Now that #17 is merged, here's how it can be improved:

1. If the document has a *single level-1 heading (=title), the TOC should be inserted right under it

   This means that the entire TOC will have to be shifted one level down (heading 2 becomes heading 3, etc.) and the original heading-1 is not rendered at all

2. If the document has a YAML frontmatter, the TOC should be inserted under it (unless rule 1 applies)

   This would require to

   • Enable YAML parsing on Markdig
   • Propagate the location of yaml block all the way to Document Index

Cool language server! :)

With marksman 2022-08-19 on Linux and Neovim I create a table of contents. I have nvim-lightbulb installed. Now the lightbulb is constantly lit (wherever my cursor is) telling me there is a code action to 'Update the Table of Contents', even when non of the headings have changed.

Is it possible to only announce this code action when a relevant item has changed? Thanks.

Ideally, marksman should follow the indentation config of the editor (or currently open file or .editorconfig), however, when that does not work, users should be able to configure indentation via, e.g., indent property which could be either tab string or a number (how many spaces should be used for indentation).

Now, I have configured indentation to three spaces in Markdown files, however, the generated TOC looks like this: Actually, this was my mistake, as I actually used heading levels 1, 3, 4, and 3 instead of 1, 2, 3, 2).

<!--toc:start-->
- [Heading 1]()
      - [Heading 2]()  <!-- 6 spaces -->
        - [Heading 3]()  <!-- 8 spaces -->
      - [Heading 2]()  <!-- 6 spaces -->
<!--toc:end-->

<!--toc:start-->
- [Heading 1]()
   - [Heading 2]()  <!-- 3 spaces -->
      - [Heading 3]()  <!-- 6 spaces -->
   - [Heading 2]()  <!-- 3 spaces -->
<!--toc:end-->

marksman always uses two spaces for indentation. The relevant line from Marksman/Toc.fs:

I would like to use marksman on my nix based system, but I don't have any experience with packaging so far and failed when I tried to do so for marksman.

I'll assist as good as I can if someone else wants to try it out.

I have tried getting this to work and I think I am almost succeeded. Unfortunately when I try to auto complete a link to another file, then I get the following error:

After this point nothing works. I use the linux binary from the latest release and use it from within neovim.

The actual link is shown in the autocomplete box as it is expected to. I select it and press enter.

Hello, so I am currently using WSL 2 Ubuntu, and I am having this error as shown in the pic when I run LspInfo on Neovim. I downloaded the latest release via curl, renamed the binary and even placed it inside my PATH folder. But for some reason, it seems like LSP is not properly detecting Marksman as completion and other features not seem to work. Here's the error message below with some additional images as reference:

In my bin folder:

In my .bashrc file:

It might be a good idea to add an action/snippet to insert emojis, just like we do on GitLab and GitHub using a colon (e.g. :wink: → 😉).

I'm trying your LSP server in nvim 0.5, and I must admit I really like it.
It globally work fine, but there is an issue about completion. I don't really know if it's a server issue or client issue. (but I don't have this issue with other lsp servers, so I'm wondering)

The issue is when I try to complete a subsection, the completion provide the correct list of possible choices, but when I select a choice it only selects mi the "##" part of the title and not the text. If I type the title by hand the reference works fine, but this really reduces the workflow speed.

Take a look at the action :

Currently, Marksman requires a project/root/folder to provide language features such as cross-file references, workspace symbols, and rename refactor that works across many files.

However, it would be convenient to provide a subset of language features in situations where there's no project/root/folder. Some features like "document symbols" will be the same as before in "single-file mode". Other will lose cross-file capabilities and will work only intra-file, e.g. goto def, references, renames, completion. But perhaps this is enough for one-off edits of random markdown files.

Hello, thank you for creating this project, it looks really useful!

My usecase is that I have a Zettelkasten repository containing a lot of interlinked notes with generated filenames. For example, my typical note might have a filename 202208240944.md and a content like this:

# Some title here

Lorem ipsum

If I try to link to this note from another file, I can type [[some and I will correctly get a list of suggestions with this note on the top. However, if I decide to use that suggestion, the result will become [[some-title-here]] instead of [[202208240944]]. In other words, marksman seems to assume that the title directly translates to the filename - but I think that is often not the case.

I've tried to use it with Kakoune (using kak-lsp) and Helix and both editors show the same behavior.

Right now, diagnostic calculation is added as a hook and happens inside a critical section: https://github.com/artempyanykh/marksman/blob/main/Marksman/Server.fs#L216; only the actual public of the updated diagnostic happens asynchronously. This means that diagnostic recalculation happens on every keystroke/state_update which is not cool and can lead to timeouts (albeit only on really huge workspaces with hundreds or thousands of md files).

The good news that it should be pretty easy to move diag recalc off of the critical section and make it happen async.

Is it possible to to configure autocompletion to insert markdown link format [Foo Bar](foo-bar.md) instead of wiki format [[foo-bar]]?

Brought up in #9
Description in clangd: https://clangd.llvm.org/extensions.html#utf-8-offsets

as Title , lsp take error when change file .

2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "Unhandled exception: System.AggregateException: One or more errors occurred. (MailboxProcessor.PostAndAsyncReply timeout。)\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- " ---> System.TimeoutException: MailboxProcessor.PostAndAsyncReply timeout。\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "   at <StartupCode$FSharp-Core>[email protected](FSharpOption`1 res) in D:\\a\\_work\\1\\s\\src\\FSharp.Core\\mailbox.fs:line 472\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "   at Microsoft.FSharp.Control.AsyncPrimitives.CallThenInvokeNoHijackCheck[a,b](AsyncActivation`1 ctxt, b result1, FSharpFunc`2 userCode) in D:\\a\\_work\\1\\s\\src\\FSharp.Core\\async.fs:line 525\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "   at Microsoft.FSharp.Control.Trampoline.Execute(FSharpFunc`2 firstAction) in D:\\a\\_work\\1\\s\\src\\FSharp.Core\\async.fs:line 112\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "   --- End of inner exception stack trace ---\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "   at Ionide.LanguageServerProtocol.Server.startWithSetup[client](FSharpFunc`2 setupRequestHandlings, Stream input, Stream output, FSharpFunc`2 clientCreator) in /Users/runner/work/marksman/marksman/LanguageServerProtocol/LanguageServerProtocol.fs:line 170\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "   at Marksman.Program.startLSP(Int32 verbosity, Boolean waitForDebugger) in /Users/runner/work/marksman/marksman/Marksman/Program.fs:line 58\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "   at [email protected](Tuple`2 tupledArg)\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "   at FSharp.SystemCommandLine.CommandBuilders.SetHandlerInt@207-2.Invoke(InvocationContext ctx)\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "   at System.CommandLine.Invocation.AnonymousCommandHandler.Invoke(InvocationContext )\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "   at System.CommandLine.Invocation.InvocationPipeline.<>c__DisplayClass4_0.<<BuildInvocationChain>b__0>d.MoveNext()\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "--- End of stack trace from previous location ---\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "   at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c__DisplayClass17_0.<<UseParseErrorReporting>b__0>d.MoveNext()\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "--- End of stack trace from previous location ---\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "   at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c__DisplayClass12_0.<<UseHelp>b__0>d.MoveNext()\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "--- End of stack trace from previous location ---\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "   at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c__DisplayClass22_0.<<UseVersionOption>b__0>d.MoveNext()\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "--- End of stack trace from previous location ---\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "   at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c__DisplayClass19_0.<<UseTypoCorrections>b__0>d.MoveNext()\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "--- End of stack trace from previous location ---\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "   at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c.<<UseSuggestDirective>b__18_0>d.MoveNext()\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "--- End of stack trace from previous location ---\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "   at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c__DisplayClass16_0.<<UseParseDirective>b__0>d.MoveNext()\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "--- End of stack trace from previous location ---\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "   at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c.<<RegisterWithDotnetSuggest>b__5_0>d.MoveNext()\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "--- End of stack trace from previous location ---\n"
2022-08-18T19:34:09.381 helix_lsp::transport [ERROR] err <- "   at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c__DisplayClass8_0.<<UseExceptionHandler>b__0>d.MoveNext()\n"
2022-08-18T19:34:09.394 helix_lsp::transport [ERROR] err: <- StreamClosed
2022-08-18T19:34:09.394 helix_lsp::transport [ERROR] err: <- StreamClosed
2022-08-18T19:34:11.320 helix_lsp::transport [ERROR] err: <- IO(Os { code: 32, kind: BrokenPipe, message: "Broken pipe" })
2022-08-18T19:34:11.320 helix_lsp::transport [ERROR] err: <- IO(Os { code: 32, kind: BrokenPipe, message: "Broken pipe" })
2022-08-18T19:34:11.882 helix_lsp::transport [ERROR] err <- "[19:34:11 INF] <LSP Entry> Starting Marksman LSP server: {}\n"

os: macos 12.5
termial: alacritty
editor: helix
marksman: 1.0.0

lsp config: ~/.config/helix/languages.toml

[[language]]
name = "markdown"
indent = { tab-width = 4, unit = "    " }
file-types = ["md", "markdown"]
roots = [".git"]
language-server = { command = "marksman", args=["server"], language-id = "markdown" }

hx -vvv take log ~/.cache/helix/helix.log

I've tried configuring marksman as per the README but it doesn't work for me.

❯ hx --health markdown
Configured language server: marksman
Binary for language server: /home/zim/.local/bin/marksman
Configured debug adapter: None
Highlight queries: ✓
Textobject queries: ✘
Indent queries: ✘

I have an empty .marksman.toml file in the cwd.
log(hx -vvv file)->:log-open:

2022-09-17T12:17:52.443 helix_loader [DEBUG] Located configuration folders: []
2022-09-17T12:17:52.454 helix_view::clipboard [INFO] Using xclip to interact with the system and selection (primary) clipboard
2022-09-17T12:17:52.499 mio::poll [TRACE] registering event source with poller: token=Token(1), interests=READABLE | WRITABLE
2022-09-17T12:17:52.499 mio::poll [TRACE] registering event source with poller: token=Token(2), interests=READABLE | WRITABLE
2022-09-17T12:17:52.499 mio::poll [TRACE] registering event source with poller: token=Token(3), interests=READABLE | WRITABLE
2022-09-17T12:17:52.499 mio::poll [TRACE] registering event source with poller: token=Token(4), interests=READABLE | WRITABLE
2022-09-17T12:17:52.499 mio::poll [TRACE] registering event source with poller: token=Token(5), interests=READABLE | WRITABLE
2022-09-17T12:17:52.499 helix_lsp::transport [INFO] -> {"jsonrpc":"2.0","method":"initialize","params":{"capabilities":{"textDocument":{"codeAction":{"codeActionLiteralSupport":{"codeActionKind":{"valueSet":["","quickfix","refactor","refactor.extract","refactor.inline","refactor.rewrite","source","source.organizeImports"]}}},"completion":{"completionItem":{"resolveSupport":{"properties":["documentation","detail","additionalTextEdits"]},"snippetSupport":false},"completionItemKind":{}},"hover":{"contentFormat":["markdown"]},"publishDiagnostics":{},"rename":{"dynamicRegistration":false,"honorsChangeAnnotations":false,"prepareSupport":false},"signatureHelp":{"signatureInformation":{"activeParameterSupport":true,"documentationFormat":["markdown"],"parameterInformation":{"labelOffsetSupport":true}}}},"window":{"workDoneProgress":true},"workspace":{"applyEdit":true,"configuration":true,"didChangeConfiguration":{"dynamicRegistration":false},"symbol":{"dynamicRegistration":false},"workspaceFolders":true}},"processId":693130,"rootPath":"/home/zim/.md","rootUri":"file:///home/zim/.md","workspaceFolders":[{"name":".md","uri":"file:///home/zim/.md"}]},"id":0}
2022-09-17T12:17:52.499 mio::poll [TRACE] registering event source with poller: token=Token(0), interests=READABLE
2022-09-17T12:17:52.499 mio::poll [TRACE] registering event source with poller: token=Token(1), interests=READABLE
2022-09-17T12:17:53.070 helix_lsp::transport [ERROR] err <- "[12:17:52 INF] <LSP Entry> Starting Marksman LSP server: {}\n"
2022-09-17T12:17:54.275 helix_lsp::transport [INFO] <- {"jsonrpc":"2.0","id":0,"result":{"capabilities":{"textDocumentSync":{"openClose":true,"change":2},"hoverProvider":true,"completionProvider":{"triggerCharacters":["[","#","("]},"definitionProvider":true,"referencesProvider":true,"documentSymbolProvider":true,"workspaceSymbolProvider":true,"codeActionProvider":{"resolveProvider":false},"renameProvider":true,"semanticTokensProvider":{"legend":{"tokenTypes":["property","variable"],"tokenModifiers":[]},"range":true,"full":{"delta":false}},"workspace":{"workspaceFolders":{"supported":true,"changeNotifications":true},"fileOperations":{"didCreate":{"filters":[{"pattern":{"glob":"**/*.md","matches":"file","options":{"ignoreCase":true}}}]},"didDelete":{"filters":[{"pattern":{"glob":"**/*.md","matches":"file","options":{"ignoreCase":true}}}]}}}}}}
2022-09-17T12:17:54.275 helix_lsp::transport [INFO] <- {"capabilities":{"codeActionProvider":{"resolveProvider":false},"completionProvider":{"triggerCharacters":["[","#","("]},"definitionProvider":true,"documentSymbolProvider":true,"hoverProvider":true,"referencesProvider":true,"renameProvider":true,"semanticTokensProvider":{"full":{"delta":false},"legend":{"tokenModifiers":[],"tokenTypes":["property","variable"]},"range":true},"textDocumentSync":{"change":2,"openClose":true},"workspace":{"fileOperations":{"didCreate":{"filters":[{"pattern":{"glob":"**/*.md","matches":"file","options":{"ignoreCase":true}}}]},"didDelete":{"filters":[{"pattern":{"glob":"**/*.md","matches":"file","options":{"ignoreCase":true}}}]}},"workspaceFolders":{"changeNotifications":true,"supported":true}},"workspaceSymbolProvider":true}}
2022-09-17T12:17:54.275 helix_lsp::transport [INFO] -> {"jsonrpc":"2.0","method":"initialized","params":{}}
2022-09-17T12:17:54.276 helix_lsp::transport [INFO] -> {"jsonrpc":"2.0","method":"textDocument/didOpen","params":{"textDocument":{"languageId":"markdown","text":"# Zim's Setup Guide\n\n## Backups\n\n*   Browser plugins\n*   `~/.config` dir, rcs, scripts, etc.\n*   Could even use `rsync` to take backups\n\n## Installing Arch Linux\n\n[Arch Installation Guide](https://wiki.archlinux.org/title/installation_guide)\n\n### Prerequisite Knowledge:\n\n*   Selecting mirrors with [ reflector ](https://wiki.archlinux.org/title/Reflector)\n\n*   Chrooting: In case something goes wrong with your installation you will have to chroot into your installation for which you will have to first mount your \"root\" partition and **then** the \"EFI system\" partition. Althought sometimes you might not need to mount the \"EFI system\" partition. We made the installation at `/root` and our EFI mount point(according to the installation guide) is at `/boot`\n\n*   Install gvim/nvim during `pacstrap`\n\n*   To install other packages or package groups, append the names to the pacstrap command above (space separated) or use pacman while [ chrooted ](https://wiki.archlinux.org/title/installation_guide#Chroot) into the new system.\n\n*   Necessary packages:\n    ```console\n    pacman -Sy xorg alsa network-manager-applet polybar haskell-random acpi brightnessctl zellij fish nushell kitty alacritty rofi feh xorg-xinit starship man-db man-pages slock\n    ```\n\n*   Rust installation(for paru): `curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh`\n    ```console\n    sudo pacman -S --needed base-devel\n    git clone https://aur.archlinux.org/paru.git\n    cd paru\n    makepkg -si\n\n    paru -Sy otf-font-awesome-5 xmonad-git xmonad-contrib-git brave-bin microsoft-edge-stable-bin kmonad-git\n    ```\n\n### Installing Bootloader\n\nIn the installation guide our boot partition was mounted at `/mnt/boot` and `/boot` while chrooted. Let's install Gummiboot now:\n\n*   Install using: `bootctl install` which might result in an error, usually complaining about the efi dir not found.\n*   [Configure gummiboot](https://wiki.archlinux.org/title/Systemd-boot#Configuration). Also, checkout [github gists](https://gist.github.com/zim0369) for \"arch.conf\" and \"loader.conf\"\n\n## Setting up the System\n\n### Necessary Knowledge\n\n[System Maintenance](https://wiki.archlinux.org/title/system_maintenance)\n\n[General Recommendations](https://wiki.archlinux.org/title/General_recommendations)\n\n### Placing Required Files\n","uri":"file:///home/zim/.md/arch_install.md","version":0}}}

Please add a guide to README.md on how to install, place binaries in the system or compile from source.

Marksman returns code action of the source kind (titled Create a Table of Contents) when editor asks for code actions without explicitly specifying any kinds.

For example editor make this request:

{
  "context": {
    "diagnostics": [],
    "triggerKind": 2
  },
  "range": {
    "end": {
      "character": 0,
      "line": 0
    },
    "start": {
      "character": 0,
      "line": 0
    }
  },
  "textDocument": {
    "uri": "file:///usr/local/workspace/sublime-packages/LSP/README.md"
  }
}

and marksman returns:

[
  {
    "edit": {
      "changes": {
        "file:///usr/local/workspace/sublime-packages/LSP/README.md": [
          {
            "newText": "<!--toc:start-->\n- [Installation](#installation)\n  - [Stable Version](#stable-version)\n  - [Development Version](#development-version)\n- [Getting started](#getting-started)\n- [Getting help](#getting-help)\n<!--toc:end-->\n\n",
            "range": {
              "end": {
                "character": 0,
                "line": 0
              },
              "start": {
                "character": 0,
                "line": 0
              }
            }
          }
        ]
      }
    },
    "isPreferred": false,
    "kind": "source",
    "title": "Create a Table of Contents"
  }
]

I don't have any explicit description of expected behavior for this but based on VSCode primarily (and what we have mirrored in Sublime Text) it seems that the source code actions should not be returned unless specifically asked for by the editor. Not following this logic results in editor providing source diagnostics for every location in the file (since editors request diagnostics for the current selection/cursor position) which is non-ideal for those kind of code actions.

(This issue originated from the discussion in sublimelsp/LSP#1846 (comment))

Formatting tables manually is a pain. Does table formatting sound like a plausible feature? I can use a third party formatter in the meantime but it does sound convenient.

Looking at my hakyll blog posts (e.g. this) I see the following structure

---
title: [title]
...
---

## Heading 1
## Heading 2

Since there's no 1st level heading there's no clash between yaml's title and a potential markdown # Title.

I guess the question for me is: in the presence of a front matter do we want to
a) treat front matter's title as a level 1 heading and therefore prohibit the use of # ... in the body of the document.

---
title: Front matter
---

# I am prohibited <-- this will be flagged as a duplicate title

## Why

## When

b) treat front matter's title as a level 0 heading and level 1 headings will be just regular document headings and there can be multiple of those, e.g.

---
title: Front matter
---

# What is front matter <-- OK

# Why front matter <-- Also OK

Is it possible to provide a native mac arm64 build?

The README says the following, but how to enable?

Marksman has a custom single-file mode that provides a subset of language features for markdown files open outside of any project. This works well for small one-off edits or when opening random markdown files.

First, marksman treat link labels case sensitive.
For exmaple, the following markdown triggers a false positive warning.

[A][]

[a]: https://example.com

According to John Gruber's canonical description of Markdown's syntax,
link labels are case insensitive.

Link definition names may consist of letters, numbers, spaces, and punctuation — but they are not case sensitive. E.g. these two links:

[link text][a]
[link text][A]

are equivalent.

Second, marksman does not recognize shortcut reference links.
For example, the following markdown does not trigger any warnings.

[b]

Shortcut reference links were introduced by Gruber himself in a beta version of Markdown.pl, but never included in the official syntax description.
Most popular markdown parsers support shortcut referneces.
Also, CommonMark includes shortcut reference links in its specification.

; marksman --version
1.0.0-19de239

Currently, the version is static 1.0.0 which is not cool for debugging. Need to teach the build to fetch the version from the environment.

I have a directory full of all sorts of random things, including markdown files. It has a .git repo initialized at the root.

If I cd into it and open a markdown file with headings, the space e - open symbol picker works like a charm with the headings!

Unfortunately if I choose space E - open workspace symbol picker then the language server hangs, and I get Async job failed, request timed out appearing randomly in red at the bottom of the screen. I need to restart Helix to get marksman working again.

Thanks for the crate.

https://github.com/artempyanykh/zeta-note/blob/main/lsp-document/Cargo.toml#L12

I'd suggest updating the lsp-types to a newer version, also because tower-lsp uses a newer version, which makes it 'harder' to use this crate as this crate uses an older version.

https://github.com/ebkalderon/tower-lsp/blob/master/Cargo.toml#L23

i want it to ignore the h1 , or

i want it to ignore headings that are specially tagged/annotated with some non renderable syntax.

I have such a requirement, I need to know what AST node is currently selected by the mouse.

My AST uses offset to store range information.

Because binary search based on offset is easier to write.

The error reporting library (ariadne) I use is also offset based.

So I want an method like the following:

fn lsp_pos_to_offset(&self, lsp_pos: &lsp_types::Position) -> Option<usize>

Any reason for choosing [:another-note] over the more widely used [[another-note]] syntax? Maybe I'm missing the trade-offs behind this decision, but using wiki-links means your notebook works with a wide variety of editors and tools.

Hi,
thanks for great LSP server. I noticed that when I use link in header like in the example bellow the server autocompletes incorrect link like in the example:

# Google

test

# [Google link](https://google.com)

test

[google link](#google)
[google link with link](#google-linkhttpsgooglecom)

I believe the second link should omit the url part like this#google-link.

Here is screenshot from neovim how I see autocomplete options for this example:

It would be nice if ARM linux binaries could be provided

E.g.

Here comes the [foo]!

[foo]: url

We should be able to rename [foo] and all its occurrences in the document.

Same goes for headings:

# Doc 1
# Doc 12

[[doc-1]]
[[doc-1#doc-12]

Should be able to rename title and headings and update all references.

• Change the code to suggest 'Create ToC' when either
  • the requested range overlaps with the 1st line of the document,
  • code action kind is 'source'.
• 'Update ToC' should be suggested anywhere in the document still (as is now).
• Update the docs

There are many extensions to markdown that define there own flavors of links. Lack of support for these extensions may trigger diagnostics that can be distracting. The idea is to provide an option to disable various types of diagnostics while still be able to use other pieces of functionality from the supported set.

Currently, we calculate line map from scratch after every update. With incremental text sync we could do better than this.

The idea is to implement an incremental recalculation of a line map, e.g.

val incrementalUpdate : initMap:LineMap -> changedRange:Range -> changeMap:LineMap -> LineMap

Hi, thanks for the great lang server for markdown.
I am wondering if marksman support markdownlint or other lint tools/plugins?

It seems that some LSP server capabilities such as diagnostic, find references etc. are missing when using marksman with Neovim:

My Neovim config:

local M = {}

M.config = function()
  local lsp_configs = require "lspconfig.configs"

  local cmd
  cmd = { "marksman", "server" }

  lsp_configs.marksman = {
    default_config = {
      cmd = cmd,
      filetypes = { "markdown" },
      root_dir = function(fname)
        local util = require "lspconfig.util"
        local root_files = {
          ".marksman.toml",
        }
        return util.root_pattern(unpack(root_files))(fname) or util.root_pattern ".git"(fname) or util.path.dirname(fname)
      end,
      settings = {},
    },
  }

  local status_ok, lsp = pcall(require, "lspconfig")
  if not status_ok then
    return
  end

  if lsp.marksman.setup then
    lsp.marksman.setup {
      on_attach = function(client, bufnr)
      end,
      on_init = require("lvim.lsp").common_on_init,
      capabilities = require("lvim.lsp").common_capabilities(),
    }
  end
end
return M

I created an empty .marksman.toml file at the root of my notebooks path.

This is a non-standard feature https://www.markdownguide.org/extended-syntax/#heading-ids
However it can be useful! For instance, docusaurus (docusaurus.io) understands custom heading IDs.

Cont to #86 . See https://pandoc.org/MANUAL.html#extension-implicit_header_references

The server never responds to the initialize request from the client, if it contains workspaceFolders: null.

From the LSP specs:

        /**
	 * The workspace folders configured in the client when the server starts.
	 * This property is only available if the client supports workspace folders.
	 * It can be `null` if the client supports workspace folders but none are
	 * configured.
	 *
	 * @since 3.6.0
	 */
	workspaceFolders?: [WorkspaceFolder](https://microsoft.github.io/language-server-protocol/specifications/lsp/3.17/specification/#workspaceFolder)[] | null;

Currently, marksman supports go-to definition for inline links with absolute paths only (absolute relative to the repo/folder root), e.g.

[title](/path-to-file.md)

We should also support relative paths (at least in terms of go-to definition/references) e.g.

[title](./path-to-file.md)

1. Have some diagnostics visible in a markdown file
2. Close the file
3. Open the same file again

Expected: Same diagnostics shown as before closing the file.

Actual: The diagnostics are not shown after re-opening the file.

Since 0.5 release neovim supports lsp. So, there is a project lsp-config where your lsp server is added as supported.

But when I try to integrate it in my workflow I always got errors. Authors of that plugin advice me to wright to you about issues. So my questions are:

1. May I open any issues about integration your lsp- server in nvim or your work is just around vsCode?
2. May you try to integrate zeta-note to nvim and wright a little tutorial if it works fine for you?

Thx for your work!)

Cont. to #41 https://github.com/jgm/commonmark-hs/blob/master/commonmark-extensions/test/bracketed_spans.md

In a document like:

# Changelog

## [14.0]
### Breaking
- foo

## [13.0]
### Breaking
- bar

The second Breaking header is reported as duplicate. I think it doesn't make sense since it's under different parent header.

I have a file like so:

# Heading 1

## Subheading 1.1

## Subheading 1.2

# Heading 2 

## Subheading 2.1

## Subheading 2.2

When I open the symbol picker I expect the ordering to be top to bottom, but it instead lists H1s first, then H2s:

The idea is to look for:

1. either .gitignore
2. or just .ignore
   at the root of the project and use the globs there to exclude dirs/files from scanning.

E.g. https://github.com/rust-lang/rust-analyzer has editor subfolder which has an ignore for node_modules. Without support for non-root ignores Marksman reads all md files from node_modules which is not needed and not helpful.

Code lenses require custom code for each LSP client.

We already have the client functionality in https://github.com/artempyanykh/marksman-vscode but not other editors.

So the plan is:

1. Add the functionality to produce code lenses with # of references that would work with marksman-vscode.
2. Add support for nvim.
3. Add support for Emacs LSP.

A continuation of #52 (comment)

The problem is that the current TOC formatting doesn't play nicely with prettier.

There is an issue in prettier prettier/prettier#10128 about this, but right now it's causing friction.

A potential workaround is to add newlines after/before <--!toc:xx--> marker.

/cc @keynmol @David-Else