Finding files in sub-folders by name

[cool-RR]

I have a certain need, and I'm wondering whether there are existing VisiData commands that can help with it.

I'm in a DirSheet which has a lot of folders in it. Each of these folders has in it a file called rollout.jsonl. I want to go through these files one-by-one. I wish there was a way to say "In this DirSheet, please search for files with this name in all subfolders recursively, and open something like a DirSheet where each row is one of these files, so I could open each of them by pressing Enter."

Is there something like that?

[saulpw]

Use the -r/--dir-recurse flag. Select all rollout.jsonl files. " to pull them into their own sheet. Or g Enter to open them up, each in their own sheet.

[saulpw]

given is the url/path that was given to the constructor, verbatim. So ., not expanded, or the url, unparsed.

As for RepeatFile inside Path, what's the non-weird way of caching a file-like so that an HTTP stream can be opened multiple times without making multiple requests?

[cool-RR]

As for RepeatFile inside Path, what's the non-weird way of caching a file-like so that an HTTP stream can be opened multiple times without making multiple requests?

I assume you're inviting constructive criticism here.

Whatever's going on with the processing of file names out of an HTTP stream shouldn't have anything to do with the sheet that displays a list of files. Your process for managing that HTTP stream should output an iterator of paths (which could come as either strings or Path objects), which then FileListSheet should consume. I figured out by trial-and-error that you designed it so the stream of paths should be strings, with newlines at the end for some reason, and then they need to be fed into the fptext of a Path object for some unclear reason, and that Path needs some string for its given attribute which the docs don't say what its significance is (and your reply didn't help because you described it from an implementation perspective rather than a user's / product perspective, i.e. I have no idea what actually is done with this given.)

Here's a saner API:

paths = (path for path in pathlib.Path(dir_sheet.source).rglob('*')
         if re.search(regex, path.name))
file_list_sheet = FileListSheet(name='rfind', paths=paths)

If you don't want to change the signature of __init__ you could also do:

file_list_sheet = FileListSheet.from_paths(name='rfind', paths=paths)

By the way, even though in my code above I initialize the FileListSheet with name='rfind', its name shows up as ''. Any idea why?

[saulpw]

By the way, even though in my code above I initialize the FileListSheet with name='rfind', its name shows up as ''. Any idea why?

Name is not a kwarg; for a few versions, it's been a list of name components as *args. This API may be confusing, maybe it should be name_or_names and the constructor can figure out how to deal with it whether it's a single string name or a list of name components.

[cool-RR]

Works now, thank you. I also changed it to a more reasonable f'{dir_sheet}_rfind'.

[saulpw]

Ah, this is the purpose of those *args names! Call it as FileListSheet(*dir_sheet.names, 'rfind', ...) and then there will be two name components, which will look just like your {dir_sheet}_rfind, but will be broken down into name components. This isn't immediately useful, but may well factor into how the sheet tabs look on the status bar very shortly.

[cool-RR]

That makes sense, I changed my code accordingly now. Thank you.

…

On Mon, Nov 27, 2023 at 1:26 AM Saul Pwanson ***@***.***> wrote: Ah, this is the purpose of those *args names! Call it as FileListSheet(*dir_sheet.names, 'rfind', ...) and then there will be two name components, which will look just like your {dir_sheet}_rfind, but will be broken down into name components. This isn't immediately useful, but may well factor into how the sheet tabs look on the status bar very shortly. — Reply to this email directly, view it on GitHub <#2032 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAN3SVU2TMWXDSHCAL7O73YGPF3NAVCNFSM6AAAAAA5ECNOJOVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM3TMNZVGIYTS> . You are receiving this because you authored the thread.Message ID: ***@***.***>