|
| 1 | +from __future__ import annotations |
| 2 | + |
| 3 | +from dataclasses import dataclass |
| 4 | +from typing import Any |
| 5 | + |
| 6 | +from memory_store import MemoryStore |
| 7 | + |
| 8 | + |
| 9 | +class StopRun(Exception): |
| 10 | + def __init__(self, reason: str): |
| 11 | + super().__init__(reason) |
| 12 | + self.reason = reason |
| 13 | + |
| 14 | + |
| 15 | +@dataclass(frozen=True) |
| 16 | +class Budget: |
| 17 | + max_capture_items: int = 6 |
| 18 | + max_retrieve_top_k: int = 6 |
| 19 | + max_query_chars: int = 240 |
| 20 | + max_answer_chars: int = 700 |
| 21 | + max_value_chars: int = 120 |
| 22 | + max_seconds: int = 25 |
| 23 | + |
| 24 | + |
| 25 | +def _is_number(value: Any) -> bool: |
| 26 | + return isinstance(value, (int, float)) and not isinstance(value, bool) |
| 27 | + |
| 28 | + |
| 29 | +def validate_memory_candidates( |
| 30 | + raw: Any, |
| 31 | + *, |
| 32 | + allowed_keys_policy: set[str], |
| 33 | + allowed_scopes_policy: set[str], |
| 34 | + max_items: int, |
| 35 | + max_value_chars: int, |
| 36 | +) -> dict[str, Any]: |
| 37 | + if not isinstance(raw, dict): |
| 38 | + raise StopRun("invalid_memory_candidates:not_object") |
| 39 | + |
| 40 | + items = raw.get("items") |
| 41 | + if not isinstance(items, list): |
| 42 | + raise StopRun("invalid_memory_candidates:items") |
| 43 | + |
| 44 | + normalized: list[dict[str, Any]] = [] |
| 45 | + for item in items: |
| 46 | + if not isinstance(item, dict): |
| 47 | + raise StopRun("invalid_memory_candidates:item") |
| 48 | + |
| 49 | + required_keys = {"key", "value"} |
| 50 | + if not required_keys.issubset(item.keys()): |
| 51 | + raise StopRun("invalid_memory_candidates:missing_keys") |
| 52 | + |
| 53 | + key = item.get("key") |
| 54 | + value = item.get("value") |
| 55 | + scope = item.get("scope", "user") |
| 56 | + ttl_days = item.get("ttl_days", 180) |
| 57 | + confidence = item.get("confidence", 0.8) |
| 58 | + |
| 59 | + if not isinstance(key, str) or not key.strip(): |
| 60 | + raise StopRun("invalid_memory_candidates:key") |
| 61 | + key = key.strip() |
| 62 | + if key not in allowed_keys_policy: |
| 63 | + raise StopRun(f"memory_key_not_allowed_policy:{key}") |
| 64 | + |
| 65 | + if not isinstance(value, str) or not value.strip(): |
| 66 | + raise StopRun("invalid_memory_candidates:value") |
| 67 | + value = value.strip() |
| 68 | + if len(value) > max_value_chars: |
| 69 | + raise StopRun("invalid_memory_candidates:value_too_long") |
| 70 | + |
| 71 | + if not isinstance(scope, str) or not scope.strip(): |
| 72 | + raise StopRun("invalid_memory_candidates:scope") |
| 73 | + scope = scope.strip() |
| 74 | + if scope not in allowed_scopes_policy: |
| 75 | + raise StopRun(f"memory_scope_not_allowed_policy:{scope}") |
| 76 | + |
| 77 | + if not _is_number(ttl_days): |
| 78 | + raise StopRun("invalid_memory_candidates:ttl_days") |
| 79 | + ttl_days = int(float(ttl_days)) |
| 80 | + ttl_days = max(1, min(365, ttl_days)) |
| 81 | + |
| 82 | + if not _is_number(confidence): |
| 83 | + raise StopRun("invalid_memory_candidates:confidence") |
| 84 | + confidence = float(confidence) |
| 85 | + confidence = max(0.0, min(1.0, confidence)) |
| 86 | + |
| 87 | + normalized.append( |
| 88 | + { |
| 89 | + "key": key, |
| 90 | + "value": value, |
| 91 | + "scope": scope, |
| 92 | + "ttl_days": ttl_days, |
| 93 | + "confidence": round(confidence, 3), |
| 94 | + } |
| 95 | + ) |
| 96 | + |
| 97 | + if len(normalized) > max_items: |
| 98 | + raise StopRun("invalid_memory_candidates:too_many_items") |
| 99 | + |
| 100 | + return {"items": normalized} |
| 101 | + |
| 102 | + |
| 103 | +def validate_retrieval_intent( |
| 104 | + raw: Any, |
| 105 | + *, |
| 106 | + allowed_scopes_policy: set[str], |
| 107 | + max_top_k: int, |
| 108 | +) -> dict[str, Any]: |
| 109 | + if not isinstance(raw, dict): |
| 110 | + raise StopRun("invalid_retrieval_intent:not_object") |
| 111 | + |
| 112 | + if raw.get("kind") != "retrieve_memory": |
| 113 | + raise StopRun("invalid_retrieval_intent:kind") |
| 114 | + |
| 115 | + query = raw.get("query") |
| 116 | + if not isinstance(query, str) or not query.strip(): |
| 117 | + raise StopRun("invalid_retrieval_intent:query") |
| 118 | + |
| 119 | + top_k = raw.get("top_k", 4) |
| 120 | + if not isinstance(top_k, int) or not (1 <= top_k <= max_top_k): |
| 121 | + raise StopRun("invalid_retrieval_intent:top_k") |
| 122 | + |
| 123 | + scopes_raw = raw.get("scopes") |
| 124 | + normalized_scopes: list[str] = [] |
| 125 | + if scopes_raw is not None: |
| 126 | + if not isinstance(scopes_raw, list) or not scopes_raw: |
| 127 | + raise StopRun("invalid_retrieval_intent:scopes") |
| 128 | + for scope in scopes_raw: |
| 129 | + if not isinstance(scope, str) or not scope.strip(): |
| 130 | + raise StopRun("invalid_retrieval_intent:scope_item") |
| 131 | + normalized_scope = scope.strip() |
| 132 | + if normalized_scope not in allowed_scopes_policy: |
| 133 | + raise StopRun(f"invalid_retrieval_intent:scope_not_allowed:{normalized_scope}") |
| 134 | + normalized_scopes.append(normalized_scope) |
| 135 | + |
| 136 | + payload = { |
| 137 | + "kind": "retrieve_memory", |
| 138 | + "query": query.strip(), |
| 139 | + "top_k": top_k, |
| 140 | + } |
| 141 | + if normalized_scopes: |
| 142 | + payload["scopes"] = normalized_scopes |
| 143 | + return payload |
| 144 | + |
| 145 | + |
| 146 | +class MemoryGateway: |
| 147 | + def __init__( |
| 148 | + self, |
| 149 | + *, |
| 150 | + store: MemoryStore, |
| 151 | + budget: Budget, |
| 152 | + allow_execution_keys: set[str], |
| 153 | + allow_execution_scopes: set[str], |
| 154 | + ): |
| 155 | + self.store = store |
| 156 | + self.budget = budget |
| 157 | + self.allow_execution_keys = set(allow_execution_keys) |
| 158 | + self.allow_execution_scopes = set(allow_execution_scopes) |
| 159 | + |
| 160 | + def write( |
| 161 | + self, |
| 162 | + *, |
| 163 | + user_id: int, |
| 164 | + items: list[dict[str, Any]], |
| 165 | + source: str, |
| 166 | + ) -> dict[str, Any]: |
| 167 | + if len(items) > self.budget.max_capture_items: |
| 168 | + raise StopRun("max_capture_items") |
| 169 | + |
| 170 | + writable: list[dict[str, Any]] = [] |
| 171 | + blocked: list[dict[str, Any]] = [] |
| 172 | + |
| 173 | + for item in items: |
| 174 | + key = item["key"] |
| 175 | + scope = item["scope"] |
| 176 | + |
| 177 | + if key not in self.allow_execution_keys: |
| 178 | + blocked.append({"key": key, "reason": "key_denied_execution"}) |
| 179 | + continue |
| 180 | + if scope not in self.allow_execution_scopes: |
| 181 | + blocked.append( |
| 182 | + { |
| 183 | + "key": key, |
| 184 | + "scope": scope, |
| 185 | + "reason": "scope_denied_execution", |
| 186 | + } |
| 187 | + ) |
| 188 | + continue |
| 189 | + |
| 190 | + writable.append(item) |
| 191 | + |
| 192 | + written = [] |
| 193 | + if writable: |
| 194 | + written = self.store.upsert_items(user_id=user_id, items=writable, source=source) |
| 195 | + |
| 196 | + return { |
| 197 | + "written": written, |
| 198 | + "blocked": blocked, |
| 199 | + } |
| 200 | + |
| 201 | + def retrieve( |
| 202 | + self, |
| 203 | + *, |
| 204 | + user_id: int, |
| 205 | + intent: dict[str, Any], |
| 206 | + include_preference_keys: bool = False, |
| 207 | + ) -> dict[str, Any]: |
| 208 | + query = intent["query"] |
| 209 | + if len(query) > self.budget.max_query_chars: |
| 210 | + raise StopRun("invalid_retrieval_intent:query_too_long") |
| 211 | + |
| 212 | + requested_scopes = set(intent.get("scopes") or self.allow_execution_scopes) |
| 213 | + denied = sorted(requested_scopes - self.allow_execution_scopes) |
| 214 | + if denied: |
| 215 | + raise StopRun(f"scope_denied:{denied[0]}") |
| 216 | + |
| 217 | + items = self.store.search( |
| 218 | + user_id=user_id, |
| 219 | + query=query, |
| 220 | + top_k=intent["top_k"], |
| 221 | + scopes=requested_scopes, |
| 222 | + include_preference_keys=include_preference_keys, |
| 223 | + ) |
| 224 | + |
| 225 | + return { |
| 226 | + "query": query, |
| 227 | + "requested_scopes": sorted(requested_scopes), |
| 228 | + "include_preference_keys": include_preference_keys, |
| 229 | + "items": items, |
| 230 | + } |
0 commit comments