qemu: wait_hit and wait_region poll a settling page

A hit or region check fired once, so a check one second after `nav`
raced a page still laying itself out (flare-edge #175). wait_hit X Y CLASS
TIMEOUT_S [box=...] [TEXT] and wait_region NAME TOLERANCE TIMEOUT_S run
the same judgement every 0.5 s until it holds or the deadline passes, and
report how long they waited. The assert_hit judgement moves into
judge_hit so both verbs mean the same thing by a match; a missing
reference or foreign tolerance stays fatal on the first answer, since
waiting cannot fix either. Offline tests cover both verbs and the timeout.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01N3G6m9Aw5RyVY4ZowtKzEj
This commit is contained in:
Noah
2026-09-08 17:33:19 -06:00
co-authored by Claude Fable 5.1
parent 40de48092f
commit cab6c25c7e
2 changed files with 88 additions and 21 deletions
+79 -18
View File
@@ -261,6 +261,40 @@ class Ctl:
HIT_RE = re.compile(r'^hit \d+,\d+: (\S+)(?: text="(.*)" box=(-?\d+),(-?\d+),(\d+)x(\d+))?$') HIT_RE = re.compile(r'^hit \d+,\d+: (\S+)(?: text="(.*)" box=(-?\d+),(-?\d+),(\d+)x(\d+))?$')
def judge_hit(ctl, x, y, cls, rest):
"""One `hit` round trip judged the way assert_hit documents: -> (ok, detail).
REST is the optional box=X1,Y1,WxH followed by the caption words."""
rest = list(rest)
want_box = None
if rest and rest[0].startswith("box="):
want_box = rest[0][4:]
rest = rest[1:]
want_text = " ".join(rest)
got = parse_hit(ctl.send(f"hit {x} {y}"))
if got is None:
return False, "nothing clickable there"
if got[0] != cls:
return False, f"{got[0]} text={got[1]!r} box={got[2]}"
if want_box and got[2] != want_box:
return False, f"{cls} moved: box={got[2]}"
if want_text and want_text not in got[1]:
return False, f"{cls} text={got[1]!r}"
return True, ""
def poll_until(check, timeout_s, period=0.5):
"""Run CHECK (-> (ok, detail)) until it holds or TIMEOUT_S passes;
-> (ok, detail, seconds waited). The wait_* verbs share this so they
all mean the same thing by a timeout."""
start = time.monotonic()
deadline = start + timeout_s
while True:
ok, detail = check()
if ok or time.monotonic() >= deadline:
return ok, detail, time.monotonic() - start
time.sleep(period)
def parse_hit(reply): def parse_hit(reply):
"""-> (cls, text, box) with box as 'x1,y1,WxH', or None for 'nothing'; """-> (cls, text, box) with box as 'x1,y1,WxH', or None for 'nothing';
raises on an unparseable reply.""" raises on an unparseable reply."""
@@ -578,24 +612,22 @@ def drive(s, f, script_path, outdir, size, ctl_path=None, console_path=None, ref
# geometry the UI itself reports is what tells one instance from # geometry the UI itself reports is what tells one instance from
# another. # another.
need_ctl(lineno, cmd) need_ctl(lineno, cmd)
x, y, cls = args[0], args[1], args[2] ok, detail = judge_hit(ctl, args[0], args[1], args[2], args[3:])
rest = args[3:] record(lineno, line, "ok" if ok else "fail", detail)
want_box = None elif cmd == "wait_hit":
if rest and rest[0].startswith("box="): # wait_hit X Y CLASS TIMEOUT_S [box=X1,Y1,WxH] [TEXT...]: assert_hit
want_box = rest[0][4:] # polled every 0.5s until it holds or TIMEOUT_S passes. A page
rest = rest[1:] # still building after `nav` answers `hit` with whatever is there
want_text = " ".join(rest) # at that instant, so a check one second in raced the layout
got = parse_hit(ctl.send(f"hit {x} {y}")) # (flare-edge #175). The timeout is the author's bound on how long
if got is None: # settling may take, not a sleep: a check that holds early returns
record(lineno, line, "fail", "nothing clickable there") # early, and the detail says how long it took.
elif got[0] != cls: need_ctl(lineno, cmd)
record(lineno, line, "fail", f"{got[0]} text={got[1]!r} box={got[2]}") x, y, cls, timeout_s = args[0], args[1], args[2], float(args[3])
elif want_box and got[2] != want_box: rest = args[4:]
record(lineno, line, "fail", f"{cls} moved: box={got[2]}") ok, detail, waited = poll_until(lambda: judge_hit(ctl, x, y, cls, rest), timeout_s)
elif want_text and want_text not in got[1]: record(lineno, line, "ok" if ok else "fail",
record(lineno, line, "fail", f"{cls} text={got[1]!r}") f"waited {waited:.1f}s" if ok else detail)
else:
record(lineno, line, "ok")
elif cmd == "assert_json": elif cmd == "assert_json":
# assert_json PATH OP VALUE: one read of the guest's status JSON # assert_json PATH OP VALUE: one read of the guest's status JSON
# (see eval_json). PATH not (yet) in the doc is a `fail`, since # (see eval_json). PATH not (yet) in the doc is a `fail`, since
@@ -692,6 +724,35 @@ def drive(s, f, script_path, outdir, size, ctl_path=None, console_path=None, ref
else: else:
ok, detail = imgtools.compare(ref, cropped, ref["tolerance"]) ok, detail = imgtools.compare(ref, cropped, ref["tolerance"])
record(lineno, line, "ok" if ok else "fail", detail) record(lineno, line, "ok" if ok else "fail", detail)
elif cmd == "wait_region":
# wait_region NAME TOLERANCE TIMEOUT_S: assert_region polled every
# 0.5s, a fresh screendump each time, until the crop matches or
# the deadline passes (flare-edge #175). A missing reference or a
# foreign tolerance is fatal exactly as for assert_region: waiting
# cannot fix either, so polling stops on the first such answer.
need_imgtools(lineno, cmd)
name, want_tol, timeout_s = args[0], args[1], float(args[2])
ref = refs.get(name)
if ref is not None and want_tol != ref.get("tolerance"):
record(lineno, line, "fatal",
f"reference {name!r} was captured as {ref.get('tolerance')}, "
f"script expects {want_tol}: recapture or fix the spec")
else:
fault = []
def check_region():
r, cropped, err = fresh_region(name, "assert")
if err:
fault.append(err)
return True, err
return imgtools.compare(r, cropped, r["tolerance"])
ok, detail, waited = poll_until(check_region, timeout_s)
if fault:
record(lineno, line, "fatal", fault[0])
else:
record(lineno, line, "ok" if ok else "fail",
f"waited {waited:.1f}s" if ok else detail)
elif cmd == "assert_ocr": elif cmd == "assert_ocr":
# assert_ocr NAME REGEX: same crop as assert_region, OCR'd # assert_ocr NAME REGEX: same crop as assert_region, OCR'd
# through `tesseract`, REGEX searched (re.search, spaces allowed # through `tesseract`, REGEX searched (re.search, spaces allowed
+9 -3
View File
@@ -133,11 +133,14 @@ class DriveVerbs(unittest.TestCase):
"capture_region r1 0 0 8 8 exact\n" "capture_region r1 0 0 8 8 exact\n"
"assert_region r1 exact\n" "assert_region r1 exact\n"
"assert_region r1\n" "assert_region r1\n"
"wait_hit 47 676 obj 2 box=12,640,72x72\n"
"wait_region r1 exact 2\n"
"scroll 360 400 300\n" "scroll 360 400 300\n"
"home\n" "home\n"
) )
self.assertIsNone(rc, [r for r in rows if r["status"] != "ok"]) self.assertIsNone(rc, [r for r in rows if r["status"] != "ok"])
self.assertEqual(len(rows), 14) self.assertEqual(len(rows), 16)
self.assertIn("waited", by["wait_hit 47 676 obj 2 box=12,640,72x72"]["detail"])
self.assertTrue(all(r["status"] == "ok" for r in rows)) self.assertTrue(all(r["status"] == "ok" for r in rows))
def test_mismatches_are_fails_not_stops(self): def test_mismatches_are_fails_not_stops(self):
@@ -149,13 +152,16 @@ class DriveVerbs(unittest.TestCase):
"assert_json a.zz eq 1\n" "assert_json a.zz eq 1\n"
"assert_stat fps lt 0\n" "assert_stat fps lt 0\n"
"scroll 360 400 0\n" "scroll 360 400 0\n"
"wait_hit 47 676 obj 1 box=0,0,1x1\n"
"assert_page Demo/Rows\n" "assert_page Demo/Rows\n"
) )
self.assertEqual(rc, 1) self.assertEqual(rc, 1)
self.assertEqual([r["status"] for r in rows], self.assertEqual([r["status"] for r in rows],
["fail", "fail", "fail", "fail", "fail", "fail", "ok"]) ["fail", "fail", "fail", "fail", "fail", "fail", "fail", "ok"])
self.assertIn("moved", by["assert_hit 47 676 obj box=0,0,1x1"]["detail"]) self.assertIn("moved", by["assert_hit 47 676 obj box=0,0,1x1"]["detail"])
self.assertGreaterEqual(time.monotonic() - t0, 1.0, "wait_json must honour its timeout") self.assertIn("moved", by["wait_hit 47 676 obj 1 box=0,0,1x1"]["detail"])
self.assertGreaterEqual(time.monotonic() - t0, 2.0,
"wait_json and wait_hit must each honour their timeout")
def test_region_faults_are_per_step_fatal(self): def test_region_faults_are_per_step_fatal(self):
# A pre-seeded reference whose box does not fit a 64x64 screendump, # A pre-seeded reference whose box does not fit a 64x64 screendump,