Redis server-side protocol parser for Ragel

redis-parser.rl

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

%%{
    machine redis_parser;

    action reset_line {
        if(line.argv) {
            free(line.argv);
            line.argv = NULL;
        }
        memset(&line, 0, sizeof(struct batchline));
    }

    action mark {
        mark = fpc - data;
    }

    action write_arg {
        if( line.argv == NULL ) {
            line.argv = malloc(sizeof(struct slice));
            line.argc = 1;
        }
        else {
            line.argc += 1;
            line.argv = realloc(line.argv, sizeof(struct slice) * line.argc);
        }
        line.argv[line.argc-1].data  = data + mark;
        line.argv[line.argc-1].len  = (fpc - data) - mark;
    }

    action arg_err {
        printf("error parsing argument! %.*s %zu/%zu continuing happily\n\n", (int)((fpc - data) - mark), data + mark, fpc - data, mark);
        fhold; fgoto gobble_line;
    }

    action write_message {
        if( ! line.ignore ) {
            if( line.argc ) {
                for(size_t i = 0; i < line.argc; i++) {
                    printf("arg %zu:%.*s\n", i, (int)line.argv[i].len, line.argv[i].data);
                }
            }
            printf("--------------------\n");

        }
    }

    action message_err {
        printf("error parsing message! %zu/%zu continuing happily\n\n", fpc - data, mark);
        fhold; fgoto gobble_line;
    }

    action write_arg_len {
        printf("arg len: '%.*s' %zu/%zu\n\n", (fpc - data) - mark, data + mark, fpc - data, mark);
    }

    action argc_reset {
        line.parse_arg_count = 0;
    }

    action argc_add_digit {
        line.parse_arg_count = line.argc * 10 + (fc - '0');
    }

    action argsz_reset {
        line.parse_arg_len = 0;
    }

    action argsz_add_digit {
        line.parse_arg_len = line.parse_arg_len * 10 + (fc - '0');
    }

    action test_arg_len {
       ((fpc - data) - mark) < line.parse_arg_len
    }

    action set_unified {
        line.unified = true;
    }

    action set_inline {
        line.unified = false;
    }

    action arg_count_ok {
        ( ! line.unified || (line.argc < line.parse_arg_count))
    }

    action set_ignore {
        line.ignore = true;
    }

    separator = " ";
    int = [0-9]+;
    crlf = ("\r\n" | "\n");
    gobble_line := (any -- crlf)* crlf @{ fcall main; };

    uni_arg_len = "$" @argsz_reset (digit @argsz_add_digit)+ crlf >mark;
    uni_arg = uni_arg_len (any when test_arg_len)+ >mark %write_arg;
    uni_arg_count = "*" @argc_reset (digit @argc_add_digit)+ crlf;
    uni_request = uni_arg_count (uni_arg <: crlf)+ @set_unified;

    arg_valid = any -- (separator | "\z" | "\r" | crlf);
    cmd_valid = any -- ("*" | "$" | "#" | separator | "\z" | "\r" | crlf);
    inl_cmd = cmd_valid+ >mark %write_arg $err(arg_err);
    inl_arg = arg_valid+ >mark %write_arg $err(arg_err);
    inl_request = inl_cmd (separator inl_arg when arg_count_ok)* crlf @set_inline;

    comment_line = space* "#" @set_ignore (any -- crlf)+ crlf;
    message = (comment_line|inl_request|uni_request) %write_message >reset_line <err(message_err);
    main := message+;
}%%


%% write data;

const char *data =
"*3\r\n$3\r\nSET\r\n$5\r\nmykey\r\n$9\r\nmyva\r\nlue\r\n"
"*2\r\n$3\r\nSET\r\n$5\r\nmykey\r\n$7\r\nmyvalue\r\n"
"+ DERP\r\n"
" #test\n"
"#test\n"
"#derp\n"
"PING $(DERP) 3\r\n"
"PING 1234\r\n"
"1234\r\n"
"PIN\rG\r\n"
"PRIVMSG #luahelp :say something\r\n"
"PRIVMSG :say something\r\n"
":user!~ccfreak2k@4chan.fm PRIVMSG #wiremod :No.\r\n"
":ccfreak2k@@!~ccfreak2k@4chan.fm PRIVMSG #wiremod :No.\r\n"
":ccfreak2k!~ccfreak2k@4chan.fm PRIVMSG #wi\rremod :No.\r\n"
"PRIVMSG abc :def hij\r\n";

struct slice
{
    const char* data;
    int len;
};

struct batchline
{
  struct slice* argv;
  size_t argc;

  size_t parse_arg_count;
  size_t parse_arg_len;
  bool unified;
  bool ignore;
};

int main(int argc, char **argv)
{
    int cs;
    size_t mark;
    struct batchline line;
    memset(&line, 0, sizeof(struct batchline));

    const char *p = data;
    const char *pe = p + strlen(p);
    const char *eof = pe;

    int stack[10] = {0};
    int top = 0;

    %% write init;
    %% write exec;

    printf("Finished at state %i\n", cs);

    return 0;
}