Use string concatentation instead of spreading to prevent stack overflow

kayahr · Feb 4, 2024 · 3a82edd · 3a82edd
1 parent fdd79ed
commit 3a82edd
Show file tree

Hide file tree

Showing 2 changed files with 20 additions and 12 deletions.
diff --git a/src/main/TextDecoder.ts b/src/main/TextDecoder.ts
@@ -64,18 +64,18 @@ export class TextDecoder implements globalThis.TextDecoder {
 
         // Decode the input bytes
         const inputStream = new ByteBuffer(bytes);
-        const output = [];
+        let output = "";
         let result: number | number[] | null;
         while (!inputStream.isEndOfBuffer()) {
             result = this.decoder.decode(inputStream);
             if (result === FINISHED) {
                 break;
             }
             if (result != null) {
-                if (Array.isArray(result)) {
-                    output.push(...result);
+                if (typeof result === "number") {
+                    output += String.fromCodePoint(result);
                 } else {
-                    output.push(result);
+                    output += String.fromCodePoint(...result);
                 }
             }
         }
@@ -86,10 +86,10 @@ export class TextDecoder implements globalThis.TextDecoder {
                     break;
                 }
                 if (result != null) {
-                    if (Array.isArray(result)) {
-                        output.push(...result);
+                    if (typeof result === "number") {
+                        output += String.fromCodePoint(result);
                     } else {
-                        output.push(result);
+                        output += String.fromCodePoint(...result);
                     }
                 }
             } while(!inputStream.isEndOfBuffer());
@@ -99,14 +99,14 @@ export class TextDecoder implements globalThis.TextDecoder {
         // Remove BOM header from output if ignoreBOM flag is not set
         if ([ "utf-8", "utf-16le", "utf-16be" ].includes(this.encoding) && !this.ignoreBOM && !this.seenBOM) {
             if (output.length > 0) {
-                if (output[0] === 0xFEFF) {
-                    output.shift();
-                }
                 this.seenBOM = true;
+                if (output[0] === "\uFEFF") {
+                    return output.substring(1);
+                }
             }
         }
 
-        // Create and return string from decoded code points
-        return String.fromCodePoint(...output);
+        // Join the decoded code points into a full string and return it
+        return output;
     }
 }
diff --git a/src/test/TextDecoder.test.ts b/src/test/TextDecoder.test.ts
@@ -328,4 +328,12 @@ describe("TextDecoder", () => {
     it("correctly maps 0xCA to U+05BA in windows-1255 encoding", () => {
         expect(new TextDecoder("windows-1255").decode(new Uint8Array([ 0xca ]))).toBe("\u05BA");
     });
+    it("correctly decodes a very large input", () => {
+        const data = new Uint16Array(1_000_000);
+        data.fill(0x5400);
+        const utf16 = new Uint8Array(data.buffer);
+        const decoded = new TextDecoder("utf-16be").decode(utf16);
+        expect(decoded.length).toBe(1_000_000);
+        expect(decoded).toContain("TTTT");
+    });
 });