Recursive regular expression (or alternative)

[André Hänsel]

Hi,

I'm looking for a pattern to parse the following subject:

blabla(cond1)?{abc}:{def}blabla(cond2)?{qwe(cond3) ?{poi}:{iop}}

Probably I can't do that by executing a regexp _once_, so I need some
kind of recursive call to get those two lists (e.g. in an array):
array(
0 => array("cond1","abc","def"),
1 => array("cond2","qwe(cond3)?{poi}:{iop}","")
)
and for the second one then I need the array:
array(
0 => array("cond3","poi","iop")
)

Or in words: In a configuration file I have _nested_ conditional blocks
in the format (condition)?{content_when_true}:{content_when_fals e}
while the false content ist optional.

I built the following pattern, but thats just a bad guess:
/\(([a-zA-Z0-9_ &|]+)\)\?\{(.*)\}(\:\{(.*)\})?/
It matches both blocks when it's greedy and ignores the nested block
when it's ungreedy.

If you have any completly different solution I'll be glad to hear it,
too. :)

Best regards,
André

In comp.lang.perl.misc André Hänsel <an***@webkr.de> wrote:

: Or in words: In a configuration file I have _nested_ conditional blocks
: in the format (condition)?{content_when_true}:{content_when_fals e}
: while the false content ist optional.

Hint "configuration file": CPAN has a number of configuration file modules,
did you have a look at them?

Hint "nested": Nested data are preferrably tackled by a parser, not a regex.

: If you have any completly different solution I'll be glad to hear it,
: too. :)

Yes. The XML::Simple module. You have full control over your nested data
structures and can easily evaluate branches of your data conditionally.
Provided, however, that your regex in the beginning didn't describe your
final data format but rather a model of what you want to achieve.

Oliver.

--
Dr. Oliver Corff e-mail: co***@zedat.fu-berlin.de

[fletch]

CPAN is Perl not PHP, not sure if previous poster has got his
newsgroups confused. Everything else the previous poster said applies.

[André Hänsel]

fletch schrieb:

CPAN is Perl not PHP, not sure if previous poster has got his
newsgroups confused. Everything else the previous poster said applies.
That's ok, the principles probably are the same.
co***@zedat.fu-berlin.de schrieb:
Hint "nested": Nested data are preferrably tackled by a parser, not a regex.

Well, how does such a parser look like? Can you give me a hint how to
implement it.

Your other hints don't help me since the format of the file is not to
be changed.

Best regards,
André

[fletch]

Give us some actual data to work with, and what the parser should
produce, your original post hasn't really made that clear to me.

[Tad McClellan]

[ Followup set to a newsgroup that I read ]
André Hänsel <an***@webkr.de> wrote:

In a configuration file I have _nested_ conditional blocks If you have any completly different solution I'll be glad to hear it,
too. :)

perldoc -q nest

How do I find matching/nesting anything?
--
Tad McClellan SGML consulting
ta***@augustmail.com Perl programming
Fort Worth, Texas