = ff_file_path; }
+       else if (Glib::file_test(X_("C:\\Program Files\\harvid\\ffmpeg.exe"), Glib::FILE_TEST_EXISTS)) {
+               ffmpeg_exe = X_("C:\\Program Files\\ffmpeg\\ffmpeg.exe");
+       }
        else if (Glib::file_test(X_("C:\\Program Files\\ffmpeg\\ffmpeg.exe"), Glib::FILE_TEST_EXISTS)) {
                ffmpeg_exe = X_("C:\\Program Files\\ffmpeg\\ffmpeg.exe");
        }
 
-       if (find_file_in_search_path (PBD::SearchPath(Glib::getenv("PATH")), X_("ffprobe"), ff_file_path)) { ffprobe_exe = ff_file_path; }
-//else if (find_file_in_search_path (PBD::SearchPath(Glib::getenv("PATH")), X_("avprobe"), ff_file_path)) { ffprobe_exe = ff_file_path; }
-       else if (find_file_in_search_path (PBD::SearchPath(std::string("/usr/local/bin/")), X_("ffprobe_harvid"), ff_file_path)) { ffprobe_exe = ff_file_path; }
+       if (find_file_in_search_path (PBD::SearchPath(Glib::getenv("PATH")), X_("ffprobe_harvid"), ff_file_path)) { ffprobe_exe = ff_file_path; }
+       else if (Glib::file_test(X_("C:\\Program Files\\harvid\\ffprobe.exe"), Glib::FILE_TEST_EXISTS)) {
+               ffprobe_exe = X_("C:\\Program Files\\ffmpeg\\ffprobe.exe");
+       }
        else if (Glib::file_test(X_("C:\\Program Files\\ffmpeg\\ffprobe.exe"), Glib::FILE_TEST_EXISTS)) {
                ffprobe_exe = X_("C:\\Program Files\\ffmpeg\\ffprobe.exe");
        }
 
        if (ffmpeg_exe.empty() || ffprobe_exe.empty()) {
-               PBD::warning << _("No ffprobe or ffmpeg executables could be found on this system. Transcoding is not possible until you install those tools.") << endmsg;
+               PBD::warning << _(
+                               "No ffprobe or ffmpeg executables could be found on this system.\n"
+                               "Video import and export is not possible until you install those tools.\n"
+                               "Ardour requires ffmpeg and ffprobe from ffmpeg.org - version 1.1 or newer.\n"
+                               "\n"
+                               "The tools are included with the Ardour releases from ardour.org "
+                               "and also available with the video-server at http://x42.github.com/harvid/\n"
+                               "\n"
+                               "Important: the files need to be installed in $PATH and named ffmpeg_harvid and ffprobe_harvid.\n"
+                               "If you already have a suitable ffmpeg installation on your system, we recommend creating "
+                               "symbolic links from ffmpeg to ffmpeg_harvid and from ffprobe to ffprobe_harvid.\n"
+                               ) << endmsg;
                return;
        }
        ffexecok = true;
@@ -85,8 +99,8 @@ TranscodeFfmpeg::probe ()
        char **argp;
        argp=(char**) calloc(7,sizeof(char*));
        argp[0] = strdup(ffprobe_exe.c_str());
-       argp[1] = strdup("-print_format"); // "-of"  ; new version and avprobe compat but avprobe does not yet support csv
-       argp[2] = strdup("csv"); // TODO use "csv=nk=0" and parse key/value pairs -> ffprobe version agnostic or parse XML or JSON key/value
+       argp[1] = strdup("-print_format");
+       argp[2] = strdup("csv=nk=0");
        argp[3] = strdup("-show_format");
        argp[4] = strdup("-show_streams");
        argp[5] = strdup(infile.c_str());
@@ -104,119 +118,111 @@ TranscodeFfmpeg::probe ()
 
        std::vector<std::vector<std::string> > lines;
        ParseCSV(ffoutput, lines);
+       double timebase = 0;
        m_width = m_height = 0;
        m_fps = m_aspect = 0;
        m_duration = 0;
        m_codec.clear();
        m_audio.clear();
 
+#define PARSE_FRACTIONAL_FPS(VAR) \
+       { \
+               std::string::size_type pos; \
+               VAR = atof(value.c_str()); \
+               pos = value.find_first_of('/'); \
+               if (pos != std::string::npos) { \
+                       VAR = atof(value.substr(0, pos).c_str()) / atof(value.substr(pos+1).c_str()); \
+               } \
+       }
+
        for (std::vector<std::vector<std::string> >::iterator i = lines.begin(); i != lines.end(); ++i) {
                if (i->at(0) == X_("format")) {
                        /* format,filename,#streams,format-name,format-long-name,start-time,duration,size,bitrate */
                } else
                if (i->at(0) == X_("stream")) {
-                       /*--------- Stream format
-                        * stream,index,codec-name,codec-name-long,PROFILE,
-                        *   codec_time_base,codec_tag_string,codec_tag[hex],
-                        * VIDEO:
-                        *   width,height,has_b_frames,sample_aspect_ratio,display_aspect_ratio
-                        *   pix_fmt,level,
-                        *   timecode
-                        * AUDIO:
-                        *   sample_fmt,sample_rate,channels,bits_per_sample
-                        *
-                        * all cont'd;
-                        *   r_frame_rate,avg_frame_rate,time_base,start_time,duration,
-                        *   bit_rate,nb_frames,nb_read_frames,nb_read_packets
-                        *
-                        *---------- Example
-                        * stream,0,mpeg2video,MPEG-2 video,video,1/50,[0][0][0][0],0x0000,720,576,1,16:15,4:3,yuv420p,8,00:02:30:00,0x1e0,25/1,25/1,1/90000,0.360000,N/A,7000000,N/A,N/A,N/A
-                        * stream,1,ac3,ATSC A/52A (AC-3),audio,1/48000,[0][0][0][0],0x0000,s16,48000,6,0,-1,-1.000000,-1.000000,-1.000000,-1.000000,0x80,0/0,0/0,1/90000,0.280000,312.992000,448000,N/A,N/A,N/A
-                        * stream,2,ac3,ATSC A/52A (AC-3),audio,1/48000,[0][0][0][0],0x0000,s16,48000,2,0,-1,-1.000000,-1.000000,-1.000000,-1.000000,0x82,0/0,0/0,1/90000,0.280000,312.992000,384000,N/A,N/A,N/A
-                        * stream,3,ac3,ATSC A/52A (AC-3),audio,1/48000,[0][0][0][0],0x0000,s16,48000,2,0,-1,-1.000000,-1.000000,-1.000000,-1.000000,0x81,0/0,0/0,1/90000,0.280000,312.992000,192000,N/A,N/A,N/A
-                        */
-                       if (i->at(4) == X_("video") && m_width == 0) {
-                               std::string::size_type pos;
-
-                               m_width = atoi(i->at(8).c_str());
-                               m_height = atoi(i->at(9).c_str());
-                               m_codec = i->at(3) + " -- " + i->at(2);
-                               m_fps = atof(i->at(17).c_str());
-
-                               pos = i->at(17).find_first_of('/');
-                               if (pos != std::string::npos) {
-                                       m_fps = atof(i->at(17).substr(0, pos).c_str()) / atof(i->at(17).substr(pos+1).c_str());
+                       if (i->at(5) == X_("codec_type=video") && m_width == 0) {
+
+                               for (std::vector<std::string>::iterator kv = i->begin(); kv != i->end(); ++kv) {
+                                       const size_t kvsep = kv->find('=');
+                                       if(kvsep == std::string::npos) continue;
+                                       std::string key = kv->substr(0, kvsep);
+                                       std::string value = kv->substr(kvsep + 1);
+
+                                       if (key == X_("index")) {
+                                               m_videoidx = atoi(value.c_str());
+                                       } else if (key == X_("width")) {
+                                               m_width = atoi(value.c_str());
+                                       } else if (key == X_("height")) {
+                                               m_height = atoi(value.c_str());
+                                       } else if (key == X_("codec_name")) {
+                                               if (!m_codec.empty()) m_codec += " ";
+                                               m_codec += value;
+                                       } else if (key == X_("codec_long_name")) {
+                                               if (!m_codec.empty()) m_codec += " ";
+                                               m_codec += "[" + value + "]";
+                                       } else if (key == X_("codec_tag_string")) {
+                                               if (!m_codec.empty()) m_codec += " ";
+                                               m_codec += "(" + value + ")";
+                                       } else if (key == X_("r_frame_rate")) {
+                                               PARSE_FRACTIONAL_FPS(m_fps)
+                                       } else if (key == X_("avg_frame_rate") && m_fps == 0) {
+                                               PARSE_FRACTIONAL_FPS(m_fps)
+                                       } else if (key == X_("time_base")) {
+                                               PARSE_FRACTIONAL_FPS(timebase)
+                                       } else if (key == X_("timecode") && m_duration == 0) {
+                                               int h,m,s; char f[7];
+                                               if (sscanf(i->at(16).c_str(), "%d:%d:%d:%s",&h,&m,&s,f) == 4) {
+                                                       m_duration = (ARDOUR::framecnt_t) floor(m_fps * (
+                                                                       h * 3600.0
+                                                               + m * 60.0
+                                                               + s * 1.0
+                                                               + atoi(f) / pow(10, strlen(f))
+                                                       ));
+                                               }
+                                       } else if (key == X_("duration_ts") && m_fps == 0 && timebase !=0 ) {
+                                               m_duration = atof(value.c_str()) * m_fps * timebase;
+                                       } else if (key == X_("duration") && m_fps != 0 && m_duration == 0) {
+                                               m_duration = atof(value.c_str()) * m_fps;
+                                       } else if (key == X_("display_aspect_ratio")) {
+                                               std::string::size_type pos;
+                                               pos = value.find_first_of(':');
+                                               if (pos != std::string::npos && atof(value.substr(pos+1).c_str()) != 0) {
+                                                       m_aspect = atof(value.substr(0, pos).c_str()) / atof(value.substr(pos+1).c_str());
+                                               }
+                                       }
                                }
 
-                               pos = i->at(12).find_first_of(':');
-                               m_aspect = 0;
-                               if (pos != std::string::npos && atof(i->at(12).substr(pos+1).c_str()) != 0) {
-                                       m_aspect = atof(i->at(12).substr(0, pos).c_str()) / atof(i->at(12).substr(pos+1).c_str());
-                               }
                                if (m_aspect == 0) {
                                        m_aspect = (double)m_width / (double)m_height;
                                }
 
-                               int h,m,s; char f[7];
-                               if (sscanf(i->at(15).c_str(), "%d:%d:%d:%s",&h,&m,&s,f) == 4) {
-                                       m_duration = (ARDOUR::framecnt_t) floor(m_fps * (
-                                                       h * 3600.0
-                                               + m * 60.0
-                                               + s * 1.0
-                                               + atoi(f) / pow(10, strlen(f))
-                                       ));
-                               } else {
-                                       m_duration = atof(i->at(21).c_str()) * m_fps;
-                               }
-
-                       } else if (i->at(4) == X_("audio")) {
+                       } else if (i->at(5) == X_("codec_type=audio")) { /* new ffprobe */
                                AudioStream as;
-                               as.name = i->at(3) + " " + i->at(2) + " " + i->at(8) + " " + i->at(9);
-                               as.stream_id  = i->at(1);
-                               as.channels   = atoi(i->at(10).c_str());
-                               m_audio.push_back(as);
-
-                       } else if (i->at(5) == X_("video") && m_width == 0) { /* new ffprobe */
-                               std::string::size_type pos;
-
-                               m_width = atoi(i->at(9).c_str());
-                               m_height = atoi(i->at(10).c_str());
-                               m_codec = i->at(3) + " -- " + i->at(2);
-                               m_fps = atof(i->at(18).c_str());
-
-                               pos = i->at(18).find_first_of('/');
-                               if (pos != std::string::npos) {
-                                       m_fps = atof(i->at(18).substr(0, pos).c_str()) / atof(i->at(18).substr(pos+1).c_str());
-                               }
-
-                               pos = i->at(13).find_first_of(':');
-                               m_aspect = 0;
-                               if (pos != std::string::npos && atof(i->at(13).substr(pos+1).c_str()) != 0) {
-                                       m_aspect = atof(i->at(13).substr(0, pos).c_str()) / atof(i->at(13).substr(pos+1).c_str());
-                               }
-                               if (m_aspect == 0) {
-                                       m_aspect = (double)m_width / (double)m_height;
-                               }
+                               for (std::vector<std::string>::iterator kv = i->begin(); kv != i->end(); ++kv) {
+                                       const size_t kvsep = kv->find('=');
+                                       if(kvsep == std::string::npos) continue;
+                                       std::string key = kv->substr(0, kvsep);
+                                       std::string value = kv->substr(kvsep + 1);
+
+                                       if (key == X_("channels")) {
+                                               as.channels   = atoi(value.c_str());
+                                       } else if (key == X_("index")) {
+                                               as.stream_id  = value;
+                                       } else if (key == X_("codec_long_name")) {
+                                               if (!as.name.empty()) as.name += " ";
+                                               as.name += value;
+                                       } else if (key == X_("codec_name")) {
+                                               if (!as.name.empty()) as.name += " ";
+                                               as.name += value;
+                                       } else if (key == X_("sample_fmt")) {
+                                               if (!as.name.empty()) as.name += " ";
+                                               as.name += "FMT:" + value;
+                                       } else if (key == X_("sample_rate")) {
+                                               if (!as.name.empty()) as.name += " ";
+                                               as.name += "SR:" + value;
+                                       }
 
-                               int h,m,s; char f[7];
-                               if (sscanf(i->at(17).c_str(), "%d:%d:%d:%s",&h,&m,&s,f) == 4) {
-                                       m_duration = (ARDOUR::framecnt_t) floor(m_fps * (
-                                                       h * 3600.0
-                                               + m * 60.0
-                                               + s * 1.0
-                                               + atoi(f) / pow(10, strlen(f))
-                                       ));
-                               } else if (atof(i->at(23).c_str()) != 0) {
-                                       m_duration = atof(i->at(23).c_str());
-                               } else {
-                                       m_duration = atof(i->at(24).c_str()) * m_fps;
                                }
-
-                       } else if (i->at(5) == X_("audio")) { /* new ffprobe */
-                               AudioStream as;
-                               as.name = i->at(3) + " " + i->at(2) + " " + i->at(9) + " " + i->at(10);
-                               as.stream_id  = i->at(1);
-                               as.channels   = atoi(i->at(11).c_str());
                                m_audio.push_back(as);
                        }
                }
@@ -253,8 +259,8 @@ TranscodeFfmpeg::default_encoder_settings ()
        ffs.clear();
        ffs["-vcodec"] = "mpeg4";
        ffs["-acodec"] = "ac3";
-       ffs["-b"] = "5000k";
-       ffs["-ab"] = "160k";
+       ffs["-b:v"] = "5000k";
+       ffs["-b:a"] = "160k";
        return ffs;
 }
 
@@ -311,6 +317,7 @@ TranscodeFfmpeg::encode (std::string outfile, std::string inf_a, std::string inf
 
        argp[a++] = strdup("-i");
        argp[a++] = strdup(inf_a.c_str());
+
        for(FFSettings::const_iterator it = ffs.begin(); it != ffs.end(); ++it) {
                argp[a++] = strdup(it->first.c_str());
                argp[a++] = strdup(it->second.c_str());
@@ -319,12 +326,36 @@ TranscodeFfmpeg::encode (std::string outfile, std::string inf_a, std::string inf
                argp[a++] = strdup("-metadata");
                argp[a++] = format_metadata(it->first.c_str(), it->second.c_str());
        }
+       if (m_lead_in != 0 && m_lead_out != 0) {
+               std::ostringstream osstream;
+               argp[a++] = strdup("-vf");
+               osstream << X_("color=c=black:s=") << m_width << X_("x") << m_height << X_(":d=") << m_lead_in << X_(" [pre]; ");
+               osstream << X_("color=c=black:s=") << m_width << X_("x") << m_height << X_(":d=") << m_lead_out << X_(" [post]; ");
+               osstream << X_("[pre] [in] [post] concat=n=3");
+               argp[a++] = strdup(osstream.str().c_str());
+       } else if (m_lead_in != 0) {
+               std::ostringstream osstream;
+               argp[a++] = strdup("-vf");
+               osstream << X_("color=c=black:s=") << m_width << X_("x") << m_height << X_(":d=") << m_lead_in << X_(" [pre]; ");
+               osstream << X_("[pre] [in] concat=n=2");
+               argp[a++] = strdup(osstream.str().c_str());
+       } else if (m_lead_out != 0) {
+               std::ostringstream osstream;
+               argp[a++] = strdup("-vf");
+               osstream << X_("color=c=black:s=") << m_width << X_("x") << m_height << X_(":d=") << m_lead_out << X_(" [post]; ");
+               osstream << X_("[in] [post] concat=n=2");
+               argp[a++] = strdup(osstream.str().c_str());
+       }
+
        if (map) {
+               std::ostringstream osstream;
                argp[a++] = strdup("-map");
-               argp[a++] = strdup("0:0");
+               osstream << X_("0:") << m_videoidx;
+               argp[a++] = strdup(osstream.str().c_str());
                argp[a++] = strdup("-map");
                argp[a++] = strdup("1:0");
        }
+
        argp[a++] = strdup("-y");
        argp[a++] = strdup(outfile.c_str());
        argp[a] = (char *)0;
@@ -357,23 +388,26 @@ TranscodeFfmpeg::extract_audio (std::string outfile, ARDOUR::framecnt_t samplera
   if (stream >= m_audio.size()) return false;
 
        char **argp;
+       int i = 0;
 
        argp=(char**) calloc(15,sizeof(char*));
-       argp[0] = strdup(ffmpeg_exe.c_str());
-       argp[1] = strdup("-i");
-       argp[2] = strdup(infile.c_str());
-       argp[3] = strdup("-ar");
-       argp[4] = (char*) calloc(7,sizeof(char)); snprintf(argp[4], 7, "%"PRId64, samplerate);
-       argp[5] = strdup("-ac");
-       argp[6] = (char*) calloc(3,sizeof(char)); snprintf(argp[6], 3, "%i", m_audio.at(stream).channels);
-       argp[7] = strdup("-map");
-       argp[8] = (char*) calloc(8,sizeof(char)); snprintf(argp[8], 8, "0:%s", m_audio.at(stream).stream_id.c_str());
-       argp[9] = strdup("-vn");
-       argp[10] = strdup("-acodec");
-       argp[11] = strdup("pcm_f32le");
-       argp[12] = strdup("-y");
-       argp[13] = strdup(outfile.c_str());
-       argp[14] = (char *)0;
+       argp[i++] = strdup(ffmpeg_exe.c_str());
+       argp[i++] = strdup("-i");
+       argp[i++] = strdup(infile.c_str());
+#if 0 /* ffmpeg write original samplerate, use a3/SRC to resample */
+       argp[i++] = strdup("-ar");
+       argp[i] = (char*) calloc(7,sizeof(char)); snprintf(argp[i++], 7, "%"PRId64, samplerate);
+#endif
+       argp[i++] = strdup("-ac");
+       argp[i] = (char*) calloc(3,sizeof(char)); snprintf(argp[i++], 3, "%i", m_audio.at(stream).channels);
+       argp[i++] = strdup("-map");
+       argp[i] = (char*) calloc(8,sizeof(char)); snprintf(argp[i++], 8, "0:%s", m_audio.at(stream).stream_id.c_str());
+       argp[i++] = strdup("-vn");
+       argp[i++] = strdup("-acodec");
+       argp[i++] = strdup("pcm_f32le");
+       argp[i++] = strdup("-y");
+       argp[i++] = strdup(outfile.c_str());
+       argp[i++] = (char *)0;
        /* Note: argp is free()d in ~SystemExec */
 #if 1 /* DEBUG */
        if (debug_enable) { /* tentative debug mode */
@@ -422,13 +456,13 @@ TranscodeFfmpeg::transcode (std::string outfile, const int outw, const int outh,
        argp[0] = strdup(ffmpeg_exe.c_str());
        argp[1] = strdup("-i");
        argp[2] = strdup(infile.c_str());
-       argp[3] = strdup("-b");
+       argp[3] = strdup("-b:v");
        argp[4] = (char*) calloc(7,sizeof(char)); snprintf(argp[4], 7, "%i0k", bitrate);
        argp[5] = strdup("-s");
        argp[6] = (char*) calloc(10,sizeof(char)); snprintf(argp[6], 10, "%ix%i", width, height);
        argp[7] = strdup("-y");
        argp[8] = strdup("-vcodec");
-       argp[9] = strdup("mjpeg");
+       argp[9] = strdup("mpeg4");
        argp[10] = strdup("-an");
        argp[11] = strdup("-intra");
        argp[12] = strdup("-g");
@@ -484,10 +518,24 @@ void
 TranscodeFfmpeg::ffmpegparse_a (std::string d, size_t /* s */)
 {
        const char *t;
+       int h,m,s; char f[7];
+       ARDOUR::framecnt_t p = -1;
+
        if (!(t=strstr(d.c_str(), "time="))) { return; }
-       ARDOUR::framecnt_t f = (ARDOUR::framecnt_t) floorf (atof(t+5) * m_fps);
-       if (f > m_duration ) { f = m_duration; }
-       Progress(f, m_duration); /* EMIT SIGNAL */
+
+       if (sscanf(t+5, "%d:%d:%d.%s",&h,&m,&s,f) == 4) {
+               p = (ARDOUR::framecnt_t) floor( 100.0 * (
+                     h * 3600.0
+                   + m * 60.0
+                   + s * 1.0
+                   + atoi(f) / pow(10, strlen(f))
+               ));
+               p = p * m_fps / 100.0;
+               if (p > m_duration ) { p = m_duration; }
+               Progress(p, m_duration); /* EMIT SIGNAL */
+       } else {
+               Progress(0, 0); /* EMIT SIGNAL */
+       }
 }
 
 void
@@ -503,10 +551,13 @@ TranscodeFfmpeg::ffmpegparse_v (std::string d, size_t /* s */)
                  printf("ffmpeg: '%s'\n", d.c_str());
                }
 #endif
+               Progress(0, 0); /* EMIT SIGNAL */
                return;
        }
        ARDOUR::framecnt_t f = atol(d.substr(6).c_str());
-       Progress(f, m_duration); /* EMIT SIGNAL */
+       if (f == 0) {
+               Progress(0, 0); /* EMIT SIGNAL */
+       } else {
+               Progress(f, m_duration); /* EMIT SIGNAL */
+       }
 }
-
-#endif /* WITH_VIDEOTIMELINE */