Vim Macro

Myphre · 11-24-2019, 08:12 AM

Hi,
I need to edit some HTML pages and i want to create a macro in vim to make that. But i cant make properly the macro. Im lost in how to make the commands and make the macro work.

What I need is to:
(1) Locate and erase the "<";
(2) advanced one word;
(3) put "(" ;
(4) go to the next ">" and substitute this for ")";
(5) go to the next line

The commands i have seen that i can use are:
:s/<// "search for "<" and erase him
w "go to the next word
press i "enter in edition mode
insert "(" "insert the "("
press ESC " exit edition mode
:s/>/)/ " go to the next ">" and substitute this for an ")"
j "go to the next line

I have tried to just start the macro and put those commands but i receive the error " E465: :winsize requires two number arguments" and cant figure out what is happening.

I need too erase all the things that are between the "</ >", including the "</ >".

I was seeing that i can use vim scripts to make an if condition but i dont know either how to make and run the script.

Example:
Original code:

Code:

  
        <a href="index.html" class="col-3">
            <div class="verticalCenter">
                <img id="logo" src="logo.webp">
            </div>
        </a>

What I need to make:

Code:

  
        a (href="index.html" class="col-3")
            div (class="verticalCenter")
                img (id="logo" src="logo.webp")

thanks for the help

pan64 · 11-24-2019, 08:18 AM

you need to solve it without using insert mode, try to solve 1-2-3 with one single: s/<(wordspace)/$1(/
but obviously you need to fine tune it

Myphre · 11-24-2019, 08:25 AM

Quote:

Originally Posted by pan64

you need to solve it without using insert mode, try to solve 1-2-3 with one single: s/<(wordspace)/$1(/
but obviously you need to fine tune it

There is a place where I can find some examples and what each command does? I have seeing a lot of sites that try to explain, some say about pressing " i " and " Esc " while you are recording but any of them appear to not solve my problem.
Others use expressions like: 03wdei^R=2012-^R"^M^[0j that appear to use the " i " and " Esc " (^[). But, anyway, i cant properly run my macro :/

pan64 · 11-24-2019, 08:37 AM

again, forget that i and Esc, try to solve it using a regexp
word is \w\w* space is \s, so this is the correct command for 1-2-3:
:s/<$\w\w*\s$/\1(/
you can obviously add replacing > to ) too, and you will have s single command to process a line

Code:

:s/<\(\w\w*\s\)\(.*\)>/\1(\2)/

Myphre · 11-24-2019, 09:37 AM

Quote:

Originally Posted by pan64

again, forget that i and Esc, try to solve it using a regexp
word is \w\w* space is \s, so this is the correct command for 1-2-3:
:s/<$\w\w*\s$/\1(/
you can obviously add replacing > to ) too, and you will have s single command to process a line

Code:

:s/<\(\w\w*\s\)\(.*\)>/\1(\2)/

There is a place where i can find more info? Im used to with some common commands of VIM but this next step to macros appear to put a lot of new stuffs and a new pattern of thinking.
This command you put appear to work but i cant understand it properly and it erase the final of the line, so the formatting are getting messy.

As i could understand:
:s/ - search
$ - escape character and new command ?
\w - escape character, go to next word ?
\w* - escape character, all the worlds?
\s$ - escape character search ?

Im reading now this site: http://vimregex.com/

pan64 · 11-24-2019, 09:42 AM

$ and $ are grouping, the first one is group 1, the second is group 2.
\w is a word character (mor or less: a-z)
\s is a space character (space or tab).
\1 and \2 are backreferences to the groups in the search expression
* means any number of \w (or whatever located just before *)

it is more or less identical to sed, so you need to check how sed works.

Myphre · 11-24-2019, 11:41 AM

Quote:

Originally Posted by pan64

$ and $ are grouping, the first one is group 1, the second is group 2.
\w is a word character (mor or less: a-z)
\s is a space character (space or tab).
\1 and \2 are backreferences to the groups in the search expression
* means any number of \w (or whatever located just before *)

it is more or less identical to sed, so you need to check how sed works.

Those examples I have understood: its like the same as another programming languages. What i'm not is how macros in vim work, more exactly how i need to think to make them.
For the first time I'm reading about sed, what appear to be the basic of how the macros work in vim - as i understood.
Ok, but the way to think to make sed( or macro in vim) work appear to be different from what I'm used to.

In the case of the command :s/, i read in the Gnu manual for sed:
:s/regexp/replacement/flags.

In the code you sent me:
:s
/Regexp: <$\w\w*\s$$.*$>
/Replacement: \1(\2)
/Flags:

I cant understand yet the syntax or how it works, how to read this.

In the Regexp I'm reading something like : expression "<", the first group ( one word, all the words - why? in thesis I just need to search the ">" - , and space (?) ) and the second group (match everything), expression ">".
In the Replacement: Use first group - Why the "(" ?, isnt for groups? - the second group?
??? How the "(" and ")" are being put?

I have tried to make the macro, after the end, jump to the next line. But i have no success at all

pan64 · 11-24-2019, 12:58 PM

I don't understand what you are talking about. :s in vi works like sed. Both understand [almost] the same syntax. So

Code:

:  go to command mode
s/ substitute command
<  first letter is a literal < in the search expression
() is a group, but need to be escaped, therefore \( is the beginning of the [first] group
\w a single word char
\w* zero or more of \w
\s whitespace character
\) closing ) of the first group
\(.*\) second group, contains anything ( until > )
>  a literal >
/  replacement separator (= delete what was found and use this instead)
\1 put first group
(  put a (
\2 put second group
)  put a )
/  end of replacement string

The macro you trying to write is just a key sequence you want to repeat.
so you will type:

Code:

#to set macro "a"
qa:s/<\(\w\w*\s\)\(.*\)>/\1(\2)/q
# to call macro "a"
@a

and that will just replay the key sequence as you specified.
https://vim.fandom.com/wiki/Macros

Myphre · 11-24-2019, 07:22 PM

Quote:

Originally Posted by pan64

I don't understand what you are talking about. :s in vi works like sed. Both understand [almost] the same syntax. So

Code:

:  go to command mode
s/ substitute command
<  first letter is a literal < in the search expression
() is a group, but need to be escaped, therefore \( is the beginning of the [first] group
\w a single word char
\w* zero or more of \w
\s whitespace character
\) closing ) of the first group
\(.*\) second group, contains anything ( until > )
>  a literal >
/  replacement separator (= delete what was found and use this instead)
\1 put first group
(  put a (
\2 put second group
)  put a )
/  end of replacement string

The macro you trying to write is just a key sequence you want to repeat.
so you will type:

Code:

#to set macro "a"
qa:s/<\(\w\w*\s\)\(.*\)>/\1(\2)/q
# to call macro "a"
@a

and that will just replay the key sequence as you specified.
https://vim.fandom.com/wiki/Macros

Thanks for the help and sorry for something, English isn't my mother language and I read better than write in it.

I think I'm getting. I have write something weird: what i wasn't understanding was why you used the "\w" and "\w*" ( in the first group). I'm making some tests here and seeing that doesn't make any difference - apparently.

I'm not using vim now, I'm using the sed command like this site describe: https://unix.stackexchange.com/quest...tern-using-sed With vim macro the code wont go to the next line and with this i doesn't need to go the next line, the command does all I need.

But I'm not getting how to delete all the "</word>". I write this code:

Code:

 sed -i 's/<\/\(\w\)\>//' index.html
And this too:  sed -i 's/<\/\(\w\)>//' index.html

As I have understood, I'm referencing to the "</" with a word inside and that ends with ">". Like the "</div>".
But wont work as I was expecting: the code only delete the codes with "</word" and not those "</word>" - I'm running manually for each type, because the code below work:

Code:

sed -i 's/<\/div>//' index.html

pan64 · 11-25-2019, 01:12 AM

\w is exactly one character,
\w* zero or more chars
\w\w* is one ore more chars, same as:
\w+ one or more chars, but + is sometimes not understood

< and > are used as is, do not need to escape them (so \ is not required before >)

Geist · 12-04-2019, 07:38 AM

HTML with regex is generally awful.

I would go with a script here, really, not a macro, since some tags are not terminated with a </tag>, if they all were then a macro would be relatively easy,
it would be something like (I'm writing down a key sequence now, convert text in brackets to keys):

Code:

0
f<
w
"tyiw
w
"gyiw
/\/[control+r]t[enter]
dd
[control+o]
F<
x
/[control+r]g[enter]
i([escape]
f>
r)

Which you could record as a macro and would do those cases, otherwise you need a lookup for the edge cases, etc.
It's messy.