Thread: [HELP] writing my own compiler

OK, so i'm writing my own compilier that will compile a script that controls a robot. An example of that script would be.

Code:

#Example Program
RESET
Forward 3
pause 1
headlights ON
Headlights on
reverse 3
pause 2
headlights off
TurnRight 4
TURNLEFT 4
pAuSe 3
# Do this three times in a row
LOOP 3
     FORWARD 2
     Headlights ON
     Pause 2
     RESET
End Loop
RESET

I kind of know how to go about doing this so far I have this much.

- reads the file(script) to a series of strings
- changes the entire script to lower case
- tokenize the string
- debug it by checking to make sure everything is in the right place

my current problem....well it's not really a problem, but it's annoying. my code is kind of repatitive and i can't seem to think of a good function to replace the duplicating parts. The repeating part is:

Code:

             if (strcmp(tokenPtr, "reset") == 0)
               printf ("RESET\n");
             else if (strcmp(tokenPtr, "forward") == 0) 
               printf ("FORWARD\n");
             else if (strcmp(tokenPtr, "reverse") == 0)
               printf ("REVERSE\n");
             else if (strcmp(tokenPtr, "turnleft") == 0)
               printf ("TURN LEFT\n");
             else if (strcmp(tokenPtr, "turnright") == 0)
               printf ("TURN RIGHT\n");
             else if (strcmp(tokenPtr, "headlights") == 0)
               printf ("HEADLIGHTS\n");
             else if (strcmp(tokenPtr, "pause") == 0)
               printf ("PAUSE\n");
             else if (strcmp(tokenPtr, "loop") == 0)
               printf ("LOOP\n");
             tokenPtr = strtok(NULL, " ");

I use it once when checking the script for errors, and i'll be using it again when i actually run the script once it is error free. Any suggestions?

Strtok can help in tokenizing strings. However, it also damages the original input string. If that is a concern to you, then I have my own customized strtok you can try.

Well, do you understand pointers and arrays?
Those are essential to making your own tokenizer.

The way strtok works is that it searches a string until it finds, say, a space. Then it replaces the space with '\0', thus truncating the string. Then you can copy the string using strcpy or appropriate string to number functions to get your data.
Knowing pointers and arrays, it should not be difficult.

My own strtok is an example of how it's done. You can have a look at it if you're stuck.

No problem. If you have problems with the code or get stuck on something you don't understand, check back and we'll give you some help!

I'd also suggest two things: Use a switch statement instead of lots of else ifs and there is a function called toupper() which will make a char into it's uppercase equivalent (or leave it as it is if it's already uppercase). Why not do that one first (it'll take care of anything like "RESET" and "FORWARD") then use a switch and search for the unusual ones like "TURN LEFT" -> "turnleft"), where you're adding or removing spaces.

QuantumPete

Ah, but switches doesn't work on string literals.
A way to do that would be hash tables, but perhaps that's a little too much (too complex)?

Or, make a list or your "commands", e.g.:

Code:

const char *commands[] = { "forward", ... }; 

for(i = 0; i < sizeof(commands)/sizeof(commands[0]); i++)
   if (strcmp(commands[i], your_str) == 0) break;
...

The next extension of that is to have more of the parsing built into the table:

Code:

enum argtype { None, OnOff, Steps };
struct command {
   char *cmd;
   enum argtype args;
};

const struct commands[] = {
{ "reset", None },
{ "headlights", OnOff }, 
{ "forward", Steps },
... 
};

--
Mats