moltbot/src/infra/windows-encoding.test.ts

import { describe, expect, it } from "vitest";
import {
  createWindowsOutputDecoder,
  decodeWindowsOutputBuffer,
  parseWindowsCodePage,
} from "./windows-encoding.js";

describe("windows output encoding", () => {
  it("parses code pages from chcp output text", () => {
    expect(parseWindowsCodePage("Active code page: 936")).toBe(936);
    expect(parseWindowsCodePage("活动代码页: 65001")).toBe(65001);
    expect(parseWindowsCodePage("no code page")).toBeNull();
  });

  it("decodes GBK output on Windows when UTF-8 is invalid and code page is known", () => {
    const raw = Buffer.from([0xb2, 0xe2, 0xca, 0xd4, 0xa1, 0xab, 0xa3, 0xbb]);

    expect(
      decodeWindowsOutputBuffer({
        buffer: raw,
        platform: "win32",
        windowsEncoding: "gbk",
      }),
    ).toBe("测试～；");
  });

  it("prefers valid UTF-8 output on Windows even when the console code page is legacy", () => {
    const raw = Buffer.from("测试", "utf8");

    expect(
      decodeWindowsOutputBuffer({
        buffer: raw,
        platform: "win32",
        windowsEncoding: "gbk",
      }),
    ).toBe("测试");
  });

  it("keeps multibyte Windows codepage characters intact across chunk boundaries", () => {
    const decoder = createWindowsOutputDecoder({
      platform: "win32",
      windowsEncoding: "gbk",
    });

    expect(decoder.decode(Buffer.from([0xb2]))).toBe("");
    expect(decoder.decode(Buffer.from([0xe2, 0xca]))).toBe("测");
    expect(decoder.decode(Buffer.from([0xd4]))).toBe("试");
    expect(decoder.flush()).toBe("");
  });

  it("replays buffered UTF-8 lead bytes when split GBK output falls back to the console code page", () => {
    const decoder = createWindowsOutputDecoder({
      platform: "win32",
      windowsEncoding: "gbk",
    });

    expect(decoder.decode(Buffer.from([0xc4]))).toBe("");
    expect(decoder.decode(Buffer.from([0xe3]))).toBe("你");
    expect(decoder.flush()).toBe("");
  });

  it("keeps split valid UTF-8 output on the UTF-8 path for streaming decode", () => {
    const decoder = createWindowsOutputDecoder({
      platform: "win32",
      windowsEncoding: "gbk",
    });
    const raw = Buffer.from("测试", "utf8");

    expect(decoder.decode(raw.subarray(0, 1))).toBe("");
    expect(decoder.decode(raw.subarray(1, 3))).toBe("测");
    expect(decoder.decode(raw.subarray(3))).toBe("试");
    expect(decoder.flush()).toBe("");
  });
});