[SOLVED] More question about regular expressions

ppmoore · 11-25-2010, 04:16 AM

Hello,

As a learning exercise, I'm trying to use regcomp() and regexec() to match the string
"H,2,1,1,"

My code is:

Code:

const char* test = "H,2,1,1,";
char* pattern = "H,[0-9],[0-9],[0-9]";
regex_t rexpr;
regmatch_t match[3];

if( ( result = regcomp( &rexpr, pattern, 0 ) ) )
{
  printf("Cannot compile pattern");
  exit();
}

if( regexec(&rexpr, test, 1, match, 0 ) ) 
{
  printf("Cannot parse string");
  exit();
}

The pattern matches, as expected.

When I change the pattern to any of the following, it fails

Code:

"H,[0-9]+,[0-9]+,[0-9]+"   // match 1 or more of each digit
"H,[0-9]*,[0-9]*,[0-9]*"   // match 0 or more of each digit
"H(,[0-9])+"               // match 1 or more groups of ,<digit>
"H(,[0-9]){3}"             // match 3 groups of ,<digit>

I tested each of the above on several regex test sites, and they work. Is the regex implementation included in the standard C library more limited than for other implementations?

Many thanks,
Paul

grail · 11-25-2010, 06:10 AM

Quote:

When I change the pattern to any of the following, it fails

Does this mean you receive an error?
It does not compile?
Simply does not display that it matched?

As you can see, failure could mean any number of things so a little more detail might help?

Also, based on the snippet you have provided there does not appear to be any way to tell if success or not, assuming it passes the 2 included tests.

Sergei Steshenko · 11-25-2010, 06:42 AM

Quote:

Originally Posted by ppmoore

Hello,

As a learning exercise, I'm trying to use regcomp() and regexec() to match the string
"H,2,1,1,"

My code is:

Code:

const char* test = "H,2,1,1,";
char* pattern = "H,[0-9],[0-9],[0-9]";
regex_t rexpr;
regmatch_t match[3];

if( ( result = regcomp( &rexpr, pattern, 0 ) ) )
{
  printf("Cannot compile pattern");
  exit();
}

if( regexec(&rexpr, test, 1, match, 0 ) ) 
{
  printf("Cannot parse string");
  exit();
}

The pattern matches, as expected.

When I change the pattern to any of the following, it fails

Code:

"H,[0-9]+,[0-9]+,[0-9]+"   // match 1 or more of each digit
"H,[0-9]*,[0-9]*,[0-9]*"   // match 0 or more of each digit
"H(,[0-9])+"               // match 1 or more groups of ,<digit>
"H(,[0-9]){3}"             // match 3 groups of ,<digit>

I tested each of the above on several regex test sites, and they work. Is the regex implementation included in the standard C library more limited than for other implementations?

Many thanks,
Paul

Why do you have '1' here:

Code:

if( regexec(&rexpr, test, 1, match, 0 ) )

while you have '3' here:

Code:

regmatch_t match[3];

...

And how about '0' here:

Code:

if( regexec(&rexpr, test, 1, match, 0 ) )

- maybe you need REG_EXTENDED ?

?

ppmoore · 11-25-2010, 08:15 AM

Thanks Sergei,

I had also independently found that not using REG_EXTENDED was the problem, and was going to update the thread when I found your reply.

Best,
Paul

grail · 11-25-2010, 09:37 AM

Glad you got there in the end

Please mark as SOLVED. You may even want to copy your solution in for others to understand where you went wrong