K&R exercise 1-24

k&r-1-24.c

#include <stdio.h>
//I really want to use exit(), so sue me.
#include <stdlib.h>

#define true 1
#define false 0

//machine states
#define SOURCE 1 //regular code
#define COMMENT 2 //currently in a comment
#define LINE_COMMENT 3 //currently in a single-line comment
#define QUOTE 4 //currently in a quote
#define PRECOMMENT 5 //MAY BE entering a comment (encountered "/", waiting for "*")
#define POSTCOMMENT 6 //MAY BE exiting a comment (encountered "*", waiting for "/")
#define ESCAPING 7 //escape next character

//uncomment to activating debugging output
//#define debug true;

int stack[256];
int stackPosition = 0;
int states[256];
int statePosition = 0;

//manipulate the state stack
int getState() { return states[statePosition]; }
void pushState(int state) { states[++statePosition] = state; }
void popState() { statePosition--; }

//manipulate the nested syntax stack
int getSyntax() { return stack[stackPosition]; }
void pushSyntax(int c) { stack[++stackPosition] = c; }
void popSyntax() { stackPosition--; }

//Print stacks for debugging
void die() {
  #ifdef debug
  printf("Current syntax stack:\n");
  for (int i = 0; i <= stackPosition; i++) {
    putchar(stack[i]);
  }
  printf("\nCurrent state stack:\n");
  for (int i = 0; i <= statePosition; i++) {
    printf("%d", states[i]);
  }
  putchar('\n');
  #endif
  exit(0);
}

//returns -1 for error, otherwise 1
int evaluate(int c) {
  int state = getState();
  switch (state) {
    case SOURCE:
      switch (c) {
        case '{':
        case '(':
        case '[':
          pushSyntax(c);
        break;
        
        case '}':
          if (getSyntax() != '{') {
            printf("Unexpected }\n");
            return -1;
          }
          popSyntax();
        break;
        
        case ')':
          if (getSyntax() != '(') {
            printf("Unexpected )\n");
            return -1;
          }
          popSyntax();
        break;
        
        case ']':
          if (getSyntax() != '[') {
            printf("Unexpected ]\n");
            return -1;
          }
          popSyntax();
        break;
        
        case '/':
          pushState(PRECOMMENT);
        break;
        
        case '"':
        case '\'':
          pushSyntax(c);
          pushState(QUOTE);
        break;
      }
      break;
    
    case PRECOMMENT:
      popState(); 
      if (c == '*') {
        pushState(COMMENT); //replace PRECOMMENT with COMMENT
      } else if (c == '/') {
        pushState(LINE_COMMENT); //replce PRECOMMENT with LINE_COMMENT
      }
      //if neither matches, it's just division
    break;
      
    case POSTCOMMENT:
      popState(); //remove POSTCOMMENT state
      if (c == '/') {
        popState(); //remove COMMENT as well
      }
    break;
      
    case COMMENT:
      if (c == '*') {
        pushState(POSTCOMMENT);
      }
    break;

    case LINE_COMMENT:
      //end line comment at end of line
      if (c == '\n') {
        popState();
      }
    break;
      
    case QUOTE:
      if (getSyntax() == c) {
        popSyntax();
        popState();
      }
      if (c == '\\') {
        pushState(ESCAPING);
      }
    break;
    
    case ESCAPING:
      popState(); //just skips a character, really
  }
  return true;
}

void main() {
  stack[stackPosition] = '_'; //gotta start somewhere
  states[statePosition] = SOURCE;  
  int c;
  int pos = 1;
  int line = 1;
  while ((c = getchar()) != EOF) {
    
    #ifdef debug
    putchar(c);
    #endif
    
    int result = evaluate(c);
    if (result == -1) {
      printf("Syntax error at %d:%d\n", line, pos);
      die();
    }
    if (c == '\n') {
      pos = 1;
      line++;
    } else {
      pos++;
    }
  }
  //You can't have syntax or state left over
  if (stackPosition > 0 || statePosition > 0) {
    printf("Unexpected end of file!\n");
    die();
  }
  //You made it!
  printf("No syntax errors found!\n");
}

Requires C99 mode if debug is defined, due to loop variable declaration.