test(uri): ✅ add tests for file:// URIs with non-local authority

Author: kikkomep

tests/unit/test_rocrate.py

@@ -604,6 +604,9 @@ def _metadata_dict_with_id(entity_id: str) -> dict:
         "s3://bucket/key",
         "https://example.org/data.txt",
         "arcp://name,foo/bar",
+        # `file://` URIs with a (non-local) authority denote files living on
+        # another host (RFC 8089), so they are remote too.
+        "file://gs02r3b58-ib0/scratch/tmp/5190874/tmp_rf_samples_slt86rc0",
         # scheme-only absolute URIs (no authority; RO-Crate 1.1 § 4.2.2 + RFC 3986)
         "urn:doi:10.5281/zenodo.1234",
         "doi:10.5281/zenodo.1234",
@@ -627,3 +630,30 @@ def test_absolute_uri_data_entity_is_classified_as_remote(entity_id):
     assert entity not in crate.metadata.get_data_entities(exclude_web_data_entities=True), (
         f"Entity '{entity_id}' should be excluded from local-only data entities"
     )
+
+
+@pytest.mark.parametrize(
+    "entity_id",
+    [
+        # `file://` URIs without an authority (RFC 8089) or with the special
+        # `localhost` authority refer to the local machine, so they describe
+        # local payload members that the must/4 check must still verify.
+        "file:///absolute/path/to/file.txt",
+        "file://localhost/absolute/path/to/file.txt",
+    ],
+)
+def test_local_file_uri_data_entity_is_not_remote(entity_id):
+    """
+    `file://` Data Entity identifiers that point to the local machine (empty or
+    `localhost` authority) MUST NOT be treated as remote/web-based: only
+    `file://<host>/...` URIs with a real host are remote (issue #176 follow-up).
+    """
+    crate = ROCrate.from_metadata_dict(_metadata_dict_with_id(entity_id))
+    entity = crate.metadata.get_entity(entity_id)
+    assert entity is not None, "Entity should be present in the metadata"
+    assert not entity.is_remote(), (
+        f"Entity with local file URI '{entity_id}' should NOT be classified as remote"
+    )
+    assert entity not in crate.metadata.get_web_data_entities(), (
+        f"Entity '{entity_id}' should not be listed as a web data entity"
+    )

tests/unit/test_uri.py

@@ -73,6 +73,39 @@ def test_scheme_only_absolute_uri_is_remote(uri_str, expected_scheme):
     assert not uri.is_natively_checkable()
 
 
+def test_file_uri_with_remote_host_is_remote():
+    # A `file://` URI carrying a (non-local) authority points to a file on
+    # another host (RFC 8089) and must be treated as remote, not as a local
+    # payload member (regression for issue #176 with `file://` schemes).
+    uri = URI("file://gs02r3b58-ib0/scratch/tmp/5190874/tmp_rf_samples_slt86rc0")
+    assert uri.scheme == "file"
+    assert uri.is_remote_resource()
+    assert not uri.is_local_resource()
+    assert not uri.is_natively_checkable()
+
+
+@pytest.mark.parametrize("uri_str", [
+    "file:///absolute/path/file.txt",
+    "file://localhost/absolute/path/file.txt",
+])
+def test_file_uri_to_local_host_is_local(uri_str):
+    # An empty or `localhost` authority denotes the local machine.
+    uri = URI(uri_str)
+    assert uri.scheme == "file"
+    assert uri.is_local_resource()
+    assert not uri.is_remote_resource()
+
+
+@pytest.mark.parametrize("path", ["README.md", "data/file.txt", "./", "/abs/dir"])
+def test_local_path_never_gains_a_spurious_host(path):
+    # Plain filesystem paths are normalized to authority-less `file:` URIs, so
+    # the first path segment is never mistaken for a remote host.
+    uri = URI(path)
+    assert uri.is_local_resource()
+    assert not uri.is_remote_resource()
+    assert uri.get_netloc() == ""
+
+
 def test_url_with_query_params():
     uri = URI("http://example.com?param1=value1&param2=value2")
     assert uri.get_query_param("param1") == "value1"