«

How can I match 'n' number of neighbouring words of a pattern using regular
expressions?

For example, suppose I am looking for the pattern "length xyz cm" in some
text. where xyz is a number - integer or fraction or decimal point. How can
I also grab about 3-5 words on either side of the pattern "length xyz cm"?
The surrounding words are not always constant & may be variable. Also, the
original text to be matched is not just a single sentence, but lines from a
file concatenated together - so the text has many newline characters too. I
only want the words on the same line as the pattern.

I have tried using regex of the form
/\b(\w*)\b(\w*)\b(\w*)\b($pattern)\b(\w*)\b(\w*)\b( \w*), but this doesn't
work for some reason. Could someone please offer some suggestions?

thanks!

[ Reply not posted to the defunct group comp.lang.perl ]

CV wrote:
> How can I match 'n' number of neighbouring words of a pattern using
> regular expressions?
>
> For example, suppose I am looking for the pattern "length xyz cm"
> in some text. where xyz is a number - integer or fraction or
> decimal point. How can I also grab about 3-5 words on either side
> of the pattern "length xyz cm"? The surrounding words are not
> always constant & may be variable. Also, the original text to be
> matched is not just a single sentence, but lines from a file
> concatenated together - so the text has many newline characters
> too. I only want the words on the same line as the pattern.
>
> I have tried using regex of the form
> /\b(\w*)\b(\w*)\b(\w*)\b($pattern)\b(\w*)\b(\w*)\b( \w*), but this
> doesn't work for some reason.

It doesn't work for several reasons, such as:

- No space characters.
- '\w*\b\w*' is an impossible combination that can never match (check
out the description of \b in "perldoc perlre" to learn why).
- The \w character class does not include e.g. the '$' character,
while you mentioned that a "word" may be a variable.

> Could someone please offer some suggestions?

Try something like this:

/((?:\S+ +){0,3})\b($pattern)\b((?: +\S+){0,3})/

--
Gunnar Hjalmarsson
Email: http://www.gunnar.cc/cgi-bin/contact.pl

Gunnar Hjalmarsson <> wrote:

> - '\w*\b\w*' is an impossible combination that can never match


It will match any string with at least one \w character in it:

$_ = 'hi';
print "matched '$&'\n" if /\w*\b\w*/;


> (check
> out the description of \b in "perldoc perlre" to learn why).


Check out this part too:

... counting the imaginary characters off the
beginning and end of the string as matching a \W




\W could be the beginning of string in the OP's regex.


--
Tad McClellan SGML consulting
Perl programming
Fort Worth, Texas

In article <S7OdnYSjY_-DGa7cRVn->,
CV <> wrote:
>How can I match 'n' number of neighbouring words of a pattern using regular
>expressions?
>
>For example, suppose I am looking for the pattern "length xyz cm" in some
>text. where xyz is a number - integer or fraction or decimal point. How can
>I also grab about 3-5 words on either side of the pattern "length xyz cm"?
>The surrounding words are not always constant & may be variable. Also, the
>original text to be matched is not just a single sentence, but lines from a
>file concatenated together - so the text has many newline characters too. I
>only want the words on the same line as the pattern.
>
>I have tried using regex of the form
>/\b(\w*)\b(\w*)\b(\w*)\b($pattern)\b(\w*)\b(\w*)\b( \w*), but this doesn't
>work for some reason. Could someone please offer some suggestions?
>

You may be confused about the \b assertion. Did you intend for
something with \w and \W..? Also, what if the pattern falls
at the beginning or end of the line... do you want to capture
the patterns that may not have 3-5 surrounding words?

One possibility presuming you intend to capture 3-5 surrounding
words:


my $text = "...";
my $pattern = 'length ... cm ';

my $words = '(?:\w+[^\w\n]+){3,5}';
#my $words = '(?:\w+[^\w\n]+){0,5}'; # to catch every pattern

print $1 while /($words$pattern$words)/g;


[ Note the 3-5 surrounding words may consume another
adjacent $pattern instance but you don't specify what
to do in that case. }


hth,
--
Charles DeRykus

Tad McClellan wrote:
> Gunnar Hjalmarsson wrote:
>>
>> - '\w*\b\w*' is an impossible combination that can never match
>
> It will match any string with at least one \w character in it:
>
> $_ = 'hi';
> print "matched '$&'\n" if /\w*\b\w*/;
>
>> (check
>> out the description of \b in "perldoc perlre" to learn why).
>
> Check out this part too:
>
> ... counting the imaginary characters off the
> beginning and end of the string as matching a \W
>
>
>
> \W could be the beginning of string in the OP's regex.

Thanks, Tad, I stand corrected (even if it doesn't do what the OP
wanted it to do...).

--
Gunnar Hjalmarsson
Email: http://www.gunnar.cc/cgi-bin/contact.pl