lexer.rl

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

%%{
    machine test_lexer;

    newline  = [\n|\r\n]+;
    integer  = ('+'|'-')?[0-9]+;
    operator = ['+''-''*''/''=''%''^']+;

    main := |*
        newline => { line += 1; column = 0; };

        integer => {
            Lexer_emit("integer", ts, te, line, column);

            column = Lexer_advance_column(ts, te, column);
        };

        operator => {
            Lexer_emit("operator", ts, te, line, column);

            column = Lexer_advance_column(ts, te, column);
        };

        space => {
            column = Lexer_advance_column(ts, te, column);
        };
    *|;
}%%

%% write data;

typedef struct {
    const char *name;
    const char *value;
    int column_start;
    int column_end;
    int line;
} Lexer_Token;

int Lexer_advance_column(const char *ts, const char *te, int column)
{
    int length = te - ts;

    return column + length;
}

Lexer_Token *Lexer_emit(
    const char *name,
    const char *ts,
    const char *te,
    int line,
    int column
)
{
    int length = te - ts;

    Lexer_Token *token = malloc(sizeof(Lexer_Token));

    token->name         = name;
    token->value        = strndup(ts, length);
    token->column_start = column;
    token->column_end   = column + length;
    token->line         = line;

    return token;
}

void Lexer_lex(char *buffer)
{
    const char *p  = buffer;
    const char *pe = buffer + strlen(buffer);
    const char *ts, *te;

    char *eof = 0;

    int line   = 1;
    int column = 0;
    int act    = 0;
    int cs     = 0;

    Lexer_Token tokens[10];

    %% write init;
    %% write exec noend;
}

int main()
{
    Lexer_lex("10 + 20");

    return 0;
}