From a3ba321e5e630bda5ed8a047c5e02db23d5cf5f1 Mon Sep 17 00:00:00 2001
From: mokurin000 <1348292515a@gmail.com>
Date: Mon, 4 Aug 2025 01:55:44 +0800
Subject: [PATCH] refactor: drop players.json support

---
 .gitignore                                 |  2 +-
 utils/{export_players.py => export_b50.py} | 28 +++-------------------
 utils/hasher.py                            |  8 +++++++
 3 files changed, 12 insertions(+), 26 deletions(-)
 rename utils/{export_players.py => export_b50.py} (87%)
 create mode 100644 utils/hasher.py

diff --git a/.gitignore b/.gitignore
index f943528..92f9633 100644
--- a/.gitignore
+++ b/.gitignore
@@ -6,7 +6,7 @@
 
 /players*.parquet
 /b50*.json*
-/region*.json*
+/region*.parquet
 
 /.python-version
 /uv.lock
diff --git a/utils/export_players.py b/utils/export_b50.py
similarity index 87%
rename from utils/export_players.py
rename to utils/export_b50.py
index c0ee5fc..2592dd8 100644
--- a/utils/export_players.py
+++ b/utils/export_b50.py
@@ -1,10 +1,10 @@
 import orjson as json
-import hashlib
 from typing import Callable
 from datetime import datetime
 from decimal import Decimal, getcontext
 
-SALT = b"Lt2N5xgjJOqRsT5qVt7wWYw6SqOPZDI7"
+from hasher import salted_hash_userid
+
 
 with open("musicDB.json", "r", encoding="utf-8") as f:
     music_db = json.loads(f.read())
@@ -80,12 +80,6 @@ def dx_rating(difficulty: Decimal, achievement: int) -> int:
     return int(result)
 
 
-def salted_hash_userid(player: dict):
-    uid = player["userId"]
-    hash_uid = hashlib.sha256(f"{uid}".encode("utf-8") + SALT)
-    player["userId"] = hash_uid.hexdigest()[:16]
-
-
 def clean_b50(b50: dict[str, str | dict]):
     urating: dict[str, list[dict[str, int]]] = b50["userRating"]
 
@@ -141,17 +135,6 @@ def clean_b50(b50: dict[str, str | dict]):
     )
 
 
-def clean_player(player: dict):
-    player.pop("isLogin")
-    player.pop("lastLoginDate")
-    player.pop("lastPlayDate")
-    player.pop("isNetMember")
-    player.pop("dailyBonusDate")
-    player.pop("banState")
-    player.pop("nameplateId")
-    player.pop("trophyId")
-
-
 def record_time(*, _: list[datetime] = []):
     last_time = _
     if not last_time:
@@ -174,7 +157,7 @@ def process(
     print(f"loaded, cost {record_time():.2f}s")
 
     for entry in data:
-        salted_hash_userid(entry)
+        entry["userId"] = salted_hash_userid(entry["userId"])
         clean_fields(entry)
     print(f"processed, cost {record_time():.2f}s")
 
@@ -186,11 +169,6 @@ def process(
 
 
 def main():
-    # process(
-    #     clean_player,
-    #     "players.json",
-    #     "players_pub.json",
-    # )
     process(
         clean_b50,
         "b50.json",
diff --git a/utils/hasher.py b/utils/hasher.py
new file mode 100644
index 0000000..fb5b7a3
--- /dev/null
+++ b/utils/hasher.py
@@ -0,0 +1,8 @@
+import hashlib
+
+
+def salted_hash_userid(user_id: int):
+    SALT = b"Lt2N5xgjJOqRsT5qVt7wWYw6SqOPZDI7"
+
+    hash_uid = hashlib.sha256(f"{user_id}".encode("utf-8") + SALT)
+    return hash_uid.hexdigest()[:16]