Possible encoding issue with ObsidianFollowLink

[ErenKyojin]

🐛 Describe the bug

I'm not sure if this is a bug, an inetnded feature, or something which might be editable in settings but i have a really big vault with a lot of links i wrote "manually" with different cases compared to the actual file name, the issue is that while obsidian opens the correct file without any issue, in neovim i get a new file generated in the root folder every time.

Versions

~ nvim --version
NVIM v0.9.1
Build type: Release
LuaJIT 2.1.0-beta3

~ nvim --headless -c 'lua print("Obsidian v" .. require("obsidian").VERSION)' -c q
Obsidian v1.12.0

[epwalsh]

It's not intended to be case sensitive, and it works as expected for me. Can you give me a particular example (note name and link) that doesn't work for you?

[ErenKyojin]

Actually the issue might be with aliases, not with case sensitivity, and maybe there's also the fact that they are in sub-folders
Running ObsidianFollowLink on the link [[Tensione|differenza di potenziale]], which should bring me to the file vaultroot/elettrotecnica/tensione.md, creates a new file vaultroot/tensione.md.
The same happens for, as another example, the link [[campi di forze conservative|campo conservativo]] creates vaultroot/campi di forze conservative instead of opening vaultroot/fisica/campi di forze conservative.

Sorry about the confusion and the late reply.

[epwalsh]

Hey @ErenKyojin, thanks for the additional info. Unfortunately I still wasn't able to reproduce the issue. If you get a chance to try to debug this yourself, the relevant logic is all in here:

obsidian.nvim/lua/obsidian/command.lua

Lines 544 to 597 in 60f8da4

	local note_name = current_line:sub(open, close)
	if note_name:match "^%[.-%]%(.*%)$" then
	-- transform markdown link to wiki link
	note_name = note_name:gsub("^%[(.-)%]%((.*)%)$", "%2|%1")
	else
	-- wiki link
	note_name = note_name:sub(3, #note_name - 2)
	end
	-- Split note file name/path from note name/alias.
	local note_file_name = note_name
	if note_name:match "|[^%]]*" then
	note_file_name = note_file_name:sub(1, note_file_name:find "|" - 1)
	note_name = note_name:sub(note_name:find "|" + 1, note_name:len())
	end
	-- Check if it's a URL.
	if note_file_name:match "^[%a%d]*%:%/%/" then
	if client.opts.follow_url_func ~= nil then
	client.opts.follow_url_func(note_file_name)
	else
	echo.warn(
	"This looks like a URL. You can customize the behavior of URLs with the 'follow_url_func' option.",
	client.opts.log_level
	)
	end
	return
	end
	-- Remove header link from the end if there is one.
	local header_link = note_file_name:match "#[%a%d-_]+$"
	if header_link ~= nil then
	note_file_name = note_file_name:sub(1, -header_link:len() - 1)
	end
	-- Ensure file name ends with suffix.
	if not note_file_name:match "%.md" then
	note_file_name = note_file_name .. ".md"
	end
	-- Search for matching notes.
	local notes = util.find_note(client.dir, note_file_name)
	if #notes < 1 then
	local aliases = note_name == note_file_name and {} or { note_name }
	local note = client:new_note(note_file_name, nil, nil, aliases)
	vim.api.nvim_command("e " .. tostring(note.path))
	elseif #notes == 1 then
	local path = notes[1]
	vim.api.nvim_command("e " .. tostring(path))
	else
	echo.err("Multiple notes with this name exist", client.opts.log_level)
	return
	end

[lpanebr]

Indeed the second example debunks the case sensitivity add a culprit...

I remember having this issue also but I'm very behind on my taking notes habit..

Also I'm avoiding the computer due to lower back pain ATM.

From the behavior and code above it's clear that the util.find_note(client.dir, note_file_name) is not returning anything as #notes < 1 is true.

I've reviewed the code from util.find_note to plenary s scan_dir and is_file and all seems perfect. I reached my limit at Path:_stat() in
https://github.com/nvim-lua/plenary.nvim/blob/0dbe561ae023f02c2fb772b879e905055b939ce3/lua/plenary/path.lua#L283C35-L283C35

Then I kind of remember thinking that my issue also happened when my notes filename had accented characters like essência.md (maybe it's something to do with encoding?)

[epwalsh]

If this is still an issue I'll reopen.