Check for valid catalog directories.

src/hipscat_import/index/arguments.py

@@ -5,6 +5,7 @@
 
 from hipscat.catalog import Catalog
 from hipscat.catalog.index.index_catalog_info import IndexCatalogInfo
+from hipscat.io.validation import is_valid_catalog
 
 from hipscat_import.runtime_arguments import RuntimeArguments
 
@@ -38,6 +39,8 @@ def _check_arguments(self):
         if not self.include_hipscat_index and not self.include_order_pixel:
             raise ValueError("At least one of include_hipscat_index or include_order_pixel must be True")
 
+        if not is_valid_catalog(self.input_catalog_path):
+            raise ValueError("input_catalog_path not a valid catalog")
         self.input_catalog = Catalog.read_from_hipscat(catalog_path=self.input_catalog_path)
 
         if self.compute_partition_size < 100_000:

src/hipscat_import/margin_cache/margin_cache_arguments.py

@@ -5,7 +5,7 @@
 import numpy as np
 from hipscat.catalog import Catalog
 from hipscat.catalog.margin_cache.margin_cache_catalog_info import MarginCacheCatalogInfo
-from hipscat.io import file_io
+from hipscat.io.validation import is_valid_catalog
 
 from hipscat_import.runtime_arguments import RuntimeArguments
 
@@ -33,8 +33,10 @@ def __post_init__(self):
 
     def _check_arguments(self):
         super()._check_arguments()
-        if not file_io.does_file_or_directory_exist(self.input_catalog_path):
-            raise FileNotFoundError("input_catalog_path not found on local storage")
+        if not self.input_catalog_path:
+            raise ValueError("input_catalog_path is required")
+        if not is_valid_catalog(self.input_catalog_path):
+            raise ValueError("input_catalog_path not a valid catalog")
 
         self.catalog = Catalog.read_from_hipscat(self.input_catalog_path)
 

src/hipscat_import/soap/arguments.py

@@ -2,6 +2,7 @@
 
 from hipscat.catalog.association_catalog.association_catalog import AssociationCatalogInfo
 from hipscat.catalog.catalog_type import CatalogType
+from hipscat.io.validation import is_valid_catalog
 
 from hipscat_import.runtime_arguments import RuntimeArguments
 
@@ -33,11 +34,15 @@ def _check_arguments(self):
             raise ValueError("object_catalog_dir is required")
         if not self.object_id_column:
             raise ValueError("object_id_column is required")
+        if not is_valid_catalog(self.object_catalog_dir):
+            raise ValueError("object_catalog_dir not a valid catalog")
 
         if not self.source_catalog_dir:
             raise ValueError("source_catalog_dir is required")
         if not self.source_object_id_column:
             raise ValueError("source_object_id_column is required")
+        if not is_valid_catalog(self.source_catalog_dir):
+            raise ValueError("source_catalog_dir not a valid catalog")
 
         if self.compute_partition_size < 100_000:
             raise ValueError("compute_partition_size must be at least 100_000")

tests/hipscat_import/catalog/test_file_readers.py

@@ -33,6 +33,14 @@ def test_unknown_file_type():
         get_file_reader("unknown")
 
 
+def test_file_exists(small_sky_dir):
+    """File reader factory method should fail for missing files or directories"""
+    with pytest.raises(FileNotFoundError, match="File not found"):
+        next(CsvReader().read("foo_not_really_a_path"))
+    with pytest.raises(FileNotFoundError, match="Directory found at path"):
+        next(CsvReader().read(small_sky_dir))
+
+
 def test_csv_reader(small_sky_single_file):
     """Verify we can read the csv file into a single data frame."""
     total_chunks = 0

tests/hipscat_import/index/test_index_argument.py

@@ -22,6 +22,17 @@ def test_empty_required(tmp_path, small_sky_object_catalog):
             output_catalog_name="small_sky_object_index",
         )
 
+    ## Input path is bad
+    with pytest.raises(ValueError, match="input_catalog_path"):
+        IndexArguments(
+            input_catalog_path="/foo",
+            indexing_column="id",
+            output_path=tmp_path,
+            output_catalog_name="small_sky_object_index",
+            overwrite=True,
+        )
+
+    ## Indexing column is required.
     with pytest.raises(ValueError, match="indexing_column "):
         IndexArguments(
             input_catalog_path=small_sky_object_catalog,

tests/hipscat_import/margin_cache/test_arguments_margin_cache.py

@@ -11,13 +11,23 @@
 def test_empty_required(tmp_path):
     """*Most* required arguments are provided."""
     ## Input catalog path is missing
-    with pytest.raises(FileNotFoundError, match="input_catalog_path"):
+    with pytest.raises(ValueError, match="input_catalog_path"):
         MarginCacheArguments(
             margin_threshold=5.0,
             output_path=tmp_path,
             output_catalog_name="catalog_cache",
         )
 
+    ## Input catalog path is bad
+    with pytest.raises(ValueError, match="input_catalog_path"):
+        MarginCacheArguments(
+            input_catalog_path="/foo",
+            margin_threshold=5.0,
+            output_path=tmp_path,
+            output_catalog_name="catalog_cache",
+            overwrite=True,
+        )
+
 
 def test_margin_order_dynamic(small_sky_source_catalog, tmp_path):
     """Ensure we can dynamically set the margin_order"""

tests/hipscat_import/soap/test_soap_arguments.py

@@ -46,6 +46,35 @@ def test_empty_required(tmp_path, small_sky_object_catalog, small_sky_source_cat
             )
 
 
+def test_catalog_paths(tmp_path, small_sky_object_catalog, small_sky_source_catalog):
+    """*Most* required arguments are provided."""
+    ## Object catalog path is bad.
+    with pytest.raises(ValueError, match="object_catalog_dir"):
+        SoapArguments(
+            object_catalog_dir="/foo",
+            object_id_column="id",
+            source_catalog_dir=small_sky_source_catalog,
+            source_object_id_column="object_id",
+            output_catalog_name="small_sky_association",
+            output_path=tmp_path,
+            progress_bar=False,
+            overwrite=True,
+        )
+
+    ## Source catalog path is bad.
+    with pytest.raises(ValueError, match="source_catalog_dir"):
+        SoapArguments(
+            object_catalog_dir=small_sky_object_catalog,
+            object_id_column="id",
+            source_catalog_dir="/foo",
+            source_object_id_column="object_id",
+            output_catalog_name="small_sky_association",
+            output_path=tmp_path,
+            progress_bar=False,
+            overwrite=True,
+        )
+
+
 def test_compute_partition_size(tmp_path, small_sky_object_catalog, small_sky_source_catalog):
     """Test validation of compute_partition_size."""
     with pytest.raises(ValueError, match="compute_partition_size"):