fix: improve catalog and data loading locally (#69)

Author: hansmohrmann

src/nnja_ai/catalog.py

@@ -42,7 +42,7 @@ class DataCatalog:
 
     def __init__(
         self,
-        mirror: Optional[str] = DEFAULT_MIRROR,
+        mirror: Optional[str] = None,
         base_path: Optional[str] = None,
         catalog_json: Optional[str] = None,
     ):
@@ -60,7 +60,9 @@ def __init__(
         Raises:
             ValueError: If both mirror and custom parameters are specified.
         """
-        # Validate parameters - no mix and match
+        # Validate parameters and specify a default mirror if not provided - no mix and match
+        if mirror is None and base_path is None and catalog_json is None:
+            mirror = DEFAULT_MIRROR
         if mirror is not None and (base_path is not None or catalog_json is not None):
             raise ValueError(
                 "Cannot specify both 'mirror' and custom parameters ('base_path', 'catalog_json'). "

src/nnja_ai/io.py

@@ -145,17 +145,21 @@ def load_parquet(
         case "pandas":
             import pandas as pd
 
-            return pd.concat(
-                [
-                    pd.read_parquet(
-                        uri,
-                        columns=columns,
-                        storage_options=auth_args,
-                        **backend_kwargs,
-                    )
-                    for uri in parquet_uris
-                ]
-            )
+            if not auth_args:
+                # This is a local file system, so we can just read the parquet files directly
+                return pd.read_parquet(parquet_uris, columns=columns, **backend_kwargs)
+            else:
+                return pd.concat(
+                    [
+                        pd.read_parquet(
+                            uri,
+                            columns=columns,
+                            storage_options=auth_args,
+                            **backend_kwargs,
+                        )
+                        for uri in parquet_uris
+                    ]
+                )
         case "polars":
             import polars as pl