Ragel for classifying log lines

line_classifier.rl

#include <assert.h>
#include <string.h>
#include <stdio.h>

#define LINE_IS_SIGNAL 1
#define LINE_IS_NOISE 2
#define LINE_IS_UNKNOWN -1

%%{
    machine line_classifier;
    
    action is_signal { return LINE_IS_SIGNAL; }
    action is_noise { return LINE_IS_NOISE; }
    action is_unknown { return LINE_IS_UNKNOWN; }
    
    signal = (
        /ERROR/ |
        /WARNING/
        ) @(is_signal);
    
    noise = (
        /blah/
        ) @(is_noise);
    
    main := ( signal | noise ) @(is_unknown) $err(is_unknown);

}%%

%% write data;

int classify_line(const char * line, int length) {
    const char *p = line;
    const char *pe = line + length;
    const char *eof = pe;
    int cs;
    
    %% write init;
    %% write exec;
    
    /* this should never happen, since the state machine should return a value in all cases */
    assert(0);
}


/* to run it */
void classify(const char * line) {
    const char * kind;
    switch (classify_line(line, strlen(line))) {
    case LINE_IS_NOISE: kind = "noise"; break;
    case LINE_IS_SIGNAL: kind = "signal"; break;
    default: kind = "unknown"; break;
    }
    
    printf("<%s> \"%s\"\n", kind, line);
}

int main(int argc, char ** argv) {
    classify("ERROR: this is important");
    classify("who knows what this is?");
    classify("blah blah blah");
    classify("WARNING FROM THE FUTURE: DON'T EAT THAT ICE CREAM");
    return 0;
}

Makefile

line_classifier: line_classifier.o

%.c: %.rl
	ragel -G2 $^ -o $@

result.txt

<signal> "ERROR: this is important"
<unknown> "who knows what this is?"
<noise> "blah blah blah"
<signal> "WARNING FROM THE FUTURE: DON'T EAT THAT ICE CREAM"