util/base64: skip any invalid char for RFC2045

RFC 2045 states that any invalid character should be skipped over, this
is the RFC used by mime handler in Suricata code to deal with base64
encoded data.
So far, only spaces were skipped as a part of implementation of this
RFC, extend it to also skip over any other invalid character. Add
corresponding test.

src/util-base64.c

@@ -107,9 +107,9 @@ Base64Ecode DecodeBase64(uint8_t *dest, uint32_t dest_size, const uint8_t *src,
         /* Get decimal representation */
         val = GetBase64Value(src[i]);
         if (val < 0) {
-            if ((mode == BASE64_MODE_RFC2045) && (src[i] == ' ')) {
+            if (mode == BASE64_MODE_RFC2045 && src[i] != '=') {
                 if (bbidx == 0) {
-                    /* Special case where last block of data has a leading space */
+                    /* Special case where last block of data has a leading space or invalid char */
                     leading_sp++;
                 }
                 sp++;
@@ -156,7 +156,7 @@ Base64Ecode DecodeBase64(uint8_t *dest, uint32_t dest_size, const uint8_t *src,
         }
     }
 
-    if (!valid && mode == BASE64_MODE_RFC4648) {
+    if (bbidx > 0 && bbidx < 4 && ((!valid && mode == BASE64_MODE_RFC4648))) {
         padding = B64_BLOCK - bbidx;
         *decoded_bytes += ASCII_BLOCK - padding;
         DecodeBase64Block(dptr, b64);
@@ -304,18 +304,22 @@ static int B64TestVectorsRFC2045(void)
     const char *fin_str8 = "foobar";
 
     const char *src9 = "Zm$9vYm.Fy";
-    const char *fin_str9 = "";
+    const char *fin_str9 = "foobar";
 
-    TEST_RFC2045(src1, fin_str1, strlen(fin_str1), strlen(fin_str1), strlen(src1), BASE64_ECODE_OK);
+    const char *src10 = "Y21Wd2IzSjBaVzFoYVd4bWNtRjFaRUJoZEc4dVoyOTJMbUYxOmpqcHh4b3Rhb2w%5d";
+    const char *fin_str10 = "cmVwb3J0ZW1haWxmcmF1ZEBhdG8uZ292LmF1:jjpxxotaol9t";
+
+    TEST_RFC2045(src1, fin_str1, ASCII_BLOCK * 2, strlen(fin_str1), strlen(src1), BASE64_ECODE_OK);
     TEST_RFC2045(src2, fin_str2, ASCII_BLOCK * 2, strlen(fin_str2), strlen(src2), BASE64_ECODE_OK);
     TEST_RFC2045(src3, fin_str3, ASCII_BLOCK * 2, strlen(fin_str3), strlen(src3), BASE64_ECODE_OK);
     TEST_RFC2045(src4, fin_str4, ASCII_BLOCK * 2, strlen(fin_str4), strlen(src4), BASE64_ECODE_OK);
     TEST_RFC2045(src5, fin_str5, ASCII_BLOCK * 2, strlen(fin_str5), strlen(src5), BASE64_ECODE_OK);
     TEST_RFC2045(src6, fin_str6, ASCII_BLOCK * 2, strlen(fin_str6), strlen(src6), BASE64_ECODE_OK);
     TEST_RFC2045(src7, fin_str7, ASCII_BLOCK * 2, strlen(fin_str7), strlen(src7), BASE64_ECODE_OK);
     TEST_RFC2045(src8, fin_str8, ASCII_BLOCK * 2, strlen(fin_str8), strlen(src8), BASE64_ECODE_OK);
-    TEST_RFC2045(src9, fin_str9, ASCII_BLOCK * 2, 0, 0,
-            BASE64_ECODE_ERR); // TODO this should be accepted just like the previous string
+    TEST_RFC2045(src9, fin_str9, ASCII_BLOCK * 2, strlen(fin_str9), strlen(src9), BASE64_ECODE_OK);
+    TEST_RFC2045(src10, fin_str10, strlen(fin_str10) + 3, strlen(fin_str10), strlen(src10),
+            BASE64_ECODE_OK);
     PASS;
 }
 

src/util-base64.h

@@ -45,8 +45,8 @@ typedef enum {
      * BASE64("fooba") = "Zm9vYmE="
      * BASE64("foobar") = "Zm9vYmFy"
      * BASE64("foobar") = "Zm 9v Ym Fy"   <-- Notice how the spaces are ignored
-     * BASE64("f") = "Zm$9vYm.Fy"    # TODO according to RFC, All line breaks or *other characters*
-     * not found in base64 alphabet must be ignored by decoding software
+     * BASE64("foobar") = "Zm$9vYm.Fy"    # According to RFC 2045, All line breaks or *other
+     * characters* not found in base64 alphabet must be ignored by decoding software
      * */
     BASE64_MODE_RFC2045, /* SPs are allowed during transfer but must be skipped by Decoder */
     BASE64_MODE_STRICT,