[hls] New option --hls-segment-ignore-names #1432
Conversation
You can ignore segment names that might corrupt or lag your output - this will freeze the player until there is a valid segment name - this will not add the segment to an output file, so it won't be damaged and can be edited. Note: The `--hls-timeout` must be increased, to a time that is longer than the ignored break. Example: `streamlink URL best --hls-segment-ignore-names 000,001,002 --hls-timeout 1200` This will ignore every segment that ends with 000.ts, 001.ts and 002.ts
Codecov Report
@@ Coverage Diff @@
## master #1432 +/- ##
==========================================
+ Coverage 50.48% 50.48% +<.01%
==========================================
Files 104 104
Lines 7472 7485 +13
==========================================
+ Hits 3772 3779 +7
- Misses 3700 3706 +6 |
|
@back-to I got streamlink setup with the pull request but now I seem to have found a completely different issue now. [stream.hls][debug] Segments in this playlist are encrypted I don't remember getting that before at all. Though I am still playing with ffplay just fine. It seems like streamlink just keeps saying the encrypted message and the normal adding segments to queue and skipping segments. I don't seem to actually be getting any output ts file though. Though I tested really quick with out the new option and it did write a file that was playable, so it appears it's not "encrypted" in a sense like DRM or something that can't be recorded. |
|
the the there are maybe dead minutes in the output file, but you can just cut it. you need to run the command longer than the Pause break, i only tested tbd, the other channels might have different filenames |
|
Oh thanks for the info, going to try it again. I really wish they really had a better commercial break method. Watching a screen just saying "We'll Be Right Back" for 3 minutes straight with no music is even more boring than actually watching a normal commercial break. LoL |
|
So far it appears to work for TBD. Originally I had an outdated URL, that just gave me a commerical break the entire time. I noticed on the website I guess it supplements with ads. The only thing I noticed is, I still had to do a quick ffmpeg copy audio/video to fix the header, because even though the file looks great being just the program with no commercials, it oddly wouldn't rewind correctly in ffplay at first. Also VLC won't play the original ts until the header is fixed. |
|
VLC 3.0 handle TBD stream good enough. VLC 3.0 is at RC stage as of now (downloadable in nightly builds) (see also my comment just added on the linked issue) |
|
@bastimeyer, @beardypig Any concerns over this new option? |
|
Is it a work around for discontinuity issues? Seems generally useful though :) |
|
Is there any other feedback here from anyone? If not I want to get this merged since it's a good addition. |
| # this will be used to ignore segments. | ||
| self.ignore_names = list(set(self.ignore_names)) | ||
| self.ignore_names = "|".join(list(map(re.escape, self.ignore_names))) | ||
| self.ignore_names_re = re.compile(r"(?:{blacklist})\.ts".format( |
There was a problem hiding this comment.
This will ignore every segment that ends with ...
Just a minor suggestion: you should add the dollar anchor to the end of the regex, just to be sure that a segment doesn't get matched by accident if a URL somehow contains another .ts, you'll never know 😃
There was a problem hiding this comment.
ends with
i search for the end of the filename not for the uri end,
if there are params after 00.ts?foo=bar it's not the end of the filename
$ won't work for this
if i would parse the uri and just work with the path, then sure
but using urlparse might be to advanced.
There was a problem hiding this comment.
True... I had static segment URLs in mind without query data.
\.ts(?:\?.*)?$ should work, but we could also just leave it as it is, it's good enough. :)
|
|
||
| Default is None. | ||
|
|
||
| Note: The --hls-timeout must be increased, to a time that is longer than the ignored break. |
There was a problem hiding this comment.
I guess this can't be automated, right? Segment lengths are dependent on the streaming service (may even be variable) and there could also be multiple segments in a row which get removed.
There was a problem hiding this comment.
i could change timeout=self.timeout to a bigger value
if this command is used
but that's not really automated
|
@bastimeyer Any other feedback? If not I would like to get this merged. |
|
No, let's merge this... Further improvements, if needed, can be added in another PR |
[hls] New option --hls-segment-ignore-names
It has some downsides such as `--hls-timeout`, Plugin specified FilteredHLSStream should be used instead. streamlink#3187 --- Afreeca Plugin as an Example streamlink#3408 --- Revert streamlink#1432
You can ignore segment names that might corrupt or lag your output
so it won't be damaged and can be edited.
Note: The
--hls-timeoutmust be increased, to a time that is longerthan the ignored break.
Example:
streamlink URL best --hls-segment-ignore-names 000,001,002 --hls-timeout 1200This will ignore every segment that ends with 000.ts, 001.ts and 002.ts
some names/text can be changed, if someone has a better idea.
Hey @RedPenguin2
it would be nice if you could test this,
if you don't know how to, i can help you.
streamlink http://www.tbd.com best -l debug --hls-segment-ignore-names 0000,0001,0002 -o test.ts --hls-timeout 2400change the url with your hls link
this PR would fix #687