[Boost-users] C++ Parsing - oops

From: Foster, Gareth (gareth.foster_at_[hidden])
Date: 2004-09-30 08:22:18

Hello John, Joel,

I didn't mean to knock your work with that last email John, it might have
come across badly, it was meant more as a compliment to spirit and a
reference to the fact that it 'appears' to fit the bill for what I want to
do better, not to bash regex, I like regex too. For the record it needn't be
'perfect' indexing of the C++ code, I don't mind if I get two hits for the
same classname for example. I used regex for a tool to index the C++ and C
files themselves, so I could quickly search for file names using regular
expressions, it works great!

So thank you both, no offence caused I hope.

So, will spirit be the best way then? Glad to hear its already a part of
boost btw, congratulations mate.

Cheers!

Gaz

-----Original Message-----
From: boost-users-bounces_at_[hidden]
[mailto:boost-users-bounces_at_[hidden]] On Behalf Of John Maddock
Sent: 30 September 2004 11:33
To: 	
Subject: Re: [Boost-users] Boost:regex and C++ Parsing - Clarification

> Thank you Caleb and Hartmut for your replies. You both seem to think regex
> is a bad way to go, I will explain better what I want to write just to be
> clear.
>
> I want to write a tool (cli probably) where I can say, here you go, here 
> is
> a large folder full of code, go and parse it. I will store the results in
> XML format somewhere, then, I can do say, "<myapp> class someclass" and 
> the
> program will go and find where that class is declared/defined using its
> database, saving me headache.
>
> So I thought I could use one of the C++ expat wrappers, and boost regex
> looked powerful enough to do the parsing if only I were handy enough with
> regular expression syntax.
>
> Anyway, I don't know if that better explanation will make any difference 
> to
> you recommendations, I look forward to reading you opinions.
>
> Oh, and the example I looked at is here Hartmut:
> http://boost.org/libs/regex/example/snippets/regex_search_example.cpp - 
> that
> is what got me thinking I might actually be able to take on this 
> challenge.

It depends what you want to do: if you want to use a "real" C++ parser then 
you will also have to preprocess the code (including the includes) and then 
parse the code.  In theory this gives you a "perfect" result, but only if 
you know what include paths to use, and what predefined macros should be set

(think about conditional code blocks).

Regexes on the other hand, don't require you to preprocess the code, but can

get confused by macros and the like.

So you have to choose the way that best meets your expectations, and live 
with the defects either which way ;-)

To solve your problem BTW, why not scan through the file for line starts 
(keeping count obviously!), and at each line start see if it's also the 
start of the regex you are interested in (one that matches a class 
definition for example), if you do this don't forget to either:

prefix your expression with \A

or

Pass the match_continuous flag to regex_search,

Either will anchor the search at the start of the line you are checking, and

prevent the whole text being searched.

John. 

_______________________________________________
Boost-users mailing list
Boost-users_at_[hidden]
http://lists.boost.org/mailman/listinfo.cgi/boost-users