Rework symbols classes in url parser.

1 files changed, 160 insertions, 40 deletions

diff --git a/src/libserver/url.c b/src/libserver/url.c
index fb3e961e6..23b052b50 100644
--- a/src/libserver/url.c
+++ b/src/libserver/url.c
@@ -152,47 +152,167 @@ struct url_match_scanner {
 
 struct url_match_scanner *url_scanner = NULL;
 
-static guchar url_scanner_table[256] = {
-		1, 1, 1, 1, 1, 1, 1, 1, 1, 9, 9, 1, 1, 9, 1, 1,
-		1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
-		24, 128, 160, 128, 128, 128, 128, 128, 160, 160, 128, 128, 160, 192,
-		160, 160,
-		68, 68, 68, 68, 68, 68, 68, 68, 68, 68, 160, 160, 32, 128, 32, 128,
-		160, 66, 66, 66, 66, 66, 66, 66, 66, 66, 66, 66, 66, 66, 66, 66,
-		66, 66, 66, 66, 66, 66, 66, 66, 66, 66, 66, 160, 160, 160, 128, 192,
-		128, 66, 66, 66, 66, 66, 66, 66, 66, 66, 66, 66, 66, 66, 66, 66,
-		66, 66, 66, 66, 66, 66, 66, 66, 66, 66, 66, 128, 128, 128, 128, 1,
-		1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
-		1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
-		1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
-		1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
-		1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
-		1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
-		1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
-		1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1
+enum {
+	IS_LWSP = (1 << 0),
+	IS_DOMAIN = (1 << 1),
+	IS_URLSAFE = (1 << 2),
+	IS_MAILSAFE = (1 << 3),
+	IS_DOMAIN_END = (1 << 4)
 };
 
-enum {
-	IS_CTRL = (1 << 0),
-	IS_ALPHA = (1 << 1),
-	IS_DIGIT = (1 << 2),
-	IS_LWSP = (1 << 3),
-	IS_SPACE = (1 << 4),
-	IS_SPECIAL = (1 << 5),
-	IS_DOMAIN = (1 << 6),
-	IS_URLSAFE = (1 << 7)
+static const unsigned int url_scanner_table[256] = {
+		0, 0, 0, 0, 0, 0, 0, 0, 0, IS_LWSP, IS_LWSP, IS_LWSP, IS_LWSP, IS_LWSP, 0, 0,
+		0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, IS_LWSP /*   */,
+		IS_MAILSAFE /* ! */, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* " */,
+		IS_MAILSAFE /* # */, IS_MAILSAFE /* $ */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* % */, 0 /* & */, IS_MAILSAFE /* ' */,
+		IS_MAILSAFE /* ( */, IS_MAILSAFE /* ) */, IS_MAILSAFE /* * */,
+		IS_MAILSAFE /* + */, IS_MAILSAFE /* , */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* - */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* . */, IS_DOMAIN_END|IS_MAILSAFE /* / */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* 0 */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* 1 */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* 2 */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* 3 */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* 4 */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* 5 */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* 6 */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* 7 */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* 8 */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* 9 */, IS_DOMAIN_END|IS_MAILSAFE /* : */,
+		IS_MAILSAFE /* ; */, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* < */, 0 /* = */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* > */, IS_DOMAIN_END /* ? */, 0 /* @ */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* A */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* B */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* C */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* D */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* E */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* F */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* G */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* H */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* I */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* J */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* K */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* L */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* M */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* N */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* O */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* P */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* Q */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* R */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* S */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* T */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* U */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* V */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* W */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* X */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* Y */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* Z */, IS_MAILSAFE /* [ */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* \ */, IS_MAILSAFE /* ] */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* ^ */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* _ */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* ` */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* a */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* b */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* c */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* d */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* e */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* f */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* g */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* h */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* i */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* j */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* k */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* l */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* m */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* n */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* o */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* p */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* q */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* r */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* s */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* t */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* u */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* v */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* w */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* x */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* y */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* z */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* { */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* | */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* } */,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE /* ~ */,
+		0, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE, IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE,
+		IS_URLSAFE|IS_DOMAIN|IS_MAILSAFE
 };
 
-#define is_ctrl(x) ((url_scanner_table[(guchar)(x)] & IS_CTRL) != 0)
-#define is_lwsp(x) ((url_scanner_table[(guchar)(x)] & IS_LWSP) != 0)
-#define is_atom(x) ((url_scanner_table[(guchar)(x)] & (IS_SPECIAL | IS_SPACE | \
-    IS_CTRL)) == 0)
-#define is_usersafe(x) ((url_scanner_table[(guchar)(x)] & (IS_CTRL | IS_SPACE)) == 0)
-#define is_alpha(x) ((url_scanner_table[(guchar)(x)] & IS_ALPHA) != 0)
-#define is_digit(x) ((url_scanner_table[(guchar)(x)] & IS_DIGIT) != 0)
+#define is_lwsp(x) ((url_scanner_table[(guint)(x)] & IS_LWSP) != 0)
+#define is_mailsafe(x) ((url_scanner_table[(guint)(x)] & (IS_MAILSAFE)) != 0)
 #define is_domain(x) ((url_scanner_table[(guchar)(x)] & IS_DOMAIN) != 0)
-#define is_urlsafe(x) ((url_scanner_table[(guchar)(x)] & (IS_ALPHA | IS_DIGIT | \
-    IS_URLSAFE)) != 0)
+#define is_urlsafe(x) ((url_scanner_table[(guchar)(x)] & (IS_URLSAFE)) != 0)
 
 const gchar *
 rspamd_url_strerror (enum uri_errno err)
@@ -425,7 +545,7 @@ rspamd_mailto_parse (struct http_parser_url *u, const gchar *str, gsize len,
 					SET_U (u, UF_USERINFO);
 					st = parse_at;
 				}
-				else if (!is_usersafe (t)) {
+				else if (!is_mailsafe (t)) {
 					goto out;
 				}
 				p++;
@@ -449,7 +569,7 @@ rspamd_mailto_parse (struct http_parser_url *u, const gchar *str, gsize len,
 				st = parse_query;
 				break;
 			case parse_query:
-				if (!is_atom (t)) {
+				if (!is_mailsafe (t)) {
 					goto out;
 				}
 				p++;
@@ -1509,7 +1629,7 @@ url_email_end (struct url_callback_data *cb,
 		}
 
 		c = pos - 1;
-		while (c > cb->begin && is_usersafe (*c)) {
+		while (c > cb->begin && is_mailsafe (*c)) {
 			c --;
 		}
 		/* Rewind to the first alphanumeric character */
@@ -1523,7 +1643,7 @@ url_email_end (struct url_callback_data *cb,
 			p ++;
 		}
 		/* Rewind it again to avoid bad emails to be detected */
-		while (p > pos && !g_ascii_isalnum (*p)) {
+		while (p > pos && p < cb->end && !g_ascii_isalnum (*p)) {
 			p --;
 		}