zinid.rl

/*
 * Compile:
 * $ ragel -G2 zinid.rl -o zinid.c
 * $ gcc -O2 zinid.c -o zinid
 */

#include <string.h>
#include <stdlib.h>
#include <stdio.h>
#include <unistd.h>

#define BUFSIZE 1024

char **idx;
size_t size = 1024*1024;
size_t ptr = 0;
int pos = 0;
char *line_stop = NULL;
char *line_start = NULL;
char *key_stop = NULL;
char *val_start = NULL;
char *val_stop = NULL;

%%{
  machine foo;

  action key_start {line_start = p;}
  action key_stop {key_stop = p;}
  action val_start {val_start = p;}
  action val_stop {val_stop = p;}
  action line_stop {
    line_stop = p;
    idx[pos] = line_start;
    idx[pos+1] = key_stop;
    idx[pos+2] = val_start;
    idx[pos+3] = val_stop;
    pos += 4;
  }
  action parse_error {error = 1;}

  word = [a-z0-9]+;
  
  main := (
	   (word > key_start % key_stop) ' '
	   (word > val_start % val_stop)
	   (('\r'?'\n') % line_stop) any*
	 ) $! parse_error;
}%%

%% write data;

int parse(char *buf, char *pe) {
  int error = 0, cs = 0;
  char *space = NULL, *crlf = NULL;
  char *p = buf;
  char *eof = pe;

  %% write init;
  %% write exec;

  return error;
}

void do_write(char *out, const char *in, size_t len) {
  if ((ptr + len) >= size) {
    write(STDOUT_FILENO, out, ptr);
    ptr = 0;
  }
  memcpy(out + ptr, in, len);
  ptr += len;
}

int main(int argc, char *argv[]) {
  int i = 0, j = 0, got = 0;
  idx = malloc(1024*1024);
  char *in = NULL, *buf = NULL;
  char *end = NULL;
  in = malloc(1024*1024);
  char *out = malloc(size);
  line_start = in;

  while (1) {
    if ((got = read(STDIN_FILENO, in, BUFSIZE)) > 0) {
      in = in + got;
      while (!parse(line_start, in)) {
	line_start = line_stop;
      }
    } else
	break;
  }

  for (i = 0; i<pos; i += 4) {
    int len_i = idx[i+1] - idx[i];
    for (j = 0; j<pos; j += 4) {
      int len_j = idx[j+3] - idx[j+2];
      do_write(out, idx[i], len_i);
      do_write(out, idx[j+2], len_j);
      do_write(out, "\n", 1);
    }
  }

  write(STDOUT_FILENO, out, ptr);
  return 0;
}