With regex, how to match misused blankspace inside tab indent

Question

Say the line is:

^<space1><space2><tab1><tab2><space3><space4><tab3><tab4><space5><space6><tab5><tab6><space7><space8>blablabla

So how to match all these <space>?

(FYI, I use call matchadd("yqi_faulty_indent", '<THE_REGEX>') to highlight it)

• With ^\t*\zs\ \+, it matches only <space1><space2>
• With ^\s*\zs\ \+, it matches only <space7><space8> due to the greedy mode I think
• With ^\s\{-}\zs\ \+, it matches only <space1><space2> due to the non-greedy mode I think
• With \v((^\s*)@<= +), it matches what I want, BUT, with lookaround regex, there's unacceptable high CPU when there are long lines like seq 1 30000 | tr '\n' ' ' | fold -w 3000 | vim -

UPDATE

According to the important hint from Vivian's answer, below is my latest solution:

• call matchadd("yqi_faulty_indent", '\v((^\s*)@80<= +)') highlights blankspaces in TAB indent
• call matchadd("yqi_faulty_indent", '\v((^\s*)@80<=\t+)') highlights TABs in blankspace indent

While as mentioned in comment, there's size limitation, looking forwards furthur better solution...

Answer 1

I would do:

/\v((^\s*\t)@<= +|(^\s*)@<= +\t@=)/

For performance reason you could do (limiting the lookahead to 80 bytes):

/\v((^\s*\t)@80<= +|(^\s*)@80<= +\t@=)/

Answer 2

I would just search for ^\t*\zs \+\ze\t

This avoids a performance-killing look-behind: match any number of leading tabs, then at least one space, then at least one more tab.

The run of space characters are wrapped with start-of-replacement \zs and end-of-replacement \ze, so that you can use this inside a s/// to remove or replace just the space characters.

If you're wanting a pattern to use in syntax highlighting, you'd want to anchor at both start-of-line and end-of-previous-match: \%(^\|\zG\)\t*\zs \+\ze\t. Unfortunately \zG is not be available in all versions of Vim. (It's the equivalent of Perl's \G.)

If you don't have that then perhaps it would be acceptable simply to highlight all whitespace between the first and last misused spaces:

syntax match BadSpace "^\t*\zs \%(\s* \)\=\ze\t"
highlight BadSpace term=inverse gui=inverse

However if replacement is your intention, I suggest using the :retab command instead.

If you really want to search for any blanks rather than just ASCII space, then it gets slightly more complicated: ^\t*[^\t]\@=\s\+\t

The key is a single-character that simultaneously matches not-a-tab and matches blank.