asked on

Problem matching a list with multi-line items

I'm trying to break a chunk of text containing a numbered list into individual items, but am having trouble with items that continue to the second line. I'm hoping more pairs of eyes will see what I'm doing wrong.

Data looks like this:

1. This is the first item
2. This is a longer second item that wraps down
to the next line, but it's still one item.
3. This is the last time which is also long and
wraps to the next line as well

Open in new window

Perl code looks like this:

$/ = undef;
$_ = <>;

while (/^ \d+ [ \.]* (.*?) $/gmsx) {
  print "line[$1]\n";
}

Open in new window

Current output looks like this:
l

ine[This is the first item]
line[This is a longer second item that wraps down]
line[This is the last time which is also long and ]

Open in new window

As you can see I'm missing the second line of items two and three. I've tried various combinations of options and pattern greediness to get what I want, but so far no success.

farzanj

Obviously. You don't have a number in the second line.

Try this.

$/ = undef;
$_ = <>;

while (/^ \d+ [ \.]* (.*?) $/gmsx) {
  print "line[$1]\n";
}

Open in new window

jkfrench

ASKER

farzanj,

Thanks for the reply, but I don't see any difference between your example and the code I posted.

And I realize there is no number on the second line. I was trying to get the (.*?) pattern to match across lines to pick up the rest of the item. It looks like the $ is anchoring the pattern to the end of the line, although I was trying to use the 's' option to allow . to match newline.

farzanj

Sorry, I replaced + with *

farzanj

$/ = undef;
$_ = <>;

while (/^ \d+ [ \.]* (.*?) $/gmsx) {
  print "line[$1]\n";
}

Open in new window

farzanj

SORRY ONCE AGAIN. Problem with my clipboard.

$/ = undef;
$_ = <>;

while (/^\d* [ \.]* (.*?) $/gmsx) {
print "line[$1]\n";
}

jkfrench

ASKER

No problem. That gets the additional text, but as separate items. So I'm now getting:

line[This is the first item]
line[This is a longer second item that wraps down]
line[to the next line, but it's still one item.]
line[This is the last time which is also long and ]
line[wraps to the next line as well]

Open in new window

but am trying to get:

line[This is the first item]
line[This is a longer second item that wraps down to the next line, but it's still one item.]
line[This is the last time which is also long and wraps to the next line as well]

Open in new window

farzanj

Ok.

here is the idea. Change input record separator to

$/= '\d+';

This solves part of the problem

For the second part. You have to delete new line character from the line before printing.

farzanj

And yes, you can keep \d+ the one I had changed to \d*

ozo

$/ is a string, not a regex.

s/\n/ /g,print "line[$_]\n" for /^\d+ [ .]* ([\s\S]*?) (?=\n\d|\Z)/gmx;

SOLUTION

tel2

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial

tel2

PS: Where I wrote:
$line =~ s/\n/ /sg;
the 's' modifier is not required. It could simply be:
$line =~ s/\n/ /g;

ozo

> Have you tested that code, ozo?
Yes, it produces

line[This is the first item]
line[This is a longer second item that wraps down to the next line, but it's still one item.]
line[This is the last time which is also long and wraps to the next line as well]

Have you tested it?

tel2

Hi ozo,

Well, when I test it like this:
perl -ne 's/\n/ /g,print "line[$_]\n" for /^\d+ [ .]* ([\s\S]*?) (?=\n\d|\Z)/gmx' inputfile
I get this:
line[This is the first item]
line[This is a longer second item that wraps down]
line[This is the last time which is also long and]

What am I doing wrong?

ozo

perl -0777 -ne 's/\n/ /g,print "line[$_]\n" for /^\d+ [ .]* ([\s\S]*?) (?=\n\d|\Z)/gmx'

-0777 has the effect of $/ = undef;

tel2

Thanks ozo,

Sorry - wasn't thinking. Should have realised that's what was needed.

BTW, in your code, should this:
[ .]*
be this:
[ \.]*
?

ASKER CERTIFIED SOLUTION

ozo

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial

tel2

Thanks ozo. Sorry - I forgot about that, too!

jkfrench, pls take note of that. If you decide to use my solution, you can also change the:
@lines = split(/\d+[\. ]*/, <>);
to:
@lines = split(/\d+[. ]*/, <>);

jkfrench

ASKER

Thank you all for your help. I'm using ozo's one-liner, because it also works if there is a number embedded in the text. Since tel2's solution also worked for the data I posted, I'm awarding assist points.

tel2

Hi ozo,

Can you please explain your use of "([\s\S]*?)". It obviously works, but it just looks a tad strange to say "match zero or more spaces or non-spaces". I'm guessing it's something to do with avoiding newlines?

Hi jkf,

Thanks for the points. I suggest you be careful in future about giving people lower marks for failing to meet unspecified requirements, but if you are interested in an adjusted version which caters for that new requirement, here's one at no extra charge:

$/ = undef;
@lines = split(/\d+\. */, <>);

foreach $line (@lines)
{
  next unless $started++;
  chop $line;
  $line =~ s/\n/ /g;
  print "line[$line]\n";
}

Open in new window

For the future, I would also suggest you provide your expected output in your original post, so that experts like farzanj don't waste your time or theirs posting solutions until they know they work.

Thanks.

ozo

/[\s\S]/ matches any character including \n
I could have used /./s
but I also wanted to use a . that doesn't match \n elsewhere in the same regex
I could have used (?s:.) and (?-s:.) but [\s\S] and . seemed simpler

tel2

Thanks heaps, ozo!