Working base32, atleast on linux

encoding.c

@@ -15,31 +15,117 @@
  */
 
 #include <stdio.h>
+#include <strings.h>
+#include <string.h>
 
 // For FreeBSD
 #ifndef MIN
 #define MIN(a,b) ((a)<(b)?(a):(b))
 #endif
 
-static const char to_hex[] = "0123456789ABCDEF";
+#define SPACING 63
+#define ENC_CHUNK 8
+#define RAW_CHUNK 5
+
+static const char base32[] = "ABCDEFGHIJKLMNOPQRSTUVWXYZ98765-";
+static const char padder[] = " 1234";
+static char reverse32[128];
+static int reverse_init = 0;
+
+/* Eat 5 bytes from src, write 8 bytes to dest */
+static void
+encode_chunk(char *dest, char *src)
+{
+	unsigned char c;
+
+	*dest++ = base32[(*src & 0xF8) >> 3];	// 1111 1000 first byte
+
+	c = (*src++ & 0x07) << 2;		// 0000 0111 first byte
+	c |=  ((*src & 0xC0) >> 6);		// 1100 0000 second byte
+	*dest++ = base32[(int) c];
+
+	*dest++ = base32[(*src & 0x3E) >> 1];	// 0011 1110 second byte
+
+	c = (*src++ & 0x01) << 4;		// 0000 0001 second byte
+	c |=  ((*src & 0xF0) >> 4);		// 1111 0000 third byte
+	*dest++ = base32[(int) c];
+	
+	c = (*src++ & 0x0F) << 1;		// 0000 1111 third byte
+	c |=  ((*src & 0x80) >> 7);		// 1000 0000 fourth byte
+	*dest++ = base32[(int) c];
+	
+	*dest++ = base32[(*src & 0x7C) >> 2];	// 0111 1100 fourth byte
+
+	c = (*src++ & 0x03) << 3;		// 0000 0011 fourth byte
+	c |=  ((*src & 0xE0) >> 5);		// 1110 0000 fifth byte
+	*dest++ = base32[(int) c];
+
+	*dest++ = base32[*src++ & 0x1F];	// 0001 1111 fifth byte
+}
+
+/* Eat 8 bytes from src, write 5 bytes to dest */
+static void
+decode_chunk(char *dest, char *src)
+{
+	unsigned char c;
+	int i;
+
+	if (!reverse_init) {
+		for (i = 0; i < 32; i++) {
+			c = base32[i];
+			reverse32[(int) c] = i;
+		}
+		reverse_init = 1;
+	}
+
+	c = reverse32[(int) *src++] << 3;		// Take bits 11111 from byte 1
+	c |= (reverse32[(int) *src] & 0x1C) >> 2;	// Take bits 11100 from byte 2
+	*dest++ = c;
+	
+	c = (reverse32[(int) *src++] & 0x3) << 6;	// Take bits 00011 from byte 2
+	c |= reverse32[(int) *src++] << 1;		// Take bits 11111 from byte 3
+	c |= (reverse32[(int) *src] & 0x10) >> 4;	// Take bits 10000 from byte 4
+	*dest++ = c;
+	
+	c = (reverse32[(int) *src++] & 0xF) << 4;	// Take bits 01111 from byte 4
+	c |= (reverse32[(int) *src] & 0x1E) >> 1;	// Take bits 11110 from byte 5
+	*dest++ = c;
+	
+	c = reverse32[(int) *src++] << 7;		// Take bits 00001 from byte 5
+	c |= reverse32[(int) *src++] << 2;		// Take bits 11111 from byte 6
+	c |= (reverse32[(int) *src] & 0x18) >> 3;	// Take bits 11000 from byte 7
+	*dest++ = c;
+	
+	c = (reverse32[(int) *src++] & 0x7) << 5;	// Take bits 00111 from byte 7
+	c |= reverse32[(int) *src++];			// Take bits 11111 from byte 8
+	*dest++ = c;
+}
 
 int
 encode_data(char *buf, int len, int space, char *dest, char flag)
 {
 	int final;
 	int write;
+	int realwrite;
+	int chunks;
+	int leftovers;
 	int i;
-	int t;
-
-#define CHUNK 31
-// 31 bytes expands to 62 chars in domain
-// We just use hex as encoding right now
-
-	write = space / 2; // use two chars per byte in encoding
-	write -= (write/CHUNK); // make space for parts
+	char encoded[255];
+	char padding[5];
+	char *pp;
+	char *dp;
+	char *ep;
 
+	space -= space / SPACING;
+	chunks = (space - 1) / ENC_CHUNK;
+	while ((chunks + 1) * ENC_CHUNK + 1 > space) {
+		chunks--;
+	}
+	write = RAW_CHUNK * chunks;
 	write = MIN(write, len); // do not use more bytes than is available;
 	final = (write == len);	// is this the last block?
+	chunks = write / RAW_CHUNK;
+	leftovers = write % RAW_CHUNK;
 
 	if (flag != 0) {
 		*dest = flag;
@@ -49,43 +135,91 @@ encode_data(char *buf, int len, int space, char *dest, char flag)
 	}
 	dest++;
 
+	bzero(encoded, sizeof(encoded));
+	ep = encoded;
+	dp = buf;
+	for (i = 0; i < chunks; i++) {
+		encode_chunk(ep, dp);
+		ep += ENC_CHUNK;
+		dp += RAW_CHUNK;
+	}
+	realwrite = ENC_CHUNK * chunks;
+	bzero(padding, sizeof(padding));
+	pp = padding;
+	if (leftovers) {
+		pp += RAW_CHUNK - leftovers;
+		memcpy(pp, dp, leftovers);
+
+		pp = padding;
+		*ep++ = padder[leftovers];
+		encode_chunk(ep, pp);
+		
+		realwrite += ENC_CHUNK + 1;	// plus padding character
+	}
+	ep = encoded;
 	if (len > 0) {
-		for (i = 0; i < write; i++) {
-			if (i > 0 && i % CHUNK == 0) {
-				*dest = '.';
-				dest++;
+		for (i = 1; i <= realwrite; i++) {
+			if (i % SPACING == 0) {
+				*dest++ = '.';
 			}
-			t = (buf[i] & 0xF0) >> 4;
-			*dest++ = to_hex[t];
-			t = buf[i] & 0x0F;
-			*dest++ = to_hex[t];
+			*dest++ = *ep++;
 		}
 	}
+	
 	return write;
 }
 
 int
 decode_data(char *dest, int size, const char *src, char *srcend)
 {
-	int r;
 	int len;
+	int i;
+	int chunks;
+	int padded;
+	char encoded[255];
+	char padding[5];
+	char *pp;
+	char *ep;
 
+	// Copy flag
 	len = 1;
 	*dest = *src;
 	dest++;
 	src++;
 
+	bzero(encoded, sizeof(encoded));
+	ep = encoded;
 	while(len < size && src < srcend) {
 		if(*src == '.') {
 			src++;
 			continue;
 		}
 
-		sscanf(src, "%02X", &r);
-		*dest++ = (char)r;
-		src+=2;	
-		len++;
+		*ep++ = *src++;
 	}
+	chunks = strlen(encoded) / 8;
+	padded = strlen(encoded) % 8;
+
+	ep = encoded;
+	for (i = 0; i < chunks-1; i++) {
+		decode_chunk(dest, ep);
+		dest += RAW_CHUNK;
+		ep += ENC_CHUNK;
+		len += RAW_CHUNK;
+	}
+	// Read last chunk:
+	if (padded) {
+		pp = padding;
+		padded = *ep++ - '0';
+		decode_chunk(pp, ep);
+		pp += RAW_CHUNK - padded;
+		memcpy(dest, pp, padded);
+		len += padded;
+	} else {
+		decode_chunk(dest, ep);
+		len += RAW_CHUNK;
+	}
+
 	return len;
 }