[ie/common] Support ranges in MPD

[kinolaev]

Description of your pull request and other information

Support for Initialization@range and SegmentURL@mediaRange attributes in MPD files. When these attributes are present, yt-dlp should download only the specified portion of the associated media file.

Related PR #5661. segment_urls were renamed to segments as suggested there.

Template

Before submitting a pull request make sure you have:

• At least skimmed through contributing guidelines including yt-dlp coding conventions
• Searched the bugtracker for similar pull requests
• Checked the code with flake8 and ran relevant tests

In order to be accepted and merged into yt-dlp each piece of code must be in public domain or released under Unlicense. Check all of the following options that apply:

• I am the original author of this code and I am willing to release it under Unlicense
• I am not the original author of this code but it is in public domain or released under Unlicense (provide reliable evidence)

What is the purpose of your pull request?

• Fix or improvement to an extractor (Make sure to add/update tests)
• New extractor (Piracy websites will not be accepted)
• Core bug fix/improvement
• New feature (It is strongly recommended to open an issue first)

[iFUCKINGHATEcomputers]

This MPD issue affects Vbox7 and needs to be fixed urgently - on February 22nd Vbox7 will be privating nearly all their videos. Archivists and regular users need an easy way to back up videos/channels, whether individually or en masse.

https://www.reddit.com/r/Archiveteam/comments/19d6ou7/video_platform_vbox7_is_about_to_restrict_access/

[dirkf]

The PR parses the byte range in order to fit what is expected by the fragment downloader (where the 1 added to the range end is subtracted). However, the DASH spec (ISO 23009) specifically says that the range values are strings that conform to the HTTP specification for use in the Range header, so the processing here is not strictly necessary. Instead, the range could be passed into the HTTP header in the DASH downloader, as in ytdl-org/youtube-dl#30279.

On a style note, splitted isn't a normal English formation. The past participle of split is also split, and so is the past tense: "the range was split in two", "a colon never split the start and end of the range",

[kinolaev]

Hello @dirkf! Thank you for your feedback!

Although it's possible to pass a range string directly to http headers, I decided to parse it not only because the fragment downloader doesn't support range strings. I found this in some MPD files:

<SegmentList timescale='1000' duration='4000'>
  <Initialization sourceURL='https://example.com/2160p.mp4' range='36-800'/>
  <SegmentURL media='https://example.com/2160p.mp4' mediaRange='801-379516'/>
  <SegmentURL media='https://example.com/2160p.mp4' mediaRange='379517-656206'/>
  <!-- SegmentURLs -->
</SegmentList>

So the same file is referenced several times with different ranges. In this case we can optimize the extractor by merging consecutive fragments where the URLs are the same and the start of the range is equal to the end of the previous range. In my case one file was referenced about 100 times and I believe by replacing 100 small requests with 1 large we could significantly reduce the downloading time. I hope to submit a PR with this optimization sometime, that's why I would like to stay with current byte_range format. What do you think?

Thank you for your note on language! I'm really appreciate it because I'm actively learning English now) I copied that variable name from downloader/hls.py and it seems there are several more variables with splitted_ in their names. Maybe we can merge it as is and then correct all such spelling mistakes?) I can do it a bit later.

[kinolaev]

Hello @dirkf! I added the optimization described above and fixed the spelling mistake. Could you please review the updated PR?

[dirkf]

Have a look at the new yt-dl PR if you like. Maybe the test cases will be useful, even if nothing else.