FEAT: enbase/debase with base 36 (alphanumeric)

Author: Oldes

make/rebol3.nest

@@ -613,6 +613,9 @@ include-base85-encoding: [
 	; adds support for enbase/debase with base 85 (ASCII85)
 	config: INCLUDE_BASE85
 ]
+include-base36-encoding: [
+	config: INCLUDE_BASE36
+]
 
 ;- native additional compressions:                                              
 include-lzma-compression: [
@@ -745,6 +748,7 @@ include-rebol-bulk: [
 	:include-image-natives
 	:include-lzma-compression
 	:include-crush-compression
+	:include-base36-encoding
 	:include-base85-encoding
 	:include-view
 	:include-midi

src/boot/natives.reb

@@ -393,7 +393,7 @@ construct: native [
 debase: native [
 	{Decodes binary-coded string to binary value.}
 	value [binary! any-string!] {The string to decode}
-	base  [integer!] {Binary base to use: 85, 64, 16, or 2}
+	base  [integer!] {Binary base to use: 85, 64, 36, 16, or 2}
 	/url  {Base 64 Decoding with URL and Filename Safe Alphabet}
 	/part {Limit the length of the input}
 	limit [integer!]
@@ -402,7 +402,7 @@ debase: native [
 enbase: native [
 	{Encodes a string into a binary-coded string.}
 	value [binary! string!] {If string, will be UTF8 encoded}
-	base  [integer!] {Binary base to use: 85, 64, 16, or 2}
+	base  [integer!] {Binary base to use: 85, 64, 36, 16, or 2}
 	/url  {Base 64 Encoding with URL and Filename Safe Alphabet}
 	/part {Limit the length of the input}
 	limit [integer!]

src/core/f-enbase.c

@@ -441,6 +441,37 @@
 #endif
 
 
+#ifdef INCLUDE_BASE36
+#define BASE36_LENGTH 13
+/***********************************************************************
+**
+*/	static const REBYTE Enbase36[36] =
+/*
+**		Base-36 binary encoder table.
+**
+***********************************************************************/
+{
+	"0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ"
+};
+
+static REBU64 base36_powers[BASE36_LENGTH] = {
+	1ULL,
+	36ULL,
+	1296ULL,
+	46656ULL,
+	1679616ULL,
+	60466176ULL,
+	2176782336ULL,
+	78364164096ULL,
+	2821109907456ULL,
+	101559956668416ULL,
+	3656158440062976ULL,
+	131621703842267136ULL,
+	4738381338321616896ULL
+};
+#endif
+
+
 /***********************************************************************
 **
 */	static REBSER *Decode_Base2(const REBYTE **src, REBCNT len, REBYTE delim)
@@ -727,6 +758,61 @@
 }
 #endif
 
+#ifdef INCLUDE_BASE36
+/***********************************************************************
+**
+*/	static REBSER* Decode_Base36(const REBYTE** src, REBCNT len, REBYTE delim)
+/*
+***********************************************************************/
+{
+	REBYTE* bp;
+	const REBYTE* cp;
+	REBSER* ser;
+	REBCNT ser_size;
+	REBCNT chunk;
+	REBINT pad = 0;
+	REBU64 c = 0;
+	REBINT i, d = 0;
+
+	cp = *src;
+
+	if (len > BASE36_LENGTH) goto err;
+	else if (len == 0) {
+		ser = Make_Binary(1);
+		ser->tail = 0;
+		return ser;
+	}
+
+	for (i = 0; i < len; i++) {
+		if (cp[i] >= '0' && cp[i] <= '9')
+			d = cp[i] - '0';
+		else if (cp[i] >= 'A' && cp[i] <= 'Z')
+			d = 10 + cp[i] - 'A';
+		else if (cp[i] >= 'a' && cp[i] <= 'z')
+			d = 10 + cp[i] - 'a';
+		else goto err;
+
+		c += d * base36_powers[len - i - 1];
+		if (c < 0) goto err;
+	}
+	// Allocate buffer large enough to hold result:
+	ser = Make_Binary(8);
+	ser_size = SERIES_AVAIL(ser);
+	bp = STR_HEAD(ser);
+
+	for (i = 7; i >= 0; i--) {
+		bp[i] = (REBYTE)(c & 0xFF);
+		c >>= 8;
+	}
+	ser->tail = 8;
+	return ser;
+
+err:
+	*src = cp;
+	return 0;
+}
+#endif
+
 
 /***********************************************************************
 **
@@ -753,6 +839,13 @@
 		ser = Decode_Base85 (&src, len, delim);
 #else
 		Trap0(RE_FEATURE_NA);
+#endif
+		break;
+	case 36:
+#ifdef INCLUDE_BASE36
+		ser = Decode_Base36(&src, len, delim);
+#else
+		Trap0(RE_FEATURE_NA);
 #endif
 		break;
 	}
@@ -954,4 +1047,49 @@
 
 	return series;
 }
+#endif
+
+
+#ifdef INCLUDE_BASE36
+/***********************************************************************
+**
+*/  REBSER* Encode_Base36(REBVAL* value, REBSER* series, REBCNT len, REBFLG brk)
+/*
+**		Base36 encode a given series. Must be BYTES, not UNICODE.
+**
+***********************************************************************/
+{
+	REBYTE* bp;
+	REBYTE* src;
+	REBOOL discard = TRUE;
+	REBU64 d, m = 0;
+	REBCNT n, p = 0;
+	REBINT i;
+
+	if (len > VAL_LEN(value)) len = VAL_LEN(value);
+	if (len > sizeof(REBI64)) {
+		Trap1(RE_OUT_OF_RANGE, value);
+	}
+	else if (len == 0) {
+		series = Prep_String(series, &bp, 1);
+		SERIES_TAIL(series) = 0;
+		return series;
+	}
+	src = VAL_BIN_DATA(value);
+
+	for (bp = src, n = len; n; n--, bp++)
+		m = (m << 8) | *bp;
+
+	series = Prep_String(series, &bp, BASE36_LENGTH + 1);
+
+	for (i = BASE36_LENGTH - 1; i >= 0; i--) {
+		d = m / base36_powers[i];
+		m = m - base36_powers[i] * d;
+		discard = discard && (d == 0 && i > 0);
+		if (!discard)
+			bp[p++] = Enbase36[d];
+	}
+	SERIES_TAIL(series) = p;
+	return series;
+}
 #endif

src/core/n-strings.c

@@ -544,7 +544,13 @@ static struct digest {
 #else
 		Trap0(RE_FEATURE_NA);
 #endif
-		
+		break;
+	case 36:
+#ifdef INCLUDE_BASE36
+		ser = Encode_Base36(arg, 0, limit, FALSE);
+#else
+		Trap0(RE_FEATURE_NA);
+#endif
 		break;
 	default:
 		Trap_Arg(D_ARG(2));

src/tests/units/enbase-test.r3

@@ -239,4 +239,52 @@ if any [
 	===end-group===
 ]
 
+if any [
+	not error? err: try [enbase #{} 36]
+	err/id <> 'feature-na
+][
+	base36-int-tests: [
+		0          "0"
+		1          "1"
+		36         "10"
+		64         "1S"
+		1024       "SG"
+		19930503   "BV6H3"
+		1843067821 "UHBC8D"
+		3951668550778163018 "U0TPLAQIV70Q"
+	]
+	
+	===start-group=== "de/enbase-36"
+		--test-- "enbase debase 36"
+			--assert ""  = enbase debase ""  36 36
+			--assert "0" = enbase debase "0" 36 36
+			--assert "1" = enbase debase "1" 36 36
+			--assert "0" = enbase debase "000" 36 36
+			--assert "1" = enbase debase "001" 36 36
+
+		--test-- "enbase int 36"    
+			foreach [inp out] base36-int-tests [
+				--assert out = enbase to binary! inp 36
+			]
+
+		--test-- "debase int 36"    
+			foreach [out inp] base36-int-tests [
+				--assert out = to integer! debase inp 36
+			]
+
+		--test-- "enbase 36"
+			--assert "0" = enbase #{00} 36
+			--assert "0" = enbase #{00000000} 36
+			--assert "0" = enbase #{0000000000000000} 36
+			--assert "3W5E11264SGSF" = enbase #{FFFFFFFFFFFFFFFF} 36
+
+		--test-- "debase 36 errors"
+			--assert all [error? e: try [debase "00000000000000" 36] e/id = 'invalid-data]
+			--assert all [error? e: try [debase "0.0" 36] e/id = 'invalid-data]
+		--test-- "enbase 36 errors"
+			--assert all [error? e: try [enbase #{000000000000000000} 36] e/id = 'out-of-range]
+
+	===end-group===
+]
+
 ~~~end-file~~~