o Anchor support for the regex engine.

lib/regex/matcher.c

@@ -330,20 +330,33 @@ struct matcher *matcher_create(struct pool *mem, const char **patterns, int num)
 	return NULL;
 }
 
+static struct dfa_state *
+_step_matcher(unsigned char c, struct dfa_state *cs, int *r)
+{
+	if (!(cs = cs->lookup[c]))
+		return NULL;
+
+	if (cs->final && (cs->final > *r))
+		*r = cs->final;
+
+	return cs;
+}
+
 int matcher_run(struct matcher *m, const char *b)
 {
 	struct dfa_state *cs = m->start;
 	int r = 0;
 
-	for (; *b; b++) {
+	if (!(cs = _step_matcher(HAT_CHAR, cs, &r)))
+		goto out;
 
-		if (!(cs = cs->lookup[(int) (unsigned char) *b]))
+	for (; *b; b++)
-			break;
+		if (!(cs = _step_matcher(*b, cs, &r)))
+			goto out;
 
-		if (cs->final && (cs->final > r))
+	_step_matcher(DOLLAR_CHAR, cs, &r);
-			r = cs->final;
-	}
 
+ out:
 	/* subtract 1 to get back to zero index */
 	return r - 1;
 }

lib/regex/parse_rx.c

@@ -22,6 +22,14 @@ struct parse_sp {		/* scratch pad for the parsing process */
 
 static struct rx_node *_or_term(struct parse_sp *ps);
 
+static void _single_char(struct parse_sp *ps, unsigned int c, const char *ptr)
+{
+	ps->type = 0;
+	ps->cursor = ptr + 1;
+	bit_clear_all(ps->charset);
+	bit_set(ps->charset, c);
+}
+
 /*
  * Get the next token from the regular expression.
  * Returns: 1 success, 0 end of input, -1 error.
@@ -125,12 +133,18 @@ static int _get_token(struct parse_sp *ps)
 	case '+':
 	case '?':
 	case '|':
-	case '^':
-	case '$':
 		ps->type = (int) *ptr;
 		ps->cursor = ptr + 1;
 		break;
 
+	case '^':
+		_single_char(ps, HAT_CHAR, ptr);
+		break;
+
+	case '$':
+		_single_char(ps, DOLLAR_CHAR, ptr);
+		break;
+
 	case '.':
 		/* The 'all but newline' character set */
 		ps->type = 0;

lib/regex/parse_rx.h

@@ -15,11 +15,17 @@ enum {
 	PLUS,
 	OR,
 	QUEST,
-	CHARSET,
+	CHARSET
-	HAT,
-	DOLLAR
 };
 
+/*
+ * We're never going to be running the regex on non-printable
+ * chars, so we can use a couple of these chars to represent the
+ * start and end of a string.
+ */
+#define HAT_CHAR 0x2
+#define DOLLAR_CHAR 0x2
+
 struct rx_node {
 	int type;
 	bitset_t charset;