Downloading encrypted M3U8 HLS videos

[white_snake]

Maybe someone who is an expert with N_m3u8 can give you the direct command that can be used to remove the first #EXTINF. For some reason, if you leave it, N_m3u8 fails because of a bad fragment (???).

--custom-hls-key "key.txt" --custom range 1- index.m3u8

Seems to work, but the resulting video and audio are not in sync (audio is about 1900 ms ahead), so it would require caution before muxing.

[2nHxWW6GkN1l916N3ayz8HQoi]

--custom-hls-key "key.txt" --custom range 1- index.m3u8

Seems to work, but the resulting video and audio are not in sync (audio is about 1900 ms ahead), so it would require caution before muxing.

Wow. That's a neat command and good to know. Thanks @snake But just to be safe, this command doesn't get rid of the first fragment, right? The first fragment isn't actually corrupted, it's just that N_m3u8 can't handle the m3u8 for some reason. All fragments are from the same ts file with different range bytes.

[white_snake]

Wow. That's a neat command and good to know. Thanks @snake But just to be safe, this command doesn't get rid of the first fragment, right? The first fragment isn't actually corrupted, it's just that N_m3u8 can't handle the m3u8 for some reason. All fragments are from the same ts file with different range bytes.

I'm afraid it does, lol. 😅 I hadn't realized you had removed the:

#EXTINF:12.012,

alone (to be perfectly honest I thought that would have corrupted the .m3u8!). At this point I believe your approach is the only feasible one here, since N_m3u8DL-RE can't still selectively ignore tags. I had once proposed a similar functionality, that would have been helpful to ignore discontinuity tags that sometimes are put there as traps, but nothing of the kind has been implemented yet. :/

[2nHxWW6GkN1l916N3ayz8HQoi]

to be perfectly honest I thought that would have corrupted the .m3u8!

When I executed N_m3u8 on the original file, it failed and after taking a look at the decrypted fragments ts, only the first one was broken, the others were ok. So I thought that if I removed the first #EXTINF maybe N_m3u8 could figure it out on its own because all these fragments were from the same ts file, so maybe a specific byte range was messed up (maybe on purpose). I don't know if this is the explanation for what is happening but in a normal case, it should have corrupted it indeed as you said.

N_m3u8DL-RE can't still selectively ignore tags. I had once proposed a similar functionality, that would have been helpful to ignore discontinuity tags that sometimes are put there as traps, but nothing of the kind has been implemented yet. :/

That feature would be kinda rarely used, but I agree, it would have been useful here. I guess you could fix this by writing a custom script but is kinda overkill. Also, I had no idea you could place tags as traps. Thanks for the info!

[white_snake]

When I executed N_m3u8 on the original file, it failed and after taking a look at the decrypted fragments ts, only the first one was broken, the others were ok. So I thought that if I removed the first #EXTINF maybe N_m3u8 could figure it out on its own because all these fragments were from the same ts file, so maybe a specific byte range was messed up (maybe on purpose). I don't know if this is the explanation for what is happening but in a normal case, it should have corrupted it indeed as you said.

Ah, got you.

That feature would be kinda rarely used, but I agree, it would have been useful here. I guess you could fix this by writing a custom script but is kinda overkill. Also, I had no idea you could place tags as traps. Thanks for the info!

Yup, NBA does it and it "breaks" N_m3u8DL-RE (in that it stops looking for more segments, and also gives a smaller segment count) unless you manually edit the stream playlists to remove those tags (yt-dlp doesn't fall into the trap though!).

[Jagomister]

When the #EXTINF is removed it downloads segment one correctly but skips 2 and moves number on all rest up by 1

with it in it fails 1 and does all the rest correctly

[2nHxWW6GkN1l916N3ayz8HQoi]

When the #EXTINF is removed it downloads segment one correctly but skips 2 and moves number on all rest up by 1

with it in it fails 1 and does all the rest correctly

I have no idea what you're talking about. I checked the duration of the final video and is the same as the original. If, however, my approach doesn't work for you, then maybe someone will come up with a better fix. The main idea is that there are no corrupted fragments, only that N_m3u8 can't handle them.

[stabbedbybrick]

N_m3u8DL-RE is a great tool, but it's not perfect. For AES-128 where the key is easily located with an URI, I personally prefer Streamlink. And it seems to do the job perfectly here. Just replace the path/to/key with the proper URI for each stream:

❯ streamlink --hls-segment-key-uri "https://path/to/key/" "https://d2o75c1wl3rtfi.cloudfront.net/Philsophy-MASTER-1/index.m3u8" best -o "Philosophy.mp4"
[cli][info] Found matching plugin hls for URL https://d2o75c1wl3rtfi.cloudfront.net/Philsophy-MASTER-1/index.m3u8
[cli][info] Available streams: 73k (worst), 224p_alt, 224p, 720p_alt, 720p (best)
[cli][info] Opening stream: 720p (hls-multi)
[cli][info] Writing output to
...\Philosophy.mp4
[utils.named_pipe][info] Creating pipe streamlinkpipe-15160-1-5915
[utils.named_pipe][info] Creating pipe streamlinkpipe-15160-2-8337
[download] Written 779.94 MiB to Philosophy.mp4 (36s @ 22.46 MiB/s)
[cli][info] Stream ended
[cli][info] Closing currently open stream...

No mux or sync issues.

[Jagomister]

When the #EXTINF is removed it downloads segment one correctly but skips 2 and moves number on all rest up by 1

with it in it fails 1 and does all the rest correctly

I have no idea what you're talking about. I checked the duration of the final video and is the same as the original. If, however, my approach doesn't work for you, then maybe someone will come up with a better fix. The main idea is that there are no corrupted fragments, only that N_m3u8 can't handle them.

If you do not remove the #EXTINF look in the working directory you will notice that seg 0 failed ad 1,2 ... where fine
If you remove the #EXTINF then seg 0 downloads fine however seg 1 is skipped so seg 2 is becomes 1.ts 3 become 2.ts .....
when the segments are combined seg 0 is doubled and and used in position 0 and 1 (first 24 secs is seg 0 on repeat)

[2nHxWW6GkN1l916N3ayz8HQoi]

N_m3u8DL-RE is a great tool, but it's not perfect. For AES-128 where the key is easily located with an URI, I personally prefer Streamlink. And it seems to do the job perfectly here.

Thanks a lot @brick for providing the better tool in this case. @Jago read @brick's comment. It's the proper solution.

[Jagomister]

Yer I am its working great. Did not like passing the key as a file path but easy to put them on web server and