Perl regex multiline match without dot

Question

There are numerous questions on how to do a multiline regex in Perl. Most of them mention the s switch that makes a dot match a newline. However, I want to match an exact phrase (so, not a pattern) and I don't know where the newlines will be. So the question is: can you ignore newlines, instead of matching them with .?

MWE:

$pattern = "Match this exact phrase across newlines";

$text1 = "Match\nthis exact\nphrase across newlines";
$text2 = "Match this\nexact phra\nse across\nnewlines";

$text3 = "Keep any newlines\nMatch this exact\nphrase across newlines\noutside\nof the match";

$text1 =~ s/$pattern/replacement text/s;
$text2 =~ s/$pattern/replacement text/s;
$text3 =~ s/$pattern/replacement text/s;

print "$text1\n---\n$text2\n---\n$text3\n";

I can put dots in the pattern instead of spaces ("Match.this.exact.phrase") but that does not work for the second example. I can delete all newlines as preprocessing but I would like to keep newlines that are not part of the match (as in the third example).

Desired output:

replacement text
---
replacement text
---
Keep any newlines
replacement text
outside
of the match

Most of the time, you are treating newlines as spaces. Then there's the one time you want to ignore it. Doing either is easy. Doing both is next to impossible. — ikegami, May 24 '16 at 14:09

score 3 · Answer 1 · answered May 24 '16 at 13:56

3

Just replace the literal spaces with a character class that matches a space or a newline:

$pattern = "Match[ \n]this[ \n]exact[ \n]phrase[ \n]across[ \n]newlines";

Or, if you want to be more lenient, use \s or \s+ instead, since \s also matches newlines.

answered May 24 '16 at 13:56

Mark Reed

91,912
16
138
175

how about my second example (`phra\nse`)? – Marijn May 24 '16 at 14:04

ikegami · Answer 2 · 2016-05-24T14:49:26.413

3

Most of the time, you are treating newlines as spaces. If that's all you wanted to do, all you'd need is

$text =~ s/\n/ /g;
$text =~ /\Q$text_to_find/    # or $text =~ /$regex_pattern_to_match/

Then there's the one time you want to ignore it. If that's all you wanted to do, all you'd need is

$text =~ s/\n//g;
$text =~ /\Q$text_to_find/    # or $text =~ /$regex_pattern_to_match/

Doing both is next to impossible if you have a regex pattern to match. But you seem to want to match literal text, so that opens up some possibilities.

( my $pattern = $text_to_find )
   =~ s/(.)/ $1 eq " " ? "[ \\n]" : "\\n?" . quotemeta($1) /seg;
$pattern =~ s/^\\n\?//;
$text =~ /$pattern/

edited May 24 '16 at 14:49

answered May 24 '16 at 14:38

ikegami

367,544
15
269
518

Tested. Made a couple of fixes. – ikegami May 24 '16 at 14:50
I think that strips a newline just before the desired match – ysth May 24 '16 at 15:06

score 2 · Accepted Answer · answered May 24 '16 at 14:23

2

It sounds like you want to change your "exact" pattern to match newlines anywhere, and also to allow newlines instead of spaces. So change your pattern to do so:

$pattern = "Match this exact phrase across newlines";
$pattern =~ s/\S\K\B/\n?/g;
$pattern =~ s/ /[ \n]/g;

answered May 24 '16 at 14:23

ysth

96,171
6
121
214

Note: This assumes `$pattern` is not a regex pattern, but a string literal to find exactly. – ikegami May 24 '16 at 14:39
This won't match is `$pattern = "*"` and `$text = "*"`. – ikegami May 24 '16 at 14:41
This won't match if `$pattern = "foo-bar"` and `$text = "foo\n-bar"`. – ikegami May 24 '16 at 14:50
@ikegami I was assuming no regex metacharacters, yes. But not thinking about other non-word chars. Use ikegami's answer instead – ysth May 24 '16 at 15:08

score 1 · Answer 4 · answered May 24 '16 at 14:20

It certainly is ugly, but it works:

M\n?a\n?t\n?c\n?h\st\n?h\n?i\n?s\se\n?x\n?a\n?ct\sp\n?h\n?r\n?a\n?s\n?e\sa\n?c\n?r\n?o\n?s\n?s\sn\n?e\n?w\n?l\n?i\n?n\n?e\n?s

For every pair of letters inside a word, allow a newline between them with \n?. And replace each space in your regex with \s.

May not be usable, but it gets the job done ;)

Check it out at regex101.

Perl regex multiline match without dot

4 Answers4