fix(优先组): #18 密钥明文直配 config(弃 .env 依赖)；#5 appearances 改 distinct 视频数覆盖校准(防 reconcile 累加膨胀)；#9 oracle_sync 拉取加锁防 trigger_now 与后台并发双拉；#11 LLM 合并命名解析 target 已有 canonical(统一显示名)

fam-core/src/fam_core/oracle_sync.py

@@ -49,10 +49,16 @@ class OracleSync:
         self._last_sync_at = None
         self._last_error = None
         self._last_count = None
+        # 拉取互斥锁：trigger_now（命名后即时拉回）与后台 _run 并发时只允许一个执行
+        self._pull_lock = threading.Lock()
 
     # ------------------------------------------------------------------
     def _pull_once(self) -> bool:
-        """执行一次增量拉取。返回是否成功。"""
+        """执行一次增量拉取（加锁防并发双拉）。返回是否成功。"""
+        with self._pull_lock:
+            return self._pull_once_locked()
+
+    def _pull_once_locked(self) -> bool:
         since = db_layer.get_sync_cursor() or ''
         params = {'since': since, 'token': self.token}
         try:

fam-edge/config/config.yaml

@@ -26,9 +26,9 @@ gdrive_sync:
 oracle_db:
   path: "/opt/fam-edge/data/oracle.db"
 
-# NAS 拉取同步接口鉴权 token（与 NAS oracle_sync.token 一致）
+# NAS 拉取同步接口鉴权 token（与 NAS oracle_sync.token 一致；明文直配，不再依赖 .env）
 sync_api:
-  token: "${ORACLE_SYNC_TOKEN}"
+  token: "MLH92wv5jSDdQtHfcWJgKt-YaStn3IttjlrYxW0DwXA"
 
 # 人物识别服务
 person_service:
@@ -56,7 +56,7 @@ models:
     model_name: "gemini-flash-latest"   # 主模型（每日免费配额 20 请求，按模型独立）
     fallback_models:
       - "gemini-flash-lite-latest"
-    api_key: "${GEMINI_API_KEY}"
+    api_key: "AQ.Ab8RN6I0l8hC7hLnNHRY6qOXdch5CTWczDNlS4c1XrneGHipUQ"
     timeout: 600
     # 模型级独立超时（最终值，不参与编排层 ×2 放大）
     # gemini-flash-lite 实测 ~22-34s，按用户要求放宽至 8 分钟（480s），避免大视频/排队时过早切断
@@ -79,7 +79,7 @@ models:
       - "nvidia/nemotron-nano-12b-v2-vl"
       - "meta/llama-3.2-11b-vision-instruct"
     base_url: "https://integrate.api.nvidia.com/v1"
-    api_key: "${NVIDIA_API_KEY}"
+    api_key: "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"
     timeout: 600
     switch_interval_sec: 5      # 模型切换间隔：一个失败后等待再试下一个
     model_timeouts:             # 模型级独立超时（最终值，不参与 ×2）

fam-edge/src/fam_edge/oracle_db.py

@@ -243,6 +243,26 @@ class OracleDB:
         """手动命名：设置规范名（label 可视为别名）。"""
         self.upsert_person(label, canonical_name, source='manual')
 
+    def set_person_appearances(self, label: str, count: int, source: str = 'llm'):
+        """覆盖设置出现次数（reconcile 时用 distinct 视频数校准，避免累加膨胀）。"""
+        now = _now_iso()
+        row = self._conn.execute("SELECT * FROM people WHERE label=?", (label,)).fetchone()
+        if row:
+            if source == 'manual' or row['source'] != 'manual':
+                self._conn.execute(
+                    "UPDATE people SET appearances=?, source=?, updated_at=? WHERE label=?",
+                    (int(count), source, now, label))
+            else:
+                self._conn.execute(
+                    "UPDATE people SET appearances=?, updated_at=? WHERE label=?",
+                    (int(count), now, label))
+        else:
+            self._conn.execute(
+                "INSERT INTO people (label, canonical_name, first_seen, appearances, "
+                "source, updated_at) VALUES (?,?,?,?,?,?)",
+                (label, '', now, int(count), source, now))
+        self._conn.commit()
+
     def get_people(self) -> List[sqlite3.Row]:
         return self._conn.execute("SELECT * FROM people ORDER BY id ASC").fetchall()
 

fam-edge/src/fam_edge/person_service.py

@@ -43,7 +43,8 @@ class PersonService:
     # ------------------------------------------------------------------
     def reconcile(self):
         """汇总 + LLM 合并一次。可由定时或手动触发。"""
-        # 1. 先把所有视频的 people_mentioned 同步进 people 表（标签级）
+        # 1. 统计每个人物标签出现过的视频数（去重），校准 appearances（防每次 reconcile 累加膨胀）
+        label_videos: Dict[str, set] = {}
         for v in self.db.get_all_videos():
             try:
                 people = json.loads(v['people_json'] or '[]')
@@ -51,7 +52,9 @@ class PersonService:
                 people = []
             for p in people:
                 if p and p not in ('无人', '无'):
-                    self.db.upsert_person(p, source='llm')
+                    label_videos.setdefault(p, set()).add(v['id'])
+        for label, vids in label_videos.items():
+            self.db.set_person_appearances(label, len(vids), source='llm')
 
         # 2. 收集未命名（无 canonical 或 canonical==label）的标签 + 描述样本
         rows = self.db.get_people()
@@ -66,12 +69,17 @@ class PersonService:
         if not mapping:
             return
 
+        # 3. 落库：canonical 若是另一个 label（target），解析为其已有 canonical，保证同一身份统一显示名
+        label_to_canonical = {r['label']: (r['canonical_name'] or r['label']) for r in rows}
+        updated = 0
         for label, canonical in mapping.items():
             if label in manual:
                 continue  # 手动命名优先
             if canonical and canonical != label:
-                self.db.set_canonical(label, canonical, source='llm')
+                resolved = label_to_canonical.get(canonical, canonical)
-        logger.info(f"PersonService: LLM 合并完成，更新 {len(mapping)} 条")
+                self.db.set_canonical(label, resolved, source='llm')
+                updated += 1
+        logger.info(f"PersonService: LLM 合并完成，更新 {updated} 条")
 
     def _collect_descriptions(self, labels: List[str]) -> Dict[str, List[str]]:
         """从 events 表收集每个标签出现时的描述样本。"""