define registry dtypes

cioos-siooc · Nov 16, 2023 · 63d79d7 · 63d79d7
1 parent f0daed7
commit 63d79d7
Showing 1 changed file with 11 additions and 2 deletions.
diff --git a/ocean_data_parser/batch/registry.py b/ocean_data_parser/batch/registry.py
@@ -16,6 +16,15 @@
 ).set_index("source")
 
 
+REGSITRY_DTYPE = {
+    "source": str,
+    "mtime": float,
+    "hash": str,
+    "error_message": str,
+    "output_path": str,
+}
+
+
 class FileConversionRegistry:
     def __init__(
         self,
@@ -42,9 +51,9 @@ def load(self, overwrite=False):
         elif self.path is None or not self.path.exists():
             self.data = pd.DataFrame()
         elif self.path.suffix == ".csv":
-            self.data = pd.read_csv(self.path)
+            self.data = pd.read_csv(self.path, dtype=REGSITRY_DTYPE)
         elif self.path.suffix == ".parquet":
-            self.data = pd.read_parquet(self.path)
+            self.data = pd.read_parquet(self.path).astype(REGSITRY_DTYPE)
         else:
             raise TypeError("Unknown registry type")