Qtube

**Projection filtering**

Shortly after releasing [v1.7.0](https://github.com/Killian42/QTube/releases/tag/v1.7.0), I noticed that I missed a video property regarding their projection in the YT API documentation. This is corrected in this release with in the introduction of the `preferred_projections` parameter. This parameter filters out videos based on their projection, i.e., if it is *rectangular* (regular videos) or *360* (best enjoyed with a VR headset).

**Video Quality filtering**

Videos can now be filtered out based on their quality (Issue 11). This feature uses 4 new parameters: `lowest_resolution`, `lowest_framerate`, `lowest_definition` and `preferred_dimensions`.
The first two rely on another package (pytube) and can take quite some time (30s on average for 5 videos on my machine) to run compared to "native" parameters. I would use them only if they are absolutely necessary for your needs (that's why they are set to *null* in the template).
The last two rely directly on the YT API like previous parameters.

-`lowest_resolution`: Lowest resolution you consider acceptable. For example, if set to *1080p*, videos 720p or lower will not be added to your playlist.
-`lowest_framerate`: Lowest framerate you consider acceptable. For example, if set to *60*, 30fps videos will not be added to your playlist.
-`lowest_definition`: Lowest definition you consider acceptable. For example, if set to *HD*, SD videos will not be added to your playlist. Setting this parameter to *HD* does the same thing as setting `lowest_resolution` to *720p*, but is much faster to run.
-`preferred_dimensions`: 2D videos are regular videos and 3D videos are videos using anaglyph 3D (red-cyan videos). This parameter currently cannot search for 360° videos.

**Side note**: The names of the parameters used for title filtering have been changed to be more consistent with the other parameters.

**Upload date filtering**

Videos can now be filtered out based on the day they were uploaded to YouTube (Issue 6). This new functionality uses the `run_frequency` parameter, which handles two types of input: a selection of strings or any positive integer.
The string can be any of the following:
- *daily* (keeps only videos from the previous day)
- *weekly* (keeps only videos from the last 7 days)
- *monthly* (keeps only videos from the last 30 days)

The integer is interpreted as the cutoff number of days that the software considers (i.e., using *7* would give the same result as using *weekly*).

**Important note**: For now, only the day is relevant, and not the full upload time with hours and minutes. This means that a day here is not a rolling 24-hour period. Let's say that the software runs at 8pm on April 1st with the *run_frequency=daily* parameter. In this case, only videos uploaded between March 31st 0am and March 31st 12pm would be taken into account, and not videos uploaded between March 31st 8pm and April 1st 8pm.

**Description and tags filtering**

Videos can now be filtered out based on the content of their description (Issue 8) and of their associated tags (Issue 9). Both of these work the same way as video title filtering: two lists of strings define what is required and banned for each.
See the [ReadMe](README.md) for more information.

The release also includes a revamped [ReadMe](README.md), with a better looking design and additional sections containing new information.

**Duplicates checking and language filtering**

Prior to adding the videos to the playlist, the software now performs a duplicate check (i.e., checks if the video is already in the playlist or not) depending on the values of the `keep_duplicates` parameter (Issue 7). This prevents the same video from appearing multiple times in a playlist, for example if it was added manually before the software runs.

Videos can now be filtered out based on their language (Issue 5). Languages are specified by their ISO 639-1 codes (see this [wikipedia page](https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes)) in the `preferred_languages` parameter. The language of the video is determined first by the audio language (if provided) and then by the title language (if provided). If neither is provided, the video language is marked as *unknown*. Videos with an unknown language are not filtered out, unless they break any of the other rules.
In the future, a way to algorithmically recover the language of the title may be implemented to circumvent the issue of videos with unknown languages.

**Video Duration Filtering**

Videos can now be filtered out based on their duration!
To do so, use the `allowed_durations` parameter by specifying a lower and upper limit for video durations (see the [ReadMe](README.md) for more information).

**Important note**: Since the YT Data API unfortunately doesn't provide any information about the type of content ("normal" videos, shorts, livestreams...), the code to detect if a video is a short or not (enabled by the `keep_shorts` parameter) relies entirely on its duration (if the video is shorter than 65 seconds, it is flagged as a short). This leads to unwanted results if the `keep_shorts` parameter is set to *True* and the lower bound of the `allowed_durations` parameter is less than this 65 seconds threshold. Namely, all videos shorter than 65 seconds will be discarded, whatever that lower bound is. When (if ?) an alternative way of identifying shorts becomes available, it will be implemented to avoid such problems.