* Add option `--live-from-start` to enable downloading live videos from start
* Add key `is_from_start` in formats to identify formats (of live videos) that downloads from start
* [dash] Create protocol `http_dash_segments_generator` that allows a function to be passed instead of fragments
* [fragment] Allow multiple live dash formats to download simultaneously
* [youtube] Implement fragment re-fetching for the live dash formats
* [youtube] Re-extract dash manifest every 5 hours (manifest expires in 6hrs)
* [postprocessor/ffmpeg] Add `FFmpegFixupDuplicateMoovPP` to fixup duplicated moov atoms
Known issue: Ctrl+C doesn't work on Windows when downloading multiple formats
Closes#1521
Authored by: nao20010128nao, pukkandan
* Separate `--console-title` and `--no-progress`
* Add option `--progress` to show progress-bar even in quiet mode
* Fix and refactor `minicurses`
* Use `minicurses` for all progress reporting
* Standardize use of terminal sequences and enable color support for windows 10
* Add option `--progress-template` to customize progress-bar and console-title
* Add postprocessor hooks and progress reporting
Closes: #906, #901, #1085, #1170
* Unpad HLS fragments with PKCS#7 according to datatracker.ietf.org/doc/html/rfc8216
* media_sequence should only be incremented in for media fragments
* The native decryption should only be used if ffmpeg is unavailable since it is significantly slower. Closes#1086
Authored by: shirt-dev, pukkandan
When using
```py
os.remove(encodeFilename(filename))
os.rename(encodeFilename(temp_filename), encodeFilename(filename))
```
the `os.remove` need not be atomic and so can be executed arbitrarily compared to the immediately following rename call. It is better to use `os.replace` instead
Authored by: paulwrubel
In some streams, empty segments may appear with a bogus, non-monotone MPEG timestamp.
This should not be considered as an overflow
Authored by: fstirlitz
This reverts commit 6c907eb33f
The use of the Content-Length value here is erroneous and may lead
to truncated downloads if a compression scheme is specified in the
Content-Encoding header, as the Content-Length header refers to the
size of encoded data, not of the raw bytestream. This has been noticed
in the wild with WebVTT subtitle segments.
Authored by: fstirlitz
Fixes: https://github.com/yt-dlp/yt-dlp/issues/631#issuecomment-893338552
Previous deduplication algorithm only removed duplicate cues with
identical text, styles and timestamps. This change also merges
cues that come in ‘daisy chains’, where sequences of cues with
identical text and styles appear in which the ending timestamp of
one equals the starting timestamp of the next.
This deduplication algorithm has the somewhat unfortunate side effect
that NOTE blocks between cues, if found, will be emitted in a different
order relative to their original cues. This may be unwanted if perfect
fidelity is desired, but then so is daisy-chain deduplication itself.
NOTE blocks ought to be ignored by WebVTT players in any case.
Authored by: fstirlitz
