fix(cli): reject invalid UTF-8 JSON file bytes (#613)

* fix(cli): reject invalid UTF-8 JSON file bytes

Signed-off-by: Rudy Celekli <47457359+rudycelekli@users.noreply.github.com>

* test(browser): preserve unknown with-scope loader bindings

Signed-off-by: Rudy Celekli <47457359+rudycelekli@users.noreply.github.com>

---------

Signed-off-by: Rudy Celekli <47457359+rudycelekli@users.noreply.github.com>

src/application/JsonFiles.ts

@@ -17,7 +17,11 @@ export const readJsonFile = async (
     const encoded = await readFile(canonicalPath);
     let decoded: unknown;
     try {
-      decoded = JSON.parse(encoded.toString("utf8"));
+      decoded = JSON.parse(
+        new TextDecoder("utf-8", { fatal: true, ignoreBOM: true }).decode(
+          encoded,
+        ),
+      );
     } catch (cause: unknown) {
       return err(new EvidenceFileError("read", "invalid-json", { cause }));
     }

src/cliJsonInput.ts

@@ -16,7 +16,7 @@ export const parseCliJsonInput = async (
   if (["{", "["].includes(value.trimStart()[0] ?? ""))
     return { ok: false, error: inputError(operation) };
   try {
-    const parsed = parseJson(await readFile(value, "utf8"));
+    const parsed = parseJson(await readFile(value));
     return parsed === undefined
       ? jsonFileError(value, operation, "invalid-json")
       : { ok: true, value: parsed };
@@ -25,9 +25,15 @@ export const parseCliJsonInput = async (
   }
 };
 
-const parseJson = (value: string): unknown => {
+const parseJson = (value: string | Uint8Array): unknown => {
   try {
-    return JSON.parse(value);
+    return JSON.parse(
+      typeof value === "string"
+        ? value
+        : new TextDecoder("utf-8", { fatal: true, ignoreBOM: true }).decode(
+            value,
+          ),
+    );
   } catch {
     return undefined;
   }

tests/boundary/filesystem/jsonUtf8Input.test.ts

@@ -0,0 +1,43 @@
+import { writeFile } from "node:fs/promises";
+import { join } from "node:path";
+import { describe, expect, it } from "vitest";
+import { readJsonFile } from "../../../src/application/JsonFiles.js";
+import { parseCliJsonInput } from "../../../src/cliJsonInput.js";
+import { createTestTempDirectory } from "../../fixtures/temporaryDirectory.js";
+describe("JSON file UTF-8 byte integrity", () => {
+  it.each([0x80, 0xe9, 0xff])(
+    "rejects invalid UTF-8 byte %s through both file input boundaries",
+    async (invalidByte) => {
+      const root = await createTestTempDirectory("rea-json-utf8-");
+      const path = join(root, "input.json");
+      await writeFile(
+        path,
+        Buffer.concat([
+          Buffer.from('{"message":"'),
+          Buffer.from([invalidByte]),
+          Buffer.from('"}'),
+        ]),
+      );
+      const fileInput = await readJsonFile(path);
+      const cliInput = await parseCliJsonInput(path, "compare_web_captures");
+      expect(fileInput).toMatchObject({
+        ok: false,
+        error: { reason: "invalid-json" },
+      });
+      expect(cliInput).toMatchObject({
+        ok: false,
+        error: { input_reason: "invalid-json" },
+      });
+    },
+  );
+  it("preserves a legitimate UTF-8 replacement character", async () => {
+    const root = await createTestTempDirectory("rea-json-utf8-control-");
+    const path = join(root, "input.json");
+    const value = { message: "valid <20> and é and 😀" };
+    await writeFile(path, JSON.stringify(value), "utf8");
+    const fileInput = await readJsonFile(path);
+    const cliInput = await parseCliJsonInput(path, "compare_web_captures");
+    expect(fileInput).toMatchObject({ ok: true, value });
+    expect(cliInput).toMatchObject({ ok: true, value });
+  });
+});