Vim's "super-star" operation (search and highlight for the word under the cursor) is super-handy. But it has the limitation of searching for the exact match of the whole word.
Is it possible to have similar functionality, but w/o the enclosing word-boundary brackets?
Update: Apparently I was too quick to post the question. Looking here, there's the answer.
using g * is probably the best solution, but for new vim users it might be interesting to know how this could be done pretty much from scratch:
:nnoremap <expr> * "/".expand("<cword>")."/\<CR>"
map <expr> means that instead of mapping to a literal key sequence, the right-hand side is an expression that is evaluated every time the mapping occurs and the resulting string is treated like the RHS of a normal mapping.
expand("<cword>") returns a string containing the word under the cursor, and "\<CR>" basically just means the enter key.
So in the end we get a mapping to /<word under cursor>/enter, where the word under cursor isn't inserted until the mapping is used.
Answer is in this page: Instead of pressing *, use g*.
Related
Is there a way in Vim to paste the search register on the command-line, but without the surrounding \< brackets \>? I often find myself doing a search in a buffer, and then wanting to use the matched pattern as an argument to grep (more specifically, ack.vim).
Here's what happens if you search for foo and then enter :Ack '<C-r>/':
:Ack '\<foo\>'
This will fail to find anything. What I want instead is:
:Ack 'foo'
This is of course a simplistic example. Where this would be more valuable is with more complex search results.
I am aware of <C-r><C-w> and <C-r><C-a> for pasting the word/WORD under the cursor, and these often suffice, but not always.
<C-r>/ inserts your search pattern as you typed it: if you did /foo, the search register contains foo.
Note that <C-r>/ inserts the search pattern, not the match: if you search for foo\d, <C-r>/ will insert foo\d, not foo9.
As you found out, \< and \> are added if you used * or # in order to limit the search to whole words. Use g* and g# to not search whole words and thus avoid the \<\>.
Here is a possibly useful mapping:
nnoremap <F6> :Ack '<C-r>=expand("<cword>")<CR>'
Unless you're doing a purely literal search, there's a difference between the search pattern (a regular expression) and the (list of) matches (strings from the buffer).
If you're interested in the latter, my PatternComplete plugin offers insert-mode completion of all matches, and it can also insert the first match of the last search pattern with <C-R>&.
How to move the cursor before or after the first uppercase letter of a word in Vim?
My motivation is removing or selecting the first word of a camel-case identifier in code. For example, if the cursor is on the m character in the word camelCase, I can use the FcdtC sequence of Normal-mode commands to delete the camel prefix.
Is there a general way to jump to the next occurrence of an uppercase letter in an identifier?
In situations where approaches using only built-in Vim instruments are
preferred, the following search commands can be used.
For jumping to the next uppercase character:
/\u
For moving the cursor one character to the right of the next uppercase
character:
/\u/s+
or
/\u\zs
If one expects to use a movement like that often, one can always
define a key mapping for it as a shorthand, e.g.:
:nnoremap <leader>u /\u/s+<cr>
I don't think there is anything built-in.
As #ib. indicates, you can use a regular expression motion, but it’s not particularly easy to type. However, there is camelcasemotion plugin that adds the necessary motions, for this, as well as underscore seperated identifiers.
Updated Answer (using #ib.'s contribution)
"select from first char up to First uppercase letter ( after first char )
map ,b bv/[A-Z]<cr>h
Original Answer
Regarding jumping before and after the first uppercase letter—
You can map it if you want to.
"Before next uppercase letter
map ,A /[A-Z]<cr>l
"After next uppercase letter
map ,B /[A-Z]<cr>h
:D. Hope this helps. I'm reading your second question now.
Ok, read it. Now you can do this
bv,A
:D
I think I thought maybe "the vim way" to do it :
Vim allow us to define our own operator !
" movement mapping {
" Delete yank or change until next UpperCase
" o waits for you to enter a movement command : http://learnvimscriptthehardway.stevelosh.com/chapters/15.html
" M is for Maj (as in french)
" :<c-u>execute -> special way to run multiple normal commande in a map : learnvimscriptthehardway.stevelosh.com/chapters/16.html
onoremap M :<c-u>execute "normal! /[A-Z]\r:nohlsearch\r"<cr>
That way giving
DailyAverage.new(FooBarBaz)
If my cursor is on a (from DailyMesure) and I press dM It delete to A and give
Average.new(FooBarBaz)
It works with all command waiting for a movement (c y ........)
This snippet need to be improved because of bad highlight.
There is a phrase that I want to look for in Vim. When found, I want to delete that occurrence of the phrase. What is the easiest way to cycle through all the occurrences (via n), and delete a match one by one (I do not want to delete all of them at once).
Note: I know I can delete a certain number of characters or a number of words, but I want to specifically remove the match of my search. Is this possible?
http://vim.wikia.com/wiki/Search_and_replace
Try this search and replace:
:%s/foo/bar/gc
Change each 'foo' to 'bar', but ask for confirmation first.
Press y or n to change or keep your text.
There are 3 ways I can think of:
The way that is easiest to explain is
:%s/phrase to delete//gc
but you can also (personally I use this second one more often) do a regular search for the phrase to delete
/phrase to delete
Vim will take you to the beginning of the next occurrence of the phrase.
Go into insert mode (hit i) and use the Delete key to remove the phrase.
Hit escape when you have deleted all of the phrase.
Now that you have done this one time, you can hit n to go to the next occurrence of the phrase and then hit the dot/period "." key to perform the delete action you just performed
Continue hitting n and dot until you are done.
Lastly you can do a search for the phrase to delete (like in second method) but this time, instead of going into insert mode, you
Count the number of characters you want to delete
Type that number in (with number keys)
Hit the x key - characters should get deleted
Continue through with n and dot like in the second method.
PS - And if you didn't know already you can do a capital n to move backwards through the search matches.
1. In my opinion, the most convenient way is to search for one
occurrence first, and then invoke the following :substitute command:
:%s///gc
Since the pattern is empty, this :substitute command will look for
the occurrences of the last-used search pattern, and will then replace
them with the empty string, each time asking for user confirmation,
realizing exactly the desired behavior.
2. If it is a common pattern in one’s editing habits, one can
further define a couple of text-object selection mappings to operate
specifically on the match of the last search pattern under the cursor.
The following two mappings can be used in both Visual and
Operator-pending modes to select the text of the preceding match
of the last search pattern.
vnoremap <silent> i/ :<c-u>call SelectMatch()<cr>
onoremap <silent> i/ :call SelectMatch()<cr>
function! SelectMatch()
if search(#/, 'bcW')
norm! v
call search(#/, 'ceW')
else
norm! gv
endif
endfunction
Using these mappings one can delete the match under the cursor with
di/, or apply any other operator or visually select it with vi/.
The best way is probably to use:
:%s/phrase//gc
c asks for confirmation before each deletion. g allows multiple replacements to occur on the same line.
You can also just search using /phrase, select the next match with gn, and delete it with d.
I know that to search for a whole word in vim you need to type:
/\<word\><CR>
Now, what I would like to do is to map this behaviour to ? (as I never search backwards, and if needed I could search forward and then NN). I.e. I'd like to type:
?word<CR>
and have the same result as above (vim searches the whole word). I've been fiddling around with vim commands and mappings for some weeks now, but I'm not sure about how to accomplish this one. Thank you for any help.
Update: (insead of ? I use \ now).
I tend to use * and # (as suggested by Brian Agnew), but if you want a method that involves typing
?word<CR>
you could do something like this:
function! SearchWord(word)
let #/ = '\<' . a:word . '\>'
normal n
endfunction
command! -nargs=1 SearchWord call SearchWord(<f-args>)
nmap ? :SearchWord
Note there is a space after SearchWord on the last line.
Explanation:
The mapping will make ? open up a command prompt and type SearchWord (including the space). The command makes SearchWord myword do the equivalent of call SearchWord('myword') (i.e. it puts the quotes round the argument in order to make it into a string). The function sets the search register #/ equal to your word surrounded by \< and \> and then does a normal-mode n to find the next instance of the contents of the search register.
Of course you lose the benefits of incremental searching if you do this, but hopefully it's useful anyway.
You can search for words under the cursor using * and # (forwards and back). g* and g# will do the same but finding words containing what's under your cursor initially.
Obviously (!) you need to have found the first instance already for this to work, but it's very effective for successive searches.
Al's solution is very nice, but wouldn't it be simpler to just enter the \<\> pattern, then move the cursor back twice? This works for me:
nmap ? /\<\><Left><Left>
This way you still get incremental search (kinda).
I know that when you define a function in vim, you can use the range keyword so that users can say:
:-5,+5call MyFunction()
And then your function gets a:firstline and a:lastline.
What I want is something similar, but more like the traditional vi way of combining a command with a movement, so that 'd ' deletes a space, 'dw' deletes a word, 'd2w' deletes two words, 'd2j' deletes three lines, etc. Assuming my function gets mapped to some input-mode character sequence, is there any way to make it accept similar variable-length inputs, and then modify that text?
Just to be a little more clear, suppose I want to define a function to wrap <b> tags around existing text. We'll say that function is mapped to ;b. I want users to be able to say ';bw' to bold one word, or ';bf.' to bold everything to the end of the sentence, or whatever, with all the flexibility that vim provides to built-in commands.
If I understand what you're asking, then all you do is include the char argument in your mapping. For example:
map d :call MyFunction('d')<cr>
map dw :call MyFunction('dw')<cr>
map d2w :call MyFunction('d2w')<cr>
" Of course these would be terrible mappings because they
" are already mapped to important Vim functions
The way mappings work is that if you "overspecify" a char like 'd' above so that it is usable either by itself or as a prefix for longer command, Vim will wait briefly (for timeoutlen)after you press 'd' to see if you're going to press another character. (This depends on thetimeout option being set to true, which is the default.) If you don't press another character, Vim will execute the 'd' mapping. If you do it will call the more complex mapping. See :h map-commands generally and :h map-typing for more.
Note: After your clarification I think what you want is to create a custom 'operator' function that you can use to operate on buffer areas defined by Vim motions. See :h map-operator for info on how to do this.