feat: stock_name_hygiene 周频任务——akshare官方名剥XD标记比对,stocks/candidates/holdings截断名与ST变更自愈

2026-08-22 12:22:09 +08:00
parent 0afff445b6
commit 761706ca8f
1 changed files with 65 additions and 0 deletions
@@ -0,0 +1,65 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""stock_name_hygiene.py — 股票名称卫生（每周）
+2026-08-22 名称治理：XD/XR/DR 除权息日交易所简称被截断（紫金矿业→XD紫金矿），
+各行情源当天均无干净名。本任务每周拉 akshare 官方全量名称，剥 XD 标记后比对，
+凡 DB 名是官方名的前缀（截断）即以官方名为准修复，自愈截断/改名/ST 变更。
+"""
+import os, sys, re
+from datetime import datetime
+
+# ── 消息通道统一路由(broadcast/xmpp by delivery) ──
+try:
+    from messenger import install_stdio_hook as _msh
+    _msh()
+except Exception:
+    pass
+
+DB_PATH = os.environ.get("MOFIN_DB", "/home/hmo/MoFin/data/mofin.db")
+
+
+def strip_marker(n: str) -> str:
+    m = re.match(r"^(XD|XR|DR)(.+)$", n)
+    return m.group(2) if m else n
+
+
+def main():
+    import sqlite3
+    import akshare as ak
+
+    print(f"[stock_name_hygiene] {datetime.now():%Y-%m-%d %H:%M:%S} 开始", flush=True)
+    df = ak.stock_info_a_code_name()
+    name_map = {str(c): strip_marker(str(n)) for c, n in zip(df["code"], df["name"])}
+    print(f"  官方名称源: {len(name_map)} 条", flush=True)
+
+    conn = sqlite3.connect(DB_PATH, timeout=30)
+    conn.row_factory = sqlite3.Row
+    total = 0
+    samples = []
+    for table in ["stocks", "candidates", "holdings"]:
+        try:
+            rows = conn.execute(f"SELECT code, name FROM {table}").fetchall()
+        except Exception:
+            continue
+        fixed = 0
+        for r in rows:
+            code, cur = str(r["code"]), str(r["name"])
+            full = name_map.get(code)
+            if not full or full == cur:
+                continue
+            # 只修明确劣化：DB名是官方名前缀（XD截断）、或DB名=代码、或仅ST标记差异
+            if full.startswith(cur) or cur == code or cur.lstrip("*ST") == full.lstrip("*ST"):
+                conn.execute(f"UPDATE {table} SET name=? WHERE code=? AND name=?", (full, code, cur))
+                fixed += 1
+                if len(samples) < 8:
+                    samples.append(f"{code} {cur}→{full}")
+        if fixed:
+            print(f"  {table}: 修复 {fixed} 行", flush=True)
+        total += fixed
+    conn.commit()
+    conn.close()
+    print(f"[stock_name_hygiene] 完成，共修复 {total} 行" + (f"（样本: {'; '.join(samples)}）" if samples else ""), flush=True)
+
+
+if __name__ == "__main__":
+    main()