Add checking to ASN.1 IA5, UTF8, Printable strings

include/gmssl/asn1.h

@@ -169,7 +169,6 @@ int asn1_object_identifier_print(FILE *fp, int fmt, int ind, const char *label,
 #define asn1_implicit_enumerated_to_der(i,val,out,outlen) asn1_int_to_der_ex(ASN1_TAG_IMPLICIT(i),val,out,outlen)
 #define asn1_implicit_enumerated_from_der(i,val,in,inlen) asn1_int_from_der_ex(ASN1_TAG_IMPLICIT(i),val,in,inlen)
 
-int asn1_utf8_string_check(const char *d, size_t dlen);
 int asn1_utf8_string_to_der_ex(int tag, const char *d, size_t dlen, uint8_t **out, size_t *outlen);
 int asn1_utf8_string_from_der_ex(int tag, const char **d, size_t *dlen, const uint8_t **in, size_t *inlen);
 #define asn1_utf8_string_to_der(d,dlen,out,outlen) asn1_utf8_string_to_der_ex(ASN1_TAG_UTF8String,d,dlen,out,outlen)
@@ -177,7 +176,6 @@ int asn1_utf8_string_from_der_ex(int tag, const char **d, size_t *dlen, const ui
 #define asn1_implicit_utf8_string_to_der(i,d,dlen,out,outlen) asn1_utf8_string_to_der_ex(ASN1_TAG_IMPLICIT(i),d,dlen,out,outlen)
 #define asn1_implicit_utf8_string_from_der(i,d,dlen,in,inlen) asn1_utf8_string_from_der_ex(ASN1_TAG_IMPLICIT(i),d,dlen,in,inlen)
 
-int asn1_printable_string_check(const char *d, size_t dlen);
 int asn1_printable_string_case_ignore_match(const char *a, size_t alen, const char *b, size_t blen);
 int asn1_printable_string_to_der_ex(int tag, const char *d, size_t dlen, uint8_t **out, size_t *outlen);
 int asn1_printable_string_from_der_ex(int tag, const char **d, size_t *dlen, const uint8_t **in, size_t *inlen);
@@ -186,7 +184,6 @@ int asn1_printable_string_from_der_ex(int tag, const char **d, size_t *dlen, con
 #define asn1_implicit_printable_string_to_der(i,d,dlen,out,outlen) asn1_printable_string_to_der_ex(ASN1_TAG_IMPLICIT(i),d,dlen,out,outlen)
 #define asn1_implicit_printable_string_from_der(i,d,dlen,in,inlen) asn1_printable_string_from_der_ex(ASN1_TAG_IMPLICIT(i),d,dlen,in,inlen)
 
-int asn1_ia5_string_check(const char *d, size_t dlen);
 int asn1_ia5_string_to_der_ex(int tag, const char *d, size_t dlen, uint8_t **out, size_t *outlen);
 int asn1_ia5_string_from_der_ex(int tag, const char **d, size_t *dlen, const uint8_t **in, size_t *inlen);
 #define asn1_ia5_string_to_der(d,dlen,out,outlen) asn1_ia5_string_to_der_ex(ASN1_TAG_IA5String,d,dlen,out,outlen)
@@ -194,6 +191,10 @@ int asn1_ia5_string_from_der_ex(int tag, const char **d, size_t *dlen, const uin
 #define asn1_implicit_ia5_string_to_der(i,d,dlen,out,outlen) asn1_ia5_string_to_der_ex(ASN1_TAG_IMPLICIT(i),d,dlen,out,outlen)
 #define asn1_implicit_ia5_string_from_der(i,d,dlen,in,inlen) asn1_ia5_string_from_der_ex(ASN1_TAG_IMPLICIT(i),d,dlen,in,inlen)
 
+
+int asn1_string_is_printable_string(const char *d, size_t dlen);
+int asn1_string_is_ia5_string(const char *d, size_t dlen);
+int asn1_string_is_utf8_string(const char *d, size_t dlen);
 int asn1_string_print(FILE *fp, int fmt, int ind, const char *label, int tag, const uint8_t *d, size_t dlen);
 
 #define ASN1_UTC_TIME_LEN		(sizeof("YYMMDDHHMMSSZ")-1)

src/asn1.c

@@ -113,12 +113,73 @@ int asn1_tag_is_cstring(int tag)
 	return 0;
 }
 
-int asn1_utf8_string_check(const char *a, size_t alen)
+/*
+utf-8 character encoding
+ 	1-byte: 0xxxxxxx
+	2-byte: 110xxxxx 10xxxxxx
+	3-byte: 1110xxxx 10xxxxxx 10xxxxxx
+	4-byte: 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
+*/
+int asn1_utf8char_from_bytes(uint32_t *c, const uint8_t **pin, size_t *pinlen)
 {
+	uint32_t utf8char;
+	const uint8_t *in = *pin;
+	size_t inlen = *pinlen;
+	uint32_t utf8char_len, i;
+
+	if (!inlen) {
+		return 0;
+	}
+
+	if ((in[0] & 0x80) == 0x00) {
+		utf8char_len = 1;
+	} else if ((in[0] & 0xe0) == 0xc0) {
+		utf8char_len = 2;
+	} else if ((in[0] & 0xf0) == 0xe0) {
+		utf8char_len = 3;
+	} else if ((in[0] & 0xf8) == 0xf0) {
+		utf8char_len = 4;
+	} else {
+		//error_print();
+		return -1;
+	}
+
+	if (inlen < utf8char_len) {
+		//error_print();
+		return -1;
+	}
+
+	utf8char = in[0];
+	for (i = 1; i < utf8char_len; i++) {
+		if ((in[i] & 0x60) != 0x80) {
+			//error_print();
+			return -1;
+		}
+		utf8char = (utf8char << 8) | in[i];
+	}
+
+	*c = utf8char;
+	(*pin) += utf8char_len;
+	(*pinlen) -= utf8char_len;
 	return 1;
 }
 
-static int asn1_char_is_printable(int a)
+int asn1_string_is_utf8_string(const char *a, size_t alen)
+{
+	uint32_t utf8char;
+
+	if (!a || !alen) {
+		return 0;
+	}
+	while (alen) {
+		if (asn1_utf8char_from_bytes(&utf8char, (const uint8_t **)&a, &alen) != 1) {
+			return 0;
+		}
+	}
+	return 1;
+}
+
+int asn1_char_is_printable(int a)
 {
 	if (isalpha(a) || isdigit(a)) {
 		return 1;
@@ -132,6 +193,7 @@ static int asn1_char_is_printable(int a)
 	return 0;
 }
 
+/*
 int asn1_printable_string_check(const char *a, size_t alen)
 {
 	size_t i;
@@ -143,6 +205,18 @@ int asn1_printable_string_check(const char *a, size_t alen)
 	}
 	return 1;
 }
+*/
+
+int asn1_string_is_printable_string(const char *a, size_t alen)
+{
+	size_t i;
+	for (i = 0; i < alen; i++) {
+		if (asn1_char_is_printable(a[i]) != 1) {
+			return 0;
+		}
+	}
+	return 1;
+}
 
 int asn1_printable_string_case_ignore_match(const char *a, size_t alen,
 	const char *b, size_t blen)
@@ -177,8 +251,14 @@ int asn1_printable_string_case_ignore_match(const char *a, size_t alen,
 	return 1;
 }
 
-int asn1_ia5_string_check(const char *a, size_t alen)
+int asn1_string_is_ia5_string(const char *a, size_t alen)
 {
+	size_t i;
+	for (i = 0; i < alen; i++) {
+		if (!isascii(a[i])) {
+			return 0;
+		}
+	}
 	return 1;
 }
 
@@ -1314,17 +1394,56 @@ int asn1_string_from_der(int tag, const char **a, size_t *alen, const uint8_t **
 
 int asn1_utf8_string_from_der_ex(int tag, const char **a, size_t *alen, const uint8_t **in, size_t *inlen)
 {
-	return asn1_type_from_der(tag, (const uint8_t **)a, alen, in, inlen);
+	int ret;
+	if ((ret = asn1_type_from_der(tag, (const uint8_t **)a, alen, in, inlen)) != 1) {
+		if (ret < 0) error_print();
+		return ret;
+	}
+	if (!(*a) || !(*alen)) {
+		error_print();
+		return -1;
+	}
+	if (asn1_string_is_utf8_string(*a, *alen) != 1) {
+		error_print();
+		return -1;
+	}
+	return 1;
 }
 
 int asn1_printable_string_from_der_ex(int tag, const char **a, size_t *alen, const uint8_t **in, size_t *inlen)
 {
-	return asn1_type_from_der(tag, (const uint8_t **)a, alen, in, inlen);
+	int ret;
+	if ((ret = asn1_type_from_der(tag, (const uint8_t **)a, alen, in, inlen)) != 1) {
+		if (ret < 0) error_print();
+		return ret;
+	}
+	if (!(*a) || !(*alen)) {
+		error_print();
+		return -1;
+	}
+	if (asn1_string_is_printable_string(*a, *alen) != 1) {
+		error_print();
+		return -1;
+	}
+	return 1;
 }
 
 int asn1_ia5_string_from_der_ex(int tag, const char **a, size_t *alen, const uint8_t **in, size_t *inlen)
 {
-	return asn1_type_from_der(tag, (const uint8_t **)a, alen, in, inlen);
+	int ret;
+	if ((ret = asn1_type_from_der(tag, (const uint8_t **)a, alen, in, inlen)) != 1) {
+		if (ret < 0) error_print();
+		return ret;
+	}
+	if (!(*a) || !(*alen)) {
+		error_print();
+		return -1;
+	}
+	if (asn1_string_is_ia5_string(*a, *alen) != 1) {
+		error_print();
+		return -1;
+	}
+	return 1;
 }
 
 int asn1_utc_time_from_der_ex(int tag, time_t *t, const uint8_t **pin, size_t *pinlen)

src/x509_cer.c

@@ -528,7 +528,7 @@ static size_t _strlen(const char *s) { return s ? strlen(s) : 0; }
 static int x509_name_tag(const char *str)
 {
 	if (str) {
-		if (asn1_printable_string_check(str, strlen(str)) == 1)
+		if (asn1_string_is_printable_string(str, strlen(str)) == 1)
 			return ASN1_TAG_PrintableString;
 		else	return ASN1_TAG_UTF8String;
 	}

src/x509_ext.c

@@ -1,5 +1,5 @@
 /*
- *  Copyright 2014-2022 The GmSSL Project. All Rights Reserved.
+ *  Copyright 2014-2023 The GmSSL Project. All Rights Reserved.
  *
  *  Licensed under the Apache License, Version 2.0 (the License); you may
  *  not use this file except in compliance with the License.
@@ -527,7 +527,7 @@ int x509_general_names_add_general_name(uint8_t *gns, size_t *gnslen, size_t max
 	case X509_gn_rfc822_name:
 	case X509_gn_dns_name:
 	case X509_gn_uniform_resource_identifier:
-		if (asn1_ia5_string_check((char *)d, dlen) != 1) {
+		if (asn1_string_is_ia5_string((char *)d, dlen) != 1) {
 			error_print();
 			return -1;
 		}

tests/asn1test.c

@@ -370,7 +370,7 @@ static int test_asn1_printable_string(void)
 	char *tests[] = {
 		"hello",
 		"world",
-		"Just do it!",
+		"Just do it", // string "Just do it!" include invalid '!'
 	};
 	uint8_t buf[256];
 	uint8_t *p = buf;
@@ -408,12 +408,12 @@ static int test_asn1_printable_string(void)
 static int test_asn1_printable_string_check(void)
 {
 	char *printable_str = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789 '()+,-./:=?";
-	if (asn1_printable_string_check(printable_str, strlen(printable_str)) != 1) {
+	if (asn1_string_is_printable_string(printable_str, strlen(printable_str)) != 1) {
 		error_print();
 		return -1;
 	}
 
-	if (asn1_printable_string_check("a*b", 3) == 1) {
+	if (asn1_string_is_printable_string("a*b", 3) == 1) {
 		error_print();
 		return -1;
 	}