Import problem ("deck" field is ignored since version 24.11)

[RaZoR]

Hello.

I’m using Anki to maintain several different decks with my private English words and phrases collection.

Usually, I do import my files using command line, for instance:

Command line: anki deck_2025.txt

This file starts with:

#notetype:Basic
#deck:"2025"
...

So, the deck to import cards from this file is specified in the file.

It should work according to documentation and actually it worked successfully before update to 24.11

In 24.11 this “deck” field in txt file is ignored and Anki offers to import file to the desk I’ve selected manually at previous run.

I’ve downgraded version to 24.06.3 and now it works again.

Screenshots:

24.06.3 (correct deck “2025” is automatically selected):

24.11 (wrong deck “abc” is selected…):

[abdo]

Can you share a sample file? I tried this:

#notetype:Basic
#deck:"2025"
a,b

The “2025” is not being selected, so I can reproduce the issue (even in 24.06.3) (turned out the deck needs to be created first), but just want to test with your file to find the source of that “abc” deck.

[RaZoR]

Surely, both decks (2025 and abc) are created beforehand.
Acutally, my decks are already filled with hundreds of cards.

Both files are in the same format:

#notetype:Basic
#deck:"abc"
a;b

#notetype:Basic
#deck:"2025"
aa;bb

anki-24.06.3 = OK
anki-24.11 = ignores deck: field in import files

[abdo]

Anki 24.11 started allowing quotation marks in deck/tag names (see Do not strip quotation marks from deck/tag names by abdnh · Pull Request #3407 · ankitects/anki · GitHub), so "abc" and "2025" are interpreted literally now (the quotes were silently stripped in older versions). You can fix the issue by removing the quotes from the deck field.

[RaZoR]

I’ve already tried deck:2025 without quotes. The same resultant: anki ignores the field and tries to import cards to another deck.

#notetype:Basic
#deck:2025
a;b

$ anki deck_2025.txt

Wrong “abc” is selected:

[abdo]

Looks like there’s a separate issue here with 2025: Anki is actually interpreting the year as a deck ID and is defaulting to a different deck when no deck with such ID is found.

[abdo]

@abdo

(turned out the deck needs to be created first)

Sidenote: The manual is misleading on this point: Text Files - Anki Manual

[RaZoR]

I could say that manual is correct:

"deck" Deck name or id Presets the deck, if it exists.

(Note that deck and deck column are different keywords)

And yes! You are right - if I load deck_abc.txt, the abc is treated as deck Name, but in the problematic file, 2025 is treated as Id, not Name.

Any ideas how to deal with it without renaming my decks to ^[^0-9]* ?

It would be good to add a special condition such as toString().equals(deckName) || toInt().equals(deckId) (pseudo code).

[dae]

@Rumo I seem to recall this has come up before - do we have an existing issue about it or stance on it? I did a quick search but couldn’t find one.

[Rumo]

Not aware of it.
Seems reasonable to check if a passed number is an existing deck name, but if I recall correctly, it will require some restructuring of the code.

[llama]

I’ve been working on a pr for this. The approach taken is to enclose numeric deck and notetype* names in single-quotes ('), and escape other existing single-quotes present in non-numeric names with a backslash (\)

Problem is the breakage where old clients won’t know how to handle escaped single-quotes in deck/notetype names when trying to import these new CSVs

Thoughts? Or perhaps instead of quoting and escaping, is it preferable to just assume a numeric deck name is not an id if no such deck id exists?

*this issue also applies to notetype names btw

EDIT: opened a pr

[RaZoR]

Hello.

deck:"2025"
deck:'2025'
deck:2025

1. find “as is” "2025" in deck names
2. strip quotes, for instance: s/["']//g
3. find stripped 2025 in deck names
4. finally, find 2025 in deck IDs list

The first found result is the targeted deck.

[sorata]

I think having these as subsections of “File Header” is an issue. Your link describes “deck column” and not the header that starts with #.