's'

app/services/mp_recommend.py

@@ -1,11 +1,12 @@
 """
 公众号助手（com.tencent.mp）「收到XXX的推荐」通知入库。
 
-仅需 PocketBase 一个集合：mp_recommend_events（每条通知一行）。
+仅需 PocketBase 一个集合：mp_recommend_events（每个 mp_uid + recommender_key 一行，累计计数）。
 
 - username            Text   用户名（展示名，如从标题解析的 Eric）
-- cumulative_count    Number 该用户截至本条通知的累计触发次数（每次 webhook 写入时 = 旧总数 + 1）
-- recommender_key     Text   归一化键（casefold），与 username 对应，便于筛选，建议建索引
+ - cumulative_count    Number 该用户的累计触发次数（同一 mp_uid + recommender_key 每次 webhook 写入时 = 旧值 + 1）
+ - recommender_key     Text   归一化键（unicode 归一化 + 清洗 + casefold），与 username 对应，便于筛选，建议建索引
+ - mp_uid              Text   用户ID（用于把同名推荐人按不同账号隔离统计）
 
 字段（在 PocketBase 中新建，名称一致）：
 - recommender_key     Text
@@ -26,6 +27,7 @@ from __future__ import annotations
 
 import logging
 import re
+import unicodedata
 from typing import Any
 from urllib.parse import parse_qs
 
@@ -44,6 +46,26 @@ def _pb_escape(s: str) -> str:
     return (s or "").replace("\\", "\\\\").replace('"', '\\"')
 
 
+def _normalize_username(raw: str) -> str:
+    """
+    尽量稳定地把展示名归一化到可用于统计/过滤的形式。
+    - 兼容 emoji/颜文字/特殊字母：保留字符本身，不做 ASCII 限制
+    - 统一空白：压缩为单个空格
+    - Unicode 归一化：NFKC（把全角等兼容形态归一）
+    - 大小写：仅在 key 上 casefold
+    """
+    s = (raw or "")
+    s = unicodedata.normalize("NFKC", s)
+    s = s.replace("\u00A0", " ")
+    s = re.sub(r"\s+", " ", s).strip()
+    return s
+
+
+def _username_to_key(username: str) -> str:
+    u = _normalize_username(username)
+    return u.casefold() if u else "unknown"
+
+
 def parse_form_urlencoded_like(body: str) -> dict[str, str]:
     q = parse_qs(body, keep_blank_values=True, strict_parsing=False)
     return {
@@ -103,19 +125,35 @@ def persist_mp_recommend(
     timestamp_raw: str,
     client_ip: str,
 ) -> dict[str, Any]:
-    """写入一条事件；username、cumulative_count 写入本行（累计次数 = 同 key 已有条数 + 1）。"""
+    """按 mp_uid + recommender_key 聚合更新累计次数（存在则 +1，不存在则创建）。"""
     parsed = parse_mp_recommend_content(content)
     forward_ts = _parse_forward_ts_ms(timestamp_raw)
 
-    username = (parsed["recommender_name"] or "").strip() or "unknown"
-    key = username.casefold() if username != "unknown" else "unknown"
+    username = _normalize_username(parsed.get("recommender_name") or "") or "unknown"
+    key = _username_to_key(username if username != "unknown" else "")
+    mp_uid = (parsed.get("mp_uid") or "").strip()
 
     pb = get_pb_client()
     events = pb.collection(COL_EVENTS)
-    filt = f'recommender_key = "{_pb_escape(key)}"'
-    lst = events.get_list(1, 1, {"filter": filt})
-    total_before = int(getattr(lst, "total_items", 0) or 0)
-    cumulative_count = total_before + 1
+
+    filt_parts = [f'recommender_key = "{_pb_escape(key)}"']
+    if mp_uid:
+        filt_parts.append(f'mp_uid = "{_pb_escape(mp_uid)}"')
+    filt = " && ".join(filt_parts)
+
+    lst = events.get_list(1, 1, {"filter": filt, "sort": "-updated"})
+    items = list(getattr(lst, "items", []) or [])
+    existing = items[0] if items else None
+
+    old_count = 0
+    existing_id = None
+    if existing is not None:
+        existing_id = getattr(existing, "id", None)
+        try:
+            old_count = int(getattr(existing, "cumulative_count", 0) or 0)
+        except (TypeError, ValueError):
+            old_count = 0
+    cumulative_count = old_count + 1
 
     event_payload: dict[str, Any] = {
         "recommender_key": key,
@@ -124,7 +162,7 @@ def persist_mp_recommend(
         "package_name": parsed["package_name"],
         "title_line": parsed["title_line"],
         "summary_line": parsed["summary_line"],
-        "mp_uid": parsed["mp_uid"],
+        "mp_uid": mp_uid,
         "device_time_str": parsed["device_time_str"],
         "device_model": parsed["device_model"],
         "raw_content": parsed["raw_content"],
@@ -134,6 +172,9 @@ def persist_mp_recommend(
     if forward_ts is not None:
         event_payload["forward_timestamp_ms"] = forward_ts
 
+    if existing_id:
+        rec = events.update(existing_id, event_payload)
+    else:
         rec = events.create(event_payload)
     event_id = getattr(rec, "id", None) or str(rec)