Problem with a PCRE regexp

Question

I'm trying to create a regexp that can match this:

argument ::= define_scope [';' define_scope]*
define_scope ::= (['local'] | 'global') define_var
define_var ::= variable_name expression
variable_name ::= Name

So, something like local varName something;;world foo bar;;local foobar bar.
I've tried with:

((^|;;)?(local|world) (.+?) (.+?))+

but if I use this on the previous example I obtain these matches:

local varName s
;;world foo b
;;local foobar b

so it take only the first letter of the last word of each match.
If I remove the lazy matching from the last group, it match only:

local varName something;;world foo bar;;local foobar bar

so the last group is something;;world foo bar;;local foobar bar.

Some ideas to fix this?

Please take a look on the Markdown help: stackoverflow.com/editing-help — sidyll
– sidyll, Commented Jul 20, 2011 at 14:30

knittl · Accepted Answer · 2011-07-20 14:34:23Z

2

this is not a regular grammar and resulting sentences/words (cs speak) cannot be parsed with a regular expression. it's a context free grammar and you need a parser which utilizes recursive descent (LL-parser).

answered Jul 20, 2011 at 14:34

knittl

269k59 gold badges339 silver badges405 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

user229044 · Accepted Answer · 2011-07-20 14:52:15Z

1

Regular expressions are not the be-all end-all tool in your tool box, and they will not suffice here, but this one can be made to work for your specific limited example by telling it to match up to (but not including) the semi-colons, and removing the non-greedy ?:

/(^|;;)((local|world) (.+?) ([^;]+))/

Your problem is . matches any character. Matching . greedily was eating up the rest of the string on the first match, while non-greedily it was satisfied with the first character. The solution was to tell it to greedily match everything except semi-colons, with [^;]+. Ideally you should restrict this to the list of characters you actually expect to appear there instead of using . so freely.

edited Jul 20, 2011 at 14:52

answered Jul 20, 2011 at 14:46

user229044♦

241k41 gold badges347 silver badges350 bronze badges

Comments

Steve81 · Accepted Answer · 2011-08-11 19:25:02Z

1

That is the regexp that I needed:

((?:(local|world) )?(.*?)(?: (.+?))(?:(?<!;);(?!;)|$))+?

This one can parse without problem anything that have ;; without matching it

Thanks anyways to all.

answered Aug 11, 2011 at 19:25

Steve81

532 silver badges7 bronze badges

Collectives™ on Stack Overflow

Problem with a PCRE regexp

3 Answers 3

Comments

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

Comments

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related