courier-imap: enforce UTF8-correctness as per RFC-6855.

3 files changed, 126 insertions, 31 deletions

diff --git a/imap/imapd.c b/imap/imapd.c
index 001dfcb..cf31af4 100644
--- a/imap/imapd.c
+++ b/imap/imapd.c
@@ -603,17 +603,17 @@ char	*mailbox;
 		curtoken->tokentype != IT_ATOM &&
 		curtoken->tokentype != IT_QUOTED_STRING)
 	{
-		mailbox=0;
+		writes(tag);
+		writes(" BAD Invalid command\r\n");
+		return (0);
 	}
-	else
-	{
-		if (ok_hierarchy && (mailbox=strrchr(curtoken->tokenbuf,
-			HIERCH)) && mailbox[1] == 0)
-				*mailbox=0;
 
-		mailbox=decode_valid_mailbox(curtoken->tokenbuf,
-			autosubscribe);
-	}
+	if (ok_hierarchy && (mailbox=strrchr(curtoken->tokenbuf,
+					     HIERCH)) && mailbox[1] == 0)
+		*mailbox=0;
+
+	mailbox=decode_valid_mailbox(curtoken->tokenbuf,
+				     autosubscribe);
 
 	if ( mailbox == 0)
 	{
@@ -3162,8 +3162,9 @@ static int aclcmd(const char *tag)
 				curtoken->tokentype != IT_ATOM &&
 				curtoken->tokentype != IT_NUMBER)
 			{
-				errno=EINVAL;
-				return -1;
+				writes(tag);
+				writes(" BAD Invalid command\r\n");
+				return 0;
 			}
 
 			p=imap_foldername_to_filename(enabled_utf8,
@@ -3211,6 +3212,10 @@ static int aclcmd(const char *tag)
 			free_tempmailboxlist(mblist);
 		}
 		break;
+	case IT_ERROR:
+		writes(tag);
+		writes(" BAD Invalid command\r\n");
+		return 0;
 	}
 
 	rc=0;
@@ -4305,7 +4310,11 @@ int	uid=0;
 			if (curtoken->tokentype != IT_QUOTED_STRING &&
 				curtoken->tokentype != IT_ATOM &&
 				curtoken->tokentype != IT_NUMBER)
-				return (-1);
+			{
+				writes(tag);
+				writes(" BAD Invalid command\r\n");
+				return (0);
+			}
 			reference=imap_foldername_to_filename
 				(enabled_utf8, curtoken->tokenbuf);
 			if (!reference)
@@ -4320,12 +4329,20 @@ int	uid=0;
 			if (curtoken->tokentype != IT_QUOTED_STRING &&
 				curtoken->tokentype != IT_ATOM &&
 				curtoken->tokentype != IT_NUMBER)
-				return (-1);
+			{
+				free(reference);
+				writes(tag);
+				writes(" BAD Invalid command\r\n");
+				return(0);
+			}
 			name=imap_foldername_to_filename(enabled_utf8,
 							 curtoken->tokenbuf);
 
 			if (!name)
+			{
+				free(reference);
 				return -1;
+			}
 		}
 		if (nexttoken()->tokentype != IT_EOL)	return (-1);
 
@@ -4366,7 +4383,11 @@ int	uid=0;
 		if (tok->tokentype != IT_NUMBER &&
 			tok->tokentype != IT_ATOM &&
 			tok->tokentype != IT_QUOTED_STRING)
-			return (-1);
+		{
+			writes(tag);
+			writes(" BAD Invalid command\r\n");
+			return (0);
+		}
 
 		mailbox=imap_foldername_to_filename(enabled_utf8,
 						    tok->tokenbuf);
@@ -4439,7 +4460,11 @@ int	uid=0;
 		if (curtoken->tokentype != IT_NUMBER &&
 			curtoken->tokentype != IT_ATOM &&
 			curtoken->tokentype != IT_QUOTED_STRING)
-			return (-1);
+		{
+			writes(tag);
+			writes(" BAD Invalid command\r\n");
+			return (0);
+		}
 
 		mailbox=imap_foldername_to_filename(enabled_utf8,
 						    curtoken->tokenbuf);
@@ -4716,7 +4741,11 @@ int	uid=0;
 		if (curtoken->tokentype != IT_NUMBER &&
 			curtoken->tokentype != IT_ATOM &&
 			curtoken->tokentype != IT_QUOTED_STRING)
-			return (-1);
+		{
+			writes(tag);
+			writes(" BAD Invalid command\r\n");
+			return 0;
+		}
 
 		isdummy=0;
 
@@ -4877,7 +4906,11 @@ int	uid=0;
 		if (curtoken->tokentype != IT_NUMBER &&
 			curtoken->tokentype != IT_ATOM &&
 			curtoken->tokentype != IT_QUOTED_STRING)
-			return (-1);
+		{
+			writes(tag);
+			writes(" BAD Invalid command\r\n");
+			return (0);
+		}
 
 		p=strrchr(curtoken->tokenbuf, HIERCH);
 		if (p && p[1] == '\0')		/* Ignore hierarchy DELETE */
@@ -4981,7 +5014,7 @@ int	uid=0;
 		    curtoken->tokentype != IT_QUOTED_STRING)
 		{
 			writes(tag);
-			writes(" NO Invalid mailbox\r\n");
+			writes(" BAD Invalid command\r\n");
 			return (0);
 		}
 
@@ -5034,7 +5067,9 @@ int	uid=0;
 			curtoken->tokentype != IT_QUOTED_STRING)
 		{
 			maildir_info_destroy(&mi1);
-			return (-1);
+			writes(tag);
+			writes(" BAD Invalid command\r\n");
+			return (0);
 		}
 
 		if ((p=strrchr(curtoken->tokenbuf, HIERCH)) && p[1] == 0)
@@ -5206,7 +5241,11 @@ int	uid=0;
 		if (curtoken->tokentype != IT_NUMBER &&
 			curtoken->tokentype != IT_ATOM &&
 			curtoken->tokentype != IT_QUOTED_STRING)
-			return (-1);
+		{
+			writes(tag);
+			writes(" BAD Invalid command\r\n");
+			return (0);
+		}
 
 		p=strrchr(curtoken->tokenbuf, HIERCH);
 		if (p && p[1] == '\0')		/* Ignore hierarchy DELETE */
@@ -5286,7 +5325,11 @@ int	uid=0;
 		if (curtoken->tokentype != IT_NUMBER &&
 			curtoken->tokentype != IT_ATOM &&
 			curtoken->tokentype != IT_QUOTED_STRING)
-			return (-1);
+		{
+			writes(tag);
+			writes(" BAD Invalid command\r\n");
+			return (0);
+		}
 
 		p=strrchr(curtoken->tokenbuf, HIERCH);
 		if (p && p[1] == '\0')		/* Ignore hierarchy DELETE */
@@ -5930,6 +5973,14 @@ int	uid=0;
 		if (curtoken->tokentype == IT_ATOM &&
 			strcmp(curtoken->tokenbuf, "CHARSET") == 0)
 		{
+			if (enabled_utf8)
+			{
+				writes(tag);
+				writes(" NO CHARSET is not valid in UTF8 mode "
+				       "as per RFC 6855\r\n");
+				return (0);
+			}
+
 			curtoken=nexttoken();
 			if (curtoken->tokentype != IT_ATOM &&
 				curtoken->tokentype != IT_QUOTED_STRING)
@@ -6279,7 +6330,9 @@ int	uid=0;
 			curtoken->tokentype != IT_QUOTED_STRING)
 		{
 			free(msgset);
-			return (-1);
+			writes(tag);
+			writes(" BAD Invalid command\r\n");
+			return (0);
 		}
 
 		mailbox=decode_valid_mailbox(curtoken->tokenbuf, 1);
diff --git a/imap/imaplogin.c b/imap/imaplogin.c
index a445a35..8e2f9c9 100644
--- a/imap/imaplogin.c
+++ b/imap/imaplogin.c
@@ -37,7 +37,7 @@
 #include	"tcpd/spipe.h"
 #include	"numlib/numlib.h"
 #include	"tcpd/tlsclient.h"
-
+#include	"imapd.h"
 
 FILE *debugfile=0;
 extern void initcapability();
@@ -47,6 +47,7 @@ extern int have_starttls();
 extern int tlsrequired();
 extern int authenticate(const char *, char *, int);
 unsigned long header_count=0, body_count=0;	/* Dummies */
+int enabled_utf8=0;
 
 extern unsigned long bytes_received_count; /* counter for received bytes (imaptoken.c) */
 extern unsigned long bytes_sent_count; /* counter for sent bytes (imapwrite.c) */
diff --git a/imap/imaptoken.c b/imap/imaptoken.c
index 7df8846..82dda7c 100644
--- a/imap/imaptoken.c
+++ b/imap/imaptoken.c
@@ -1,5 +1,5 @@
 /*
-** Copyright 1998 - 2005 Double Precision, Inc.
+** Copyright 1998 - 2018 Double Precision, Inc.
 ** See COPYING for distribution information.
 */
 
@@ -16,6 +16,8 @@
 #include	<sys/types.h>
 #include	<sys/time.h>
 #include	"numlib/numlib.h"
+#include	"imapd.h"
+#include	<courier-unicode.h>
 #if	HAVE_UNISTD_H
 #include	<unistd.h>
 #endif
@@ -222,6 +224,11 @@ void smap_readline(char *buffer, size_t bufsize)
 
 #endif
 
+static int ignore_output_func(const char *ptr, size_t cnt, void *ignore)
+{
+	return 0;
+}
+
 static struct imaptoken *do_readtoken(int touc)
 {
 int	c=0;
@@ -269,6 +276,8 @@ unsigned l;
 
 	if (c == '"')
 	{
+		int bit8=0;
+
 		l=0;
 		while ((c=READ()) != '"')
 		{
@@ -280,13 +289,45 @@ unsigned l;
 				curtoken.tokentype=IT_ERROR;
 				return (&curtoken);
 			}
-			if (l < 8192)
+			if (l >= 8192)
 			{
-				appendch(c);
+				fprintf(stderr, "ERR: Quoted string literal "
+					"overflow, ip=[%s]\n",
+					getenv("TCPREMOTEIP"));
+				curtoken.tokentype=IT_ERROR;
 			}
+			if (c & 0x80)
+				bit8=1;
+			appendch(c);
 		}
 		appendch(0);
 		curtoken.tokentype=IT_QUOTED_STRING;
+
+		/*
+		** Strings must be valid UTF-8. Shortcut check.
+		*/
+
+		if (bit8)
+		{
+			int errptr=0;
+
+			unicode_convert_handle_t h=
+				unicode_convert_init("utf-8",
+						     unicode_u_ucs4_native,
+						     ignore_output_func,
+						     (void *)0);
+
+			if (unicode_convert(h, curtoken.tokenbuf,
+					    strlen(curtoken.tokenbuf)))
+			{
+				curtoken.tokentype=IT_ERROR;
+			}
+
+			if (unicode_convert_deinit(h, &errptr) || errptr)
+			{
+				curtoken.tokentype=IT_ERROR;
+			}
+		}
 		return (&curtoken);
 	}
 
@@ -339,9 +380,10 @@ unsigned l;
 	}
 
 	while (c != '\r' && c != '\n'
-		&& !isspace((int)(unsigned char)c)
-		&& c != '\\' && c != '"' && c != LPAREN_CHAR && c != RPAREN_CHAR
-		&& c != '{' && c != '}' && c != LBRACKET_CHAR && c != RBRACKET_CHAR)
+	       && !isspace((int)(unsigned char)c)
+	       && (((unsigned char)c) & 0x80) == 0
+	       && c != '\\' && c != '"' && c != LPAREN_CHAR && c != RPAREN_CHAR
+	       && c != '{' && c != '}' && c != LBRACKET_CHAR && c != RBRACKET_CHAR)
 	{
 		curtoken.tokentype=IT_ATOM;
 		if (l < IT_MAX_ATOM_SIZE)
@@ -352,7 +394,7 @@ unsigned l;
 		}
 		else
 		{
-			write_error_exit("max atom size too small");  
+			write_error_exit("max atom size too small");
 		}
 		c=READ();
 	}
@@ -564,4 +606,3 @@ int ismsgset_str(const char *p)
 	if (*p)	return (0);
 	return (1);
 }
-