neotags.nvim

A neovim plugin that generates and highlight ctags similar to easytags.

I wrote this because I didn't like that my vim froze when opening large projects.

Much of the plugin's speed boost comes from it's ability to filter tags at runtime. Only tags found in opened buffers are supplied to Neovim for highlighting. Easytags, on the other hand, supplied all tags found recursively for a whole project. For very large projects with possibly tens of thousands of tags, this made Neovim's parser grind to a halt. With the new method, the tag list is far more manageable.

Be warned that for such large projects it can take some time to process the list of tags. Fortunately, this is done asynchronously and does not slow the editor. Furthermore, it is only updated when either a new file is opened or the buffer is saved. This infrequent time cost is made in place of constantly forcing Neovim to do the same processing.

Requirements

Neotags requires Neovim with if_python3, and psutil for Python3. If :echo has("python3") returns 1 and pip3 list | grep psutil shows the psutil package, then you're done; otherwise, see below.

You can enable Python3 interface and psutil with pip:

pip3 install neovim psutil

If tags processing is taking too long it may be advisable to use pypy3 in place of python3. This is possible by adding let g:python3_host_prog = 'pypy3' to your .vimrc. Be advised that it will be necessary to have the Neovim python module and any modules required by any other Vim plugins installed for pypy3 as well (including psutil).

Configuration

There are several configuration options to tweak the processing behavior. By default, to speed things up all standard autotools files (such as configure, Makefile.in, etc) are ignored by ctags. This behavior can be disabled with by setting g:neotags_no_autoconf to 0. Further filenames may be ignored by appending --exclude=foo to the ctags argument list (see man ctags for more information).

By default, a tags file is generated in a global directory for each directory in which source files appear. If you would like to reduce the number of tags files, and thereby conglomerate a project into one file, you may designate any directory as a "project" top directory using the NeotagsAddProject command. See below for details.

Optional C Extension

Because the original implementation in python can take a modestly long time for very large projects, the section of code that does the tag filtering has been rewritten in C. If used, this can quite dramatically decrease the waiting time, often by up to 4+ times.

It can be enabled simply by compiling it. Run make in the root git directory and the small project will be automatically configured, compiled, and installed into ~/.vim_tags/bin. The requirements are cmake, libpcre2, and of course a working C compiler. All of these should be readily available on any Unix like platform. If you want to install it elsewhere feel free to configure and build the project yourself. It is also possible to configure with autotools by running the included autogen.sh if you really prefer.

The build process can be easily automated with a package manager such as dein. Just add

call dein#add('c0r73x/neotags.nvim', {'build': 'make'})

to your .vimrc or init.nvim file and dein will handle the rest. To disable it after installing either delete the binary or add let g:neotags_bin = '' to your setup.

As usual, on Windows things are more difficult. It is possible to compile with MinGW, but the resulting binary is usually slower than the original python! If you have Visual Studio installed then it is possible to generate a project with cmake, provided that you can source a copy of libpcre2-8.lib or similar from somewhere. There are no easily available pre-compiled version of this library, so you'll either have to compile it yourself or download MinGW and use its pre-compiled version (called libpcre2-8.dll.a). Put it in the same directory as the top CMakeFiles.txt file and everything should work. There isn't any shortcut around this, unfortunately.

If all of this seems like too much bother (especially for Windows users!) then as mentioned the python version will work perfectly fine, and is probably plenty fast enough for the majority of cases.

Commands

Options

Highlight Group Names

By default group name creation is set for all the different group names of all the supported languages.

C/Cpp

Default Highlight all groups:

let g:neotags#cpp#order = 'cgstuedfpm'
let g:neotags#c#order = 'cgstuedfpm'

C highlighting is identical to Cpp just remove pp from the group name. Example, cTypeTag. With the g:neotags#cpp#order function you can restrict the highlighting to selected groups. See Speed Improvements below.

Vimscript

let g:neotags#vim#order = 'acfv'

Python

let g:neotags#python#order = 'mfc'

Ruby

let g:neotags#ruby#order = 'mfc'

Shell / Zsh

let g:neotags#sh#order = 'fa'

Java

let g:neotags#java#order = 'cimegf'

Javascript

let g:neotags#javascript#order = 'cCfmpo'

Perl

let g:neotags#perl#order = 's'

Php

let g:neotags#php#order = 'cfdi'

Tips

To use the_silver_searcher or similar applications when generating tags you can do something like this.

let g:neotags_recursive = 1

" Use this option for the_silver_searcher
let g:neotags_find_tool = 'ag -g ""'
" Or this one for ripgrep. Not both.
let g:neotags_find_tool = 'ag --files'

Speed Improvements

Also on big projects syntax highlighting may become slow. To address this you can try:

set regexpengine=1

This provides significant speed improvements. In addition you set the highlight options for your language not highlight everything but maybe only the tags your interested in the most. Example:

let g:neotags#cpp#order = 'ced'

The above will only highlight cppTypeTag, cppPreProcTag, cppEnumTag.

ptags

Neotags have support for ptags by adding let g:neotags_ctags_bin = 'ptags' to your vimrc.

Do note that if you have your own custom settings for g:neotags_ctags_args you need to prepend these with -c. Also since ptags do not support -L- the g:neotags_find_tool will be ignored. (-L was added in the latest version).

This is my setup using ptags for git repositories and ctags for other folders.

function s:neotagsCtagsCheck()
    if system('git rev-parse --is-inside-work-tree') =~# 'true'
        let g:neotags_ctags_bin = 'ptags'
        echom 'Neotags using ptags'
        let g:neotags_ctags_args = [
                    \   '-c --fields=+l',
                    \   '-c --c-kinds=+p',
                    \   '-c --c++-kinds=+p',
                    \   '-c --sort=yes',
                    \   "-c --exclude='.mypy_cache'",
                    \   '-c --regex-go=''/^\s*(var)?\s*(\w*)\s*:?=\s*func/\2/f/''',
                    \   "-c --exclude='*Makefile'",
                    \   "-c --exclude='*Makefile.in'",
                    \   "-c --exclude='*aclocal.m4'",
                    \   "-c --exclude='*config.guess'",
                    \   "-c --exclude='*config.h.in'",
                    \   "-c --exclude='*config.log'",
                    \   "-c --exclude='*config.status'",
                    \   "-c --exclude='*configure'",
                    \   "-c --exclude='*depcomp'",
                    \   "-c --exclude='*install-sh'",
                    \   "-c --exclude='*missing'",
                    \ ]
    else
        let g:neotags_ctags_bin = 'ctags'
        echom 'Neotags using ctags'
        let g:neotags_ctags_args = [
                    \   '--fields=+l',
                    \   '--c-kinds=+p',
                    \   '--c++-kinds=+p',
                    \   '--sort=yes',
                    \   "--exclude='.mypy_cache'",
                    \   '--regex-go=''/^\s*(var)?\s*(\w*)\s*:?=\s*func/\2/f/''',
                    \   "--exclude='*Makefile'",
                    \   "--exclude='*Makefile.in'",
                    \   "--exclude='*aclocal.m4'",
                    \   "--exclude='*config.guess'",
                    \   "--exclude='*config.h.in'",
                    \   "--exclude='*config.log'",
                    \   "--exclude='*config.status'",
                    \   "--exclude='*configure'",
                    \   "--exclude='*depcomp'",
                    \   "--exclude='*install-sh'",
                    \   "--exclude='*missing'",
                    \ ]
    endif
endfunction

augroup Neotags
    autocmd VimEnter * call s:neotagsCtagsCheck()
augroup END

Language conversion

The neotags_ft_conv variable is used to convert for example C++ to cpp but it can be used to convert custom filetypes to ctag filetypes.

For example this is what i use for flow

let g:neotags_ft_conv = {
            \ 'C++': 'cpp',
            \ 'C#': 'cs',
            \ 'JavaScript': 'flow',
            \ }

Note that you do need to copy the javascript neotags file neotags.vim/plugin/neotags/javascript.vim to after/plugin/neotags/flow.vim and do a replace for '#javascript' to '#flow'

Custom Rules

You can create custom rules for existing languages or new languages.

let g:neotags#[ctags language]#order = 'string with ctags kinds'
let g:neotags#[ctags language]#[ctags kind] = { 'group': 'highlight' }

For more advanced rules, check the files in neotags.vim/plugin/neotags/*.vim.

You can get the list of kinds by running ctags --list-kinds=[language].

order determents priority of the highlight by first to last (tags with the same name will use the one with higher priority). Note that only kinds in the order string will be loaded.

For example, this is what I use in typescript/tsx

In ~/.ctags

--langdef=typescript
--langmap=typescript:.ts
--langmap=typescript:+.tsx
--regex-typescript=/^[ \t]*(export([ \t]+abstract)?([ \t]+default)?)?[ \t]*class[ \t]+([a-zA-Z0-9_]+)/\4/c,classes/
--regex-typescript=/^[ \t]*(declare)?[ \t]*namespace[ \t]+([a-zA-Z0-9_]+)/\2/n,modules/
--regex-typescript=/^[ \t]*(export)?[ \t]*module[ \t]+([a-zA-Z0-9_]+)/\2/M,modules/
--regex-typescript=/^[ \t]*(export)?[ \t]*function[ \t]+([a-zA-Z0-9_]+)/\2/f,functions/
--regex-typescript=/^[ \t]*export[ \t]+(var|let|const)[ \t]+([a-zA-Z0-9_]+)/\2/v,variables/
--regex-typescript=/^[ \t]*(var|let|const)[ \t]+([a-zA-Z0-9_]+)[ \t]*=[ \t]*function[ \t]*\(\)/\2/V,varlambdas/
--regex-typescript=/^[ \t]*(export)?[ \t]*(public|protected|private)[ \t]+(static)?[ \t]*([a-zA-Z0-9_]+)/\4/m,members/
--regex-typescript=/^[ \t]*(export)?[ \t]*interface[ \t]+([a-zA-Z0-9_]+)/\2/i,interfaces/
--regex-typescript=/^[ \t]*(export)?[ \t]*type[ \t]+([a-zA-Z0-9_]+)/\2/t,types/
--regex-typescript=/^[ \t]*(export)?[ \t]*enum[ \t]+([a-zA-Z0-9_]+)/\2/e,enums/
--regex-typescript=/^[ \t]*import[ \t]+([a-zA-Z0-9_]+)/\1/I,imports/
--regex-typescript=/^[ \t]*@([A-Za-z0-9._$]+)[ \t]*/\1/d,decorator/

In vimrc

let g:neotags#typescript#order = 'cnfmoited'

let g:neotags#typescript#c = { 'group': 'javascriptClassTag' }
let g:neotags#typescript#C = { 'group': 'javascriptConstantTag' }
let g:neotags#typescript#f = { 'group': 'javascriptFunctionTag' }
let g:neotags#typescript#o = { 'group': 'javascriptObjectTag' }

let g:neotags#typescript#n = g:neotags#typescript#C
let g:neotags#typescript#f = g:neotags#typescript#f
let g:neotags#typescript#m = g:neotags#typescript#f
let g:neotags#typescript#o = g:neotags#typescript#o
let g:neotags#typescript#i = g:neotags#typescript#C
let g:neotags#typescript#t = g:neotags#typescript#C
let g:neotags#typescript#e = g:neotags#typescript#C

let g:neotags#typescript#d = g:neotags#typescript#c