How to archive info for all videos that have meta data downloaded even if it was filtered?

[aaronmitchellharris]

I want to use chunk_initial_download in order to download a large channel, but I also have specific types of videos I wanted downloaded from the channel so I am using filter_exclude to exclude certain videos. Whenever a video gets excluded, it is not included in the archive information, so when I run the download again it will download the metadata for videos it had previously seen before and decided to not download. Is there a way to archive the information of any video that has had its meta data downloaded so this doesn't happen?

[jmbannon]

There is not unfortunately. What is your filter? You could try translating it to a yt-dlp match filter (https://ytdl-sub.readthedocs.io/en/latest/config_reference/plugins.html#match-filters)

These run much faster since they do not require downloading the metadata.

[aaronmitchellharris]

My filter essentially has pairs of lists containing keywords and tries to only download videos with a title that contains at least one keyword for each list in the pair. I also have some logic where I add some metadata to the downloaded video based on the keywords that were found so I'll be checking for the keywords after the match filters anyway.

That being said, I did try match filters just so I won't need to download unnecessary metadata, but it doesn't seem to be working and ends up not downloading anything.

For example, if I have two pairs of lists (["keyword1", "keyword2"], ["keyword3", "keyword4"]) and (["keyword5", "keyword6"], ["keyword7", "keyword8"]) I tried this match filter.

match_filters:
    filters:
        - "title~=(?i)(?=.*(keyword1|keyword2))(?=.*(keyword3|keyword4))"
        - "title~=(?i)(?=.*(keyword5|keyword6))(?=.*(keyword7|keyword8))"

I verified on https://regex101.com/ that this regex should work so I'm not sure what's wrong.

[jmbannon]

I'm not very familiar with yt-dlp syntax so unfortunately I'm not much help there.

How many videos are left on the channel? You could try omitting chunk_initial_download, then you'll break on existing videos

[aaronmitchellharris]

I think I figured out the issue with my match filters. I have a lot of pairs of keyword lists, so I made a script to generate the filters used in my config file. I found out one filter's regex had a "+" that was not backslash escaped. I figured this out by pasting each filter into regex101 and found that it gave the error + The preceding token is not quantifiable.

It wasn't obvious this was the problem because the download would just give this message without downloading any metadata or anything.

No files changed for test 
[ytdl-sub] Download Summary:
test     0 0 0     0 ✔
Total: 1 0 0 0     0 Success

I guess if just one match filter has a regex error, it will cause none of them to work and every video will be filtered. After fixing that user error, it seems to be working as expected so far.