can1357
4e8fa1672f
- Added a deterministic ProcessTerminal render harness with real TUI and resize simulation. - Added reflow tests for OS resize fallback, in-band precedence, and split in-band parsing. - Added a regression test ensuring stale live-region rows are cleared after rerender. - Removed the duplicate in-band authority assertion from terminal-appearance tests.
75 lines
3.0 KiB
TypeScript
75 lines
3.0 KiB
TypeScript
import { afterEach, describe, expect, it } from "bun:test";
|
|
import {
|
|
createProcessTerminalRenderHarness,
|
|
type ProcessTerminalRenderHarness,
|
|
} from "./process-terminal-render-harness";
|
|
|
|
// Geometry-reflow contract for the *real* terminal driven through the *real*
|
|
// renderer. These exercise the seam VirtualTerminal cannot model: the OS channel
|
|
// (SIGWINCH) and the DEC 2048 in-band channel disagreeing. The observable
|
|
// contract is `probe.last` — the width the transcript actually reflowed to.
|
|
describe("ProcessTerminal geometry reflow through the renderer", () => {
|
|
let harness: ProcessTerminalRenderHarness | undefined;
|
|
|
|
afterEach(() => {
|
|
harness?.dispose();
|
|
harness = undefined;
|
|
});
|
|
|
|
it("reflows to the OS width on resize when in-band resize is inactive", async () => {
|
|
harness = createProcessTerminalRenderHarness(100, 30);
|
|
await harness.settle();
|
|
expect(harness.probe.last).toBe(100);
|
|
|
|
await harness.osResize(160, 40);
|
|
|
|
expect(harness.terminal.columns).toBe(160);
|
|
expect(harness.probe.last).toBe(160);
|
|
});
|
|
|
|
it("reflows to the OS width when the post-resize in-band report is missed", async () => {
|
|
// The regression: SIGWINCH fires and process.stdout dims update, but the
|
|
// matching DEC 2048 report is dropped/malformed. Before the fix the getter
|
|
// stayed pinned to the cached report and the transcript reflowed at the old
|
|
// width — content never resized.
|
|
harness = createProcessTerminalRenderHarness(100, 30);
|
|
await harness.feed("\x1b[?2048;1$y"); // DECRPM: 2048 supported -> in-band active
|
|
await harness.inBand(30, 100, 600, 1000); // seed cached geometry at 100 cols
|
|
expect(harness.probe.last).toBe(100);
|
|
|
|
await harness.osResize(160, 40); // no follow-up in-band report
|
|
|
|
expect(harness.terminal.columns).toBe(160);
|
|
expect(harness.probe.last).toBe(160);
|
|
});
|
|
|
|
it("reflows to in-band geometry that disagrees with the OS (in-band stays authoritative)", async () => {
|
|
// No OS resize: process.stdout stays at 100, but an in-band report claims
|
|
// 140. The getter must keep preferring in-band (it can be more accurate than
|
|
// ioctl under some multiplexers), so the reconcile fix must not make the OS
|
|
// channel unconditionally win.
|
|
harness = createProcessTerminalRenderHarness(100, 30);
|
|
await harness.feed("\x1b[?2048;1$y");
|
|
await harness.inBand(30, 100, 600, 1000);
|
|
|
|
await harness.inBand(30, 140, 700, 1400);
|
|
|
|
expect(harness.terminal.columns).toBe(140);
|
|
expect(harness.probe.last).toBe(140);
|
|
});
|
|
|
|
it("reflows when an in-band report is split across stdin reads", async () => {
|
|
// A report fragmented mid-sequence must be reassembled by StdinBuffer within
|
|
// the flush window and still drive a reflow — exercises the parser through
|
|
// the full ProcessTerminal -> TUI stack.
|
|
harness = createProcessTerminalRenderHarness(100, 30);
|
|
await harness.feed("\x1b[?2048;1$y");
|
|
await harness.inBand(30, 100, 600, 1000);
|
|
|
|
await harness.feed("\x1b[48;40;160", ";800;1600t");
|
|
|
|
expect(harness.terminal.columns).toBe(160);
|
|
expect(harness.probe.last).toBe(160);
|
|
});
|
|
});
|