CCExtractor says no closed captions found in a EIA TS file #322

@cfsmp3 commented on GitHub (Dec 6, 2018):

@anshul1912 I think this was your code, can you take a look?

@cfsmp3 commented on GitHub (Dec 6, 2018): @anshul1912 I think this was your code, can you take a look?

claunia commented

@navimakarov commented on GitHub (Dec 15, 2018):

@cfsmp3 please note that we use get_best_stream function only in the place I described above. So if @anshul1912 verify this changes we can simple delete this function from ccextractor code completely. By the way, my PR is ready and I'm waiting for changes verification.

@navimakarov commented on GitHub (Dec 15, 2018): @cfsmp3 please note that we use get_best_stream function only in the place I described above. So if @anshul1912 verify this changes we can simple delete this function from ccextractor code completely. By the way, my PR is ready and I'm waiting for changes verification.

claunia commented

@cfsmp3 commented on GitHub (Dec 15, 2018):

@MakarovGCI2018 if we deleted that function then most streams wouldn't work. A requirement of FIXES is not to break more than your fix :-)

@cfsmp3 commented on GitHub (Dec 15, 2018): @MakarovGCI2018 if we deleted that function then most streams wouldn't work. A requirement of FIXES is not to break more than your fix :-)

claunia commented

@navimakarov commented on GitHub (Dec 15, 2018):

@cfsmp3 Sorry, but I don't really understand what the problem is. I've just deleted this function and tested it on all samples I had and it works great. As I understand, we have this function just in one place but ccextractor works ok without it. We don't call it from any other places so we can delete it. Am I wrong?

@navimakarov commented on GitHub (Dec 15, 2018): @cfsmp3 Sorry, but I don't really understand what the problem is. I've just deleted this function and tested it on all samples I had and it works great. As I understand, we have this function just in one place but ccextractor works ok without it. We don't call it from any other places so we can delete it. Am I wrong?

claunia commented

@anshul1912 commented on GitHub (Dec 15, 2018):

Let me explain that function.
Get_best_stream keeps the priority of subtitle when multiple subtitles are present, so removing it will work for you but then you mess with priority, you will always select whichever is on top. So I would say no to remove that function, I would suggest you to find why that function is not working. If some subtitle are wrongly reported then we always have option to pass from cmd line which one is needed for selection

int get_best_stream(struct ccx_demuxer *ctx)

{

struct cap_info* iter;

list_for_each_entry(iter, &ctx->cinfo_tree.all_stream, all_stream, struct cap_info)

{

	if(iter->codec == CCX_CODEC_TELETEXT)

		return iter->pid;

}

list_for_each_entry(iter, &ctx->cinfo_tree.all_stream, all_stream, struct cap_info)

{

	if(iter->codec == CCX_CODEC_DVB)

		return iter->pid;

}

list_for_each_entry(iter, &ctx->cinfo_tree.all_stream, all_stream, struct cap_info)

{

	if(iter->codec == CCX_CODEC_ISDB_CC)

		return iter->pid;

}

list_for_each_entry(iter, &ctx->cinfo_tree.all_stream, all_stream, struct cap_info)

{

	if(iter->codec == CCX_CODEC_ATSC_CC)

		return iter->pid;

}

return -1;

}

@anshul1912 commented on GitHub (Dec 15, 2018): Let me explain that function. Get_best_stream keeps the priority of subtitle when multiple subtitles are present, so removing it will work for you but then you mess with priority, you will always select whichever is on top. So I would say no to remove that function, I would suggest you to find why that function is not working. If some subtitle are wrongly reported then we always have option to pass from cmd line which one is needed for selection int get_best_stream(struct ccx_demuxer *ctx) { struct cap_info* iter; list_for_each_entry(iter, &ctx->cinfo_tree.all_stream, all_stream, struct cap_info) { if(iter->codec == CCX_CODEC_TELETEXT) return iter->pid; } list_for_each_entry(iter, &ctx->cinfo_tree.all_stream, all_stream, struct cap_info) { if(iter->codec == CCX_CODEC_DVB) return iter->pid; } list_for_each_entry(iter, &ctx->cinfo_tree.all_stream, all_stream, struct cap_info) { if(iter->codec == CCX_CODEC_ISDB_CC) return iter->pid; } list_for_each_entry(iter, &ctx->cinfo_tree.all_stream, all_stream, struct cap_info) { if(iter->codec == CCX_CODEC_ATSC_CC) return iter->pid; } return -1; }

claunia commented

@navimakarov commented on GitHub (Dec 15, 2018):

@anshul1912 thank you for the explanation. I will try to find out why this function doesn't work correct :-)

@navimakarov commented on GitHub (Dec 15, 2018): @anshul1912 thank you for the explanation. I will try to find out why this function doesn't work correct :-)

claunia commented

@navimakarov commented on GitHub (Dec 23, 2018):

There is an interesting thing in this file. They claim that they have some type of subtitle but actually they dont have it. And the function get_best_stream choose this pid as the one with highest priority. In this particular case we have a "fake" pid 7714 and pid, which contains captions with 5154 number.
That's why we have difference between pid and datalist->stream_pid and no captions in input as a result.
But actually, ccextractor doesn't need any changes. In those cases user has to know about this possible problem and use -datapid option.
https://github.com/CCExtractor/ccextractor/blob/master/src/lib_ccx/params.c#L508
So @rboy1 those parameters allow you to get correct output:
[input_file] -datapid 5154 -o [output_file]

@navimakarov commented on GitHub (Dec 23, 2018): There is an interesting thing in this file. They claim that they have some type of subtitle but actually they dont have it. And the function get_best_stream choose this pid as the one with highest priority. In this particular case we have a "fake" pid 7714 and pid, which contains captions with 5154 number. That's why we have difference between pid and datalist->stream_pid and no captions in input as a result. But actually, ccextractor doesn't need any changes. In those cases user has to know about this possible problem and use -datapid option. https://github.com/CCExtractor/ccextractor/blob/master/src/lib_ccx/params.c#L508 So @rboy1 those parameters allow you to get correct output: [input_file] -datapid 5154 -o [output_file]

claunia commented

@rboy1 commented on GitHub (Dec 23, 2018):

Thanks. As a user how do I know which PID? Isn’t there a way for ccextractor to pick it up automatically like you said if the preferred doesn’t exist but it found another one? In this case skip the fake and take the one indicated since it can read it?

@rboy1 commented on GitHub (Dec 23, 2018): Thanks. As a user how do I know which PID? Isn’t there a way for ccextractor to pick it up automatically like you said if the preferred doesn’t exist but it found another one? In this case skip the fake and take the one indicated since it can read it?

claunia commented

@navimakarov commented on GitHub (Dec 24, 2018):

@rboy as for now we don't support skipping fake pid. By the way, adding this function will increase get_best_stream method time .
You can get best stream pid with external programs(Take a look on mediaInfo where right pid is ID/String, VLC detected streams where pid is Original ID or ffprobe tool). Also you can get info about possible pids using -parsePMT parameter in ccextractor.

@navimakarov commented on GitHub (Dec 24, 2018): @rboy as for now we don't support skipping fake pid. By the way, adding this function will increase get_best_stream method time . You can get best stream pid with external programs(Take a look on mediaInfo where right pid is ID/String, VLC detected streams where pid is Original ID or ffprobe tool). Also you can get info about possible pids using -parsePMT parameter in ccextractor.

claunia commented

@xvkdev commented on GitHub (Dec 24, 2018):

Parsepmt lists all pids.
A possible solution could be to delay selecting the stream until data has
been found.

On Mon, Dec 24, 2018, 09:35 MakarovGCI2018 <notifications@github.com wrote:

@rboy https://github.com/rboy as for now we don't support skipping fake
pid. By the way, adding this function will increase get_best_stream method
time .
You can get best stream pid with external programs(Take a look on
mediaInfo where right pid is ID/String, VLC detected streams where pid is
Original ID or ffprobe tool). Also you can get info about possible pids
using -parsePMT parameter in ccextractor.

—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
https://github.com/CCExtractor/ccextractor/issues/757#issuecomment-449703015,
or mute the thread
https://github.com/notifications/unsubscribe-auth/ABF9zb1kQULuRdAjiOSg-gKb9eFnPOvFks5u8JHPgaJpZM4Ob9Hu
.

@xvkdev commented on GitHub (Dec 24, 2018): Parsepmt lists all pids. A possible solution could be to delay selecting the stream until data has been found. On Mon, Dec 24, 2018, 09:35 MakarovGCI2018 <notifications@github.com wrote: > @rboy <https://github.com/rboy> as for now we don't support skipping fake > pid. By the way, adding this function will increase get_best_stream method > time . > You can get best stream pid with external programs(Take a look on > mediaInfo where right pid is ID/String, VLC detected streams where pid is > Original ID or ffprobe tool). Also you can get info about possible pids > using -parsePMT parameter in ccextractor. > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub > <https://github.com/CCExtractor/ccextractor/issues/757#issuecomment-449703015>, > or mute the thread > <https://github.com/notifications/unsubscribe-auth/ABF9zb1kQULuRdAjiOSg-gKb9eFnPOvFks5u8JHPgaJpZM4Ob9Hu> > . >

claunia commented

@rboy1 commented on GitHub (Dec 24, 2018):

Would it be appropriate to automatically detect the fake ID and then search for the actual ID? It may take longer but that’s better than bypassing it since it’s much more difficult and time consuming to manually search for the actual PID and then restart it. In fact most folks won’t even realize it and will just assume that there no embedded captions since there is error either. Perhaps a cli option to automatically search for fake ID and extract/use the real ID may be make for smoother operation if the searching for the fake ID taking too long is a going to add a significant amount of time (would this be in minutes of seconds?)?

@rboy1 commented on GitHub (Dec 24, 2018): Would it be appropriate to automatically detect the fake ID and then search for the actual ID? It may take longer but that’s better than bypassing it since it’s much more difficult and time consuming to manually search for the actual PID and then restart it. In fact most folks won’t even realize it and will just assume that there no embedded captions since there is error either. Perhaps a cli option to automatically search for fake ID and extract/use the real ID may be make for smoother operation if the searching for the fake ID taking too long is a going to add a significant amount of time (would this be in minutes of seconds?)?

claunia commented