From 5df91ed775ce644f6f489563c5d0cc81ed7a7cc3 Mon Sep 17 00:00:00 2001
From: frekky <frekky@frekkworks.com>
Date: Sun, 23 Aug 2015 22:10:50 +0800
Subject: [PATCH] Added encoder length calculation functions

---
 src/base128.c  | 17 ++++++++++++++++-
 src/base32.c   | 20 +++++++++++++++++++-
 src/base64.c   | 17 ++++++++++++++++-
 src/encoding.c | 44 ++++++++++++++++++++++++++++++++++----------
 src/encoding.h |  7 +++++++
 5 files changed, 92 insertions(+), 13 deletions(-)

diff --git a/src/base128.c b/src/base128.c
index be7ba86..4e8e409 100644
--- a/src/base128.c
+++ b/src/base128.c
@@ -57,6 +57,7 @@ static int base128_decode(void *, size_t *, const char *, size_t);
 static int base128_handles_dots();
 static int base128_blksize_raw();
 static int base128_blksize_enc();
+static size_t base128_encoded_length(size_t inputlen);
 
 static struct encoder base128_encoder =
 {
@@ -66,7 +67,9 @@ static struct encoder base128_encoder =
 	base128_handles_dots,
 	base128_handles_dots,
 	base128_blksize_raw,
-	base128_blksize_enc
+	base128_blksize_enc,
+	base128_encoded_length,
+	base128_raw_length
 };
 
 struct encoder
@@ -93,6 +96,18 @@ base128_blksize_enc()
 	return BLKSIZE_ENC;
 }
 
+static size_t
+base128_encoded_length(size_t inputlen)
+{
+	return (BLKSIZE_ENC * inputlen) / BLKSIZE_RAW + ((BLKSIZE_ENC * inputlen) % BLKSIZE_RAW) ? 1 : 0;
+}
+
+static size_t
+base128_raw_length(size_t inputlen)
+{
+	return (BLKSIZE_RAW * inputlen) / BLKSIZE_ENC + ((BLKSIZE_RAW * inputlen) % BLKSIZE_ENC) ? 1 : 0;
+}
+
 inline static void
 base128_reverse_init()
 {
diff --git a/src/base32.c b/src/base32.c
index a058d38..a49963d 100644
--- a/src/base32.c
+++ b/src/base32.c
@@ -38,6 +38,9 @@ static int base32_decode(void *, size_t *, const char *, size_t);
 static int base32_handles_dots();
 static int base32_blksize_raw();
 static int base32_blksize_enc();
+static size_t base32_encoded_length(size_t inputlen);
+static size_t base32_raw_length(size_t inputlen);
+
 
 static struct encoder base32_encoder =
 {
@@ -47,7 +50,9 @@ static struct encoder base32_encoder =
 	base32_handles_dots,
 	base32_handles_dots,
 	base32_blksize_raw,
-	base32_blksize_enc
+	base32_blksize_enc,
+	base32_encoded_length,
+	base32_raw_length
 };
 
 struct encoder
@@ -74,6 +79,19 @@ base32_blksize_enc()
 	return BLKSIZE_ENC;
 }
 
+static size_t
+base32_encoded_length(size_t inputlen)
+{
+	return (BLKSIZE_ENC * inputlen) / BLKSIZE_RAW + ((BLKSIZE_ENC * inputlen) % BLKSIZE_RAW) ? 1 : 0;
+}
+
+static size_t
+base32_raw_length(size_t inputlen)
+{
+	return (BLKSIZE_RAW * inputlen) / BLKSIZE_ENC + ((BLKSIZE_RAW * inputlen) % BLKSIZE_ENC) ? 1 : 0;
+}
+
+
 inline static void
 base32_reverse_init()
 {
diff --git a/src/base64.c b/src/base64.c
index 7834731..c57a3f6 100644
--- a/src/base64.c
+++ b/src/base64.c
@@ -38,6 +38,7 @@ static int base64_decode(void *, size_t *, const char *, size_t);
 static int base64_handles_dots();
 static int base64_blksize_raw();
 static int base64_blksize_enc();
+static size_t base64_encoded_length(size_t inputlen);
 
 static struct encoder base64_encoder =
 {
@@ -47,7 +48,9 @@ static struct encoder base64_encoder =
 	base64_handles_dots,
 	base64_handles_dots,
 	base64_blksize_raw,
-	base64_blksize_enc
+	base64_blksize_enc,
+	base64_encoded_length,
+	base64_raw_length
 };
 
 struct encoder
@@ -74,6 +77,18 @@ base64_blksize_enc()
 	return BLKSIZE_ENC;
 }
 
+static size_t
+base64_encoded_length(size_t inputlen)
+{
+	return (BLKSIZE_ENC * inputlen) / BLKSIZE_RAW + ((BLKSIZE_ENC * inputlen) % BLKSIZE_RAW) ? 1 : 0;
+}
+
+static size_t
+base64_raw_length(size_t inputlen)
+{
+	return (BLKSIZE_RAW * inputlen) / BLKSIZE_ENC + ((BLKSIZE_RAW * inputlen) % BLKSIZE_ENC) ? 1 : 0;
+}
+
 inline static void
 base64_reverse_init()
 {
diff --git a/src/encoding.c b/src/encoding.c
index c954e1d..58530b4 100644
--- a/src/encoding.c
+++ b/src/encoding.c
@@ -19,19 +19,43 @@
 #include "common.h"
 #include "encoding.h"
 
+size_t
+get_raw_length(size_t enc_bytes, struct encoder *enc, const char *topdomain)
+/* Returns the maximum length of raw data that can be encoded into max_enc_bytes */
+{
+	size_t enc_datalen = enc_bytes - strlen(topdomain);
+	/* Number of dots in length of encoded data */
+	size_t dots = enc_datalen / (DNS_MAXLABEL + 1);
+	if (!enc->eats_dots()) /* Dots are not included in encoded data length */
+		enc_datalen -= dots;
+	return enc->get_raw_length(enc_datalen);
+}
+
+size_t
+get_encoded_length(size_t raw_bytes, struct encoder *enc, const char *topdomain)
+/* Returns length of encoded data from original data length orig_len; */
+{
+	size_t dots = 1; /* dot before topdomain */
+	size_t len = enc->get_encoded_length(raw_bytes) + strlen(topdomain);
+	if (!enc->places_dots())
+		dots += len / 63; /* number of dots needed in data */
+	return len;
+}
+
 int
-build_hostname(char *buf, size_t buflen,
-		const char *data, const size_t datalen,
-		const char *topdomain, struct encoder *encoder, size_t maxlen)
+build_hostname(char *buf, size_t buflen, const char *data, const size_t datalen,
+		const char *topdomain, struct encoder *encoder, size_t maxlen, size_t header_len)
+/* Builds DNS-compatible hostname for data using specified encoder and topdomain
+ * NB: Does not account for header length. Data is encoded at start of buf to
+ * (buf + MIN(maxlen, buflen)). */
 {
 	size_t space;
 	char *b;
 
-	space = MIN(maxlen, buflen) - strlen(topdomain) - DOWNSTREAM_HDR + 3;
-	/* max header length + 1 dot before topdomain + 2 safety */
-
-	if (!encoder->places_dots())
-		space -= (space / 57); /* space for dots */
+	space = get_encoded_length(MIN(maxlen, buflen), encoder, topdomain);
+	buf += header_len;
+	buflen -= header_len;
+	maxlen -= header_len;
 
 	memset(buf, 0, buflen);
 
@@ -72,7 +96,7 @@ inline_dotify(char *buf, size_t buflen)
 	char *reader, *writer;
 
 	total = strlen(buf);
-	dots = total / 57;
+	dots = total / 63;
 
 	writer = buf;
 	writer += total;
@@ -91,7 +115,7 @@ inline_dotify(char *buf, size_t buflen)
 	while (dots) {
 		*writer-- = *reader--;
 		pos--;
-		if (pos % 57 == 0) {
+		if (pos % 63 == 0) {
 			*writer-- = '.';
 			dots--;
 		}
diff --git a/src/encoding.h b/src/encoding.h
index b38d0d6..584d697 100644
--- a/src/encoding.h
+++ b/src/encoding.h
@@ -25,6 +25,8 @@
 #define DOWNCODECCHECK1      "\000\000\000\000\377\377\377\377\125\125\125\125\252\252\252\252\201\143\310\322\307\174\262\027\137\117\316\311\111\055\122\041\141\251\161\040\045\263\006\163\346\330\104\060\171\120\127\277"
 #define DOWNCODECCHECK1_LEN  48
 
+#define DNS_MAXLABEL 63
+
 struct encoder {
 	char name[8];
 	int (*encode) (char *, size_t *, const void *, size_t);
@@ -33,8 +35,13 @@ struct encoder {
 	int (*eats_dots) (void);
 	int (*blocksize_raw)(void);
 	int (*blocksize_encoded)(void);
+	size_t (*get_encoded_length)(size_t);
+	size_t (*get_raw_length)(size_t);
 };
 
+size_t get_raw_length(size_t enc_bytes, struct encoder *enc, const char *topdomain);
+size_t get_encoded_length(size_t raw_bytes, struct encoder *enc, const char *topdomain);
+
 int build_hostname(char *, size_t, const char *, const size_t, const char *, struct encoder *, size_t);
 int unpack_data(char *, size_t, char *, size_t, struct encoder *);
 int inline_dotify(char *, size_t);