Files
callstack__agent-device/scripts/perf/scenario.ts
Michał Pierzchała 3a02e514e1 refactor(ios): consolidate series batching onto the sequence runner command (#768)
* refactor(ios): consolidate series batching onto the sequence runner command

Closes #767

Routes every Apple multi-press variant (plain, double-tap, hold, jitter)
and swipe series through budget-chunked sequence requests, retiring the
daemon-side tapSeries and dragSeries senders:

- Add a doubleTap step kind to the sequence allowlist on both ends,
  mirroring the retired tapSeries doubleTapAt branch.
- The single doubleTap interactor sends a one-step sequence and parses
  the result, surfacing step failures as errors.
- Swipe series unroll ping-pong daemon-side into per-step endpoints;
  the runner's coordinate-drag path ignores durationMs exactly as the
  daemon-sent (non-synthesized) dragSeries did.
- Extract runIosSequenceChunks so press and swipe share the chunking,
  aggregation, and global step-index rebasing.
- Keep tapSeries/dragSeries runner handlers for wire compatibility with
  older daemons, annotated like interactionFrame; remove both from the
  preflight-skip allowlist (daemon never sends them) and update ADR
  0005 / protocol-optimizations docs.

This also closes the latent watchdog exposure where press --count N
--interval-ms M routed to tapSeries and executed all pauses inside one
30s-watchdog main-thread block with no chunking.

Behavior note: plain tap series now use the synthesized HID tap path on
iOS non-tv (with runner-side tapAt fallback), matching the individual
tap command instead of the retired tapSeries' XCUICoordinate taps.

https://claude.ai/code/session_01VokBZWESTDgcnbYwS4DkJo

* refactor(ios): drop dead series wire surface from the daemon

- Remove chunkRunnerSequenceSteps: superseded by the budget-aware
  chunker; no production callers remained.
- Remove tapSeries/dragSeries from the RunnerCommand union along with
  their orphaned fields (count, intervalMs, doubleTap, pauseMs,
  pattern) and protocol fixtures: this type is the send surface of the
  current daemon, which no longer sends either command. The Swift
  runner keeps serving both for wire compatibility with older daemons.
- Retarget the ready-mutation preflight test from tapSeries to
  sequence.

https://claude.ai/code/session_01VokBZWESTDgcnbYwS4DkJo

* refactor(ios): remove retired series and frame wire commands entirely

Drops the runner-side wire compatibility for tapSeries, dragSeries, and
interactionFrame now that no daemon path sends them (series fuse into
sequence since this branch; interactionFrame was fused into scroll in
#760):

- Swift: delete the three handler cases, performDragSeries, runSeries
  (no remaining callers), the CommandType enum cases, journal-retention
  and traits entries, and the Command fields (count, intervalMs,
  doubleTap, pauseMs, pattern) that existed only for them. The
  never-sent synthesized dragSeries branch goes with it.
- TS: drop interactionFrame from the RunnerCommand union and
  isReadOnlyRunnerCommand, and its protocol fixture.
- Update stale perf scenario labels referencing the retired commands.

Verified dead before removal: no dynamic command construction anywhere
(runner-command-recovery only echoes in-flight command ids), no
raw-string references in Swift, no docs references. Helpers shared with
live paths (synthesizedDragAt, doubleTapAt, keyboardAvoidingDragPoints,
sleepFor) all retain callers.

Compat: an old daemon paired with a runner built from these sources
gets a CommandType decode rejection; the source-fingerprint check
rebuilds a matching runner on the next session.

https://claude.ai/code/session_01VokBZWESTDgcnbYwS4DkJo

---------

Co-authored-by: Claude <noreply@anthropic.com>
2026-06-11 14:08:30 +02:00

172 lines
6.7 KiB
TypeScript

import path from 'node:path';
import type { ResolvedProfile } from './platform-profiles.ts';
// A legacy-form batch step: maps through the exact documented CLI grammar.
// `flags` uses internal CliFlags field names (e.g. snapshotInteractiveOnly).
export type BatchStepSpec = {
command: string;
positionals?: string[];
flags?: Record<string, unknown>;
};
type ScenarioStepBase = {
label: string;
command: string;
// When set, the harness runs an untimed `open --relaunch` (reset to root, top of list)
// before timing this step. Used for steps whose precondition is a clean root, since
// earlier commands (find/is, search) leave the list scrolled or in a different surface.
freshRoot?: boolean;
};
// Discriminated on execMode so the invoker gets the right payload without `!`/`?? []`:
// standalone carries full CLI args; batch carries one legacy batch step.
export type ScenarioStep =
| (ScenarioStepBase & { execMode: 'standalone'; args: string[] })
| (ScenarioStepBase & { execMode: 'batch'; step: BatchStepSpec; isSnapshot?: boolean });
export type StepContext = { artifactsDir: string };
function std(label: string, command: string, args: string[]): ScenarioStep {
return { label, command, execMode: 'standalone', args };
}
function bat(
label: string,
command: string,
step: BatchStepSpec,
opts: { isSnapshot?: boolean; freshRoot?: boolean } = {},
): ScenarioStep {
return { label, command, execMode: 'batch' as const, step, ...opts };
}
// One ordered pass over Settings. The harness repeats this N (+warmup) times;
// the leading `open --relaunch` resets the app to its root each round, so every
// round starts from a known state while commands run in their natural order.
export function buildSettingsTour(p: ResolvedProfile, ctx: StepContext): ScenarioStep[] {
const s = p.selectors;
const shot = path.join(ctx.artifactsDir, 'shot.png');
const rec = path.join(ctx.artifactsDir, 'rec.mp4');
const trace = path.join(ctx.artifactsDir, 'trace.log');
// Text entry differs per platform: iOS fills the root search field directly (focusing it
// first can hang); Android must open the search screen before an editable field exists.
const textEntry: ScenarioStep[] = p.selectors.searchEditableAtRoot
? [
// iOS: editable search field exists at root; fill it directly (freshRoot resets scroll).
bat(
'fill search',
'fill',
{ command: 'fill', positionals: [s.searchFieldEditable, 'general'] },
{ freshRoot: true },
),
bat('type', 'type', { command: 'type', positionals: ['wifi'] }),
bat('get editable text', 'get', {
command: 'get',
positionals: ['text', s.searchFieldEditable],
}),
bat('keyboard return', 'keyboard', { command: 'keyboard', positionals: ['return'] }),
]
: [
// Android: tap the search entry first to reveal the editable, then type/fill it.
bat(
'press search field',
'press',
{ command: 'press', positionals: [s.searchField] },
{ freshRoot: true },
),
bat('type', 'type', { command: 'type', positionals: ['wifi'] }),
bat('fill search', 'fill', {
command: 'fill',
positionals: [s.searchFieldEditable, 'general'],
}),
bat('get editable text', 'get', {
command: 'get',
positionals: ['text', s.searchFieldEditable],
}),
];
// These iOS-only repeated gesture forms fuse into `sequence` runner requests:
// press --count > 1 and swipe --count > 1 both batch their steps into one command.
const iosRunnerSeries: ScenarioStep[] =
p.platform === 'ios'
? [
bat(
'press series (sequence)',
'press',
{ command: 'press', positionals: ['200', '95'], flags: { count: 2, intervalMs: 50 } },
{ freshRoot: true },
),
bat(
'swipe series (sequence)',
'swipe',
{
command: 'swipe',
positionals: ['200', '650', '200', '450', '120'],
flags: { count: 2, pauseMs: 50, pattern: 'ping-pong' },
},
{ freshRoot: true },
),
]
: [];
return [
// --- reset to root via relaunch ---
std('open (relaunch → root)', 'open', ['open', p.appTarget, '--relaunch']),
// --- reads on the root tree (snapshots first; anchor label is visible here) ---
bat(
'snapshot -i (root)',
'snapshot',
{ command: 'snapshot', flags: { snapshotInteractiveOnly: true } },
{ isSnapshot: true },
),
bat('snapshot (root)', 'snapshot', { command: 'snapshot' }, { isSnapshot: true }),
// --- navigate into a sub-screen from a fresh root (freshRoot resets scroll so the
// deep-screen row is in view), read it, then return ---
bat(
'press → deep screen',
'press',
{ command: 'press', positionals: [s.deepScreen] },
{ freshRoot: true },
),
bat('snapshot (deep)', 'snapshot', { command: 'snapshot' }, { isSnapshot: true }),
bat(
'snapshot -i (deep)',
'snapshot',
{ command: 'snapshot', flags: { snapshotInteractiveOnly: true } },
{ isSnapshot: true },
),
bat('back', 'back', { command: 'back' }),
// --- iOS runner series commands surfaced by PR #643 ---
...iosRunnerSeries,
// --- targeted reads against the visible anchor (freshRoot so the anchor is on screen) ---
bat(
'wait text',
'wait',
{ command: 'wait', positionals: ['text', s.anchorText, '3000'] },
{ freshRoot: true },
),
bat('find', 'find', { command: 'find', positionals: [s.anchorText] }),
bat('get text', 'get', { command: 'get', positionals: ['text', s.anchorLabel] }),
bat('is visible', 'is', { command: 'is', positionals: ['visible', s.anchorLabel] }),
// --- text entry (platform-specific order; see textEntry above) then scroll results ---
...textEntry,
bat('scroll down', 'scroll', { command: 'scroll', positionals: ['down'] }),
// --- artifact-producing commands; record brackets the rest so the clip has >1s of
// footage (an instant start→stop makes simctl recordVideo fail to finalize) ---
std('record start', 'record', ['record', 'start', rec, '--hide-touches']),
bat('screenshot', 'screenshot', { command: 'screenshot', positionals: [shot] }),
bat('logs mark', 'logs', { command: 'logs', positionals: ['mark', 'perf-mark'] }),
bat('logs clear', 'logs', { command: 'logs', positionals: ['clear'] }),
std('trace start', 'trace', ['trace', 'start', trace]),
std('trace stop', 'trace', ['trace', 'stop']),
bat('perf', 'perf', { command: 'perf' }),
std('record stop', 'record', ['record', 'stop']),
];
}