some advice on installation

[SneakyShady]

Hey there!
First of all, thank you for creating such a awesome looking plugin! It made me discover Yazi too, which is really cool! Big duckdb fan, and have a whole of .csv laying around, so this is a godsend for quick data profiling and making sure I am using the right files.

Apologies if this is not the right forum for this kind of question: this is going to sound really silly, but I can't for my life seem to get it to work following the instructions ( I am on Apple Silicon, on the latest OS.)

• both duckdb and yazi are installed via homebrew
• I have duckdb 1.2.1 installed, and works in my terminal by just typing duckdb, which launches the CLI, as expected.
• Similarly, I have yazi, and typing yazi lauches Yazi, and works as expected.
• I've used the yazi plugin command to add the duckdb.yazi plugin

my ~/.config/yazi folder after (ya pack -a wylie102/duckdb)

I modified the yazi.toml config file to include the two piece which you mentioned:

[manager]
ratio = [1, 2, 5]

[plugin]  
prepend_previewers = [  
  { name = "*.csv", run = "duckdb" },  
  { name = "*.tsv", run = "duckdb" },  
  { name = "*.json", run = "duckdb" },  
  { name = "*.parquet", run = "duckdb" },  
]

prepend_preloaders = [  
  { name = "*.csv", run = "duckdb", multi = false },  
  { name = "*.tsv", run = "duckdb", multi = false },  
  { name = "*.json", run = "duckdb", multi = false },  
  { name = "*.parquet", run = "duckdb", multi = false },
  { name = "*.db", run = "duckdb" },
  { name = "*.duckdb", run = "duckdb" },
]

and duckdb cli runs as expected (nice one on those color modifications too, the default options are pretty drab!)

now here is the kicker: hovering over files in yazi with the right extensions (.csv or .parquet) does.. nothing.

• worth noting the change to the split of the UI by 1/8ths (the 1-2-5) applies correctly.

(hitting enter predictably runs into an error too, since I don't have a default app to open .parquet files)

Not sure if this is something that you can help with, but I am a bit at a loss on how to troubleshoot it. I've restarted, rebooted, tried my luck with reinstalling both Yazi and duckdb to no avail.
I'm not familiar with yazi, nor lua, and yazi doesn't seem to throw some kind of error to let me know that it doesn't know how to handle that type of file nor that an error has occurred.

Any advice on how to troubleshoot, or alternatively, if you have a different recommended way of installing yazi + duckdb to have it work, I'd really appreciate it!

In any case, really appreciate your time and efforts on this, it really does look super useful!

edit: forgot to paste the DEBUG logs, when going over a parquet or .csv file

 2025-04-07T20:47:20.272807Z ERROR yazi_plugin::isolate::peek: runtime error: ?:-1: attempt to concatenate a nil value
stack traceback:
        ?: in upvalue '?'
        ?: in function <?:287>
    at yazi-plugin/src/isolate/peek.rs:108

[wylie102]

Sure I'll try and help. It looks slightly odd as when it fails to load a file it should actively fall back to the default code readers attempt, which for parquet would just say binary file, and for csv or json would show you like a colour highlighted version of their raw text. So it looks like the plugin is hanging at some point

I think this is probably my fault as I don't think I made it clear in the installation instructions that you need the init.lua as well (I didn't really realise till now actually, even if it is empty. The setup function has defaults, but without the line in the init.lua it isn't called when yazi starts so the defaults don't get initialised.

create an init.lua in the same file as the yazi.toml and add

-- DuckDB plugin configuration
require("duckdb"):setup()

and it should work.

or you can add

require("duckdb"):setup({ mode = "standard" })

if you want it to startup in standard view rather than the summarized version.

Or you can put all the options in there set to the defauilts and then they are easy to change when you want to.

Don't forget the keymap.toml as well so that you can side scroll, again it goes in the same folder.

[[manager.prepend_keymap]]
on = "H"
run = "plugin duckdb -1"
desc = "Scroll one column to the left"

[[manager.prepend_keymap]]
on = "L"
run = "plugin duckdb +1"
desc = "Scroll one column to the right"

That's good information to know though, thanks for letting me know it wasn't working. You could also have raised it in the issues section as the plugin wasn't working correctly.

I'll move the init.lua instructions into the installation section and make sure it's clear you at least need to have the setup call in there.

Let me know if it still isn't working or if you have any other questions about the config or anything.

[wylie102]

I've made some changes to the installation section so that it now includes the minimum for the init.lua.

Let me know if you think it needs anything else.

[wylie102]

A useful tip with yazi by the way (although this may be standard functioning in the later versions, let me know) is this shell wrapper mentioned here. Once you add it if you launch with y instead of yazi when you next quit it will drop you at the location you were last browsing, rather than where you started. It makes navigating the terminal super easy. Especially with the zoxide integration.

[SneakyShady]

Hey @wylie102 , thanks a lot for taking the time to get back to me, and duly noted for the issues, I'll file it there next time!
Great to know about the init.lua, definitely had missed that small piece.

I've also implemented the y shortcut (very useful indeed!), and well as the (still not usable) navigation keybindings.

I created the init.lua file, with the basic configuration ( require("duckdb"):setup() ), and now, (some) progress:
it still fails, but this time, when trying to quit, it actually registers that something is trying to run, but hanging:

screenshot of the .config/yazi directory:

screenshot of the process freezing:

Here is a dump of the debug logs from Yazi: it no longer looks like it fails on a 'nil' value, but it seems that it doesn't like something else now.

 2025-04-08T12:10:38.586769Z DEBUG yazi_adapter::emulator: Detected background color: 1e1e/1e1e/2e2e (luma = 0.12)
    at yazi-adapter/src/emulator.rs:183

  2025-04-08T12:10:38.591063Z DEBUG yazi_adapter::emulator: Terminal responded to DA1 in 217.75µs: [27, 80, 49, 36, 114, 49, 32, 113, 27, 92, 27, 91, 63, 49, 50, 59, 49, 36, 121, 27, 91, 63, 48, 117, 27, 91, 63, 54, 50, 59, 50, 50, 99]
    at yazi-adapter/src/emulator.rs:120

  2025-04-08T12:10:38.602133Z ERROR yazi_plugin::isolate::peek: runtime error: ?:-1: attempt to call a string value (method 'name')
stack traceback:
        ?: in upvalue '?'
        ?: in upvalue '?'
        ?: in upvalue '?'
        ?: in function <?:287>
    at yazi-plugin/src/isolate/peek.rs:108

  2025-04-08T12:10:41.376894Z ERROR yazi_plugin::isolate::peek: runtime error: ?:-1: attempt to call a string value (method 'name')
stack traceback:
        ?: in upvalue '?'
        ?: in upvalue '?'
        ?: in upvalue '?'
        ?: in function <?:287>
    at yazi-plugin/src/isolate/peek.rs:108

Does that give any other clues? :/
Again, thank you for taking the time, I really appreciate it!

[SneakyShady]

out of curiosity, how does the plugin "find" the duckdb installation and use it?
I'm wondering if I can try to change the path directly to the duckdb executable, and see if that works: that might tell us that it has to do with yazi's ability to find the duckdb instance.

[wylie102]

Is the error message the same as before?

[SneakyShady]

same, but lines have changed

2025-04-08T15:15:42.099936Z ERROR yazi_plugin::isolate::peek: runtime error: ?:-1: attempt to call a string value (method 'name')
stack traceback:
        ?: in upvalue '?'
        ?: in upvalue '?'
        ?: in upvalue '?'
        ?: in function <?:446>
    at yazi-plugin/src/isolate/peek.rs:108

[wylie102]

out of curiosity, how does the plugin "find" the duckdb installation and use it? I'm wondering if I can try to change the path directly to the duckdb executable, and see if that works: that might tell us that it has to do with yazi's ability to find the duckdb instance.

it just issues it as a command, same as when using it in terminal, in fact on Mac OS it's exactly the same because to get the colouring in the output I have to trick it into thinking it's outputting to terminal so the duckdb command is actually issued by a program called script, which should be on every mac by default. You can check it that is there by doing 'which script' in terminal.

But I think if it wasn't you'd be getting a different error.

I think I might have narrowed down where the problem is coming from.

I've started a debugging branch. It just has some debug statements in it to help me track down where the issue might be.

If you download the main.lua from that and replace the main.lua in the duckdb plugin folder with it.

Then run it using YAZI_LOG=debug yazi

and take a look at what it logs.

Basically I just added function called and function successful to the main functions and one accessory one.

You'll probably still get your error but hopefully we will be able to see a bit more closely where it is failing, by a missing 'function successful.

If it is short enough just paste the log in here starting from the "setup called" one (hopefully).

[SneakyShady]

Oh, that's really nice of you. So I copy and pasted the main.lua file content, and replaced the existing main.lua with it:
running the debug step outputs the following:
interesting that it seems like it sets up multiple times in a row like that

  2025-04-08T16:52:27.504783Z DEBUG yazi_plugin::utils::log: "setup called"
    at yazi-plugin/src/utils/log.rs:10

  2025-04-08T16:52:27.504793Z DEBUG yazi_plugin::utils::log: "setup successful"
    at yazi-plugin/src/utils/log.rs:10

    at yazi-adapter/src/emulator.rs:120

  2025-04-08T16:52:27.493528Z DEBUG yazi_adapter::emulator: Detected background color: 1e1e/1e1e/2e2e (luma = 0.12)
    at yazi-adapter/src/emulator.rs:183

  2025-04-08T16:52:27.504783Z DEBUG yazi_plugin::utils::log: "setup called"
    at yazi-plugin/src/utils/log.rs:10

  2025-04-08T16:52:27.504793Z DEBUG yazi_plugin::utils::log: "setup successful"
    at yazi-plugin/src/utils/log.rs:10

    at yazi-adapter/src/emulator.rs:120

  2025-04-08T16:52:27.493528Z DEBUG yazi_adapter::emulator: Detected background color: 1e1e/1e1e/2e2e (luma = 0.12)
    at yazi-adapter/src/emulator.rs:183

  2025-04-08T16:52:27.504783Z DEBUG yazi_plugin::utils::log: "setup called"
    at yazi-plugin/src/utils/log.rs:10

  2025-04-08T16:52:27.504793Z DEBUG yazi_plugin::utils::log: "setup successful"
    at yazi-plugin/src/utils/log.rs:10

  2025-04-08T16:52:27.505041Z DEBUG yazi_adapter::emulator: Terminal responded to DA1 in 153.25µs: [27, 80, 49, 36, 114, 49, 32, 113, 27, 92, 27, 91, 63, 49, 50, 59, 49
, 36, 121, 27, 91, 63, 48, 117, 27, 91, 63, 54, 50, 59, 50, 50, 99]
    at yazi-adapter/src/emulator.rs:120

  2025-04-08T16:52:29.911246Z DEBUG yazi_plugin::utils::log: "preload called"
    at yazi-plugin/src/utils/log.rs:10

  2025-04-08T16:52:29.911353Z DEBUG yazi_plugin::utils::log: "preload called"
    at yazi-plugin/src/utils/log.rs:10

  2025-04-08T16:52:29.932156Z DEBUG yazi_plugin::utils::log: "peek called"
    at yazi-plugin/src/utils/log.rs:10

  2025-04-08T16:52:29.933177Z ERROR yazi_plugin::isolate::peek: runtime error: ?:-1: attempt to call a string value (method 'name')
stack traceback:
        ?: in upvalue '?'
        ?: in upvalue '?'
        ?: in upvalue '?'
        ?: in function <?:452>
    at yazi-plugin/src/isolate/peek.rs:108

  2025-04-08T16:52:32.024277Z DEBUG yazi_plugin::utils::log: "peek called"
    at yazi-plugin/src/utils/log.rs:10

  2025-04-08T16:52:32.025295Z ERROR yazi_plugin::isolate::peek: runtime error: ?:-1: attempt to call a string value (method 'name')
stack traceback:
        ?: in upvalue '?'
        ?: in upvalue '?'
        ?: in upvalue '?'
        ?: in function <?:452>
    at yazi-plugin/src/isolate/peek.rs:108

  2025-04-08T16:52:33.604232Z DEBUG yazi_plugin::utils::log: "peek called"
    at yazi-plugin/src/utils/log.rs:10

  2025-04-08T16:52:33.605633Z ERROR yazi_plugin::isolate::peek: runtime error: ?:-1: attempt to call a string value (method 'name')
stack traceback:
        ?: in upvalue '?'
        ?: in upvalue '?'
        ?: in upvalue '?'
        ?: in function <?:452>
    at yazi-plugin/src/isolate/peek.rs:108

[wylie102]

Try latest main. It should be fixed.

You were on a newer version of yazi than me, thats why I couldn't see the problem.

I just upgraded and it broke the same way, so I checked their release notes and they'd changed 'name' from a method of Url - Url:name() to a property Url.name. Like if in python you were trying to do path.name() instead of path.name.

Everyone else would have/will start having that issue once they upgrade to the latest yazi as well.

Interestingly I switch between the standard and the Nightly and this issue wasn't present on the nightly either.

btw, ya pack will be annoyed because you modified a file. So then annoyingly won'y let you upgrade or remove a plugin. I found the best approach is to remove the duckdb.yazi folder, remove the duckdb.yazi entry from the package.toml

then just run ya.pack -a wylie102/duckdb.yazi again and it should work.

Let me know if it works, hopefully it will this time

[SneakyShady]

Ahh, that totally makes sense. Definitely hadn't though that we could be on different versions of Yazi. Come to think of it, probably gave you a much tougher time to figure out without that information. I'll definitely try and remember to share that in the future. Breaking API changes are always an annoying one to deal with.

but here it works, in all its glory!

Just wanted to say thank you so much for taking the time to look into this, I really appreciate it, and thank you for the really cool plugin. Can't wait to get it working in my workflow / day to day!

[wylie102]

No worries. You actually probably got me ahead of it, since probably not that many people will have moved to the newer yazi and discovered it breaks my plugin yet.

Also I think the init.lua was definitely what was breaking things first, and that then hid this problem. Then after I had narrowed it down to that name part and put some debugs around it, but then none of them made into your logs and I couldn't work out why it would be a problem for you but not me.

Then I ran brew upgrade and everything made sense 😂

I should maybe set up a template for issues/discussions and ask for versions etc. But I've only had two issues raised so far.

BTW I just pushed another version that now supports the output colours on windows and linux too, if you know anyone that would find it useful. It feels a bit faster on my mac too because I was able to stop using script so it comunicates with duckdb more directly.