phone-harness 真实 iPhone 控制 harness

31def connection_state():
32    """'ready' | 'blocked' | 'no-window' | 'not-running'.
33
34    'blocked' means a connect / 'iPhone in Use' / paused interstitial is on
35    screen. Cheap to call; use it to decide whether to proceed."""
36    if mirror.running_app() is None:
37        return "not-running"
38    if mirror.find_window() is None:
39        return "no-window"
40    path, win = mirror.capture()  # window exists, so this won't launch anything
41    texts = " ".join(o["text"] for o in _ocr.recognize(path, win)).lower()
42    return "blocked" if any(m in texts for m in _BLOCKED_MARKERS) else "ready"

45def ensure_mirroring():
46    """Return the mirroring window if the phone is connected and ready.
47
48    Never launches the app, taps Connect/Continue, or polls to reconnect —
49    resuming mirroring is physical and only the user can do it. If the session
50    isn't ready, raises with instructions for the user. STOP and relay that
51    message; do not try to tap through the connect screen yourself.
52    """
53    state = connection_state()
54    if state == "ready":
55        mirror.activate()
56        return mirror.find_window()
57    if state == "not-running":
58        raise RuntimeError(
59            "iPhone Mirroring isn't running. Please open the iPhone Mirroring "
60            "app and connect your phone, then retry — reconnecting is physical, "
61            "so I can't do it for you.")
62    if state == "no-window":
63        raise RuntimeError(
64            "iPhone Mirroring is open but no phone is connected. Please connect "
65            "your phone in the app, then retry.")
66    raise RuntimeError(
67        "iPhone Mirroring is not connected — it's showing a connect / 'iPhone "
68        "in Use' screen. This needs you: open iPhone Mirroring and connect the "
69        "phone, and if it says 'iPhone in Use', LOCK your iPhone so mirroring "
70        "can resume. Then retry. I will not tap Connect for you.")

19def recognize(path, window):
20    """OCR a capture of `window` ({x, y, w, h} screen points).
21
22    Returns [{text, confidence, x, y, w, h}] where (x, y) is the box center in
23    screen points — pass straight to tap(). Vision's normalized boxes have a
24    bottom-left origin; screen points have a top-left origin, hence the flip.
25    """
26    handler = Vision.VNImageRequestHandler.alloc().initWithURL_options_(
27        NSURL.fileURLWithPath_(path), {})
28    request = Vision.VNRecognizeTextRequest.alloc().init()
29    request.setRecognitionLevel_(Vision.VNRequestTextRecognitionLevelAccurate)
30    ok, err = handler.performRequests_error_([request], None)
31    if not ok:
32        raise RuntimeError(f"Vision OCR failed: {err}")
33
34    img_w, img_h = image_size(path)
35    sx = window["w"] / img_w  # image px -> screen points
36    sy = window["h"] / img_h
37
38    out = []
39    for obs in request.results() or []:
40        cand = obs.topCandidates_(1)
41        if not cand:
42            continue
43        bb = obs.boundingBox()
44        px = bb.origin.x * img_w
45        py_top = (1.0 - bb.origin.y - bb.size.height) * img_h
46        pw = bb.size.width * img_w
47        ph = bb.size.height * img_h
48        out.append({
49            "text": str(cand[0].string()),
50            "confidence": round(float(cand[0].confidence()), 3),
51            "x": round(window["x"] + (px + pw / 2) * sx, 1),
52            "y": round(window["y"] + (py_top + ph / 2) * sy, 1),
53            "w": round(pw * sx, 1),
54            "h": round(ph * sy, 1),
55        })
56    return out

12def run_doctor():
13    print("phone-harness doctor\n")
14    ok = True
15
16    try:
17        import Quartz, Vision, AppKit  # noqa: F401
18        ok &= _check("pyobjc frameworks (Quartz, Vision, AppKit)", True)
19    except ImportError as e:
20        _check("pyobjc frameworks", False,
21               f"pip install pyobjc-framework-Quartz pyobjc-framework-Vision ({e})")
22        return 1