[SOLVED] vim regex problem :s/^\s*\(.\)\+/\\\1/g

porphyry5 · 04-20-2015, 12:05 PM

I need to prepend a backslash \ to each character in a string, excluding any leading whitespace. I believed the following would do the trick.

Code:

:s/^\s*\(.\)\+/\\\1/g

but it produces this result for the string shown

Code:

   abcde
\e

i.e. it shows only the last such replacement, not each of them.

Vim provides 2 regex engines, which you can access explicitly by prepending the search string with either '\%#=1' or '\%#=2'. Both produce the same result.

Could someone explain what I am doing wrong here

astrogeek · 04-20-2015, 12:58 PM

Quote:

Originally Posted by porphyry5

I need to prepend a backslash \ to each character in a string, excluding any leading whitespace. I believed the following would do the trick.

Code:

:s/^\s*\(.\)\+/\\\1/g

but it produces this result for the string shown

Code:

   abcde
\e

i.e. it shows only the last such replacement, not each of them.

Vim provides 2 regex engines, which you can access explicitly by prepending the search string with either '\#=1' or '\#=2'. Both produce the same result.

Could someone explain what I am doing wrong here

Your regex tells it to remove space and only capture the last character, (.), hence \e. The trailing + matches multiple characters, but only captures and replaces the last one.

Try this, assuming that you want to leave surrounding space intact:

Code:

s/\([^ \t]\)/\\\1/g

...
   \a\b\c\d\e

Or to remove leading whitespace:

Code:

s/^\s*\([^ \t]\)/\\\1/g

...
\a\b\c\d\e

(Fixed: Remove the ^, typed here in error!)

porphyry5 · 04-20-2015, 01:52 PM

Quote:

Originally Posted by astrogeek

Your regex tells it to remove space and only capture the last character, (.), hence \e. The trailing + matches multiple characters, but only captures and replaces the last one.

Try this, assuming that you want to leave surrounding space intact:

Code:

s/\([^ \t]\)/\\\1/g

...
   \a\b\c\d\e

Or to remove leading whitespace:

Code:

s/^\s*\([^ \t]\)/\\\1/g

...
\a\b\c\d\e

I tried your second suggestion, but it gave me

Code:

s/^\s*\([^ \t]\)/\\\1/g
\abcde

which seemed reasonable, because '[...]' should just access one character, so I then did

Code:

s/^\s*\([^ \t]\)\+/\\\1/g
\e

the same as my original effort, and

Code:

s/^\s*\([^ \t]\+\)/\\\1/g
\abcde

But finally this did work

Code:

s/\S/\\&/g
\a\b\c\d\e

and I guess it translates as
\S the next non-whitespace character
\\& a backslash in front of current search result
/g do every one in the line

That gave me the clue that fixing it to the start of the line was the problem, because both of these work too.

Code:

s/\([^\t ]\)/\\\1/g
s/\s*\(.\)/\\\1/g

Many thanks for your help.

astrogeek · 04-20-2015, 02:11 PM

Quote:

Originally Posted by porphyry5

More solutions

I tried your second suggestion, but it gave me

Code:

s/^\s*\([^ \t]\)/\\\1/g
\abcde

which seemed reasonable, because '[...]' should just access one character...

You are correct - I don't know why I added the caret (^) here, I did not have it in my test case - things done in haste! Remove the caret and it will work as advertized...

Glad you found a solution.