FEAT: Bincode - added command VINT (for variable-length integers)

Oldes · Jun 7, 2024 · 3cb158b · 3cb158b
1 parent e1ee49e
commit 3cb158b
Show file tree

Hide file tree

Showing 2 changed files with 62 additions and 0 deletions.
diff --git a/src/core/u-bincode.c b/src/core/u-bincode.c
@@ -218,6 +218,13 @@ static REBCNT EncodedU32_Size(u32 value) {
 	return count;
 }
 
+static REBCNT EncodedVINT_Size(REBU64 value) {
+	REBCNT count = 1;
+	REBU64 temp = value;
+	while (temp >= (1ULL << (7 * count))) count++;
+	return count;
+}
+
 /***********************************************************************
 **
 */	REBNATIVE(binary)
@@ -565,6 +572,13 @@ static REBCNT EncodedU32_Size(u32 value) {
 						}
 						goto error;
 
+					case SYM_VINT:
+						if (IS_INTEGER(next)) {
+							count += EncodedVINT_Size(VAL_UNT64(next));
+							continue;
+						}
+						goto error;
+
 					case SYM_FLOAT:
 					case SYM_F32:
 					case SYM_F32LE:
@@ -936,6 +950,16 @@ static REBCNT EncodedU32_Size(u32 value) {
 						}
 						break;
 
+					case SYM_VINT:
+						u = VAL_UNT64(next);
+						n = EncodedVINT_Size(u);
+						for (i = n-1; i > 0; i--) {
+							cp[i] = (char)(u & 0xFF);
+							u >>= 8;
+						}
+						cp[0] = (char)(u | (0x80 >> (n-1)));
+						break;
+
 					case SYM_UNIXTIME_NOW:
 						value--; // no args
 						n = 4;
@@ -1651,6 +1675,29 @@ static REBCNT EncodedU32_Size(u32 value) {
 							VAL_INDEX(buffer_write) = MAX(0, (REBI64)VAL_INDEX(buffer_write) - VAL_INDEX(buffer_read));
 							VAL_INDEX(buffer_read) = 0;
 							continue;
+
+						case SYM_VINT: {
+							// A variable-length format for positive integers
+							ASSERT_READ_SIZE(value, cp, ep, 1);
+							int mask = 0x80;
+							n = 1;
+							// Determine the length of the VINT
+							while (mask) {
+								if (cp[0] & mask) break;
+								mask >>= 1;
+								n++;
+							}
+							// Extract the VINT value
+							ASSERT_READ_SIZE(value, cp, ep, n);
+							u = (u64)(cp[0] & (0xFF >> n));
+    						for (int i = 1; i < n; i++) {
+								u = (u << 8) | cp[i];
+							}
+							VAL_SET(temp, REB_INTEGER);
+							VAL_UNT64(temp) = u;
+							break;
+						}
+
 						default:
 							Trap1(RE_INVALID_SPEC, value);
 					}

diff --git a/src/tests/units/bincode-test.r3 b/src/tests/units/bincode-test.r3
@@ -335,6 +335,21 @@ is-protected-error?: func[code][
 		--assert [0 1 128 129 130 2214768806] = binary/read b [
 			EncodedU32 EncodedU32 EncodedU32 EncodedU32 EncodedU32 EncodedU32]
 
+	--test-- "BinCode - VINT"
+	;; Another variable-length integer encoding (used in EBML/Matroska files)
+		b: binary/init none 16
+		binary/write b [
+			VINT 0
+			VINT 1
+			VINT 128
+			VINT 129
+			VINT 130
+			VINT 2214768806
+		]
+		--assert b/buffer = #{8081408040814082088402B0A6}
+		--assert [0 1 128 129 130 2214768806] = binary/read b [
+			VINT VINT VINT VINT VINT VINT]
+
 	--test-- "BinCode - BITSET8, BITSET16, BITSET32 (read)"
 		binary/read #{81800180000001} [
 			f8:  BITSET8