Some extra stuff. More pragmas, and better EOF message

lime.bootstrap

@@ -7,8 +7,11 @@ to grammar
 : {$$=array();}
 toklist sym : $$[] = $2;
 toklist lit : $$[] = $2;
+toklist num : $$[] = $2;
+toklist str : $$[] = substr($2, 1, -1);
 to toklist
 sym '=' rhs : $$ = new lime_rewrite($1); $$->add_rhs($3);
+sym ':' rhs : $$ = new lime_rewrite($1); $$->add_rhs($3);
 rewrite '|' rhs : $$->add_rhs($3);
 to rewrite
 list : $$ = new lime_rhs($1, '');

lime.php

@@ -355,9 +355,10 @@ class state {
 }
 
 class sym {
-	public function __construct($name, $id) {
+	public function __construct($name, $id, $desc = null) {
 		$this->name = $name;
 		$this->id = $id;
+		$this->desc = $desc;
 		$this->term = true;	// Until proven otherwise.
 		$this->rule = array();
 		$this->config = array();
@@ -374,6 +375,10 @@ class sym {
 
 		return $out;
 	}
+
+	public function descr() {
+		return $this->descr ? $this->descr : $this->name;
+	}
 }
 
 class rule {
@@ -689,9 +694,9 @@ class lime {
 		$sym->rule[] = $r;
 	}
 
-	function sym($str) {
+	function sym($str, $description = null) {
 		if (!isset($this->sym[$str])) {
-			$this->sym[$str] = new sym($str, count($this->sym));
+			$this->sym[$str] = new sym($str, count($this->sym), $description);
 		}
 
 		return $this->sym[$str];
@@ -999,6 +1004,17 @@ class lime {
 		case 'class':
 			$this->parser_class = $args[0];
 			break;
+		case 'expect':
+			if (!ctype_digit($args[0])) {
+				emit(sprintf('Bad expect pragma: %s', $args[0]));
+				exit(1);
+			}
+
+			$this->expect_conflicts = $args[0];
+			break;
+		case 'token':
+			$this->sym($args[0], @$args[1]);
+			break;
 		default:
 			emit(sprintf('Bad Parser Pragma: (%s)', $type));
 			exit(1);
@@ -1233,7 +1249,7 @@ function lime_bootstrap() {
 	bug_unless(is_readable($bootstrap));
 
 	foreach(file($bootstrap) as $l) {
-		$a = explode(':', $l, 2);
+		$a = preg_split('~(?<=\s|^):(?=\s|$)~', $l, 2);
 
 		if (count($a) == 2) {
 			list($pattern, $code) = $a;

lime_scan_tokens.l

@@ -34,7 +34,7 @@ ALPHA		[a-zA-Z]
 DIGIT		[0-9]
 ALNUM		{ALPHA}|{DIGIT}
 WORD		{ALNUM}|_
-STOP		"."
+STOP		"."|";"
 
 SYM			{ALPHA}{WORD}*'*
 LIT			'.'
@@ -87,6 +87,10 @@ BLOCKCMT	"/*"({CC}|{CX})*{CT}
 	[[:space:]]	{}
 	{SYM} 		tok("sym");
 	{LIT}		tok("lit");
+	{DIGIT}		tok("num");
+	\"{DCHAR}*\" tok("str");
+	'{SCHAR}*'	tok("str");
+	
 	.			lit();
 }
 

parse_engine.php

@@ -41,8 +41,8 @@ class parse_unexpected_token extends parse_error {
 }
 
 class parse_premature_eof extends parse_error {
-	public function __construct() {
-		parent::__construct('Premature EOF');
+	public function __construct(array $expect) {
+		parent::__construct('Premature EOF, expected {' . implode(', ', $expect) . '}');
 	}
 }
 
@@ -221,6 +221,8 @@ class parse_engine {
 	private function premature_eof() {
 		$seen = array();
 
+		$expect = $this->get_steps();
+
 		while ($this->enter_error_tolerant_state()) {
 			if (isset($seen[$this->state()])) {
 				// This means that it's pointless to try here.
@@ -244,7 +246,7 @@ class parse_engine {
 			}
 		}
 
-		throw new parse_premature_eof();
+		throw new parse_premature_eof($expect);
 	}
 
 	private function current_row() {