Logo Questions Linux Laravel Mysql Ubuntu Git Menu
 

vim, switching between files rapidly using vanilla Vim (no plugins)

Tags:

vim

I understand that limiting myself to vanilla Vim (not using plugins) limits the power of the editor, but as I switch between different machines frequently, it is often too much trouble to move my environment around everywhere. I want to just stay in vanilla Vim.

Something that holds me back is the ability to quickly switch between files. I (believe at least) have a good understanding of buffers, windows, tabs, as well as netrw (Vex, Ex, etc).

But in an editor such as Sublime Text, I can just type ctrl-p and instantly I am at the file.

I know that I can drop down to the shell, but I wonder if there are any other "hidden" secrets to rapidly switching between files in Vim based off more than just the filename.

like image 540
Jonathan.Brink Avatar asked Apr 18 '13 12:04

Jonathan.Brink


People also ask

How do I switch between files in Vim?

Just remember these commands to easily switch between buffers: :bf # Go to first file. :bl # Go to last file :bn # Go to next file. :bp # Go to previous file. :b number # Go to n'th file (E.g :b 2) :bw # Close current file.

What are plugins in Vim?

vim is a plugin that wraps the command-line fuzzy finder program fzf, allowing you to use it directly within Vim. Fzf is a fast and portable fuzzy finder application written in Go. It is a requirement for the Vim plugin, so make sure you install it first.

How do I use Vim plugins?

Starting with Vim version 8, you can install plugins without the need for a package manager by using the default package management tool. You can place Vim plugins in the ~/. vim/pack/vendor/start/plugin_name directory. Note that the plugin_name folder name will vary from plugin to plugin.


2 Answers

The closest equivalent ("closest", not "exact") to ST2's Ctrl+P is a plugin called, get ready… CtrlP. There are other similar plugins like Command-T or FuzzyFinder.

I use CtrlP and I love it but I wholeheartedly support your decision to go "plugin-free". It's not the easiest way to go but it will pay off in the long run.


Opening files

The most basic way to open a file is :e /path/to/filename. Thankfully, you get tab-completion and wildcards: the classic * and a special one, **, which stands for "any subdirectory".

Combining all of that, you can do:

:e **/*foo<Tab> 

to choose from all the files containing foo in their name under the working directory or:

:e **/*foo/*bar<Tab> 

to choose from all the files containing bar in their name under any subdirectory containing foo in its name, anywhere under the working directory.

Of course, that works for :tabe[dit], :sp[lit] and :vs[plit], too.

Those commands are limited to one file, though. Use :next to open multiple files:

:next **/*.js 

and take a look at :help arglist.


Jumping between buffers

:b[uffer] is the basic buffer-switching command:

:b4         " switch to buffer number 4 :bn         " switch to next buffer in the buffer list :bp         " switch to previous buffer in the buffer list :bf         " switch to first buffer in the buffer list :bl         " switch to last buffer in the buffer list :b foo<Tab> " switch by buffer name with tab-completion :b#         " switch to the alternate file 

Note that many of these commands and their relatives accept a count.

The :ls command shows you a list of loaded buffers. It is a bit "special", though: buffers are assigned a number when they are created so you can have a list that looks like 1 2 5 if you delete buffers. This is a bit awkward, yes, and that makes switching to a buffer by its number a bit too troublesome. Prefer switching by partial name, :b foo<Tab> or cycling, :bn :bp.

Anyway, here is a cool mapping that lists all loaded buffers and populates the prompt for you, waiting for you to type the number of a buffer and press <enter>:

nnoremap gb :ls<CR>:b<Space> 

With this mapping, switching to another buffer is as simple as:

gb (quickly scanning the list) 3<CR> 

or:

gb (quickly scanning the list) foo<tab><CR> 

The idea comes from this image taken from Bairui's collection of Vim infographics:

Flying vs cycling

Vim also has <C-^> (or <C-6> on some keyboards)—the normal mode equivalent of :b#—to jump between the current buffer and the previous one. Use it if you often alternate between two buffers.

Read all about buffers in :help buffers.


Go to declaration

Within a file, you can use gd or gD.

Within a project, Vim's "tags" feature is your friend but you'll need an external code indexer like ctags or cscope. The most basic commands are :tag foo and <C-]> with the cursor on a method name. Both tools are well integrated into Vim: see :help tags, :help ctags and :help cscope.

For what it's worth, I use tag navigation extensively to move within a project (using CtrlP's :CtrlPTag and :CtrlPBufTag commands, mostly, but the buit-in ones too) and my favorite "generic" buffer switching method is by name.


Deploying your config

A lot of Vim users put their config under version control which makes it very quick and easy to install your own config on a new machine. Think about it.


EDIT

A few months ago, I had to work on a remote machine with an outdated Vim. I could have installed a proper Vim and cloned my own beloved config but I decided to travel light, this time, in order to "sharpen the saw". I quickly built a minimalist .vimrc and revisited a couple of half forgotten native features. After that gig, I decided CtrlP wasn't that necessary and got rid of it: native features and custom mappings are not as sexy but they get the job done without much dependencies.


Juggling with files

set path=.,** nnoremap <leader>f :find * nnoremap <leader>s :sfind * nnoremap <leader>v :vert sfind * nnoremap <leader>t :tabfind * 

:find is a truly great command as soon as you set path correctly. With my settings, ,ffoo<Tab> will find all the files containing foo under the current directory, recursively. It's quick, intuitive and lightweight. Of course, I benefit from the same completion and wildcards as with :edit and friends.

To make the process even quicker, the following mappings allow me to skip entire parts of the project and find files recursively under the directory of the current file:

nnoremap <leader>F :find <C-R>=expand('%:h').'/*'<CR> nnoremap <leader>S :sfind <C-R>=expand('%:h').'/*'<CR> nnoremap <leader>V :vert sfind <C-R>=expand('%:h').'/*'<CR> nnoremap <leader>T :tabfind <C-R>=expand('%:h').'/*'<CR> 

WARNING! The path option is extremely powerful. The value above—.,**—works for me, as a default fallback value. In the real world, the exact value of the option will differ from project/language/framework/workflow to project/language/framework/workflow, so the proper value depends entirely on your needs. Don't blindly copy that line and expect it to solve all your problems.


Juggling with buffers

set wildcharm=<C-z> nnoremap <leader>b :buffer <C-z><S-Tab> nnoremap <leader>B :sbuffer <C-z><S-Tab> 

The mappings above list the available buffers in the "wildmenu" with an empty prompt, allowing me to either navigate the menu with <Tab> or type a few letters and <Tab> again to narrow down the list. Like with the file mappings above, the process is quick and almost friction-less.

nnoremap <PageUp>   :bprevious<CR> nnoremap <PageDown> :bnext<CR> 

Those mappings speak for themselves.


Juggling with tags

nnoremap <leader>j :tjump / 

This mapping uses regex search instead of whole word search so I can do ,jba<Tab> to find tag foobarbaz().

Yes, fuzzy matching is addictive but you can be just as productive without it. And for a fraction of the cost.


MORE EDIT

A couple of additional tips/tricks…


Wildmenu options

The "wildmenu", enabled with set wildmenu, makes file/buffer navigation easier. Its behavior is governed by a bunch of options that are worth investigating:

wildmode tells Vim how you want the "wildmenu" to behave:

set wildmode=list:full 

wildignore filters out all the cruft:

set wildignore=*.swp,*.bak set wildignore+=*.pyc,*.class,*.sln,*.Master,*.csproj,*.csproj.user,*.cache,*.dll,*.pdb,*.min.* set wildignore+=*/.git/**/*,*/.hg/**/*,*/.svn/**/* set wildignore+=tags set wildignore+=*.tar.* 

wildignorecase allows you to search for foo and find Foo:

set wildignorecase 

File marks

augroup VIMRC   autocmd!    autocmd BufLeave *.css  normal! mC   autocmd BufLeave *.html normal! mH   autocmd BufLeave *.js   normal! mJ   autocmd BufLeave *.php  normal! mP augroup END 

I recently found this gem in someone else's ~/.vimrc. It creates a file mark at the exact position of the cursor whenever you leave a buffer so that, wherever you are, 'J jumps to the latest JavaScript buffer you edited. Awesome.

like image 54
romainl Avatar answered Sep 23 '22 15:09

romainl


The answer depends a lot on your preferences and circumstances. Some examples:

  • If it's mostly two files (e.g. a C header and implementation file), <C-^> is very handy. In general, the alternate file is an important concept.
  • If you use a large editor window, window :splits turn the problem of locating a buffer from locating the window (once you've got all buffers opened). You can use [N]<C-w><C-w> to quickly switch to it.
  • If you can memorize (a few) buffer numbers, the :[N]b[uffer] and :[N]sb[uffer] commands are quite handy; :ls tells you the numbers.

Plugins (or at least custom mappings) can improve things a lot, and there's a whole variety on this topic on vim.org. There are various mechanisms to distribute your config (Pathogen + GitHub, Dropbox, ...), or you could remotely edit server files through the netrw plugin that ships with Vim.

like image 31
Ingo Karkat Avatar answered Sep 23 '22 15:09

Ingo Karkat