| 1 | from __future__ import annotations |
| 2 | |
| 3 | import re |
| 4 | from typing import Iterable |
| 5 | |
| 6 | |
| 7 | _DURABLE_SUBJECT_RE = re.compile( |
| 8 | r"\b(user|project|repo|repository|workspace|runtime|service|server|" |
| 9 | r"plugin|agent zero|a0|profile|team|organization|company)\b", |
| 10 | re.IGNORECASE, |
| 11 | ) |
| 12 | _DURABLE_RELATION_RE = re.compile( |
| 13 | r"\b(prefers?|requires?|uses?|runs?|lives?|located|configured|default|" |
| 14 | r"current(?:ly)?|must|should|always|never|constraint|path|endpoint|port)\b", |
| 15 | re.IGNORECASE, |
| 16 | ) |
| 17 | _TRANSIENT_RE = re.compile( |
| 18 | r"(/tmp\b|\btmp/|\btemporary\b|\btemp\b|\bworkdir\b|" |
| 19 | r"\bcontainer-local\b|\bmachine-local\b|\bpersonal absolute path\b|" |
| 20 | r"\blocal endpoint\b|\blocal port\b|" |
| 21 | r"\btest marker\b|\bmemory test\b|\bcleanup token\b|\bthrowaway\b|" |
| 22 | r"\bsample file\b|\bdemo file\b|\bvalidation directory\b|\blive-test\b|" |
| 23 | r"\bone-off\b|\bsession-only\b)", |
| 24 | re.IGNORECASE, |
| 25 | ) |
| 26 | _LOCAL_COORDINATE_RE = re.compile( |
| 27 | r"\b(localhost|127\.0\.0\.1|0\.0\.0\.0|::1)\b|" |
| 28 | r"https?://(?:localhost|127\.0\.0\.1|0\.0\.0\.0|\[::1\])(?::\d+)?|" |
| 29 | r"/(?:home|Users)/[^/\s]+/|" |
| 30 | r"\b[A-Za-z]:\\Users\\", |
| 31 | re.IGNORECASE, |
| 32 | ) |
| 33 | _AGENT_ACTION_RE = re.compile( |
| 34 | r"\b(agent|assistant|codex|a0)\b.{0,80}\b(" |
| 35 | r"created|implemented|built|ran|tested|verified|fixed|wrote|generated|" |
| 36 | r"saved|cleaned|inspected|synced|restarted|committed|staged|opened|" |
| 37 | r"reported|showed)\b", |
| 38 | re.IGNORECASE, |
| 39 | ) |
| 40 | _USER_REQUEST_RE = re.compile( |
| 41 | r"\b(user|the user)\s+(asked|requested|wanted|told|prompted|instructed)\b", |
| 42 | re.IGNORECASE, |
| 43 | ) |
| 44 | _COMMAND_HISTORY_RE = re.compile( |
| 45 | r"\b(ran|executed)\s+[`'\"]?[\w./-]+|\bterminal command\b|" |
| 46 | r"\bcommand output\b|\bexit code\b|\bstdout\b|\bstderr\b", |
| 47 | re.IGNORECASE, |
| 48 | ) |
| 49 | _LOW_VALUE_RE = re.compile( |
| 50 | r"^(hello|hi|greeting|conversation start|task completed|file created|" |
| 51 | r"implementation completed|test passed|memory search)$", |
| 52 | re.IGNORECASE, |
| 53 | ) |
| 54 | |
| 55 | |
| 56 | def normalize_memory_candidate(value: object) -> str: |
| 57 | return str(value).strip() |
| 58 | |
| 59 | |
| 60 | def is_auto_fragment_worth_saving(text: str) -> bool: |
| 61 | """Return True only for durable facts suitable for automatic fragments.""" |
| 62 | |
| 63 | text = normalize_memory_candidate(text) |
| 64 | has_durable_relation = bool(_DURABLE_RELATION_RE.search(text)) |
| 65 | if len(text) < 24: |
| 66 | return False |
| 67 | if _LOW_VALUE_RE.match(text): |
| 68 | return False |
| 69 | if _TRANSIENT_RE.search(text): |
| 70 | return False |
| 71 | if _LOCAL_COORDINATE_RE.search(text): |
| 72 | return False |
| 73 | if _COMMAND_HISTORY_RE.search(text): |
| 74 | return False |
| 75 | if _AGENT_ACTION_RE.search(text) and not has_durable_relation: |
| 76 | return False |
| 77 | if _USER_REQUEST_RE.search(text) and not has_durable_relation: |
| 78 | return False |
| 79 | |
| 80 | return bool(_DURABLE_SUBJECT_RE.search(text) and has_durable_relation) |
| 81 | |
| 82 | |
| 83 | def filter_auto_memory_fragments(memories: Iterable[object]) -> list[str]: |
| 84 | return [ |
| 85 | text |
| 86 | for text in (normalize_memory_candidate(memory) for memory in memories) |
| 87 | if is_auto_fragment_worth_saving(text) |
| 88 | ] |