[Feature] Add rfc2047 variant for QP decoding

2 files changed, 123 insertions, 0 deletions

diff --git a/src/libutil/str_util.c b/src/libutil/str_util.c
index 5be41f784..cbad775cb 100644
--- a/src/libutil/str_util.c
+++ b/src/libutil/str_util.c
@@ -1825,6 +1825,118 @@ decode:
 	return (o - out);
 }
 
+#define BITOP(a,b,op) \
+		((a)[(gsize)(b)/(8*sizeof *(a))] op (gsize)1<<((gsize)(b)%(8*sizeof *(a))))
+static gsize
+rspamd_memcspn (const gchar *s, const gchar *e, gsize len)
+{
+	gsize byteset[32/sizeof(gsize)];
+	const gchar *p = s, *end = s + len;
+
+	memset(byteset, 0, sizeof byteset);
+
+	for (; *e && BITOP (byteset, *(guchar *)e, |=); e++);
+	for (; p < end && !BITOP (byteset, *(guchar *)p, &); p++);
+
+	return p - s;
+}
+
+gssize
+rspamd_decode_qp2047_buf (const gchar *in, gsize inlen,
+		gchar *out, gsize outlen)
+{
+	gchar *o, *end, c;
+	const gchar *p;
+	guchar ret;
+	gsize remain, processed;
+
+	p = in;
+	o = out;
+	end = out + outlen;
+	remain = inlen;
+
+	while (remain > 0 && o < end) {
+		if (*p == '=') {
+			p ++;
+			remain --;
+
+			if (remain == 0) {
+				if (end - o > 0) {
+					*o++ = *p;
+					break;
+				}
+			}
+decode:
+			/* Decode character after '=' */
+			c = *p++;
+			remain --;
+
+			if      (c >= '0' && c <= '9') { ret = c - '0'; }
+			else if (c >= 'A' && c <= 'F') { ret = c - 'A' + 10; }
+			else if (c >= 'a' && c <= 'f') { ret = c - 'a' + 10; }
+			else if (c == '\r' || c == '\n') {
+				/* Soft line break */
+				while (remain > 0 && (*p == '\r' || *p == '\n')) {
+					remain --;
+					p ++;
+				}
+
+				continue;
+			}
+
+			if (remain > 0) {
+				c = *p++;
+				ret *= 16;
+
+				if      (c >= '0' && c <= '9') ret += c - '0';
+				else if (c >= 'A' && c <= 'F') ret += c - 'A' + 10;
+				else if (c >= 'a' && c <= 'f') ret += c - 'a' + 10;
+
+				if (end - o > 0) {
+					*o++ = (gchar)ret;
+				}
+				else {
+					return (-1);
+				}
+
+				remain --;
+			}
+		}
+		else {
+			if (end - o >= remain) {
+				processed = rspamd_memcspn (p, "=_", remain);
+				memcpy (o, p, processed);
+				o += remain;
+
+				if (processed == remain) {
+					break;
+				}
+				else {
+
+					remain -= processed;
+					p += processed;
+
+					if (*p == '=') {
+						p ++;
+						/* Skip comparison, as we know that we have found match */
+						goto decode;
+					}
+					else {
+						*o++ = ' ';
+						p ++;
+					}
+				}
+			}
+			else {
+				/* Buffer overflow */
+				return (-1);
+			}
+		}
+	}
+
+	return (o - out);
+}
+
 
 /*
  * GString ucl emitting functions
diff --git a/src/libutil/str_util.h b/src/libutil/str_util.h
index 37bd9fe82..1fdb1a857 100644
--- a/src/libutil/str_util.h
+++ b/src/libutil/str_util.h
@@ -223,6 +223,17 @@ gsize rspamd_decode_url (gchar *dst, const gchar *src, gsize size);
 gssize rspamd_decode_qp_buf (const gchar *in, gsize inlen,
 		gchar *out, gsize outlen);
 
+/**
+ * Decode quoted-printable encoded buffer using rfc2047 format, input and output must not overlap
+ * @param in input
+ * @param inlen length of input
+ * @param out output
+ * @param outlen length of output
+ * @return real size of decoded output or (-1) if outlen is not enough
+ */
+gssize rspamd_decode_qp2047_buf (const gchar *in, gsize inlen,
+		gchar *out, gsize outlen);
+
 #ifndef g_tolower
 #   define g_tolower(x) (((x) >= 'A' && (x) <= 'Z') ? (x) - 'A' + 'a' : (x))
 #endif