Fixed, will be released in 2.1.0

So I finally found at least a theoretical way to cause this issue:

If the file metadata contains a tag with empty value then sanitation will set this tag again and remove it in this process.

Now it is actually difficult to set an empty tag, because usually when you do something like metadata["sometag"] = None it will not be set, empty tags are not part of metadata. But if you do metadata.set("sometag", []) this will create an empty tag. We have other inconsistencies between these two functions, that are not obvious to a caller.

So far I could not figure out a way to actually create an empty tag with Picard, but you somehow manage that Some questions:

Some question to you:

1. What plugins do you have active?
2. Do you use the sort_multivalue_tags plugin (this is the only plugin I found uses metadata.set, but in a way that IMHO should be save)?
3. What do you have set in options for preserve tags?

So I see two potential fixes:

1. For sanitation use metadata.set, e.g.:

for name in metadata.keys():
if isinstance(metadata[name], str):
metadata.set(name, [sanitize_filename(metadata[name])])

2. Remove the difference between metadata[] and metadata.set. Not sure why this is needed, eventual side effects have to be taken into account

Thanks a lot for the feedback, interesting issue. I will take a closer look, but it might take a few days until I can get to this

Sorry for the delay. Work kept me busy for the last two weeks. I just had the chance to have a look at this issue again.

The problem has indeed nothing to do with the initial title of the issue. I finally identified the tagger script that is to blame here:

$if($eq(%_isAudiobook%,True), $set(_author, $trim($rsearch($if2(%_albumartists_sort%,%_artists_sort%),^[^;]*)) ) $if(%_author%, $set(_reader, $if($gt($sub($len(%albumartistsort%),$len($strip(%_author%))),0), $right(%albumartistsort%, $sub($len(%albumartistsort%),$len($strip(%_author%)))), ) ) ) )

The script sets the author of audiobooks and extracts the reader from the albumartistsort tag (e.g. "Tolkien, J.R.R. read by Williamson, Nicol" gives _author="Tolkien, J.R.R." and _reader="read by Williamson, Nicol"). If there is no distinct reader the _reader tag is set to an empty string. Initially this works, despite the fact that the set-method apparently can't set empty strings but calls unset instead (see func_set in script.py).

The part I don't understand happens during sanitation: Before sanitation the metadata dict contains the key "~reader", after sanitation it is missing. So it seems like _reader is unset during sanitation, but I don't get where that might happen.

All right. Feel free to just share the scripts you are using and I can test locally. I will also try to test a bit again on my side.

It seems you were right, the issue could indeed go deeper.

I did some more checks and now I think it has really nothing to do with slashes. But now I'm also more puzzled what does happen here.

Like you already tested, renaming a single file with / in one of the visible tags works just fine, the same applies to multiple files.

For me the error occurs for all audiobooks. In this case I create a hidden variable that contains slashes (To be honest, the variable is used nowhere. The script never worked the way I expected. Now that I know sanitation occurs I understand why... ). And because only for those values with slashes sanitation changes the metadata at all, the most natural thought was that the error must be connected to this hidden variable.

But when removing this part of the script the error persists though there are no fields with slashes left. The exception is thrown both for single and multiple files.

By now I suspect one of the other audiobook scripts is the culprit here. To investigate this further I have to rewrite it. I will look into this tomorrow, the scripting language is just too cumbersome for me tonight.

Looking into this right now, but I am a bit puzzled because I don't see how the size of the list should change in this loop. It's just changing the data, and for that the code is fine. I cannot reproduce it when saving a single file with renaming enabled and tags containing \ or /. The sanitation works, no exception thrown. Maybe something threading related? Can you reproduce this with a single file or only when saving multiple?

Apart from that you fix would be fine for cases where the actual size of the dict changes, nothing dirty about it. We had a couple of such issues after the Python 3 migration

I might just apply your fix anyway, but I actually prefer to be able to reproduce an issue. I have the suspicion here that the actual issue goes somewhat deeper.