Help using Regexp:

Hello,

I'm a bit confused on how to use regexp to get information that's between braces. For example, I have the following text:

str='Some.Text {a1=v1 a2=v2}';
val=regexp(str, ?, 'match', 'once')

In "?" should be a like a pattern of special characters (lookaheads, lookbehinds, non-greedys, etc) but I don't know how to make it work to just extract everything inside {}.

Can someone help me?

Thanks in advance!

Answer 1

[per isakson]

One way

>> str='Some.Text {a1=v1 a2=v2}';
>> cac = regexp( str, '(?<={)[^}]+(?=})', 'once', 'match' )
cac =
a1=v1 a2=v2
>>

[^}]+ is better than .+ because

• avoids the problem with "greedies"
• better performance; easier to evaluate

'(?<=\{)[^}]+(?=\})' it doesn't hurt to escape {} if in doubt whether the are special characters.

[Ajpaezm]

Thanks!

Any book or article that you recommend specially for these special characters and information extraction? (?<={)[^}]+(?=}) ?

[per isakson]

There are many good resources on the Internet. However, Matlab has it's own regular expression flavor. It's close to a subset of PCRE, (PERL Compatible RE). (Not the latest version of PRCE though.) Thus,

• the Matlab documentation
• Welcome to Regular-Expressions.info, The Premier website about Regular Expressions is good. It has a book review page.
• regular expression 101 is a free tool in the cloud. I use it for experimenting. But one must avoid features that are not in Matlab.
• There are some Matlab specific tools in the File Exchange
• Mastering Regular Expressions if you want to become an expert.
• and best of all is Understanding and Using Regular Expressions

[per isakson]

btw:

xpr = [
    '(?<={) '   ... % look behind for literal {
    '[^}]+  '   ... % one or more of anything, but literal }
    '(?=})  '   ... % look ahead for literal } 
    ];
xpr( isspace(xpr) ) = [];
cac = regexp( str, xpr, ... );

I find this way of writing regular expressions good. It makes me think more and experiment less; it helps me avoid falling into an infinite "trial-and-error-loops".

• "space" is a bit of a problem. I write space as \x20
• The final look ahead isn't needed, since + is greedy