← Files Portable ResumeARCHIVED FILE
skills/.portable-resume/runtime/portable_resume/adapters/openclaw.py
38.7 KB · Oct 4, 2026 · 12:33 UTC
"""Read OpenClaw per-agent SQLite sessions as inert context.
Pinned format: ``openclaw-agent-sqlite-v1`` (``PRAGMA user_version = 11`` plus
matching ``schema_meta``). Discovery walks only
<root>/agents/<agentId>/agent/openclaw-agent.sqlite
and never invokes the OpenClaw Gateway, CLI, doctor, or migration tools.
"""
from __future__ import annotations
import json
import os
import re
import sqlite3
import stat
from dataclasses import replace
from datetime import datetime, timezone
from typing import Any, Mapping
from ..bounds import DEFAULT_BOUNDS, ReadBudget
from ..diagnostics import DiagnosticError
from ..model import Query, Session, SessionSummary, Turn
from ..paths import canonical_root, canonicalize_cwd, is_within, same_cwd
from ..sanitize import sanitize_turn_record
from ..snapshot import private_sqlite_connection, query_only_live_sqlite
from .base import CapabilityReport, ResolvedRef
from .common import within_age
FORMAT_ID = "openclaw-agent-sqlite-v1"
SCHEMA_VERSION = 11
DB_BASENAME = "openclaw-agent.sqlite"
_AGENT_ID_RE = re.compile(r"^[A-Za-z0-9][A-Za-z0-9._-]{0,127}$")
_DEFAULT_EXCLUDE_VIA = frozenset({"internal", "cron", "spawn", "run", "plugin"})
_REQUIRED_COLUMNS: dict[str, frozenset[str]] = {
"schema_meta": frozenset(
{"meta_key", "role", "schema_version", "agent_id", "app_version", "created_at", "updated_at"}
),
"session_nodes": frozenset(
{
"session_key",
"current_session_id",
"entry_json",
"updated_at",
"created_via",
"display_name",
"archived_at",
"last_interaction_at",
"created_at",
}
),
"session_windows": frozenset({"session_id", "session_key", "reason", "created_at", "updated_at"}),
"transcript_events": frozenset({"session_id", "seq", "event_json", "created_at"}),
}
class _DuplicateKey(ValueError):
pass
def _object(pairs: list[tuple[str, Any]]) -> dict[str, Any]:
value: dict[str, Any] = {}
for key, item in pairs:
if key in value:
raise _DuplicateKey(key)
value[key] = item
return value
def _root_candidate(query: Query) -> str:
if query.source_root:
return query.source_root
return os.path.expanduser("~/.openclaw")
def _existing_root(query: Query) -> str | None:
candidate = _root_candidate(query)
try:
if not os.path.isdir(candidate):
return None
return canonical_root(candidate)
except DiagnosticError:
if query.source_root:
raise
return None
def _regular_dir(path: str, root: str) -> bool:
try:
mode = os.lstat(path).st_mode
except OSError:
return False
if stat.S_ISLNK(mode) or not stat.S_ISDIR(mode):
return False
try:
return is_within(path, root)
except DiagnosticError:
return False
def _regular_db_file(path: str, root: str) -> bool:
try:
mode = os.lstat(path).st_mode
except OSError:
return False
if stat.S_ISLNK(mode) or not stat.S_ISREG(mode):
return False
try:
return is_within(path, root)
except DiagnosticError:
return False
def _composite_id(agent_id: str, session_id: str) -> str:
return f"{agent_id}:{session_id}"
def _parse_ref(value: str | None) -> tuple[str | None, str | None]:
"""Return (agent_id_or_None, session_id_or_None) from a ref string.
Free-text titles (spaces / non-id tokens) return ``(None, None)`` so the
generic selector can match them after a normal bounded list.
"""
if not value:
return None, None
text = value.strip()
if not text or text == "latest":
return None, None
if ":" in text:
agent, session = text.split(":", 1)
agent = agent.strip()
session = session.strip()
if (
agent
and session
and _AGENT_ID_RE.fullmatch(agent)
and re.fullmatch(r"[A-Za-z0-9][A-Za-z0-9._-]{0,200}", session)
):
return agent, session
return None, None
if re.fullmatch(r"[A-Za-z0-9][A-Za-z0-9._-]{0,200}", text):
return None, text
return None, None
def _ms_stamp(value: object) -> str | None:
if value is None:
return None
if not isinstance(value, (int, float)) or isinstance(value, bool):
return None
number = float(value)
if number > 10_000_000_000: # epoch milliseconds
number /= 1000.0
try:
return (
datetime.fromtimestamp(number, timezone.utc)
.isoformat(timespec="microseconds")
.replace("+00:00", "Z")
)
except (OverflowError, OSError, ValueError):
return None
def _cwd_from_entry(entry_json: str | None) -> str | None:
if not entry_json:
return None
try:
payload = json.loads(entry_json, object_pairs_hook=_object)
except (json.JSONDecodeError, _DuplicateKey, RecursionError, UnicodeDecodeError):
return None
if not isinstance(payload, Mapping):
return None
raw = payload.get("cwd")
if not isinstance(raw, str) or not raw.strip():
return None
try:
return canonicalize_cwd(raw)
except DiagnosticError:
return None
def _agent_db_paths(root: str, budget: ReadBudget | None = None) -> list[tuple[str, str]]:
"""Return ``(agent_id, db_path)`` under ``agents/*/agent/openclaw-agent.sqlite``."""
limits = budget.limits if budget is not None else DEFAULT_BOUNDS
scan_limit = min(limits.scanned_records, DEFAULT_BOUNDS.scanned_records)
agents = os.path.join(root, "agents")
if not _regular_dir(agents, root):
return []
values: list[tuple[str, str]] = []
try:
with os.scandir(agents) as entries:
names: list[str] = []
for entry in entries:
if len(names) >= scan_limit:
raise DiagnosticError.limit_exceeded()
names.append(entry.name)
except DiagnosticError:
raise
except OSError as error:
raise DiagnosticError.source_busy(provider=FORMAT_ID) from error
names.sort()
for name in names:
if not _AGENT_ID_RE.fullmatch(name):
continue
agent_dir = os.path.join(agents, name)
if not _regular_dir(agent_dir, root):
continue
agent_sub = os.path.join(agent_dir, "agent")
if not _regular_dir(agent_sub, root):
continue
database = os.path.join(agent_sub, DB_BASENAME)
if _regular_db_file(database, root):
values.append((name, database))
if len(values) > scan_limit:
raise DiagnosticError.limit_exceeded()
return values
def _open_connection(database: str, root: str, budget: ReadBudget | None = None):
limits = budget.limits if budget is not None else DEFAULT_BOUNDS
try:
size = os.path.getsize(database)
except OSError as error:
raise DiagnosticError.source_busy(provider=FORMAT_ID) from error
if size > limits.sqlite_snapshot_bytes:
return query_only_live_sqlite(database, root=root, provider=FORMAT_ID)
return private_sqlite_connection(database, root=root, bounds=limits, provider=FORMAT_ID)
def _require_schema(connection: sqlite3.Connection, *, expected_agent: str | None = None) -> str:
"""Validate closed schema and return the agent_id from schema_meta."""
try:
integrity = connection.execute("PRAGMA integrity_check(1)").fetchone()
if integrity != ("ok",):
raise DiagnosticError("E_CORRUPT_RECORD", source="openclaw", provider=FORMAT_ID)
user_version = connection.execute("PRAGMA user_version").fetchone()
if user_version is None or int(user_version[0]) != SCHEMA_VERSION:
raise DiagnosticError("E_UNSUPPORTED_FORMAT", source="openclaw", provider=FORMAT_ID)
tables = {
row[0]
for row in connection.execute(
"SELECT name FROM sqlite_master WHERE type='table' AND name IN "
"('schema_meta','session_nodes','session_windows','transcript_events')"
)
}
if tables != set(_REQUIRED_COLUMNS):
raise DiagnosticError("E_UNSUPPORTED_FORMAT", source="openclaw", provider=FORMAT_ID)
for table, required in _REQUIRED_COLUMNS.items():
columns = {
str(row[1])
for row in connection.execute(f'PRAGMA table_info("{table}")')
}
if not required.issubset(columns):
raise DiagnosticError("E_UNSUPPORTED_FORMAT", source="openclaw", provider=FORMAT_ID)
meta = connection.execute(
"""
SELECT role, schema_version, agent_id
FROM schema_meta
WHERE meta_key = 'primary'
LIMIT 2
"""
).fetchall()
if len(meta) != 1:
raise DiagnosticError("E_UNSUPPORTED_FORMAT", source="openclaw", provider=FORMAT_ID)
role, schema_version, agent_id = meta[0]
if role != "agent" or int(schema_version) != SCHEMA_VERSION:
raise DiagnosticError("E_UNSUPPORTED_FORMAT", source="openclaw", provider=FORMAT_ID)
if not isinstance(agent_id, str) or not _AGENT_ID_RE.fullmatch(agent_id):
raise DiagnosticError("E_CORRUPT_RECORD", source="openclaw", provider=FORMAT_ID)
if expected_agent is not None and agent_id != expected_agent:
raise DiagnosticError("E_CORRUPT_RECORD", source="openclaw", provider=FORMAT_ID)
return agent_id
except sqlite3.DatabaseError as error:
raise DiagnosticError("E_UNSUPPORTED_FORMAT", source="openclaw", provider=FORMAT_ID) from error
except (TypeError, ValueError) as error:
raise DiagnosticError("E_UNSUPPORTED_FORMAT", source="openclaw", provider=FORMAT_ID) from error
def _row_to_summary(
*,
agent_id: str,
database: str,
session_id: str,
entry_json: object,
updated_at: object,
created_at: object,
display_name: object,
last_interaction_at: object,
query: Query,
require_age: bool,
) -> SessionSummary | None:
if not isinstance(session_id, str) or not session_id:
return None
cwd = _cwd_from_entry(entry_json if isinstance(entry_json, str) else None)
if query.cwd is not None:
if cwd is None or not same_cwd(cwd, query.cwd):
return None
stamp = _ms_stamp(last_interaction_at if last_interaction_at is not None else updated_at)
if require_age and not within_age(
stamp, query.within_min, default_minutes=DEFAULT_BOUNDS.listing_age_minutes
):
return None
title = display_name if isinstance(display_name, str) and display_name.strip() else None
return SessionSummary(
source="openclaw",
session_id=_composite_id(agent_id, session_id),
source_path=database,
title=title,
cwd=cwd,
branch=None,
created_at=_ms_stamp(created_at),
updated_at=stamp,
provider=FORMAT_ID,
warnings=(),
)
def _entry_prefix_bytes(budget: ReadBudget) -> int:
"""Return a sentinel-inclusive prefix cap for the remaining read budget."""
record_limit = min(budget.limits.record_bytes, DEFAULT_BOUNDS.record_bytes)
source_limit = min(
budget.limits.source_read_bytes, DEFAULT_BOUNDS.source_read_bytes
)
remaining = max(0, source_limit - budget.bytes_read)
return min(record_limit, remaining) + 1
def _decode_bounded_entry_json(
storage_type: object,
byte_length: object,
prefix: object,
budget: ReadBudget,
) -> str:
"""Validate and decode a SQL-bounded ``entry_json`` projection."""
budget.consume_records()
if storage_type != "text":
raise DiagnosticError("E_CORRUPT_RECORD", source="openclaw", provider=FORMAT_ID)
if type(byte_length) is not int or not isinstance(prefix, bytes):
raise DiagnosticError("E_CORRUPT_RECORD", source="openclaw", provider=FORMAT_ID)
record_limit = min(budget.limits.record_bytes, DEFAULT_BOUNDS.record_bytes)
source_limit = min(
budget.limits.source_read_bytes, DEFAULT_BOUNDS.source_read_bytes
)
remaining = max(0, source_limit - budget.bytes_read)
if byte_length < 0 or byte_length > record_limit or byte_length > remaining:
raise DiagnosticError.limit_exceeded()
if len(prefix) != byte_length:
raise DiagnosticError("E_CORRUPT_RECORD", source="openclaw", provider=FORMAT_ID)
budget.consume_bytes(byte_length)
try:
return prefix.decode("utf-8")
except UnicodeDecodeError as error:
raise DiagnosticError(
"E_CORRUPT_RECORD", source="openclaw", provider=FORMAT_ID
) from error
def _exact_session_summaries(
connection: sqlite3.Connection,
*,
agent_id: str,
database: str,
session_filter: str,
query: Query,
budget: ReadBudget,
) -> list[SessionSummary]:
"""Resolve exact composite/native ids before the normal listing cap."""
values: list[SessionSummary] = []
scan_limit = min(budget.limits.scanned_records, DEFAULT_BOUNDS.scanned_records)
entry_prefix_bytes = _entry_prefix_bytes(budget)
nodes = connection.execute(
"""
SELECT
current_session_id,
typeof(entry_json), length(CAST(entry_json AS BLOB)),
substr(CAST(entry_json AS BLOB), 1, ?),
updated_at, created_at, display_name,
last_interaction_at, archived_at, created_via
FROM session_nodes
WHERE current_session_id = ?
LIMIT ?
""",
(entry_prefix_bytes, session_filter, scan_limit + 1),
)
node_count = 0
for row in nodes:
node_count += 1
if node_count > scan_limit:
raise DiagnosticError.limit_exceeded()
entry_json = _decode_bounded_entry_json(row[1], row[2], row[3], budget)
item = _row_to_summary(
agent_id=agent_id,
database=database,
session_id=row[0],
entry_json=entry_json,
updated_at=row[4],
created_at=row[5],
display_name=row[6],
last_interaction_at=row[7],
query=query,
require_age=False,
)
if item is not None:
values.append(item)
if values:
return values
# Historical/reset window not pointed by current_session_id.
windows = connection.execute(
"""
SELECT
w.session_id,
typeof(n.entry_json), length(CAST(n.entry_json AS BLOB)),
substr(CAST(n.entry_json AS BLOB), 1, ?),
w.updated_at,
w.created_at,
COALESCE(w.display_name, n.display_name),
n.last_interaction_at
FROM session_windows w
LEFT JOIN session_nodes n ON n.session_key = w.session_key
WHERE w.session_id = ?
LIMIT ?
""",
(_entry_prefix_bytes(budget), session_filter, scan_limit + 1),
)
window_count = 0
for row in windows:
window_count += 1
if window_count > scan_limit:
raise DiagnosticError.limit_exceeded()
entry_json = _decode_bounded_entry_json(row[1], row[2], row[3], budget)
item = _row_to_summary(
agent_id=agent_id,
database=database,
session_id=row[0],
entry_json=entry_json,
updated_at=row[4],
created_at=row[5],
display_name=row[6],
last_interaction_at=row[7],
query=query,
require_age=False,
)
if item is not None:
values.append(item)
return values
def _list_nodes(
connection: sqlite3.Connection,
*,
agent_id: str,
database: str,
query: Query,
include_internal: bool,
budget: ReadBudget,
) -> list[SessionSummary]:
scan_limit = min(budget.limits.scanned_records, DEFAULT_BOUNDS.scanned_records)
list_limit = min(budget.limits.listed_sessions, DEFAULT_BOUNDS.listed_sessions)
if list_limit <= 0:
return []
rows = connection.execute(
"""
SELECT
session_key,
current_session_id,
typeof(entry_json), length(CAST(entry_json AS BLOB)),
substr(CAST(entry_json AS BLOB), 1, ?),
updated_at,
created_at,
created_via,
display_name,
archived_at,
last_interaction_at
FROM session_nodes
ORDER BY COALESCE(last_interaction_at, updated_at, created_at) DESC, current_session_id ASC
LIMIT ?
""",
(_entry_prefix_bytes(budget), scan_limit + 1),
)
values: list[SessionSummary] = []
row_count = 0
for row in rows:
row_count += 1
if row_count > scan_limit:
raise DiagnosticError.limit_exceeded()
(
_session_key,
session_id,
entry_storage_type,
entry_byte_length,
entry_prefix,
updated_at,
created_at,
created_via,
display_name,
archived_at,
last_interaction_at,
) = row
entry_json = _decode_bounded_entry_json(
entry_storage_type, entry_byte_length, entry_prefix, budget
)
if len(values) >= list_limit:
continue
if archived_at is not None and not include_internal:
continue
if (
not include_internal
and isinstance(created_via, str)
and created_via in _DEFAULT_EXCLUDE_VIA
):
continue
item = _row_to_summary(
agent_id=agent_id,
database=database,
session_id=session_id,
entry_json=entry_json,
updated_at=updated_at,
created_at=created_at,
display_name=display_name,
last_interaction_at=last_interaction_at,
query=query,
require_age=not include_internal,
)
if item is None:
continue
values.append(item)
return values
def _decode_event(raw: str) -> Mapping[str, Any]:
try:
value = json.loads(raw, object_pairs_hook=_object)
except (json.JSONDecodeError, _DuplicateKey, RecursionError, UnicodeDecodeError) as error:
raise DiagnosticError("E_CORRUPT_RECORD", source="openclaw", provider=FORMAT_ID) from error
if not isinstance(value, Mapping) or not isinstance(value.get("type"), str):
raise DiagnosticError("E_CORRUPT_RECORD", source="openclaw", provider=FORMAT_ID)
return value
def _nested_message(event: Mapping[str, Any]) -> Mapping[str, Any] | None:
raw = event.get("message")
return raw if isinstance(raw, Mapping) else None
def _event_role(event: Mapping[str, Any]) -> str | None:
role = event.get("role")
if isinstance(role, str):
return role
nested = _nested_message(event)
if nested is not None:
nested_role = nested.get("role")
if isinstance(nested_role, str):
return nested_role
return None
def _message_text(event: Mapping[str, Any]) -> str | None:
for key in ("text", "content", "summary"):
value = event.get(key)
if isinstance(value, str) and value.strip():
return value
nested = _nested_message(event)
if nested is not None:
for key in ("text", "content"):
value = nested.get(key)
if isinstance(value, str) and value.strip():
return value
# Nested content may be a list of parts with textual chunks.
parts = nested.get("content")
if isinstance(parts, list):
chunks: list[str] = []
for part in parts:
if isinstance(part, str) and part.strip():
chunks.append(part)
elif isinstance(part, Mapping):
text = part.get("text")
if isinstance(text, str) and text.strip():
chunks.append(text)
if chunks:
return "\n".join(chunks)
# Flat fixtures may store the message body under the bare "message" string key.
bare = event.get("message")
if isinstance(bare, str) and bare.strip():
return bare
return None
def _active_branch_events(decoded: list[Mapping[str, Any]]) -> list[Mapping[str, Any]]:
"""Return active-branch conversation events (message/compaction).
When events carry ``id``/``parentId``, walk from the latest leaf. Linear
fixtures without ids keep sequence order. ``branch_summary`` stays in the
ancestry index so children do not lose their parent pointer, but is not
emitted as a turn. Compaction nodes may retarget ancestry via
``firstKeptEntryId`` / ``firstKeptSeq``.
"""
graph_types = frozenset({"message", "custom_message", "compaction", "branch_summary"})
emit_types = frozenset({"message", "custom_message", "compaction"})
candidates = [event for event in decoded if event.get("type") in graph_types]
if not candidates:
return []
if not any(isinstance(event.get("id"), str) and event.get("id") for event in candidates):
return [event for event in candidates if event.get("type") in emit_types]
by_id: dict[str, Mapping[str, Any]] = {}
by_seq: dict[int, Mapping[str, Any]] = {}
for index, event in enumerate(candidates):
identifier = event.get("id")
if isinstance(identifier, str) and identifier:
if identifier in by_id:
raise DiagnosticError("E_CORRUPT_RECORD", source="openclaw", provider=FORMAT_ID)
by_id[identifier] = event
# Sequence position for firstKeptSeq is the event's own seq if present.
seq = event.get("seq")
if isinstance(seq, int):
by_seq[seq] = event
else:
by_seq[index + 1] = event
# Leaf is the latest graph node (including branch_summary). Emitting types are
# preferred only when no later graph node exists; a trailing branch_summary
# must win so we do not select an abandoned sibling (Codex P1).
leaf: Mapping[str, Any] | None = None
for event in reversed(candidates):
identifier = event.get("id")
if isinstance(identifier, str) and identifier in by_id:
leaf = event
break
if leaf is None:
return [event for event in candidates if event.get("type") in emit_types]
path: list[Mapping[str, Any]] = []
seen: set[str] = set()
current: Mapping[str, Any] | None = leaf
while current is not None:
identifier = current.get("id")
if not isinstance(identifier, str) or not identifier:
break
if identifier in seen:
raise DiagnosticError("E_CORRUPT_RECORD", source="openclaw", provider=FORMAT_ID)
seen.add(identifier)
path.append(current)
if current.get("type") == "compaction":
kept_id = current.get("firstKeptEntryId")
kept_seq = current.get("firstKeptSeq")
if isinstance(kept_id, str) and kept_id in by_id:
current = by_id[kept_id]
continue
if isinstance(kept_seq, int) and kept_seq in by_seq:
current = by_seq[kept_seq]
continue
parent = current.get("parentId")
if parent is None or parent == "":
break
if not isinstance(parent, str) or parent not in by_id:
break
current = by_id[parent]
path.reverse()
return [event for event in path if event.get("type") in emit_types]
def _show_session(
connection: sqlite3.Connection,
*,
agent_id: str,
session_id: str,
database: str,
query: Query,
budget: ReadBudget,
) -> Session:
entry_prefix_bytes = _entry_prefix_bytes(budget)
node = connection.execute(
"""
SELECT session_key, current_session_id,
typeof(entry_json), length(CAST(entry_json AS BLOB)),
substr(CAST(entry_json AS BLOB), 1, ?),
updated_at, created_at, display_name, last_interaction_at
FROM session_nodes
WHERE current_session_id = ?
LIMIT 2
""",
(entry_prefix_bytes, session_id),
).fetchall()
if len(node) != 1:
# Historical window reached by exact id (not current): allow window-only show.
window = connection.execute(
"""
SELECT session_id, session_key, display_name, created_at, updated_at
FROM session_windows
WHERE session_id = ?
LIMIT 2
""",
(session_id,),
).fetchall()
if len(window) != 1:
raise DiagnosticError("E_NO_MATCH", source="openclaw", provider=FORMAT_ID)
session_id_w, session_key, display_name, created_at, updated_at = window[0]
parent = connection.execute(
"""
SELECT typeof(entry_json), length(CAST(entry_json AS BLOB)),
substr(CAST(entry_json AS BLOB), 1, ?), last_interaction_at
FROM session_nodes
WHERE session_key = ?
LIMIT 1
""",
(entry_prefix_bytes, session_key),
).fetchone()
last_interaction_at = parent[3] if parent else None
entry_json = (
_decode_bounded_entry_json(
parent[0],
parent[1],
parent[2],
budget,
)
if parent
else None
)
title = display_name if isinstance(display_name, str) else None
stamp = _ms_stamp(last_interaction_at if last_interaction_at is not None else updated_at)
created = _ms_stamp(created_at)
else:
(
_session_key,
session_id_w,
entry_storage_type,
entry_byte_length,
entry_prefix,
updated_at,
created_at,
display_name,
last_interaction_at,
) = node[0]
entry_json = _decode_bounded_entry_json(
entry_storage_type,
entry_byte_length,
entry_prefix,
budget,
)
title = display_name if isinstance(display_name, str) else None
stamp = _ms_stamp(last_interaction_at if last_interaction_at is not None else updated_at)
created = _ms_stamp(created_at)
cwd = _cwd_from_entry(entry_json if isinstance(entry_json, str) else None)
if query.cwd is not None and (cwd is None or not same_cwd(cwd, query.cwd)):
raise DiagnosticError("E_NO_MATCH", source="openclaw", provider=FORMAT_ID)
transcript_limit = min(budget.limits.transcript_records, DEFAULT_BOUNDS.transcript_records)
limit = transcript_limit + 1
cursor = connection.execute(
"""
SELECT seq, event_json, created_at
FROM transcript_events
WHERE session_id = ?
ORDER BY seq ASC
LIMIT ?
""",
(session_id, limit),
)
turns: list[Turn] = []
warnings: list[str] = []
seen_seq: set[int] = set()
decoded: list[Mapping[str, Any]] = []
turn_bounds = replace(DEFAULT_BOUNDS, tool_output_chars=query.max_tool_chars)
row_count = 0
try:
for seq, event_json, _created in cursor:
row_count += 1
if row_count > transcript_limit:
raise DiagnosticError.limit_exceeded()
budget.consume_transcript_records()
if not isinstance(seq, int):
raise DiagnosticError("E_CORRUPT_RECORD", source="openclaw", provider=FORMAT_ID)
if seq in seen_seq:
raise DiagnosticError("E_CORRUPT_RECORD", source="openclaw", provider=FORMAT_ID)
seen_seq.add(seq)
if not isinstance(event_json, str):
raise DiagnosticError("E_CORRUPT_RECORD", source="openclaw", provider=FORMAT_ID)
encoded = event_json.encode("utf-8")
if len(encoded) > min(budget.limits.record_bytes, DEFAULT_BOUNDS.record_bytes):
raise DiagnosticError.limit_exceeded()
budget.consume_bytes(len(encoded))
event = _decode_event(event_json)
if event["type"] in {"custom", "session"}:
continue
# Attach SQL sequence for firstKeptSeq retention (Codex P1).
annotated = dict(event)
annotated["seq"] = seq
# Keep branch_summary for ancestry only; turn emission filters it out.
decoded.append(annotated)
finally:
cursor.close()
for event in _active_branch_events(decoded):
kind = event["type"]
if kind == "compaction":
text = _message_text(event)
if text is None:
continue
raw = {"role": "assistant", "content": text}
elif kind in {"message", "custom_message"}:
role = _event_role(event)
if role in {"toolResult", "bashExecution", "tool_result", "function"}:
role = "tool"
if role not in {"user", "assistant", "tool"}:
continue
text = _message_text(event)
if text is None and role == "tool":
nested = _nested_message(event)
source = nested if nested is not None else event
command = source.get("command")
output = source.get("output")
chunks = []
if isinstance(command, str) and command.strip():
chunks.append(command)
if isinstance(output, str) and output.strip():
chunks.append(output)
text = "\n".join(chunks) if chunks else None
if text is None and kind == "custom_message":
# Visible custom_message may omit role; treat as inert assistant context.
if event.get("display") is False:
continue
text = _message_text(event)
if text is None:
continue
role = "assistant"
if text is None:
continue
raw = {"role": role, "content": text}
else:
continue
turn, turn_warnings = sanitize_turn_record(
raw,
ordinal=len(turns),
bounds=turn_bounds,
)
warnings.extend(turn_warnings)
if turn is not None:
budget.consume_turns()
turns.append(turn)
last_user = next((turn.content for turn in reversed(turns) if turn.role == "user"), None)
last_assistant = next(
(turn.content for turn in reversed(turns) if turn.role == "assistant"),
None,
)
return Session(
source="openclaw",
session_id=_composite_id(agent_id, session_id),
source_path=database,
title=title,
cwd=cwd,
branch=None,
created_at=created,
updated_at=stamp,
last_user_request=last_user,
last_assistant_action=last_assistant,
turns=tuple(turns),
warnings=tuple(dict.fromkeys(warnings)),
)
class OpenClawAdapter:
key = "openclaw"
def approved_roots(self, query: Query) -> tuple[str, ...]:
root = _existing_root(query)
return (root,) if root else ()
def probe(self, query: Query) -> CapabilityReport:
try:
root = _existing_root(query)
if root is None:
return CapabilityReport(self.key, FORMAT_ID, "unavailable")
paths = _agent_db_paths(root)
if not paths:
agents = os.path.join(root, "agents")
if _regular_dir(agents, root):
return CapabilityReport(self.key, FORMAT_ID, "unsupported", root=root)
return CapabilityReport(self.key, FORMAT_ID, "unavailable", root=root)
supported = False
unsupported = False
for agent_id, database in paths:
try:
with _open_connection(database, root) as connection:
_require_schema(connection, expected_agent=agent_id)
supported = True
except DiagnosticError as error:
if error.code in {"E_UNSAFE_PATH", "E_SOURCE_BUSY", "E_SQLITE_LIVE_WAL"}:
return CapabilityReport(self.key, FORMAT_ID, "unsafe", root=root)
if error.code == "E_UNSUPPORTED_FORMAT":
unsupported = True
continue
if error.code == "E_CORRUPT_RECORD":
unsupported = True
continue
raise
if supported:
state = "partial" if unsupported else "supported"
return CapabilityReport(
self.key, FORMAT_ID, state, root=root, evidence=(FORMAT_ID,)
)
return CapabilityReport(self.key, FORMAT_ID, "unsupported", root=root)
except DiagnosticError as error:
state = "unsafe" if error.code in {"E_UNSAFE_PATH", "E_SOURCE_BUSY", "E_SQLITE_LIVE_WAL"} else "unsupported"
return CapabilityReport(self.key, FORMAT_ID, state)
def list(self, query: Query, budget: ReadBudget) -> list[SessionSummary]:
root = _existing_root(query)
if root is None:
raise DiagnosticError("E_CAPABILITY_UNAVAILABLE", source=self.key, provider=FORMAT_ID)
agent_filter, session_filter = _parse_ref(query.ref)
exact = bool(session_filter)
include_internal = exact
# Exact refs stay selectable outside default age window.
list_query = query
if exact and query.within_min is None:
list_query = Query(
source=query.source,
ref=query.ref,
cwd=query.cwd,
within_min=0,
source_root=query.source_root,
max_tool_chars=query.max_tool_chars,
)
values: list[SessionSummary] = []
any_supported = False
paths = _agent_db_paths(root, budget)
list_limit = min(budget.limits.listed_sessions, DEFAULT_BOUNDS.listed_sessions)
for agent_id, database in paths:
if agent_filter is not None and agent_id != agent_filter:
continue
try:
with _open_connection(database, root, budget) as connection:
_require_schema(connection, expected_agent=agent_id)
if session_filter is not None:
# Exact refs bypass listed_sessions prefix (Codex P1).
items = _exact_session_summaries(
connection,
agent_id=agent_id,
database=database,
session_filter=session_filter,
query=list_query,
budget=budget,
)
else:
items = _list_nodes(
connection,
agent_id=agent_id,
database=database,
query=list_query,
include_internal=include_internal,
budget=budget,
)
any_supported = True
except DiagnosticError as error:
if error.code == "E_UNSUPPORTED_FORMAT":
continue
raise
# Bound per agent, but never stop scanning other agents before the
# global timestamp sort (Codex P1 multi-agent latest).
if session_filter is not None:
values.extend(items)
else:
values.extend(items[:list_limit])
if not values and not any_supported and paths:
raise DiagnosticError("E_UNSUPPORTED_FORMAT", source=self.key, provider=FORMAT_ID)
# Newest first; ascending session_id tie-break (stable sort).
values.sort(key=lambda item: item.session_id)
values.sort(
key=lambda item: item.updated_at or "",
reverse=True,
)
# Missing timestamps last.
values.sort(key=lambda item: item.updated_at is None)
if session_filter is not None:
return values
return values[:list_limit]
def show(self, ref: ResolvedRef, query: Query, budget: ReadBudget) -> Session:
root = _existing_root(query)
if root is None:
raise DiagnosticError("E_CAPABILITY_UNAVAILABLE", source=self.key, provider=FORMAT_ID)
agent_id, session_id = _parse_ref(ref.session_id)
if agent_id is None or session_id is None:
# Composite missing — try treating whole id as native session id.
agent_id, session_id = None, ref.session_id
database = ref.source_path
if database is None:
matches: list[Session] = []
for candidate_agent, candidate_database in _agent_db_paths(root, budget):
if agent_id is not None and candidate_agent != agent_id:
continue
try:
with _open_connection(candidate_database, root, budget) as connection:
_require_schema(connection, expected_agent=candidate_agent)
matches.append(
_show_session(
connection,
agent_id=candidate_agent,
session_id=session_id,
database=candidate_database,
query=query,
budget=budget,
)
)
except DiagnosticError as error:
if error.code in {"E_NO_MATCH", "E_UNSUPPORTED_FORMAT"}:
continue
raise
if agent_id is not None:
return matches[0]
if len(matches) != 1:
raise DiagnosticError("E_NO_MATCH", source=self.key, provider=FORMAT_ID)
return matches[0]
if database is None or not _regular_db_file(database, root):
raise DiagnosticError.unsafe_path()
# Resolve agent from path when composite was incomplete.
path_agent = None
parts = os.path.normpath(database).split(os.sep)
if len(parts) >= 4 and parts[-1] == DB_BASENAME and parts[-2] == "agent":
path_agent = parts[-3]
if agent_id is None:
agent_id = path_agent
if agent_id is None or session_id is None:
raise DiagnosticError("E_NO_MATCH", source=self.key, provider=FORMAT_ID)
with _open_connection(database, root, budget) as connection:
meta_agent = _require_schema(connection, expected_agent=path_agent or agent_id)
if meta_agent != agent_id and path_agent is not None:
agent_id = meta_agent
return _show_session(
connection,
agent_id=agent_id,
session_id=session_id,
database=database,
query=query,
budget=budget,
)
ADAPTER = OpenClawAdapter()
SHA-256: b8802a16dea883a344546957f92a66b7ec3943ec72728b5e70a445a8f39f1c93