Add manifest tests

fondant/manifest.py

@@ -53,7 +53,7 @@ def remove_field(self, name: str) -> None:
         del self._specification["fields"][name]
 
     def __repr__(self) -> str:
-        return f"{self.__class__.__name__}({self._specification!r}"
+        return f"{self.__class__.__name__}({self._specification!r})"
 
 
 class Index(Subset):
@@ -149,6 +149,10 @@ def add_metadata(self, key: str, value: t.Any) -> None:
     def base_path(self) -> str:
         return self.metadata["base_path"]
 
+    @base_path.setter
+    def base_path(self, path: str) -> None:
+        self.metadata["base_path"] = path
+
     @property
     def run_id(self) -> str:
         return self.metadata["run_id"]
@@ -255,4 +259,4 @@ def evolve(self, component_spec: FondantComponentSpec) -> "Manifest":
         return evolved_manifest
 
     def __repr__(self) -> str:
-        return f"{self.__class__.__name__}({self._specification!r}"
+        return f"{self.__class__.__name__}({self._specification!r})"

tests/test_dataset.py

@@ -48,9 +48,11 @@ def test_merge_subsets(input_manifest, component_spec):
 
 def test_write_subsets(input_manifest, output_manifest, component_spec):
     # Dictionary specifying the expected subsets to write and their column names
-    subset_columns_dict = {"index": ['id', 'source'],
-                           "properties": ['Name', 'HP', 'id', 'source'],
-                           "types": ['Type 1', 'Type 2', 'id', 'source']}
+    subset_columns_dict = {
+        "index": ["id", "source"],
+        "properties": ["Name", "HP", "id", "source"],
+        "types": ["Type 1", "Type 2", "id", "source"],
+    }
 
     # Load dataframe from input manifest
     input_fds = FondantDataset(manifest=input_manifest)

tests/test_manifest.py

@@ -1,9 +1,10 @@
 import json
+import pkgutil
 from pathlib import Path
 
 import pytest
 from fondant.exceptions import InvalidManifest
-from fondant.manifest import Manifest, Type
+from fondant.manifest import Subset, Manifest, Type
 
 
 manifest_path = Path(__file__).parent / "example_specs/manifests"
@@ -28,6 +29,63 @@ def test_manifest_validation(valid_manifest, invalid_manifest):
         Manifest(invalid_manifest)
 
 
+def test_subset_init():
+    """Test initializing a subset"""
+    subset_spec = {
+        "location": "/ABC/123/images",
+        "fields": {
+            "data": {
+                "type": "binary",
+            },
+        },
+    }
+    subset = Subset(specification=subset_spec, base_path="/tmp")
+    assert subset.location == "/tmp/ABC/123/images"
+    assert (
+        subset.__repr__()
+        == "Subset({'location': '/ABC/123/images', 'fields': {'data': {'type': 'binary'}}})"
+    )
+
+
+def test_subset_fields():
+    """Test manipulating subset fields"""
+    subset_spec = {
+        "location": "/ABC/123/images",
+        "fields": {
+            "data": {
+                "type": "binary",
+            },
+        },
+    }
+    subset = Subset(specification=subset_spec, base_path="/tmp")
+
+    # add a field
+    subset.add_field(name="data2", type_=Type.binary)
+    assert "data2" in subset.fields
+
+    # add a duplicate field
+    with pytest.raises(ValueError):
+        subset.add_field(name="data2", type_=Type.binary)
+
+    # add a duplicate field but overwrite
+    subset.add_field(name="data2", type_=Type.utf8, overwrite=True)
+    assert subset.fields["data2"].type == "utf8"
+
+    # remove a field
+    subset.remove_field(name="data2")
+    assert "data2" not in subset.fields
+
+
+def test_set_base_path(valid_manifest):
+    """Test altering the base path in the manifest"""
+    manifest = Manifest(valid_manifest)
+    tmp_path = "/tmp/base_path"
+    manifest.base_path = tmp_path
+
+    assert manifest.base_path == tmp_path
+    assert manifest._specification["metadata"]["base_path"] == tmp_path
+
+
 def test_from_to_file(valid_manifest):
     """Test reading from and writing to file"""
     tmp_path = "/tmp/manifest.json"
@@ -62,7 +120,9 @@ def test_manifest_creation():
     run_id = "run_id"
     component_id = "component_id"
 
-    manifest = Manifest.create(base_path=base_path, run_id=run_id, component_id=component_id)
+    manifest = Manifest.create(
+        base_path=base_path, run_id=run_id, component_id=component_id
+    )
     manifest.add_subset("images", [("width", Type.int32), ("height", Type.int32)])
     manifest.subsets["images"].add_field("data", Type.binary)
 
@@ -72,9 +132,7 @@ def test_manifest_creation():
             "run_id": run_id,
             "component_id": component_id,
         },
-        "index": {
-            "location": f"/{run_id}/{component_id}/index"
-        },
+        "index": {"location": f"/{run_id}/{component_id}/index"},
         "subsets": {
             "images": {
                 "location": f"/{run_id}/{component_id}/images",
@@ -87,17 +145,53 @@ def test_manifest_creation():
                     },
                     "data": {
                         "type": "binary",
-                    }
-                }
+                    },
+                },
             }
-        }
+        },
     }
 
 
+def test_manifest_repr():
+    manifest = Manifest.create(base_path="/", run_id="A", component_id="1")
+    print(manifest)
+    assert (
+        manifest.__repr__()
+        == "Manifest({'metadata': {'base_path': '/', 'run_id': 'A', 'component_id': '1'}, 'index': {'location': '/A/1/index'}, 'subsets': {}})"
+    )
+
+
+def test_manifest_alteration(valid_manifest):
+    """Test alteration functionalities of a manifest via the Manifest class"""
+    manifest = Manifest(valid_manifest)
+
+    # test adding a subset
+    manifest.add_subset("images2", [("width", Type.int32), ("height", Type.int32)])
+    assert "images2" in manifest.subsets
+
+    # test adding a duplicate subset
+    with pytest.raises(ValueError):
+        manifest.add_subset("images2", [("width", Type.int32), ("height", Type.int32)])
+
+    # test removing a subset
+    manifest.remove_subset("images2")
+    assert "images2" not in manifest.subsets
+
+    # test removing a nonexistant subset
+    with pytest.raises(ValueError):
+        manifest.remove_subset("pictures")
+
+
 def test_manifest_copy_and_adapt(valid_manifest):
     """Test that a manifest can be copied and adapted without changing the original."""
     manifest = Manifest(valid_manifest)
     new_manifest = manifest.copy()
     new_manifest.remove_subset("images")
     assert manifest._specification == valid_manifest
     assert new_manifest._specification != valid_manifest
+
+
+def test_no_validate_schema(monkeypatch, valid_manifest):
+    monkeypatch.setattr(pkgutil, "get_data", lambda package, resource: None)
+    with pytest.raises(FileNotFoundError):
+        manifest = Manifest(valid_manifest)