Revert base64 decoder optimizations to support chunked base64

Fixes issue #150
jstedfast · Aug 26, 2023 · b4de388 · b4de388
1 parent 35fe6a8
commit b4de388
Show file tree

Hide file tree

Showing 2 changed files with 38 additions and 56 deletions.
diff --git a/gmime/gmime-encodings.c b/gmime/gmime-encodings.c
@@ -60,7 +60,7 @@ static unsigned char gmime_base64_rank[256] = {
 	255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
 	255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
 	255,255,255,255,255,255,255,255,255,255,255, 62,255,255,255, 63,
-	 52, 53, 54, 55, 56, 57, 58, 59, 60, 61,255,255,255,255,255,255,
+	 52, 53, 54, 55, 56, 57, 58, 59, 60, 61,255,255,255,  0,255,255,
 	255,  0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14,
 	 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25,255,255,255,255,255,
 	255, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40,
@@ -523,65 +523,53 @@ g_mime_encoding_base64_encode_step (const unsigned char *inbuf, size_t inlen, un
 size_t
 g_mime_encoding_base64_decode_step (const unsigned char *inbuf, size_t inlen, unsigned char *outbuf, int *state, guint32 *save)
 {
-	register const unsigned char *inptr = inbuf;
-	const unsigned char *inend = inptr + inlen;
-	unsigned char *outptr = outbuf;
-	unsigned int saved = *save;
+	register const unsigned char *inptr;
+	register unsigned char *outptr;
+	const unsigned char *inend;
+	register guint32 saved;
+	unsigned char last[2];
 	unsigned char c, rank;
-	int n, eq, eof = 0;
-
+	int n;
+
+	inend = inbuf + inlen;
+	outptr = outbuf;
+	inptr = inbuf;
+
+	saved = *save;
 	n = *state;
-
-	/* if n == -1, then it means we've encountered the end of the base64 stream */
-	if (n == -1)
-		return 0;
-
-	/* decode every quartet into a triplet */
+
+	if (n < 0) {
+		last[0] = '=';
+		n = -n;
+	} else {
+		last[0] = '\0';
+	}
+
+	last[1] = '\0';
+
+	/* convert 4 base64 bytes to 3 normal bytes */
 	while (inptr < inend) {
 		rank = gmime_base64_rank[(c = *inptr++)];
-
-		if (rank != 0xFF) {
+		if (rank != 0xff) {
 			saved = (saved << 6) | rank;
+			last[1] = last[0];
+			last[0] = c;
 			n++;
-
 			if (n == 4) {
-				/* flush our decoded quartet */
 				*outptr++ = saved >> 16;
-				*outptr++ = saved >> 8;
-				*outptr++ = saved;
-				saved = 0;
+				if (last[1] != '=')
+					*outptr++ = saved >> 8;
+				if (last[0] != '=')
+					*outptr++ = saved;
 				n = 0;
 			}
-		} else if (c == '=') {
-			/* Note: this marks the end of the base64 stream. The only octet that can
-			 * appear after this is another '=' (and possibly mailing-list junk). */
-			eof = 1;
-			break;
 		}
 	}
-
-	if (eof) {
-		/* Note: there shouldn't be more than 2 '=' octets at the end of a quartet,
-		 * so if n < 2, then it means the encoder is broken. */
-		if (n > 1) {
-			/* at this point, n should be either 3 or 4 */
-			eq = 4 - n;
-			saved <<= (6 * eq);
-
-			*outptr++ = saved >> 16;
-			if (n > 2)
-				*outptr++ = saved >> 8;
-		}
-
-		/* Note: If the 'eof' state value changes, be sure to update logic in gmime-utils.c:rfc2047_token_decode() */
-		n = -1;
-	}
-
-	/* save state */
+
+	*state = last[0] == '=' ? -n : n;
 	*save = saved;
-	*state = n;
-
-	return (size_t) (outptr - outbuf);
+
+	return (outptr - outbuf);
 }
 
 

diff --git a/gmime/gmime-utils.c b/gmime/gmime-utils.c
@@ -1799,16 +1799,10 @@ rfc2047_token_decode (rfc2047_token *token, unsigned char *outbuf, int *state, g
 	const unsigned char *inbuf = (const unsigned char *) token->text;
 	size_t len = token->length;
 
-	if (token->encoding == 'B') {
-		/* Note: A state of -1 means that a previous rfc2047 encoded-word token ended with an '=' (aka 'eof'),
-		 * so we need to reset the state. */
-		if (*state == -1)
-			*state = 0;
-
+	if (token->encoding == 'B')
 		return g_mime_encoding_base64_decode_step (inbuf, len, outbuf, state, save);
-	}
-
-	return quoted_decode (inbuf, len, outbuf, state, save);
+	else
+		return quoted_decode (inbuf, len, outbuf, state, save);
 }
 
 static char *