Skip to content

log_viewer: parse the whole info.log instead of the last 8MB #152

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
yichao-liang wants to merge 1 commit into from
Closed

log_viewer: parse the whole info.log instead of the last 8MB #152

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
155 changes: 87 additions & 68 deletions scripts/log_viewer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -835,9 +835,6 @@ def _parse_info_log(path: str) -> Dict[str, Any]:
instead is wrong: only some configs run an initial pre-learning test
round, so the offset between learn count and round index varies.)
"""
text, _ = read_text(path, max_bytes=8 * 1024 * 1024)
text = ANSI_RE.sub("", text)
m_vid = VIDEO_SAVED_RE.search(text)
totals: List[Tuple[int, int, Optional[float]]] = []
rounds: List[Dict[int, Dict[str, Any]]] = []
current: Dict[int, Dict[str, Any]] = {}
Expand All @@ -847,70 +844,92 @@ def _parse_info_log(path: str) -> Dict[str, Any]:
pending_test: List[int] = []
last_explore: Optional[int] = None
done = False
for line in text.splitlines():
if DONE_RE.match(line):
done = True
continue
if "Test results:" in line:
ms, mt = NUM_SOLVED_RE.search(line), NUM_TOTAL_RE.search(line)
mr = AVG_TEST_REWARD_RE.search(line)
if ms and mt:
totals.append(
(int(float(ms.group(1))), int(float(mt.group(1))),
float(mr.group(1)) if mr else None))
continue
m = TASK_VERDICT_RE.match(line)
if m:
msg = m.group(3).strip()
# A later verdict for the same task within a round (e.g. the
# impossible-goal SOLVED after an approach failure) overwrites.
current[int(m.group(1)) - 1] = {
"solved": msg.startswith("SOLVED"),
"msg": msg,
}
continue
if TASKS_SOLVED_RE.search(line):
rounds.append(current)
current = {}
for num in pending_test:
test_round[num] = len(rounds) - 1
pending_test.clear()
continue
# Logged after the "Tasks solved" line, so this decorates the
# round that line just closed.
if "Per-task rewards:" in line and rounds:
for tm in PER_TASK_REWARD_RE.finditer(line):
verdict = rounds[-1].get(int(tm.group(1)))
if verdict is not None:
verdict["reward"] = float(tm.group(2))
continue
m = SAVED_EP_RE.search(line)
if m:
if m.group(2) == "explore":
pending.append(int(m.group(1)))
elif m.group(2) == "learn":
pending.clear()
elif m.group(2) == "test":
pending_test.append(int(m.group(1)))
continue
m = INTERACTION_RE.match(line)
if m:
last_explore = pending.pop(0) if pending else None
if last_explore is not None:
explore[last_explore] = {
"reward": float(m.group(1)),
"terminated": m.group(2) == "True",
"accepted": m.group(3) == "True",
"msg": "",
}
continue
m = INTERACTION_REJECT_RE.match(line)
if m and last_explore is not None:
explore[last_explore]["msg"] = "REJECTED: " + m.group(1).strip()
continue
m = INTERACTION_BAR_RE.match(line)
if m and last_explore is not None:
explore[last_explore]["msg"] = "solved but " + m.group(1).strip()
video_rel = ""
# Stream the whole file line by line: a tail-only size cap here used
# to drop the head of long runs' logs, silently losing early env
# verdicts (grey test chips, unmarked explore episodes, mis-paired
# interaction verdicts). Memory stays O(line) and the parse is
# cached by (mtime, size) in _cached.
try:
f = open(path, "r", encoding="utf-8", errors="replace")
except OSError:
f = None
if f is not None:
with f:
for raw in f:
line = ANSI_RE.sub("", raw).rstrip("\n")
if not video_rel:
m = VIDEO_SAVED_RE.search(line)
if m:
video_rel = m.group(1)
continue
if DONE_RE.match(line):
done = True
continue
if "Test results:" in line:
ms, mt = NUM_SOLVED_RE.search(line), NUM_TOTAL_RE.search(
line)
mr = AVG_TEST_REWARD_RE.search(line)
if ms and mt:
totals.append(
(int(float(ms.group(1))), int(float(mt.group(1))),
float(mr.group(1)) if mr else None))
continue
m = TASK_VERDICT_RE.match(line)
if m:
msg = m.group(3).strip()
# A later verdict for the same task within a round (e.g.
# the impossible-goal SOLVED after an approach failure)
# overwrites.
current[int(m.group(1)) - 1] = {
"solved": msg.startswith("SOLVED"),
"msg": msg,
}
continue
if TASKS_SOLVED_RE.search(line):
rounds.append(current)
current = {}
for num in pending_test:
test_round[num] = len(rounds) - 1
pending_test.clear()
continue
# Logged after the "Tasks solved" line, so this decorates the
# round that line just closed.
if "Per-task rewards:" in line and rounds:
for tm in PER_TASK_REWARD_RE.finditer(line):
verdict = rounds[-1].get(int(tm.group(1)))
if verdict is not None:
verdict["reward"] = float(tm.group(2))
continue
m = SAVED_EP_RE.search(line)
if m:
if m.group(2) == "explore":
pending.append(int(m.group(1)))
elif m.group(2) == "learn":
pending.clear()
elif m.group(2) == "test":
pending_test.append(int(m.group(1)))
continue
m = INTERACTION_RE.match(line)
if m:
last_explore = pending.pop(0) if pending else None
if last_explore is not None:
explore[last_explore] = {
"reward": float(m.group(1)),
"terminated": m.group(2) == "True",
"accepted": m.group(3) == "True",
"msg": "",
}
continue
m = INTERACTION_REJECT_RE.match(line)
if m and last_explore is not None:
explore[last_explore]["msg"] = ("REJECTED: " +
m.group(1).strip())
continue
m = INTERACTION_BAR_RE.match(line)
if m and last_explore is not None:
explore[last_explore]["msg"] = ("solved but " +
m.group(1).strip())
if current: # run still in progress or crashed mid-round
rounds.append(current)
return {
Expand All @@ -919,7 +938,7 @@ def _parse_info_log(path: str) -> Dict[str, Any]:
"explore": explore,
"test_round": test_round,
"done": done,
"video_rel": m_vid.group(1) if m_vid else "",
"video_rel": video_rel,
}


Expand Down
Loading