* Remove struct from `embedthumbnail`
* Use bullet lists in readme where numbered list don't make sense
* Fix error introduced in 9c2b75b561 when `ie_result` is `None`
Now `playlist_index` is always the position of the video in the actual playlist and `playlist_autonumber` is the position of the item in the playlist queue
* Allow slicing lists/strings using `field.start🔚step`
* A field can also be used as offset like `field1+num+field2`
* A default value can be given using `field|default`
* Capture all format strings and set it to `None` if invalid. This prevents invalid fields from causing errors
This reverts commit 0704d2224b.
This is a commit from `youtube-dlc`. It is not clear what the original purpose of this was. It seems to be a way for extractors to pass `title` and `id` through when the entry is processed by another extractor
* But `title` can already be passed through using `url_transparent`
* `id` is never supposed to be passed through since it could cause issues with archiving
* Treat `--sub-langs` entries as regex
* `all` can be used to refer to all the subtitles
* the language code can be prefixed with `-` to exclude it
* Deprecates `--all-subs`
Closes#253
* Renamed `--external-downloader-args` to `--downloader-args`
* Added `native` as an option for the downloader
* Use similar syntax to `--downloader-args` etc. Eg: `--downloader dash:native --downloader aria2c`
* Deprecated `--hls-prefer-native` and `--hls-prefer-ffmpeg` since the same can now be done with `--downloader "m3u8:native"` and `m3u8:ffmpeg` respectively
* Split `frag_urls` protocol into `m3u8_frag_urls` and `dash_frag_urls`
* Standardize shortening of protocol names with `downloader.shorten_protocol_name`
* If `--no-clean-infojson` is given, the video ids are saved/loaded from in the infojson along with their playlist index
* If a video entry that was not saved is requested, we fallback to using `webpage_url` to re-extract the entries
Related: https://github.com/yt-dlp/yt-dlp/issues/190#issuecomment-804921024