Fix base32 unicode alphabet (#2380)

Author: loki1205

src/core/operations/ToBase32.mjs

@@ -43,7 +43,14 @@ class ToBase32 extends Operation {
         if (!input) return "";
         input = new Uint8Array(input);
 
-        const alphabet = args[0] ? Utils.expandAlphRange(args[0]).join("") : "ABCDEFGHIJKLMNOPQRSTUVWXYZ234567=";
+        const alphabet = args[0] ?
+            Utils.expandAlphRange(args[0]).join("") :
+            "ABCDEFGHIJKLMNOPQRSTUVWXYZ234567=";
+
+        // Unicode-safe alphabet handling
+        // Supports BMP + non-BMP characters (emoji, Mahjong tiles, etc.)
+        const alphabetChars = Array.from(alphabet);
+
         let output = "",
             chr1, chr2, chr3, chr4, chr5,
             enc1, enc2, enc3, enc4, enc5, enc6, enc7, enc8,
@@ -74,10 +81,19 @@ class ToBase32 extends Operation {
                 enc8 = 32;
             }
 
-            output += alphabet.charAt(enc1) + alphabet.charAt(enc2) + alphabet.charAt(enc3) +
-                alphabet.charAt(enc4) + alphabet.charAt(enc5) + alphabet.charAt(enc6) +
-                alphabet.charAt(enc7) + alphabet.charAt(enc8);
+            // Preserve original charAt() behavior:
+            // out-of-range indexes return ""
+            output +=
+                (alphabetChars[enc1] || "") +
+                (alphabetChars[enc2] || "") +
+                (alphabetChars[enc3] || "") +
+                (alphabetChars[enc4] || "") +
+                (alphabetChars[enc5] || "") +
+                (alphabetChars[enc6] || "") +
+                (alphabetChars[enc7] || "") +
+                (alphabetChars[enc8] || "");
         }
+
         return output;
     }
 

tests/node/tests/NodeDish.mjs

@@ -65,6 +65,42 @@ TestRegister.addApiTests([
         assert.strictEqual(result.toString(), "493e8136b759370a415ef2cf2f7a69690441ff86592aba082bc2e2e0");
     }),
 
+    it("Composable Dish: toBase32 should support non-BMP Unicode alphabets", () => {
+        const alphabet = "🀇🀈🀉🀊🀋🀌🀍🀎🀏🀙🀚🀛🀜🀝🀞🀟🀠🀡🀐🀑🀒🀓🀔🀕🀖🀗🀘🀀🀁🀂🀃🀅";
+
+        const result = new Dish("hello")
+            .apply(toBase32, {alphabet})
+            .toString();
+
+        // Should not contain replacement characters
+        assert.equal(result.includes("�"), false);
+
+        // Should contain only symbols from the alphabet
+        for (const ch of Array.from(result)) {
+            assert.ok(Array.from(alphabet).includes(ch));
+        }
+
+        // "hello" => 8 Base32 symbols
+        assert.equal(Array.from(result).length, 8);
+    }),
+
+    it("Composable Dish: toBase32 should omit padding for 32-character Unicode alphabets", () => {
+        const alphabet = "🀇🀈🀉🀊🀋🀌🀍🀎🀏🀙🀚🀛🀜🀝🀞🀟🀠🀡🀐🀑🀒🀓🀔🀕🀖🀗🀘🀀🀁🀂🀃🀅";
+
+        const result = new Dish("hell")
+            .apply(toBase32, {alphabet})
+            .toString();
+
+        // Should not leak undefined from array indexing
+        assert.equal(result.includes("undefined"), false);
+
+        // Should not contain replacement characters
+        assert.equal(result.includes("�"), false);
+
+        // Unpadded Base32 output for 4-byte input should be 7 symbols
+        assert.equal(Array.from(result).length, 7);
+    }),
+
     it("Dish translation: ArrayBuffer to ArrayBuffer", () => {
         const dish = new Dish(new ArrayBuffer(10), 4);
         dish.get("array buffer");

tests/node/tests/nodeApi.mjs

@@ -109,6 +109,38 @@ TestRegister.addApiTests([
         assert.equal(3 + result, 35);
     }),
 
+    it("toBase32: should support non-BMP Unicode alphabets", () => {
+        const alphabet = "🀇🀈🀉🀊🀋🀌🀍🀎🀏🀙🀚🀛🀜🀝🀞🀟🀠🀡🀐🀑🀒🀓🀔🀕🀖🀗🀘🀀🀁🀂🀃🀅";
+
+        const result = chef.toBase32("hello", {alphabet}).toString();
+
+        // Should not contain replacement characters
+        assert.equal(result.includes("�"), false);
+
+        // Should contain only symbols from the alphabet
+        for (const ch of Array.from(result)) {
+            assert.ok(Array.from(alphabet).includes(ch));
+        }
+
+        // "hello" => 8 Base32 symbols
+        assert.equal(Array.from(result).length, 8);
+    }),
+
+    it("toBase32: should omit padding for 32-character Unicode alphabets", () => {
+        const alphabet = "🀇🀈🀉🀊🀋🀌🀍🀎🀏🀙🀚🀛🀜🀝🀞🀟🀠🀡🀐🀑🀒🀓🀔🀕🀖🀗🀘🀀🀁🀂🀃🀅";
+
+        const result = chef.toBase32("hell", {alphabet}).toString();
+
+        // Should not leak undefined from array indexing
+        assert.equal(result.includes("undefined"), false);
+
+        // Should not contain replacement characters
+        assert.equal(result.includes("�"), false);
+
+        // Unpadded Base32 output for 4-byte input should be 7 symbols
+        assert.equal(Array.from(result).length, 7);
+    }),
+
     it("chef.help: should exist", () => {
         assert(chef.help);
     }),

tests/operations/tests/Base32.mjs

@@ -172,5 +172,27 @@ TestRegister.addTests([
             },
         ],
     },
+    {
+        name: "To Base32: should support non-BMP Unicode alphabets",
+        input: "hello",
+        expectedOutput: "🀝🀈🀐🀔🀖🀀🀊🀟",
+        recipeConfig: [
+            {
+                op: "To Base32",
+                args: ["🀇🀈🀉🀊🀋🀌🀍🀎🀏🀙🀚🀛🀜🀝🀞🀟🀠🀡🀐🀑🀒🀓🀔🀕🀖🀗🀘🀀🀁🀂🀃🀅"],
+            },
+        ],
+    },
+    {
+        name: "To Base32: should omit padding for 32-character Unicode alphabets",
+        input: "hell",
+        expectedOutput: "🀝🀈🀐🀔🀖🀀🀇",
+        recipeConfig: [
+            {
+                op: "To Base32",
+                args: ["🀇🀈🀉🀊🀋🀌🀍🀎🀏🀙🀚🀛🀜🀝🀞🀟🀠🀡🀐🀑🀒🀓🀔🀕🀖🀗🀘🀀🀁🀂🀃🀅"],
+            },
+        ],
+    },
 ]);