update prune and add tests

digital_land/phase/prune.py

@@ -47,9 +47,17 @@ def process(self, stream):
                 entry_number = block["entry-number"]
 
                 if self.issues:
-                    self.issues.log_issue(
-                        "entity", "unknown entity", curie, line_number=line_number
-                    )
+                    if not reference:
+                        self.issues.log_issue(
+                            "entity",
+                            "unknown entity - missing reference",
+                            curie,
+                            line_number=line_number,
+                        )
+                    else:
+                        self.issues.log_issue(
+                            "entity", "unknown entity", curie, line_number=line_number
+                        )
 
                 logging.info(
                     f"{resource} row {entry_number}: missing entity for {curie}"

tests/unit/phase/test_prune.py

@@ -0,0 +1,45 @@
+from digital_land.phase.prune import EntityPrunePhase
+from digital_land.log import IssueLog
+from digital_land.log import DatasetResourceLog
+
+
+class TestEntityPrunePhase:
+    def test_process_returns_missing_reference(self):
+        input_stream = [
+            {
+                "row": {
+                    "prefix": "ancient-woodland",
+                    "reference": "",
+                    "entity": "",
+                },
+                "entry-number": 1,
+                "resource": "123",
+                "line-number": 2,
+            }
+        ]
+        issues = IssueLog()
+        resource_log = DatasetResourceLog()
+        phase = EntityPrunePhase(issues, resource_log)
+
+        list(phase.process(input_stream))
+        assert issues.rows[0]["issue-type"] == "unknown entity - missing reference"
+
+    def test_process_returns_unknown_entity(self):
+        input_stream = [
+            {
+                "row": {
+                    "prefix": "dataset",
+                    "reference": "REF01",
+                    "entity": "",
+                },
+                "entry-number": 1,
+                "resource": "123",
+                "line-number": 2,
+            }
+        ]
+        issues = IssueLog()
+        resource_log = DatasetResourceLog()
+        phase = EntityPrunePhase(issues, resource_log)
+
+        list(phase.process(input_stream))
+        assert issues.rows[0]["issue-type"] == "unknown entity"