Start anew

[msysgit.git] / share / vim / vim58 / doc / tips.txt

*tips.txt*      For Vim version 5.8.  Last change: 2000 Feb 09


                  VIM REFERENCE MANUAL    by Bram Moolenaar


Tips and ideas for using Vim                            *tips*

Editing C programs                              |C-editing|
Finding where identifiers are used              |ident-search|
Editing local HTML files (WWW)                  |html-editing|
Editing paragraphs without a line break         |edit-no-break|
Switching screens in an xterm                   |xterm-screens|
Scrolling in Insert mode                        |scroll-insert|
Smooth scrolling                                |scroll-smooth|
Correcting common typing mistakes               |type-mistakes|
Counting words, lines, etc.                     |count-items|
Restoring the cursor position                   |restore-position|
Renaming files                                  |rename-files|
Speeding up external commands                   |speed-up|
Useful mappings                                 |useful-mappings|
Compressing the help files                      |gzip-helpfile|
Hex editing                                     |hex-editing|
Executing shell commands in a window            |shell-window|
Using <> notation in autocommands               |autocmd-<>|

==============================================================================
Editing C programs                                      *C-editing*

There are quite a few features in Vim to help you edit C program files.  Here
is an overview with tags to jump to:

|C-indenting|           Automatically set the indent of a line while typing
                        text.
|=|                     Re-indent a few lines.
|format-comments|       Format comments.

|:checkpath|            Show all recursively included files.
|[i|                    Search for identifier under cursor in current and
                        included files.
|[_CTRL-I|              Jump to match for "[i"
|[I|                    List all lines in current and included files where
                        identifier under the cursor matches.
|[d|                    Search for define under cursor in current and included
                        files.

|CTRL-]|                Jump to tag under cursor (e.g., definition of a
                        function).
|CTRL-T|                Jump back to before a CTRL-] command.
|:tselect|              Select one tag out of a list of matching tags.

|gd|                    Go to Declaration of local variable under cursor.
|gD|                    Go to Declaration of global variable under cursor.

|gf|                    Go to file name under the cursor.

|%|                     Go to matching (), {}, [], /* */, #if, #else, #endif.
|[/|                    Go to previous start of comment.
|]/|                    Go to next end of comment.
|[#|                    Go back to unclosed #if, #ifdef, or #else.
|]#|                    Go forward to unclosed #else or #endif.
|[(|                    Go back to unclosed '('
|])|                    Go forward to unclosed ')'
|[{|                    Go back to unclosed '{'
|]}|                    Go forward to unclosed '}'

|v_ab|                  Select "a block" from "[(" to "])", including braces
|v_ib|                  Select "inner block" from "[(" to "])"
|v_aB|                  Select "a block" from "[{" to "]}", including brackets
|v_iB|                  Select "inner block" from "[{" to "]}"

==============================================================================
Finding where identifiers are used                      *ident-search*

You probably already know that |tags| can be used to jump to the place where a
function or variable is defined.  But sometimes you wish you could jump to all
the places where a function or variable is being used.  This is possible in
two ways:
1. Using "grep" and quickfix commands.  This should work on most Unix systems,
   but can be slow (it reads all files) and only searches in one directory.
2. Using ID utils.  This is fast and works in multiple directories.  It uses a
   database to store locations.  You will need some additional programs for
   this to work.  And you need to keep the database up to date.

1. Using grep.

Add one long line to your .vimrc:
>  map _g :let efsave=&ef<Bar>let &ef=tempname()<Bar>exe ':!grep -n -w "<cword>" *.[cChH] *.cc *.cpp *.txt >'.&ef<CR>:cf<CR>:exe ":!rm ".&ef<CR>:let &ef=efsave<Bar>unlet efsave<CR><CR>:cc<CR>

NOTE: This requires that the '<' flag is excluded from 'cpoptions'.

That's all!  You can use this by placing the cursor on any identifier, and
hitting "_g".  Then go to further matches with ":cn".  ":cc" and ":cp" can be
used to move around in the list of matches.  Use ":cl" to see all matches.
Also see |quickfix|.

This depends on the format of what grep produces and 'errorformat'.  It should
work with the default settings.

Not all versions of "grep" accept the "-w" argument.  You can try if replacing
        -w "<cword>"
with this
>       "\<<cword>\>"
works for you.  A drawback is often that the '_' character is treated like a
non-word character.  Thus "msg" will also match "msg_row".


2. With a combination of Vim and the GNU id-tools.

This is still is a rather simple tool, but it works.

What you need:
- Vim 5.0 or later.
- The GNU id-tools installed (mkid is needed to create ID and lid is needed to
  use the macros).
- An identifier database file called "ID" in the current directory.  You can
  create it with the shell command "mkid file1 file2 ..".

Put this in your .vimrc:
>       map _u :call ID_search()<Bar>execute "/\\<" . g:word . "\\>"<CR>
>       map _n :n<Bar>execute "/\\<" . g:word . "\\>"<CR>
>
>       function ID_search()
>         let g:word = expand("<cword>")
>         let x = system("lid --key=none ". g:word)
>         let x = substitute(x, "\n", " ", "g")
>         execute "next " . x
>       endfun

To use it, place the cursor on a word, type "_u" and vim will load the file
that contains the word.  Search for the next ocurance of the word in the same
file with "n".  Go to the next file with "_n".

This has been tested with id-utils-3.2 (which is the name name of the id-tools
archive file on your closest gnu-ftp-mirror).

[the idea for this comes from Andreas Kutschera]

==============================================================================
Editing local HTML files (WWW)                          *html-editing*

Vim has some features which can help simplify the creation and maintenance of
HTML files, provided that the files you are editing are available on the local
file system.  The "]f", "gf" and "CTRL-W f" commands can be used to jump to
the file whose name appears under the cursor, thus not only checking that the
link is valid (at least the file name part of the URL) but also providing a
quick and easy way to edit many related HTML pages at once.  |gf|
A set of macros to help with editing html can be found on the Vim pages. |www|

If you want to view your HTML file, after writing a new version with Vim, have
a look at the "atchange" program.  It can be used to automatically start a
viewer when the last-changed date/time of a file changes.  You can find info
at: http://www-lmmb.ncifcrf.gov/~toms/atchange.html

==============================================================================
Editing paragraphs without a line break                 *edit-no-break*

If you are typing text in Vim that will later go to a word processor, it is
useful to keep a paragraph as a single line.  To make this more easy:
- set 'wrap' on, to make lines wrap                             |'wrap'|
- set 'linebreak' on, to make wrapping happen at a blank        |'linebreak'|
- set 'textwidth' to zero, to avoid linebreaks                  |'tw'|
- use "gk" and "gj" to move one screen line up or down          |gj|

You might also want to set 'showbreak' to recognize a line that is wrapped:
>       :set showbreak=>>>

==============================================================================
Switching screens in an xterm           *xterm-screens* *xterm-save-screen*

(From comp.editors, by Juergen Weigert, in reply to a question)

:> Another question is that after exiting vim, the screen is left as it
:> was, i.e. the contents of the file I was viewing (editting) was left on
:> the screen. The output from my previous like "ls" were lost,
:> ie. no longer in the scrolling buffer. I know that there is a way to
:> restore the screen after exiting vim or other vi like editors,
:> I just don't know how. Helps are appreciated. Thanks.
:
:I imagine someone else can answer this.  I assume though that vim and vi do
:the same thing as each other for a given xterm setup.

They not necessarily do the same thing, as this may be a termcap vs.
terminfo problem. You should be aware that there are two databases for
describing attributes of a particular type of terminal: termcap and
terminfo. This can cause differences when the entries differ AND when of
the programs in question one uses terminfo and the other uses termcap
(also see |+terminfo|).

In your particular problem, you are looking for the control sequences
^[[?47h and ^[[?47l. These switch between xterms alternate and main screen
buffer. As a quick workaround a command sequence like
>       echo -n "^[[?47h"; vim ... ; echo -n "^[[?47l"
may do what you want. (My notation ^[ means the ESC character, further down
you'll see that the databases use \E instead).

On startup, vim echoes the value of the termcap variable ti (terminfo:
smcup) to the terminal. When exiting, it echoes te (terminfo: rmcup). Thus
these two variables are the correct place where the above mentioned control
sequences should go.

Compare your xterm termcap entry (found in /etc/termcap) with your xterm
terminfo entry (retrieved with /usr/5bin/infocmp -C xterm). Both should
contain entries similar to:
>       :te=\E[2J\E[?47l\E8:ti=\E7\E[?47h:

PS: If you find any difference, someone (your sysadmin?) should better check
    the complete termcap and terminfo database for consistency.

NOTE 1: If you recompile Vim with SAVE_XTERM_SCREEN defined in feature.h, the
builtin xterm will include the mentioned "te" and "ti" entries.

NOTE 2: If you want to disable the screen switching, and you don't want to
change your termcap, you can add these lines to your .vimrc:
>       :set t_ti= t_te=

==============================================================================
Scrolling in Insert mode                                *scroll-insert*

If you are in insert mode and you want to see something that is just off the
screen, you can use CTRL-X CTRL-E and CTRL-X CTRL-Y to scroll the screen.
                                                |i_CTRL-X_CTRL-E|

To make this easier, you could use these mappings:
>       :inoremap <C-E> <C-X><C-E>
>       :inoremap <C-Y> <C-X><C-Y>
(Type this literally, make sure the '<' flag is not in 'cpoptions').
You then lose the ability to copy text from the line above/below the cursor
|i_CTRL-E|.

Also consider setting 'scrolloff' to a larger value, so that you can always see
some context around the cursor.  If 'scrolloff' is bigger than half the window
height, the cursor will always be in the middle and the text is scrolled when
the cursor is moved up/down.

==============================================================================
Smooth scrolling                                        *scroll-smooth*

If you like the scrolling to go a bit smoother, you can use these mappings:
>       :map <C-U> <C-Y><C-Y><C-Y><C-Y><C-Y><C-Y><C-Y><C-Y><C-Y><C-Y><C-Y><C-Y><C-Y><C-Y><C-Y><C-Y>
>       :map <C-D> <C-E><C-E><C-E><C-E><C-E><C-E><C-E><C-E><C-E><C-E><C-E><C-E><C-E><C-E><C-E><C-E>

(Type this literally, make sure the '<' flag is not in 'cpoptions').

==============================================================================
Correcting common typing mistakes                       *type-mistakes*

When there are a few words that you keep on typing in the wrong way, make
abbreviations that correct them.  For example:
>       :ab teh the
>       :ab fro for

==============================================================================
Counting words, lines, etc.                             *count-items*

To count how often any pattern occurs in a buffer, set 'report' to 0, and use
the substitute command to replace the pattern with itself.  The reported
number of substitutions is the number of items.  Examples:

>       :set report=0
>       :%s/./&/g               characters
>       :%s/\i\+/&/g            words
>       :%s/^//g                lines
>       :%s/the/&/g             "the" anywhere
>       :%s/\<the\>/&/g         "the" as a word

You might want to reset 'hlsearch' or do ":nohlsearch".

==============================================================================
Restoring the cursor position                           *restore-position*

Sometimes you want to write a mapping that makes a change somewhere in the
file and restores the cursor position, without scrolling the text.  For
example, to change the date mark in a file:
>   map <F2> msHmtgg/Last [cC]hange:\s*/e+1<CR>"_D"=strftime("%Y %b %d")<CR>p'tzt`s

Breaking up saving the position:
        ms      store cursor position in the 's' mark
        H       go to the first line in the window
        mt      store this position in the 't' mark

Breaking up restoring the position:
        't      go to the line previously at the top of the window
        zt      scroll to move this line to the top of the window
        `s      jump to the original position of the cursor

==============================================================================
Renaming files                                          *rename-files*

Say I have a directory with the following files in them (directory picked at
random :-):

buffer.c
charset.c
digraph.c
...

and I want to rename *.c *.bla.  I'd do it like this:

>       $ vim
>       :r! ls *.c
>       :%s/\(.*\).c/mv & \1.bla
>       :w !sh
>       :q!

==============================================================================
Speeding up external commands                           *speed-up*

In some situations, execution of an external command can be very slow.  This
can also slow down wildcard expansion on Unix.  Here are a few suggestions to
increase the speed.

If your .cshrc (or other file, depending on the shell used) is very long, you
should separate it into a section for interactive use and a section for
non-interactive use (often called secondary shells).  When you execute a
command from Vim like ":!ls", you do not need the interactive things (for
example, setting the prompt).  Put the stuff that is not needed after these
lines:

>       if ($?prompt == 0) then
>               exit 0
>       endif

Another way is to include the "-f" flag in the 'shell' option, e.g.:

>       :set shell=csh\ -f

(the backslash is needed to include the space in the option).
This will make csh completely skip the use of the .cshrc file.  This may cause
some things to stop working though.

==============================================================================
Useful mappings                                         *useful-mappings*

Here are a few mappings that some people like to use.

                                                        *map-backtick*
>       :map ' `
Make the single quote work like a backtick.  Puts the cursor on the column of
a mark, instead of going to the first non-blank character in the line.

                                                        *emacs-keys*
For Emacs-style editing on the command-line:
>       " start of line
>       :cnoremap <C-A>         <Home>
>       " back one character
>       :cnoremap <C-B>         <Left>
>       " delete character under cursor
>       :cnoremap <C-D>         <Del>
>       " end of line
>       :cnoremap <C-E>         <End>
>       " forward one character
>       :cnoremap <C-F>         <Right>
>       " recall newer command-line
>       :cnoremap <C-N>         <Down>
>       " recall previous (older) command-line
>       :cnoremap <C-P>         <Up>
>       " back one word
>       :cnoremap <Esc><C-B>    <S-Left>
>       " forward one word
>       :cnoremap <Esc><C-F>    <S-Right>

NOTE: This requires that the '<' flag is excluded from 'cpoptions'. |<>|

                                                        *format-bullet-list*
This mapping will format any bullet list.  It requires that there is an empty
line above and below each list entry.  The expression commands are used to
be able to give comments to the parts of the mapping.

>       let m =     ":map _f  :set ai<CR>"    " need 'autoindent' set
>       let m = m . "{O<Esc>"                 " add empty line above item
>       let m = m . "}{)^W"                   " move to text after bullet
>       let m = m . "i     <CR>     <Esc>"    " add space for indent
>       let m = m . "gq}"                     " format text after the bullet
>       let m = m . "{dd"                     " remove the empty line
>       let m = m . "5lDJ"                    " put text after bullet
>       execute m                             |" define the mapping

(<> notation |<>|.  Note that this is all typed literally.  ^W is "^" "W", not
CTRL-W.  You can copy/paste this into Vim if '<' is not included in
'cpoptions')

Note that the last comment starts with |", because the ":execute" command
doesn't accept a comment directly.

You also need to set 'textwidth' to a non-zero value, e.g.,
>       set tw=70

A mapping that does about the same, but takes the indent for the list from the
first line (Note: this mapping is a single long line with a lot of spaces):
>       :map _f :set ai<CR>}{a                                                          <Esc>WWmmkD`mi<CR><Esc>kkddpJgq}'mJO<Esc>j

                                                        *collapse*
These two mappings reduce a sequence of empty (;b) or blank (;n) lines into a
single line
>    :map ;b   GoZ<Esc>:g/^$/.,/./-j<CR>Gdd
>    :map ;n   GoZ<Esc>:g/^[ <Tab>]*$/.,/[^ <Tab>]/-j<CR>Gdd

==============================================================================
Compressing the help files                              *gzip-helpfile*

For those of you who are really short on disk space, you can compress the help
files and still be able to view them with Vim.  This makes accessing the help
files a bit slower and requires the "gzip" program.

(1) Compress all the help files: "gzip doc/*.txt".

(2) Edit "doc/tags" and change the ".txt" to ".txt.gz":
>       :%s=\(\t.*\.txt\)\t=\1.gz\t=

(3) Add these lines to your vimrc:
>       augroup helpgzip
>       autocmd!
>       autocmd BufReadPre  *.txt.gz set bin
>       autocmd BufReadPost *.txt.gz let ch_save = &ch|set ch=2
>       autocmd BufReadPost *.txt.gz '[,']!gunzip
>       autocmd BufReadPost *.txt.gz set nobin
>       autocmd BufReadPost *.txt.gz let &ch = ch_save|unlet ch_save
>       autocmd BufReadPost *.txt.gz execute ":doautocmd BufReadPost " . expand("%:r")
>       augroup END
>       set helpfile={dirname}/help.txt.gz

Where {dirname} is the directory where the help files are.  If you have
already included autocommands to edit ".gz" files (from
$VIMRUNTIME/vimrc_example.vim), you should omit the three "autocmd" lines.
Setting 'cmdheight' to two is to avoid the |hit-return| prompt; it is not
mandatory.  You must also make sure that $VIMRUNTIME is set to where the other
Vim files are, when they are not in the same location as the compressed "doc"
directory.  See |$VIMRUNTIME|.

==============================================================================
Executing shell commands in a window                    *shell-window*

There have been questions for the possibility to execute a shell in a window
inside Vim.  The answer: you can't!  Including this would add a lot of code to
Vim, which is a good reason not to do this.  After all, Vim is an editor, it
is not supposed to do non-editing tasks.  However, to get something like this,
you might try splitting your terminal screen or display window with the
"splitvt" program.  You can probably find it on some ftp server.  The person
that knows more about this is Sam Lantinga <slouken@cs.ucdavis.edu>.
An alternative is the "window" command, found on BSD Unix systems, which
supports multiple overlapped windows.  Or the "screen" program, found at
www.uni-erlangen.de, which supports a stack of windows.

==============================================================================
Hex editing                                     *hex-editing* *using-xxd*

When you need to edit binary data, and prefer to do that on a hexadecimal
listing, you can use the xxd program.  Use this to convert from binary to
hexadecimal:
>       :%!xxd
and use this to convert it back:
>       :%!xxd -r

If one has a particular extension that one uses for binary files (such as exe,
bin, etc), you may find it helpful to automate the process with the following
bit of autocmds for your <.vimrc>.  Change that "*.bin" to whatever
comma-separated list of extension(s) you find yourself wanting to edit:

>       " vim -b : edit binary using xxd-format!
>       augroup Binary
>         au!
>         au BufReadPre  *.bin let &bin=1
>         au BufReadPost *.bin if &bin | %!xxd
>         au BufReadPost *.bin set ft=xxd | endif
>         au BufWritePre *.bin if &bin | %!xxd -r
>         au BufWritePre *.bin endif
>         au BufWritePost *.bin if &bin | %!xxd
>         au BufWritePost *.bin set nomod | endif
>       augroup END

==============================================================================
Using <> notation in autocommands                       *autocmd-<>*

The <> notation is not recognized in the argument of an :autocmd.  To avoid
having to use special characters, you could use a self-destroying mapping to
get the <> notation and then call the mapping from the autocmd.  Example:

                                                *map-self-destroy*
> " This is for automatically adding the name of the file to the menu list.
> " It uses a self-destroying mapping!
> " 1. use a line in the buffer to convert the 'dots' in the file name to \.
> " 2. store that in register '"'
> " 3. add that name to the Buffers menu list
> " WARNING: this does have some side effects, like overwriting the
> " current register contents and removing any mapping for the "i" command.
> "
> autocmd BufNewFile,BufReadPre * nmap i :nunmap i<CR>O<C-R>%<Esc>:.g/\./s/\./\\./g<CR>0"9y$u:menu Buffers.<C-R>9 :buffer <C-R>%<C-V><CR><CR>
> autocmd BufNewFile,BufReadPre * normal i

Another method, perhaps better, is to use the ":execute" command.  In the
string you can use the <> notation by preceding it with a backslash.  Don't
forget to double the number of existing backslashes and put a backslash before
'"'.

> autocmd BufNewFile,BufReadPre * exe "normal O\<C-R>%\<Esc>:.g/\\./s/\\./\\\\./g\<CR>0\"9y$u:menu Buffers.\<C-R>9 :buffer \<C-R>%\<C-V>\<CR>\<CR>"

For a real buffer menu, user functions should be used (see |:function|), but
then the <> notation isn't used, which defeats using it as an example here.

 vim:ts=8:sw=8:tw=78: