Definitely not intended. I thought chrome doesn't write session cookies to disk 🤔. If it does, we should "fix" the cookies similar to https://github.com/yt-dlp/yt-dlp/blob/b96187a9603aaa7ead8ad94ac82433f085ea4895/yt_dlp/utils.py#L1654-L1667

I thought chrome doesn't write session cookies to disk 🤔

It does because of the Continue where you left off startup option.

Firefox is also affected by this issue: It seems like session cookies are not stored in its cookies.sqlite file.

Instead, they are in sessionstore.jsonlz4 (but that file only exists if Firefox is not running). That's a file in Mozilla's custom mozLz4 format, which essentially is just the LZ4 block format preceded by a lower-endian 32-bit unsigned int specifying the size of the uncompressed data in bytes.

Once uncompressed, the sessionstore data can be parsed as JSON, and the session cookies found in the top-level key cookies of the resulting object.

Was about to open the same issue, here is what I found:

Disclaimer at the start - I did this investigation with a recent version of brave, will need to check if this applies to all chromium browsers and if so from which version.

The brave cookies database has 2 columns related to cookie expiry: expires_utc and has_expires. yt-dlp only checks the expires_utc column:

• For session cookies expires_utc is set to 0 and has_expires to 0.
• For expiring cookies expires_utc is set to a utc timestamp and has_expires to 1

The current logic in yt-dlp takes expires_utc only and the phyton cookie will consider a value of 0 as expired:

class Cookie:
    # ...
    def is_expired(self, now=None):
        if now is None: now = time.time()
        if (self.expires is not None) and (self.expires <= now):
            return True
        return False

To support session_cookies we should therefore set expires to None, since they do not expire.

@Tblue can you elaborate on what you mean by with the prefix? I could confirm that this file is created when I close firefox: ~/Library/Application Support/Firefox/Profiles/$profile.default-esr/sessionstore.jsonlz4

It starts like this:

mozLz40�≈¡���{"version":["ses��Ú
restore",1],"windows":[{"tab	�bentrie

I tried removing the mozLz40 or the mozLz40≈¡� using a text editor (probably not the right tool 😄) but passing that to lz4 I always get Error 44 : Unrecognized header : file cannot be decoded, what do I have to remove exactly?

Actually nevermind, I found this superuser thread on it: https://superuser.com/a/1363751

And it linked this python gist that I could test sucessfully: https://gist.github.com/Tblue/62ff47bef7f894e92ed5

I could also confirm that the cookie can be found in there, I censored mine so we have a reference for the format:

jsonpath /cookies/4

    {
      "host": ".vimeo.com",
      "value": "the cookie value",
      "path": "/",
      "name": "12345678_password",
      "secure": true,
      "originAttributes": {
        "firstPartyDomain": "",
        "geckoViewSessionContextId": "",
        "inIsolatedMozBrowser": false,
        "partitionKey": "",
        "privateBrowsingId": 0,
        "userContextId": 0
      },
      "schemeMap": 2
    },

Edit: I am blind, this is even the gist of @Tblue. 😄 Before we can make a PR with this - could you just confirm that you are also willing to publish it (or parts of it) under UNLICENSE?

I think we would only need to copy this method: https://gist.github.com/Tblue/62ff47bef7f894e92ed5#file-mozlz4a-py-L76

@StefanLobbenmeier, you might also find this script useful as an inspiration (feel free to treat it as licensed under the UNLICENSE as well): dump-firefox-cookies.py

Keeping open until this is solved for Firefox as well

Yeah did not have the time to implement for Firefox, but it’s still on my radar. If someone else wants to take it I don’t mind though :)