reading data from strings

[Winston Hong]

Hi guys, I'm doing a final year project that requires some programming in C language.

+CMTI : "SM",2

+CMGR : "REC UNREAD", "+6512345678", "08/05/27"

How do I read the value "2" and the mobile phone number, +6512345678", from the data received?

The data is received serially from a GSM modem.

Thanks.

[jason_m]

Check out reading formatted input with the scanf family of functions:
http://www.opengroup.org/onlinepubs/...ns/fscanf.html

Here are some examples to get you thinking in the right direction:

#include <stdio.h>

int main() {
  char *str1 = "\"SM\"";
  char *str2 = "val1,val2";
  char buf1[10];
  char buf2[10];
  char buf3[10];

  /* read everything from an opening quote up to an end quote */
  sscanf(str1, "\"&#37;[^\"]s", buf1);
  printf("%s\n",buf1);
  
  /* break up the input string by the comma character */
  sscanf(str2, "%[^,],%s", buf2, buf3);
  printf("buf2=%s\tbuf3=%s\n", buf2, buf3);

  return 0;
}

Specifically, read the scanf documentation around character sets [...] and [^...].

That should get you going in the right direction.

Cheers,
Jason

[Winston Hong]

Thanks but I still don't understand how to get values from the string.

Maybe you could use my quotes and use it as an example?

[Elysia]

#include <stdio.h>

int main() {
  char *str1 = "\"SM\"";
  char *str2 = "val1,val2";
  char buf1[10];
  char buf2[10];
  char buf3[10];

  /* read everything from an opening quote up to an end quote */
  sscanf(str1, "\"%[^\"]s", buf1);
  printf("%s\n",buf1);
  
  /* break up the input string by the comma character */
  sscanf(str2, "%[^,],%s", buf2, buf3);
  printf("buf2=%s\tbuf3=%s\n", buf2, buf3);

  return 0;
}

I'm sorry, but this code fails too:
It's buffer overrun prone: http://cpwiki.sourceforge.net/Buffer_overrun
It uses char where const char should be used: http://cpwiki.sourceforge.net/Common...kes_and_errors
And finally a presentation about C security: http://www.cse.scu.edu/~tschwarz/COE...es/Strings.ppt

Winston Hong: If you do not understand this simple code, then I'm afraid you need to study a bit harder. C is not exactly trivial, but we cannot directly give our the answer because you wouldn't learn.

[rags_to_riches]

A final year project? Your final year of school? What are you studying?

[QuantumPete]

A final year project? Your final year of school? What are you studying?

does it matter?

[jason_m]

I'm sorry, but this code fails too:
It's buffer overrun prone: http://cpwiki.sourceforge.net/Buffer_overrun
It uses char where const char should be used: http://cpwiki.sourceforge.net/Common...kes_and_errors
And finally a presentation about C security: http://www.cse.scu.edu/~tschwarz/COE...es/Strings.ppt

You are right. This was a lazy, late night attempt on my part to quickly introduce the the conversion specifiers and parsing with scanf. The code works as is just fine, but could lead readers astray.

Maybe tonight after work I'll toss up an example that is a little more secure.

[rags_to_riches]

I'm curious about a "final year project." I'm not sure if that means the final year of a degree program, or if this is the final project of his first year.

[Winston Hong]

Well, its not that I don't understand what he's doing but..I still don't really see how it relates to my project. I think I need to look into the code more thoroughly.

And FYI, I'm a student doing a project for my final year of a degree.

[Elysia]

OK, so it may not be THE best example, but if you know how strings are composed and how to use them, then you should have no trouble with this at all.
Possibly some understanding of pointers, as well.
You know where inside that string the information you want it stored.

[jason_m]

In general, reading messages like this I find to be difficult, or at at least tedious. You need to have a full understanding of the protocol...the language that is being spoken across the medium. Once you know that, the general algorithm that you follow might be something like:

1. Read enough characters to know what type of message I'm dealing with.
2. The protocol will tell me how messages of that type are formatted. If I know the format of the message now completely, then read in the rest of the message, storing it in a buffer. Then parse the message using scanf(). If I don't know fully the format of the message, the protocol must tell me at least what to expect next, so go back to step 1, read a little more, decide what to do. From there, either read a little more or finally parse the string. Etc. etc..

The intension of my code was to introduce the %[^...] and indirectly, the %[...] conversion specifiers. These will come in handy if you need to read characters belonging to a certain set, or if you need to stop when you encounter a character from a certain delimiting set (for example, an end quote, a comma, etc. - these are the cases my examples address.) There is a lot of good stuff in the documentation file I linked to in my post above. If you read through it all, you'll be a pro at parsing strings.

The critique of my code was that it arbitrarily picks storage buffers of length 10 and goes on to read data into them without doing any bounds checking.

C does not enforce the end of an array. If you try to write beyond it, the language will let you, and you'll probably end up messing up all sorts of stuff. You could write over other variables or even write over your stack (including the return pointer!!) This leads to buffer overflows exploits and embarrassed programmers.

You can limit the number of characters scanf will read into your string, I'll leave it up to you to read through the documentation and see how that is done.

[Winston Hong]

Thanks for the link jason_m. It was really helpful and I've gotten what I wanted.