dovecot-1.2: Split quoted-printable parsing to q-p body parsing ...

[dovecot at dovecot.org]

details:   http://hg.dovecot.org/dovecot-1.2/rev/ea84eb708efc
changeset: 9248:ea84eb708efc
user:      Timo Sirainen <tss at iki.fi>
date:      Sun Jul 26 16:32:18 2009 -0400
description:
Split quoted-printable parsing to q-p body parsing and "Q" header parsing.
Fixed also several other issues in quoted-printable parsing to get it fully
RFC compliant.

diffstat:

3 files changed, 87 insertions(+), 26 deletions(-)
src/lib-mail/message-header-decode.c |    6 +-
src/lib-mail/quoted-printable.c      |  101 ++++++++++++++++++++++++++--------
src/lib-mail/quoted-printable.h      |    6 +-

diffs (167 lines):

diff -r 7168ca62c556 -r ea84eb708efc src/lib-mail/message-header-decode.c
--- a/src/lib-mail/message-header-decode.c	Sun Jul 26 15:03:30 2009 -0400
+++ b/src/lib-mail/message-header-decode.c	Sun Jul 26 16:32:18 2009 -0400
@@ -36,9 +36,9 @@ message_header_decode_encoded(const unsi
 	switch (data[start_pos[0]+1]) {
 	case 'q':
 	case 'Q':
-		quoted_printable_decode(data + start_pos[1] + 1,
-					start_pos[2] - start_pos[1] - 1,
-					NULL, decodebuf);
+		quoted_printable_q_decode(data + start_pos[1] + 1,
+					  start_pos[2] - start_pos[1] - 1,
+					  decodebuf);
 		break;
 	case 'b':
 	case 'B':
diff -r 7168ca62c556 -r ea84eb708efc src/lib-mail/quoted-printable.c
--- a/src/lib-mail/quoted-printable.c	Sun Jul 26 15:03:30 2009 -0400
+++ b/src/lib-mail/quoted-printable.c	Sun Jul 26 16:32:18 2009 -0400
@@ -5,8 +5,86 @@
 #include "hex-binary.h"
 #include "quoted-printable.h"
 
+#define QP_IS_TRAILING_SPACE(c) \
+	((c) == ' ' || (c) == '\t')
+
 void quoted_printable_decode(const unsigned char *src, size_t src_size,
 			     size_t *src_pos_r, buffer_t *dest)
+{
+	char hexbuf[3];
+	size_t src_pos, pos, next;
+
+	hexbuf[2] = '\0';
+
+	next = 0;
+	for (src_pos = 0; src_pos < src_size; src_pos++) {
+		if (src[src_pos] != '=' && src[src_pos] != '\n')
+			continue;
+
+		if (src[src_pos] == '\n') {
+			/* drop trailing whitespace */
+			pos = src_pos;
+			if (pos > 0 && src[pos-1] == '\r')
+				pos--;
+			while (pos > 0 && QP_IS_TRAILING_SPACE(src[pos-1]))
+				pos--;
+			buffer_append(dest, src + next, pos - next);
+			next = src_pos+1;
+			buffer_append(dest, "\r\n", 2);
+			continue;
+		}
+
+		/* '=' */
+		buffer_append(dest, src + next, src_pos - next);
+		next = src_pos;
+
+		if (src_pos+1 >= src_size)
+			break;
+
+		if (src[src_pos+1] == '\n') {
+			/* =\n -> skip both */
+			src_pos += 2;
+			next += 2;
+			continue;
+		}
+
+		if (src_pos+2 >= src_size)
+			break;
+
+		if (src[src_pos+1] == '\r' && src[src_pos+2] == '\n') {
+			/* =\r\n -> skip both */
+			src_pos += 3;
+			next += 3;
+			continue;
+		}
+
+		/* =<hex> */
+		hexbuf[0] = src[src_pos+1];
+		hexbuf[1] = src[src_pos+2];
+
+		if (hex_to_binary(hexbuf, dest) == 0) {
+			src_pos += 2;
+			next = src_pos + 1;
+		} else {
+			/* non-hex data, show as-is */
+			next = src_pos;
+		}
+	}
+	if (src_pos == src_size) {
+		/* add everything but trailing spaces */
+		if (src_pos > 0 && src[src_pos-1] == '\r')
+			src_pos--;
+		while (src_pos > 0 && QP_IS_TRAILING_SPACE(src[src_pos-1]))
+			src_pos--;
+		buffer_append(dest, src + next, src_pos - next);
+		next = src_pos;
+	}
+
+	*src_pos_r = next;
+}
+
+void quoted_printable_q_decode(const unsigned char *src, size_t src_size,
+			       buffer_t *dest)
 {
 	char hexbuf[3];
 	size_t src_pos, next;
@@ -19,31 +97,16 @@ void quoted_printable_decode(const unsig
 			continue;
 
 		buffer_append(dest, src + next, src_pos - next);
-		next = src_pos+1;
+		next = src_pos;
 
 		if (src[src_pos] == '_') {
 			buffer_append_c(dest, ' ');
-			continue;
-		}
-
-		if (src_pos+1 >= src_size)
-			break;
-
-		if (src[src_pos+1] == '\n') {
-			/* =\n -> skip both */
-			src_pos++;
+			next++;
 			continue;
 		}
 
 		if (src_pos+2 >= src_size)
 			break;
-
-		if (src[src_pos+1] == '\r' && src[src_pos+2] == '\n') {
-			/* =\r\n -> skip both */
-			src_pos += 2;
-			next++;
-			continue;
-		}
 
 		/* =<hex> */
 		hexbuf[0] = src[src_pos+1];
@@ -57,9 +120,5 @@ void quoted_printable_decode(const unsig
 			next = src_pos;
 		}
 	}
-
 	buffer_append(dest, src + next, src_size - next);
-
-	if (src_pos_r != NULL)
-		*src_pos_r = src_pos;
 }
diff -r 7168ca62c556 -r ea84eb708efc src/lib-mail/quoted-printable.h
--- a/src/lib-mail/quoted-printable.h	Sun Jul 26 15:03:30 2009 -0400
+++ b/src/lib-mail/quoted-printable.h	Sun Jul 26 16:32:18 2009 -0400
@@ -5,9 +5,11 @@
    size of src, and may be same as src. Decoding errors are ignored.
 
    This function may be called multiple times for parsing the same stream.
-   If src_pos is non-NULL, it's updated to first non-translated character in
-   src. */
+   src_pos is updated to first non-translated character in src. */
 void quoted_printable_decode(const unsigned char *src, size_t src_size,
 			     size_t *src_pos_r, buffer_t *dest);
+/* Decode MIME "Q" encoding. */
+void quoted_printable_q_decode(const unsigned char *src, size_t src_size,
+			       buffer_t *dest);
 
 #endif