Use code completion in vim as in sublime text - vim

I would to have a autocompletion exactly as in sublime text but in vim.
I want this behaviour :
When I start to type a word, I want to have a box which suggest completions;
In the suggested completions, I want to have last words used;
I want to have functions finded through ctags;
I want to have standard functions for the language I'm using;
I want to have snippets;
There is a lot of plugins about autocompletion for vim and I'm lost.

Currently I'm using YouCompleteMe... It's very good, though I dunno whether it can be configured to sort words based on last usage. It has general fuzzy completion capabilities that will work on any file type (you may find fuzzy completion awesome), but it also contains semantic completion for C/C++/Objective-C/Objective-C++ through libclang.
Particularly, I aways compile the lastest clang, libc++ and VIM from sources to tune and set VIM to use my environment python/ruby/etc and get a decent C++11 support. I then, compile YouCompleteMe from sources too, though I first install it through a plugin manager. My current choice is VAM, but there're others like Vundle and Pathogen.

Try this method:
for plugin in plugins
try
install plugin
read doc
test plugin
catch
remove plugin
endtry
endfor
Also, make sure you read and understand :h ins-completion before hunting down plugins.

Related

Is it possible to reload compiler plugins in Vim?

I am creating a small compiler plugin in Vim and I'm finding very slow to debug it.
It seems like there are ways to debug the errorformat but some times I just want to force Vim to reload my compiler plugin that I created in ~/.vim/after/compiler/
It doesn't look like I can just source the file as it fails because of CompilerSet. Is there any other way of reloading a compiler plugin?
Simply running :compiler {name} should reload your full compiler script.
See :help :compiler with the details of how a compiler plug-in is loaded.
An alternative is to source the compiler plug-in file directly. In order to handle the :CompilerSet command, you can either define the user command yourself before you run the plug-in file, or use a snippet in your plug-in file itself to define it if not defined already.
Many of the compiler plug-ins shipped with Vim include this snippet at the top:
if exists(":CompilerSet") != 2 " older Vim always used :setlocal
command -nargs=* CompilerSet setlocal <args>
endif
So apparently that's all you need...
But using :compiler {name} each time to fully source it should be a better approach, assuming that's a possibility to you.
(Tangentially) I find tpope’s scriptease plugin to have many benefits when working on a vim plugin (and also when writing answers to vim questions where I have to dig deep into runtime files).
Highlights:
:PP is both a pretty-printer and a repl
:Runtime helps reload files
:Breakadd is much smarter
:Vedit and cousins open up files on the runtime path

VIM uppercase autocompletion

I switched to vim after using Eclipse for many years. One feature which I really miss in vim is upper-case based autocompletion.
E.g. when I type SetMWL <ctrl-N> I would like that vim also suggests SetMaximumWordLength for camel-case and set_maximum_word_length for snake_case.
I've also looked for plugins but couldn't find one with this feature.
The easiest way to achieve your demand is using plugins and maralla/completor.vim is one of the plugins which has the functionality.
Although Vim itself doesn't have that kind of functionality by default, you can implement your own completion method using complete() function, but it is not easy.

Do I need ctags installed if I want to use YCM?

I'm new to Vim and at this moment search the ways jumping to definitions (my lang set is Go, CSS/Sass, HTML, JS). As I far I understand, ctags is the most popular solution here. However I'm going to install YCM, which is offer something similar: GoToDeclaration and GoToDefinition (unfortunately, for C-family only). I want to understand how YCM is related to ctags. So does YCM use its own mechanism to detect actual definitions or it uses ctags to provide this feature? Or in other words, do I need ctags installed if I want to use YCM?
By default, YouCompleteMe only collects identifiers from the files you visit.
If you want it to collect identifiers from your whole project then you will need Exuberant Ctags and put let g:ycm_collect_identifiers_from_tags_files = 1 in your vimrc.

How to understand code with vim?

Vim is a nice editor and I use it for many tasks.
However, when it comes to start working on a new (possibly huge) codebase, I don't feel comfortable using it to go around the code with the objective of understanding how things work.
For example, if I want to see where a C++ function is used, I can :vimgrep for that function in every **/*.cpp files and :copen the quickfix window to jump on every occurrence... of that string.
If I do the same with e.g. Eclipse (call hierarchy of a C++ method), that will not be just a string, but a C++ method defined in an object, so I will get a precise indication of the usages of that function (and not also a function with the same name defined in another class).
So the question is, how to make vim a powerful tool to analyze code?
Subquestions:
Are there any vim plugins designed for this?
Does it make sense to use vim to only analyze code? Probably external tools (e.g.: OpenGrok) can do the job?
Vim is a text editor. What you want is almost completely orthogonal to editing text and completely outside of Vim's own abilities.
However, Vim is quite good at using external tools like ctags and cscope for navigating within a project. Supposing you have created a tags file and/or a cscope.out database, Vim has a bunch of commands you can use to "jump to definition", "jump to usage", etc.: :ts[elect] foo, <C-]> over a function name and so on… You can find all the info you need in :h ctags and :h cscope.
If you are curious, GNU GLOBAL is another alternative. Another pro of cscope is that it comes with its own TUI that you can use in your shell.
The only plugin offering a cscope (+ ctags) interface I know of is CCTree which seems to be limited to C.
There are a bunch of ctags oriented plugins like TagBar or TagList you could try but note that, while ctags is limited to definitions, cscope can also do usage and callers.
You should keep in mind that these tools are code indexers: you shouldn't expect them to "understand" your code or be even remotely as precise as IDE tools. However I love Vim, I'd suggest you use a tool better suited to that task than a text editor.
When I switched from Netbeans to Vim, I felt the same as you. I missed Netbeans feature to see any function definition upon a right click. Disclaimer: I use Ruby, Javascript mainly and sometimes PHP
I tried ctags, but found it is not so accurate and not so clean. I also tried plugins Tagbar and Taglist using together with ctags. Tagbar is a bit heavy in my opinion, takes lot of CPU and memory when handling tags. Taglist is better, but the best use case is to browse long files instead of tags.
Finally I gave up using ctags.
Later I found there are better solutions specific to language. For example, for Ruby there is a plugin to show ri doc of any function within installed gems(libs), with only one key.
But I still don't use that often as my habit already changed. I like things to be lean, to be in there right places, and to be fast.
Now I feel comfortable by:
Using tmux together with Vim. Check doc and verify code in console when needed.
Use snippet plugin(Neosnippet) to store frequently used codes, methods. The snippets management in Vim is far better and flexible than any IDE I've seen.
Use brain to store more, with less touching of mouse.
Hope these help.
There is a plugin called fly.vim that is super awesome to browse source code. It makes use of a cscope database and provides simple navigation mechanism. Combine it with autotags plugin that generates and maintains cscope and ctags for a project in a central location, and you can switch between different code bases with ease.
I was using Source Insight to browse the Linux Kernel source code and when I switched over to this combination, I had nothing to complain. It may take some time and/or effort to get in speed with this setup, though. If you know how ctags and cscope work, then, probably you'll pick it up in less than an hour. But the advantages: cscope indexes code fast, vim uses cscope fast, fly.vim queries through cscope and displays it fast, and in a usable format. Plus, it maintains jump history.

Refactoring in Vim

Of course the fact that you can refactor on IDEs is priceless for many, I hardly ever do it when I am coding but I may try to do it when editing some one else's source. How do you accomplish such a trivial task across multiple files in Vim?
I found this plugin for refactoring Ruby, but how about "any" language?
I agree with the 'Vim is not an IDE' paradigm. But there are times when there isn't an IDE. Here's what I use in those situations:
Disclaimer: The ubiquity of Language Server Protocol servers, linters and fixers since I wrote this have also brought some great refactoring capabilities to Vim (and other editors). IMO they are a long way from equaling the capabilities of a purpose-built IDE (I prefer ALE and nvim-lspconfig for these kinds of features). See other answers on this question for more info!
:grep, :vimgrep, :GrepperAg, :Ggrep
Refactoring that has more to do with regular replacements I usually use :grep on my project tree and then record a macro to do the refactor - :g and :s are no brainers. Usually it'll let me quickly modify a large number of files with very little effort. Honestly, I use this method more than any other.
Depending on your workflow the built-in commands might be slow/inconvenient. If you use git, then you'll wanna use the excellent Fugitive plugin and its :Ggrep command to only search files checked into git. I also like the vim-grepper because it is search-tool-agnostic (supports ag, sift, ripgrep, etc) and speedy.
:argdo, :cdo, and :bufdo
:cdo and :argdo are handy to execute vim commands over a set of files.
command line
When it's harder to determine the list of files that need changes via :vimgrep I resort to the command line grep/find commands to more closely curate the list of files that I need to refactor. Save the list to a text file and use :e and a mashup of macro recordings to make the changes I need to make.
I find that the less rusty I keep my macro recording skills the more useful I find Vim for refactoring: feeling comfortable saving/restoring from registers, incrementing/decrementing register counter variables, cleaning/saving macro recordings to file for later use, etc.
Update
Since writing this more videocasts for the methods I describe have been published on vimcasts.org (I encourage you to watch ALL the Vimcasts!). For refactoring watch these ones:
Substitution with :Subvert
Project wide search/replace
Search multiple files with :vimgrep
Use :argdo to change multiple files
Vimgolf is also a great way to practice.
Language Server Protocol (LSP)
The Language server protocol contains the feature for smart renaming of symbols across a project:
https://microsoft.github.io//language-server-protocol/specifications/specification-3-14/#textDocument_rename
For example following language server support this:
Clangd for C++
ccls for C/C++/Objective-C
Eclipse.jdt.ls for Java
pyls (with rope) for Python
tsserver for TypeScript
Solargraph for Ruby
gopls official lsp for Go (alpha stage in Nov 2019)
texlab for LaTeX
You can find more language servers under https://langserver.org/.
Vim
A vim editor client is necessary to use them within vim. Following options exist:
LanguageClient-neovim (requires rust) suggests the mapping:
nnoremap <silent> <F2> :call LanguageClient_textDocument_rename()<CR>
coc.nvim (requires node.js) suggests the mapping:
" Remap for rename current word
nmap <leader>rn <Plug>(coc-rename)
Ale has
nnoremap <silent> <Plug>(ale_rename) :ALERename<Return>
Ale does not define any keybindings. This has to be done by the user.
vim-lsp provides following command
:LspRename
Similar to Ale no mapping is suggested. However, of course you can define one as following
nmap <leader>r <plug>(lsp-rename)
(<leader>r is to be replaced by your choice; I do not know one which most plugins agree on)
vim-lsc has a default mapping:
'Rename': 'gR'
See also YouCompleteMe which facilitates LSPs as well.
Neovim
Neovim has initial builtin support for lsp since 13.11.2019
See for common configurations of LSPs the project nvim-lspconfig which suggests <space>rn as a mapping for vim.lsp.buf.rename().
Other Refactorings
I do not know if there are plans for the LSP protocol to support more complex refactorings, such as changing class structure, adding parameters to methods/functions or moving a method to a different class. For a list of refactorings see https://refactoring.com/catalog/.
Python
For the python language following plugins provide 'smart' renaming capabilities for vim:
jedi-vim (github) <leader>r
ropevim (github) CTRL-c r r
python-mode (github) :h pymode-rope-refactoring
C-Family
Try the plugin Clighter for rename-refactoring for the c-family. It is based on clang, but there are limitations and the plugin is marked as deprecated.
Suggested mapping by Clighter is
nmap <silent> <Leader>r :call clighter#Rename()<CR>
Note, the successor plugin clighter8 has removed the renaming functionality in the commit 24927db42.
If you use neovim, you can take a look at the plugin clamp. It suggests
nmap <silent> <Leader>r :call ClampRename()<CR>
Maybe not the most elegant solution, but I found it very handy: I use ECLIM to connect VIM and Eclipse. Of course all my source code editing is done in VIM, but when it's time to refactor, one can take advantage of Eclipse's superior cababilities in this matter.
Give it a try.
I wrote this plugin for generic refactoring. It still requires many improvements. Sometime in the future I'll try to abandon ctags in favour of clang for C&C++ refactorings.
Plugin YouCompleteMe (YCM) (20k stars on github)
http://ycm-core.github.io/YouCompleteMe/#the-refactorrename-new-name-subcommand
:h RefactorRename-new-name
In supported file types, this command attempts to perform a semantic
rename of the identifier under the cursor. This includes renaming
declarations, definitions and usages of the identifier, or any other
language-appropriate action. The specific behavior is defined by the
semantic engine in use.
Similar to FixIt, this command applies automatic modifications to your
source files. Rename operations may involve changes to multiple files,
which may or may not be open in Vim buffers at the time. YouCompleteMe
handles all of this for you. The behavior is described in the
following section.
Supported in filetypes: c, cpp, objc, objcpp, cuda, java, javascript,
typescript, rust, cs
By default there is no mapping.
Plugin Factorus
There is another vim plugin dedicated for refactoring called factorus which is available on github.
Currently (2017-12), it supports the languages
c,
java, and
python.
Place cursor at name to refactor and type
gd (or gD if you're refactoring a global variable).
Then
cgn new_name esc
and
. one or more times to refactor next occurrence(s)
or
:%norm . to refactor all occurrences in the buffer at once.
I write a lot of C/C++ code in vim. The most common refactoring that I do is renaming variables, class names, etc. Usually, I use :bufdo :%s/source/dest/g to do a search/replace in files, which is almost the same as renaming provided by big IDE's.
However, in my case, I found that I usually rename similar entities, spelled in different cases (i.e CamelCase, snake_case, etc.), so I decided to write a small utility to help with this kind of "smart-case" search/replace, it is hosted here. It is a command-line utility, not a plugin for vim, I hope that you can find it useful.
Go
The tool godoctor (github) supports several refactoring capabilities
Rename
Extract Function
Extract Local Variable
Toggle var ⇔ :=
Add Godoc stubs
There is a vim plugin https://github.com/godoctor/godoctor.vim which makes them available
With cursor in thing to rename:
:Rename <newname>
Highlighting block to extract:
:Refactor extract newfunc
vim-go
Precise type-safe renaming of identifiers with :GoRename.
Language server gopls
https://github.com/golang/tools/blob/master/gopls/doc/status.md#rename
For refactoring, if you're using Unite (and you should), you can then use vim-qfreplace and make it extremely easy. Check this video that demonstrates how it works. Once your workflow is set, you can make some mappings to optimize it (instead of typing most things like in the video).
A combination of two plugins: vim-ripgrep, to find across files and put the results in the quickfix window, and quickfix-reflector to save the changes right in the quickfix window and have it automatically save each change across the files.
I would consider using the spacemacs version of emacs. It is uses the same modes and most keystrokes as Vim but has many more add-on because of it's lisp nature. If you want to program in C++ you just add the c++ layer and most of the IDE is just set up for you already. For other interpreted languages like python or bash you do not need to leave spacemacs to use them. They even have a way to run blocks of code directly within your text which works fantastic for literate programming or reproducible programming where the code and the data are in the same file. Both done as text.
Spacemacs is much more heavy handed in it's initial load but the additional stuff you can do with it is worth the few seconds of startup cost. One layer org-mode is worth checking it out. It is the best outliner, programmer, day timer / todo list I have ever used.
The CoC addon has (among other features) the ability to rename variables.
https://github.com/neoclide/coc.nvim
" Symbol renaming.
nmap <leader>rn <Plug>(coc-rename)

Resources