What is the best and most effective way to parse configuration files (standard Linux text files, # is a comment)? Is there a standard library that does that or do I have to implement my own parser (or lex for that matter)? Is it a good idea to use xml for configuration files, do you believe user would prefer that over usual text files? How would you go about parsing xml?

Thank you!
Max

I think XML is a good idea, since it takes care of all the mundane text processing. It's a little awkward if your user is used to regular files. But I'd go for it anyway because it really handles a lot of the work for you.

edit: you don't have to manually parse xml yourself. There are libraries for that (expat and libxml2 for eg). You just provide a DTD. That's the reason why I recommended xml.

max_sipos, are you going to take the XML route? I was considering creating a library that would parse an apache like conf file (name/value pairs with sections contained within <></> pairs). Actually I was hoping Apache would have a separate library for this posted but it does not appear so... the search continues.

jpbarto

Actually, I thought that, for the purposes of my application, a custom built, tiny parser is good enough. It would be probably a good idea if something like that would already exist, but the little I have spent analyzing sources of popular programs, I have noticed that all of them use integrated parsers.

Well I think I'm going to try my hand at writing one. It won't be XML but will follow coding standards such as httpd.conf (name value pairs and <></> sections). It doesn't make sense not to have one common one (like getopt).

True, if there is a way I could help, don't hesitate pm-ing me
Best,

uhm ... max ... uhm ... you have private messages turn off ... hehe.

Anyway, I've just started creating a couple of structs for holding config data and I should have something a little more testable here in a few days. Once I do I'll let you know as I'd appreciate a tester and a second pair of eyes to make sure the code isn't completely nuts.

Oops I see you're supposed to be a contributing member to use PM . I'll have to contribute some time

Anyways, I'd be very happy test the code, be sure to post it here when you're ready!

Max, I've completed some initial tests on an alpha release for a library of sorts. I'll be posting it along with a README on Tuesday. Please download it and give it a try. It's not too well tested yet but its operable (at least on NetBSD). Please take a look and make any recommendations you'd like.

jpbarto

Ok a first alpha release has been made available. You can download it from:

http://jpbarto.freeshell.org/?id=code#conf4c

It still needs a few things like better error handling, proper configuration and makefiles, and portability testing but its operational so I thought I'd make it available for testing.

jpbarto

Ok ... you remember that whole 'download it and try it' thing ... yeah ... well uhm ... unless you're going to run it on NetBSD you might have some problems. I just got a chance to try it out on a Gentoo box and the thing just segfaults right out. Some of the tests complete successfully but the big 'test-everything' test fails immediately ... so back to the coding board I go.

A better C4C has been posted and is ready for general tester consumption. A couple of detailed readmes are included with the download. To get it up and running just download the tarball, unpack, configure, make.

You can then cd to the test directory and run make. But before you execute any of the test binaries you will have to define LD_LIBRARY_PATH (unless you installed the C4C library). I also included txt files which dictate what the output from these binaries *should* look like so you can just run a diff on the actual vs predicted to see if something went wrong.

Enjoy:
http://jpbarto.freeshell.org/dl.php?id=C4C.tar.gz

OK, I've did a short test of the code. I had no problems making. BTW I'm running slackware 10.

****************** System info:
cat /proc/cpuinfo:

processor : 0
vendor_id : AuthenticAMD
cpu family : 6
model : 6
model name : AMD Athlon(tm) XP 1900+
stepping : 2
cpu MHz : 1595.492
cache size : 256 KB
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 1
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 mmx fxsr sse syscall mp mmxext 3dno
wext 3dnow
bogomips : 3185.04

cat /proc/meminfo:

total: used: free: shared: buffers: cached:
Mem: 527814656 522891264 4923392 0 38309888 336007168
Swap: 509923328 16453632 493469696
MemTotal: 515444 kB
MemFree: 4808 kB
MemShared: 0 kB
Buffers: 37412 kB
Cached: 326556 kB
SwapCached: 1576 kB
Active: 223372 kB
Inactive: 218632 kB
HighTotal: 0 kB
HighFree: 0 kB
LowTotal: 515444 kB
LowFree: 4808 kB
SwapTotal: 497972 kB
SwapFree: 481904 kB

cat /proc/version:
Linux version 2.4.26 (root@tree) (gcc version 3.3.4) #6 Mon Jun 14 19:07:27 PDT 2004
******************

I've executed the test programs one by one and compared the differences to .output files.

testc4c: OK
testc4cf: OK
testconffile: OK
testconfsection: diff output:
11c11
< stor 1- 3
---
> stor 1- 3
17c17
< stor 2- 3
---
> stor 2- 3
testconfstor: OK
testconfvalue: OK

Besides the diff at testconfsection, all programs worked properly without anything like segfault

Hope this helps, I will continue testing when I find time!

It does help, there seem to be some differences or disparities in how strings are handled between BSD and Linux. I have a couple of functions which trim non-printables and whitespace, so I'll be looking closely at those to see where improvements can be made. Also I'm going to be constructing a Conf4CReader object and functions which will allow a conf file to be easily traversed via line-by-line. After that its on to being able to easily create a configuration file and then write it to disk, instead of just reading from disk. Other improvements and suggestions are greatly appreciated of course.

Thanks,
jpbarto