shn.l

gistfile1.c

%{
#include <stdio.h>
#include <string.h>

enum yytokentype {
	COMMA = 258,
	SEMICOLON,
	LBRACKET,
	RBRACKET,
	LBRACE,
	RBRACE,
	FULLSTOP,
	COLON,

	IDENTIFIER,
	TAG,
	WORD,
	REGEXP,

	DEFSYMBOL,
	DEFPAR,
	DEFREGEXP,
	DEFDICT
};

char *yylval;
%}

ws [ \t\n\v\f\r]
ident [A-Za-z][A-Za-z0-9_]*
word [^.:);]+
regexp [^;]+
comment "#".+?"\n"

%s defsymbol
%s defpar
%s defregexp
%s defdict
%s words
%s word

%%

<*>{comment} {}
<*>{ws}+ {}


<INITIAL>"%defsymbol" {
	BEGIN(defsymbol);
	return DEFSYMBOL;
}
<INITIAL>"%defpar" {
	BEGIN(defpar);
	return DEFPAR;
}
<INITIAL>"%defdict" {
	BEGIN(defdict);
	return DEFDICT;
}


<defsymbol>{ident} {
	yylval = strdup(yytext);
	return IDENTIFIER;
}
<defsymbol>"," { return COMMA; }
<defsymbol>";" {
	BEGIN(INITIAL);
	return SEMICOLON;
}


<defpar>{ident} {
	yylval = strdup(yytext);
	return IDENTIFIER;
}
<defpar>"{" {
	BEGIN(words);
	return LBRACE;
}


<defdict>{ident} {
	yylval = strdup(yytext);
	return IDENTIFIER;
}
<defdict>"{" {
	BEGIN(words);
	return LBRACE;
}


<words>"}" {
	return RBRACE;
}
<words>";" {
	BEGIN(INITIAL);
	return SEMICOLON;
}
<words>"(" {
	BEGIN(word);
	return LBRACKET;
}
<words>"%regexp" {
	BEGIN(defregexp);
	return DEFREGEXP;
}


<word>"."{ident} {
	yylval = strdup(yytext + 1);
	return TAG;
}
<word>{word} {
	yylval = strdup(yytext);
	return WORD;
}
<word>"." { return FULLSTOP; }
<word>":" { return COLON; }
<word>")" { return RBRACKET; }
<word>";" {
	BEGIN(words);
	return SEMICOLON;
}

<defregexp>{regexp} {
	yylval = strdup(yytext);
	return REGEXP;
}
<defregexp>";" {
	BEGIN(words);
	return SEMICOLON;
}

%%

int main(int argc, char **argv) {
	int tok;
	while (tok = yylex()) {
		printf("%d", tok);
		if (tok == IDENTIFIER || tok == WORD || tok == REGEXP) {
			printf(" %s", yylval);
		}
		printf("\n");
	}
}