Garrett Hildebrand's vi Page

Regarding the vi versus emacs debate, I like this letter by Tim O'Reilly: http://www.oreilly.com/pub/a/oreilly/ask_tim/1999/unix_editor.html.

So, what will you get from reading this Web page? Well, besides the usual pointers to things written by other poeple, I have some interesting tips and tricks, and one really "geek chic" macro to share.

vi Books

A Guide to vi
Visual Editing on the UNIX System

Learning the vi editor
6th edition, November 1998
ISBN 1-56592-426-6.

Editions 1 through 5 were authored by Linda Lamb, and these are now out of print. Arnold Robbins name appears with Linda Lamb on the 6th edition's list of authors, but Tim O'Reilly helped too: "While I don't have an author credit on the cover, I actually wrote a large part of the fourth and fifth editions, including the opening paean to the virtues of vi." -- Tim O'Reilly, Jun 21 1999, in response to an "ask Tim" email.

vi Help

• William Totten's <totten@pobox.com> vi man page.
• The Vi Lovers Homepage, good for most versions of vi, including vim, etc.
• www.vi-editor.org Good links after wading through vi clones area.
• Mastering the vi editor a University of Hawaii vi tutorial by Ben Yoshino.
• another vi tutorial, by Lee Sailor and Maarten Litmaath.
• vi editor FAQ..
• Another vi editor FAQ.
• Maarten Litmaath's vi Reference, maintained by James Hu on the CS dept. Web server at WUSTL.

Vim Help

• Wikia.com Best Vim tips. Note guu and gUU!

vi Articles

Grasping more of the vi editor at http://www.itworld.com/Comp/2378/swol-0397-unix101/

Using the vi Editor at http://www.onlamp.com/pub/a/bsd/2001/10/25/FreeBSD_Basics.html

Elvis

• Elvis documentation.
• Readme and pointers to elvis.

Vim

• The main mirror, www.vim.org.
• The VIM FAQ

Using vi with other Tools

At the bash Shell prompt: set -o vi

At the Korn Shell prompt: set -o vi

At the tcsh Shell prompt: bindkey -v

To get into the vi editor to look at or modify your history, simply type <ESC>k and then you move up or down in the stack with j and k. Note that searches should be done with ? when you first go in, because you are at the bottom, not the top.

To use vi as your editor for the Pine email tool, simply go to the configuration setup (M for main menu, S for setup, C for configuration, then use w to search for 'editor') and set the editor to vi. In the .pinerc file it will appear thus:

# Specifies the program invoked by ^_ in the Composer,
# or the "enable-alternate-editor-implicitly" feature.
editor=vi

enable-alternate-editor-cmd,

enable-alternate-editor-implicitly

vi Tricks

But first I'll mention a few things. :g/search-string/ can be placed in front of several ex commands, and ex underlies vi. The two most common commands I'd use here would be m and d, with 'd' to delete lines which are matched up, and 'm' used to move the matched lines to somewhere. As an example of the latter, :g/X-/m0 will move all lines which contain 'X-' to the top of the file. But the trick here is that the 'g' may be replaced with a 'v' to cause an inverse effect. In other words, all lines which do not match are affected by the command.

Also, many people use the 'u' command to undo. But few use the 'U' command. The difference is that 'u' undoes the last change, but 'U' undoes all changes made to the current line (so long as you are still on it).

Command String	Description of Action
:v/./d or :g/^$/d	Delete all empty lines. Add match for blanks if required.
:%s/^$/^M	Add a new blank line to all existing blank lines. ^M is control-v, control-m
:v/^flort/s/^t&/	Insert a tab before any line that does _not_ with 'flort'. ^t is actually a tab character
:/flort/,/flortx/d	Delete from the first occurance of 'flort' to the first occurance of 'flortx'.
:/flort/:r filex	Find the first occurance of 'flort' and insert under that line the contents of 'filex'
:/flort/:r !date	Find the first occurance of 'flort' and insert under that line the output of the shell command 'date' (or any other shell command)
:g/./join	One of my favorite discoveries. Joins each pair of lines.
:v/^str/-1join	Created from a question by John Lenning
:v/^...........$/d	Created by John Lenning while working on Puzzle #35 ("They've got it all") of Matt Gaffney's Weekly Crossword Contest. Answers in the grid provided clues to the first and last names of 3 people who's collective names were pangrammatic. One was an ex-NHL allstar for 11 letters. John copied a list of players from a web page and pasted them into a vi session, then after removing white space used this command to delete all lines that did not have eleven letters. My version of that was to take the cleaned-up file and use "awk 'BEGIN { FS=RS } {print length, $0}' scratchfile | sort -n" (awk puts length numbers in front of each line) but I think John's is much more elegant.
:v/./.,/./-1join	Replace two or more blank lines with one blank line. :v is like :g except that the regex (which is /./ in this case, or match a character) is inverted for the global scope. Blank lines do not match.
:.,/^$/g/./.,/^$/j	This has the same scope as !} (e.g., a paragraph) but it allows the joining of all lines in it. 'j' is the same as 'join' and is an ex command. Since !} takes you out to the shell, we do this odder format to get to the ex command underlying vi.
:.,/^$/-1j	If you liked the previous :.,/^$/g/./.,/^$/j, you'll really like this shorter, alternative form, which essentially does the same thing. Adding the -1 fixes a problem the new form introduces, which is to strip out the white space as well. The minus one leaves it there. I store it in my .exrc file as map =j :.,/^$/-1j^M using control-v control-m to get the ^M.
:.,$g/./.,/^$/-1j	Here we have taken the "join paragraph lines" formula in its longer form of :.,/^$/g/./.,/^$/j and modified the file-scope from paragraph to current position to end-of-file (assuming there is a blank line at the end of the last paragraph). All paragraphs delimited by white-space lines will become one-line paragraphs separated by the white space.
:.,+2j	Join the current line with the next 2 (3 lines joined).
:g/.*/.,+2j	Like the previous join but global. The /.*/ matches all lines, and every 3 lines are joined. If you have no blank lines, you can simply use /./ instead. Inspired by a question from JJ at the University of Alcala, Spain, in August of 2006.
o%s/.<esc>x9p a/&_<esc>"add:@a	JJ again. He wanted to be able to insert a character after a particular column (which is KNOWN to exist). What this does is build a statement like '%s/........./&_' on a new open line, which is then deleted into buffer 'a' and then executed via ex as '@a'. The number 9 is arbitrary (could be any column), and the underscore could be any character or string. But the columns (its global) must exist or you get a pattern matching error. This beats using a mouse, IMHO :-). the '.<esc>x9p' part puts '.' in the buffer and then '9p' pastes it on the line 9 times (you could just type them all in).
:g/rex/s/$/str/	(where 'rex' is short for 'regex'). On any line in the file which contains the regular expression specified then append to those lines only the string 'str'. Many variations on this formula can exist and be interesting.
:.!fmt	Format the current line only (Unix)
!}fmt	Format the current paragraph, (e.g., to next whitepspace; UNIX)
!3}fmt	Format the current and next two paragraphs (or next n-1; UNIX)
:.,/regex/!fmt	Format from the current line to matching regex, say, a word (UNIX)
!Gfmt	Format everything from the current line to the end of the file (UNIX)
:%s g	Do a previous substitution on a global basis. Great for when you forget the % (which is really 1$) the first time. Doh.
:.,$s g	Do a previous substitution from the current line to the end of the file.
:s g	Do a previous substitution globally on the current line only.
:%s	Do a previous substitution on the first matching occurance of all lines, but not globally on the line.
:g/flort/s/foo/bar/g	Replace all occurances of 'foo' with 'bar' BUT ONLY on lines which contain the string 'flort'.
xp	Swap the location the character at the cursor and the one following.
ddp	Swap the order of two lines.
uu	Go to the location of the last change made. It is actually simply two undo commands, but it has that effect if you are elsewhere in the file than the last change made. This tip is from Victor Langeveld, who cautions, "'uu' doesn't work with vim: in vim you'd use 'u^R' with the default settings."
:g/.*/m0	Reverses the order of the lines in a file.
ma	Mark a location you want to come back to with marker a (or b, or c, or d...). Use in conjunction with 'a (or 'b, 'c, 'd, etc...).
'a	After moving to another part of the file, you type single-quote followed by the marker letter (a, b, c, d, etc.). to go back.
''	Typing in two single-quotes after a search or nG (where n is {1,2,3...n}) will return you to where you were before the search. Very handy for going off and double-checking something else in the file, then returning to where you are currently editing. Also, doing '''' (four single-quottes) as '' <pause> '' will allow you to toggle back-and-forth between the two locations, visually comparing. Doing an update of some kind at the searched-to location does not seem to affect this operation.
s/./\u&/g	This makes all the characters on the current line upper-case. Replacing the '\u' with \l will do the reverse. I found this on Nick Gammon's vi page, but he uses s/\<./\u&/g. I found that dropping the '\<' seemed to have no effect in vi on SunOS.
:.! tr '[a-z]' '[A-Z]'	Another way to make everything on the current line upper-case. Note, however, that this can be of the form !) or !} (i.e., !}tr '[a-z]' '[A-Z]') to change either the current sentence or the current paragraph.
. and &	Simply typing in a period '.' after doing some sort of simple change such as r or x will cause the last change to be repeated. This is very useful for changing a numeral or letter in a column of text, by doing j. over and over again (go down one line in the same column and do the last one-character change again). The '&' repeats the last substitute using s/regex/replacement/ but remember :%s g to do it globally.
d0	This is not really a trick, yet it is amazing how many people are ignorant of it. It is more of a practice. I use d0 all the time to delete text prior to my cursor position and up to the start of the line. The analog is d$, or delete until end-of-line.
dG (and dnG)	This also is not a trick, and like d0 it is amazing how many people are ignorant of it as novices. And in fact, I looked at three fairly good vi "cheat sheets" that came off of some pretty thorough vi web sites that did not mention it at all. But anyway, this will delete from the currrent line to the end of the file. Bam! Just like Emmeril adding spices to a dish. dnG deletes from the current cursor position to line 'n'. So d1G will delete to the top of the file.
ctd	... where 'd' is any delimiter. Use this for changing text between two quotes, for example. Place your cursor on the 1st character after the open quote, do ct", then type in the new text (such as a URL). Then hit <ESC> to pop out of change mode and back into edit mode. The trailing " will be preserved.
H:.,+22list	If you execute this you will first move to the top line on the current screen, then all lines on the screen will have a dollar-sign '$' placed at the end, and you will be prompted by the message, [Hit return to continue]. When you hit return, (enter key), you will be back in normal mode again. Useful for seeing if there are trailing blanks at the end of the lines. To check just the line you are on, you can do :.list or simply go to the end of the line with $.
:.= and :=	:.= gives the current line number, := shows total lines in file. So does <ctrl>G.

set showmode
set ignorecase
" set directory=/var/tmp
"
map , :n^M map =i :r $HOME/.include.headers^M
map =dd :!rm %^M^M:n^M
map K $a ^[Bi^M^[y$$pBhxi^M^[$ma^!`awc -m^Md0i(^[$a)^[kJkJ$
map =f !}fmt^M map =s :r ~/.signature^M
map =w :w^M:n^M
map =c >>d0:co.^Mk:s/./x/g^M40A x^[079lD:s/x//g^MJ
map V :w^M:!ispell -x %^M:e!^M^M