消灭更多bug，修改K线颜色，macd颜色

datasvc/app/storage.py

@@ -1,10 +1,15 @@
+import logging
 import os
+import shutil
 import threading
+from datetime import datetime
 from typing import List, Optional
 
 import pandas as pd
 import pyarrow.dataset as ds
 
+logger = logging.getLogger("datasvc")
+
 
 _lock = threading.Lock()
 
@@ -24,7 +29,20 @@ def read_candles(base_dir: str, symbol: str, timeframe: str, start: Optional[int
     p = _path(base_dir, symbol, timeframe)
     if not os.path.exists(p):
         return pd.DataFrame(columns=["timestamp", "open", "high", "low", "close", "volume"])  # empty
-    df = pd.read_parquet(p)
+    try:
+        df = pd.read_parquet(p)
+    except Exception as exc:
+        with _lock:
+            backup = _backup_corrupted_file(p)
+        extra = f"，已备份至 {backup}" if backup else ""
+        logger.warning(
+            "读取缓存失败，将视为空数据 [%s %s]%s：%s",
+            symbol,
+            timeframe,
+            extra,
+            exc,
+        )
+        return pd.DataFrame(columns=["timestamp", "open", "high", "low", "close", "volume"])
     if start is not None:
         df = df[df["timestamp"] >= int(start)]
     if end is not None:
@@ -38,19 +56,53 @@ def upsert_candles(base_dir: str, symbol: str, timeframe: str, candles: List[Lis
     new_df = pd.DataFrame(candles, columns=["timestamp", "open", "high", "low", "close", "volume"])
     with _lock:
         if os.path.exists(p):
-            old = pd.read_parquet(p)
+            try:
+                old = pd.read_parquet(p)
+            except Exception as exc:
+                backup = _backup_corrupted_file(p)
+                extra = f"，已备份至 {backup}" if backup else ""
+                logger.warning(
+                    "读取缓存失败，准备重建文件 [%s %s]%s：%s",
+                    symbol,
+                    timeframe,
+                    extra,
+                    exc,
+                )
+                old = pd.DataFrame(columns=["timestamp", "open", "high", "low", "close", "volume"])
             merged = pd.concat([old, new_df], ignore_index=True)
             merged = merged.drop_duplicates(subset=["timestamp"], keep="last").sort_values("timestamp")
         else:
             merged = new_df.sort_values("timestamp")
-        merged.to_parquet(p, index=False)
+        temp_path = f"{p}.tmp"
+        try:
+            merged.to_parquet(temp_path, index=False)
+            os.replace(temp_path, p)
+        finally:
+            if os.path.exists(temp_path):
+                try:
+                    os.remove(temp_path)
+                except OSError:
+                    pass
 
 
 def get_last_timestamp(base_dir: str, symbol: str, timeframe: str) -> Optional[int]:
     p = _path(base_dir, symbol, timeframe)
     if not os.path.exists(p):
         return None
-    df = pd.read_parquet(p)
+    try:
+        df = pd.read_parquet(p)
+    except Exception as exc:
+        with _lock:
+            backup = _backup_corrupted_file(p)
+        extra = f"，已备份至 {backup}" if backup else ""
+        logger.warning(
+            "获取最后时间戳失败 [%s %s]%s：%s",
+            symbol,
+            timeframe,
+            extra,
+            exc,
+        )
+        return None
     if df.empty:
         return None
     return int(df["timestamp"].iloc[-1])
@@ -60,10 +112,36 @@ def read_candle_exact(base_dir: str, symbol: str, timeframe: str, timestamp: int
     p = _path(base_dir, symbol, timeframe)
     if not os.path.exists(p):
         return pd.DataFrame(columns=["timestamp", "open", "high", "low", "close", "volume"])
-    dataset = ds.dataset(p, format="parquet")
-    table = dataset.to_table(filter=ds.field("timestamp") == int(timestamp))
+    try:
+        dataset = ds.dataset(p, format="parquet")
+        table = dataset.to_table(filter=ds.field("timestamp") == int(timestamp))
+    except Exception as exc:
+        with _lock:
+            backup = _backup_corrupted_file(p)
+        extra = f"，已备份至 {backup}" if backup else ""
+        logger.warning(
+            "读取指定时间 K 线失败 [%s %s]%s：%s",
+            symbol,
+            timeframe,
+            extra,
+            exc,
+        )
+        return pd.DataFrame(columns=["timestamp", "open", "high", "low", "close", "volume"])
     if table.num_rows == 0:
         return pd.DataFrame(columns=["timestamp", "open", "high", "low", "close", "volume"])
     return table.to_pandas()
 
 
+def _backup_corrupted_file(path: str) -> Optional[str]:
+    try:
+        if not os.path.exists(path):
+            return None
+        timestamp = datetime.utcnow().strftime("%Y%m%d%H%M%S")
+        backup_path = f"{path}.corrupted.{timestamp}"
+        shutil.move(path, backup_path)
+        return backup_path
+    except Exception as exc:
+        logger.warning("备份损坏文件失败 (%s)：%s", path, exc)
+        return None
+
+