feat: use jsonl to store race data

sub-task of #107

.vscode/cspell.json

@@ -4,6 +4,7 @@
         "astype",
         "cygames",
         "dmmgameplayer",
+        "jsonl",
         "LANCZOS",
         "nargs",
         "ndarray",

auto_derby/__main__.py

@@ -4,16 +4,18 @@
 
 
 import argparse
-from auto_derby import plugin
 import logging
 import logging.handlers
 import os
 import time
+import warnings
 import webbrowser
 
 import win32con
 import win32gui
 
+from auto_derby import plugin
+
 from . import clients, config, jobs, templates, version
 
 LOGGER = logging.getLogger(__name__)
@@ -113,6 +115,7 @@ def main():
             continue
         logging.getLogger(i).setLevel(logging.DEBUG)
 
+    warnings.filterwarnings("once", module="auto_derby(\\..*)?")
     try:
         main()
     except SystemExit:

auto_derby/_config.py

@@ -44,7 +44,7 @@ class config:
     CHECK_UPDATE = os.getenv("AUTO_DERBY_CHECK_UPDATE", "").lower() == "true"
 
     single_mode_race_data_path = os.getenv(
-        "AUTO_DERBY_SINGLE_MODE_RACE_DATA_PATH", "single_mode_races.json"
+        "AUTO_DERBY_SINGLE_MODE_RACE_DATA_PATH", "single_mode_races.jsonl"
     )
     ocr_data_path = os.getenv("AUTO_DERBY_OCR_LABEL_PATH", "ocr_labels.json")
     ocr_image_path = os.getenv("AUTO_DERBY_OCR_IMAGE_PATH", "")

auto_derby/single_mode/race.py

@@ -6,6 +6,7 @@
 import logging
 import math
 import os
+import warnings
 from typing import Any, Dict, Iterator, Set, Text, Tuple, Type
 
 import cast_unknown as cast
@@ -26,11 +27,28 @@ class g:
     race_class: Type[Race]
 
 
-def reload() -> None:
+def _iter_races():
+    with open(g.data_path, "r", encoding="utf-8") as f:
+        for line in f:
+            yield Race.new().from_dict(json.loads(line))
+
+
+def _load_legacy_json():
+    warnings.warn(
+        "json race data support will be removed at next major version, use jsonl instead",
+        DeprecationWarning,
+    )
     with open(g.data_path, "r", encoding="utf-8") as f:
         g.races = tuple(Race.new().from_dict(i) for i in json.load(f))
 
 
+def reload() -> None:
+    if g.data_path.endswith(".json"):
+        _load_legacy_json()
+        return
+    g.races = tuple(_iter_races())
+
+
 def _running_style_single_score(
     ctx: Context,
     race1: Race,
@@ -423,15 +441,15 @@ def __init__(self):
 
     def to_dict(self) -> Dict[Text, Any]:
         return {
-            "name": self.name,
             "stadium": self.stadium,
+            "name": self.name,
+            "grade": self.grade,
+            "ground": self.ground,
+            "distance": self.distance,
             "permission": self.permission,
             "month": self.month,
             "half": self.half,
-            "grade": self.grade,
             "entryCount": self.entry_count,
-            "distance": self.distance,
-            "ground": self.ground,
             "track": self.track,
             "turn": self.turn,
             "targetStatuses": self.target_statuses,

scripts/extract_single_mode_races.py

@@ -116,20 +116,21 @@ def _read_master_mdb(path: Text) -> Iterator[Race]:
 
 
 def main():
-
     parser = argparse.ArgumentParser()
     parser.add_argument(
         "path",
         nargs="?",
-        default=os.getenv("LocalAppData", "")
-        + "Low/cygames/umamusume/master/master.mdb",
+        default=os.path.expandvars(
+            "${LocalAppData}Low/cygames/umamusume/master/master.mdb"
+        ),
     )
     args = parser.parse_args()
     path: Text = args.path
 
-    data = [i.to_dict() for i in _read_master_mdb(path)]
     with pathlib.Path(g.data_path).open("w", encoding="utf-8") as f:
-        json.dump(data, f, ensure_ascii=False, indent=2)
+        for race in _read_master_mdb(path):
+            json.dump(race.to_dict(), f, ensure_ascii=False)
+            f.write("\n")
 
 
 if __name__ == "__main__":