Thread: Parsing a string

- A have a hardware board with a 80186 processor.
- I communicate with the board with telnet.
- The server is a delphi program
- I have developed a protocol for the communication so I can connect many hardware boards.

I send strings with telnet, The string need to be parsed at the hardware board, which is programmed in C.

I have never build a parser before, so i would like to know what's the best method for this? I new to C programming so I'm not very good with C.

I have an example with the strtok function. It works, but I don't know how to detect errors in the strings. The protocol string have the following lay-out:

command^comment|date|time|MAC|field|field|field|

the number of fields is variable.

Thanks

it would be a lot easier to develop the program with a desktop program and compiler. Once this is tested you can transfer to the final target compiler. Your desktop program doesn't need to use telnet, just hardcode some strings for testing/debugging purposes. I would also use strtok() to tokenize the string. Then parse each of the strings for possible errors -- what errors you will have to decide because you havn't provided enough detailed information (that I probably don't need or want to know anyway).

You have understand me wrong i think. I develop the source code with borland C++. I have special plugins for the hardware so i can debug via telnet.

The biggest problem is the variable fields. I find it hard to make some code that can easily handle this.

What type of data does each field contain? If this is known beforehand, you may be able to just parse the whole line with sscanf() in a loop. Otherwise, there are some good parsing tools out there: yacc is one I recommend - it's not difficult to use once you know what your grammar is (and in this case you seem to have that pretty well sorted).

If neither of those methods are applicable to your situation, you will probably have to write a custom parser. This, again, isn't too difficult once you understand the concepts involved. Basically you need to work out what your grammar is, then what you want to do when you encounter each element. If you decide to go with writing your own, you should go read up on the topic. Here is a good place to start:
http://en.wikipedia.org/wiki/Parsing

Maybe a primitive but applicable way: what about using some control characters in the message for each field and ,after receving the message string, counting all the control characters until running into a \0. So, at the end of the loop you will have the number of fields.

I used this technique for a simple CGI URL parser.

Thnx for the replies!

I tried the last example, but i get an error...probably my lack of C knowledge

I want to make a function with a string as input and an array of fields as output.

The protocol is definied as following:

command^comment|field|field|field|0x0A

command = 3 numbers
0x0A = end of string (telnet)

Code:

char parser(char *line)
{
   int i;
   char *token = line;
   char result;
   for ( i = 0; *line; ++i )
   {
   	size_t len = strcspn(token, "|\n");
   	printf("token[%2d] = \"%*.*s\"\n", i, (int)len, (int)len, token);
   	token += len + 1;
      result[i] = token;
   }
   return result;
}

I get the error: Invalid indirection at the line: result[i] = token;

That's because you've declared result as a char, so result[i] doesn't make much sense.

how can i return a array of strings ??

I started testing and I can't get it working

I have made the following function:

Code:

void parse(char *line,char *output)
{
  char buf[100],ch;
  int i,count=0;

  //count | tokens in string
  for(i=0;ch!='\0';i++)
     {
       ch=line[i];
       if(ch=='|')
           count++;
  }
  printf("count=%d\n",count);

  //parse string into fields
   for (i=0;i==count;++i)
   	{
   	 size_t len = strcspn(line,"|");
                 sprintf(buf,"%*.*s\n", (int)len, (int)len, line);
                 printf("%s\n",buf);
   	 line += len + 1;
   	}
}

the token count works fine. The program prints count = 6 when i use the following as input:

Code:

002^Resultaat op aanmelden|20050915|121533|INT^TEMP^1^S|BOOL^LED^1^A|BOOL^LED^2^A|0x0A

the format is: command^comment|date| time|field|field|field|0x0A

unfortunately it doesnt print the fields

i want the function to output an array of fields, but i havent implemented that yet

I work with telnet, where 0x0A is the end of line character

It just won't work

I tried the examples and i can print the fields, but I want to store them in variables, so i can work with them in my program.

Yacc looks suitable for my problem, but i don't know how to work with it.



I will explain my problem in more detail:

I am building a home automation system. Every room has a roomcontroller with several sensors and actuators. The core of the roomcontroller is a Beck SC12 chip (complete tcp/ip stack + 80186 proc.) and a I/O controller.

I can debug the program with a program via telnet.

There is a server which runs a delphi program that communicates with the roomcontrollers via telnet.

For this I have the following protocol:

Code:

command^comment|MAC|field|field|field|0x0A

the command field is a command of 3 numbers
tha MAC field is for identification (for example 0A_0A_0A_0A-0A_0A)
the number of fields is variable and there are subfields separated by ^
0x0A is the telnet end of line

an example string:

Code:

001^Aanmelden|00_00_00_00_00_00|INT^TEMP^1^S|INT^LIGHT^1^S|BOOL^RELAY^2^A|0x0A

this string should do the following:
- login to the server (command 001)
- tell the server what sensors and actuators the roomcontroller has.

In this case:
tempature sensor INT = integer TEMP = name 1 = number S = sensor)

light sensor INT = integer LIGHT = name 1 = number S = sensor)

relay BOOL = boolean RELAY = name 2 = number A = actuator


Anyone got any suggestions?? I'm really stuck on this

I can't get that working. The problem is the variable fields, so I want to store the data in a array, but I can't get that working. Someone got an idea ??