CODEWITHSUNDEEP
phpregexpcre
Yrtiop
Yrtiop

Reputation: 51

Multi line negative lookahead

I'm not really good with regex (i'm on this one for hours) and I struggle to replace all empty lines between 2 identifier ("{|" and "|}")

My regex look like that (sorry for your eyes) : (\{\|)((?:(?!\|\}).)+)(?:\n\n)((?:(?!\|\}).)+)(\|\})

Demo

It works, but it delete only the last empty line, can you help me to make it work with all the empty lines ?

I tryed to add a negative lookahead on \n\n with a repeating group on everything but it did not work.

Upvotes: 1

Views: 457

Answers (3)

Casimir et Hippolyte
Casimir et Hippolyte

Reputation: 89557

Several ways:

The \G based pattern: (only one pattern is needed)

$txt = preg_replace('~ (?: \G (?!\A) | \Q{|\E ) [^|\n]*+ (?s: (?! \Q|}\E | \n\n) . [^|\n]*)*+ \n \K \n+ ~x', '', $txt);

The \G matches the start of the string or the position in the string after the last successful match. This ensures that several matches are contigous.

What I call a \G based pattern can be schematized like that:

(?: \G position after a successful match | first match beginning ) reach the target \K target

The "reach the target" part is designed to never match the closing sequence |}. So once the last target is found, the \G part will fail until the first match part succeeds again.

~ 
### The beginning
(?:
    \G (?!\A) # contigous to a successful match
  |
    \Q{|\E # opening sequence
           #; note that you can add `[^{]* (*SKIP)` before to quickly avoid 
           #; all failing positions

           #; note that if you want to check that the opening sequence is followed by 
           #; a closing sequence (without an other opening sequence), you can do it
           #; here using a lookahead
)

### lets reach the target
#; note that all this part can also be written like that `(?s:(?!\|}|\n\n).)*`
#; or `(?s:[^|\n]|(?!\|}|\n\n).)*`, but I choosed the unrolled pattern that is
#; more efficient.

[^|\n]*+ # all that isn't a pipe or a newline

# eventually a character that isn't the start of |} or \n\n
(?s:   
    (?! \Q|}\E | \n\n ) # negative lookahead
    . # the character
    [^|\n]*
)*+
#; adding a `(*SKIP)` here can also be usefull if there's no more empty lines
#; until the closing sequence

### The target

\n \K \n+ # the \K is a conveniant way to define the start of the returned match
          # result, this way, only \n+ is replaced (with nothing)
~x

or preg_replace_callback: (more simple)

$txt = preg_replace_callback('~\Q{|\E .*? \Q|}\E~sx', function ($m) {
    return preg_replace('~\n+~', "\n", $m[0]);
}, $txt);

demos

Upvotes: 3

blhsing
blhsing

Reputation: 106553

You can use a positive lookahead pattern to ensure that a matching blank line is followed by |}, but also use a negative lookahead pattern to ensure that none of the characters between the blank line and the |} is the starting position of a {|:

\n{2,}(?=(?:(?!\{\|).)*?\|\})

Demo: https://regex101.com/r/oWfkg1/8

Upvotes: 2

MonkeyZeus
MonkeyZeus

Reputation: 20737

If you use:

(?<={\|)(\n{2,}|(\r\n){2,}|\s+)(?=\|})

Then it will match new lines and empty space found between {| and |}

Upvotes: 1

Related Questions