Merge pull request #97 from GenomicDataInfrastructure/fix/identifier-value-parsing

fix(identifier): correctly parse values containing '=' in Identifier

Author: brunopacheco1

ckanext/fairdatapoint/harvesters/domain/identifier.py

@@ -10,7 +10,6 @@
 class IdentifierException(Exception):
     pass
 
-
 class Identifier:
 
     def __init__(self, guid: str):
@@ -31,22 +30,17 @@ def get_id_value(self) -> str:
     def get_part(self, index: int) -> str:
         key_values = self.guid.split(SEPARATOR)
 
-        if len(key_values) > 0:
-            # Get the last one, that's the one we are interested in
-            key_value = key_values[-1].split(KEY_VALUE_SEPARATOR)
-            if len(key_value) == 2:
-                result = key_value[index]
-            else:
-                raise IdentifierException(
-                    "Unexpected number of parts in key_value [{}]: [{}]",
-                    key_values[1],
-                    len(key_value),
-                )
-        else:
+        if not self.guid.strip() or key_values == ['']:
+            raise IdentifierException(
+                f"Empty or improperly formatted record identifier: [{self.guid}]"
+            )
+
+        key_value = key_values[-1].split(KEY_VALUE_SEPARATOR, 1)
+
+        if len(key_value) != 2:
             raise IdentifierException(
-                "Unexpected number of parts in record identifier [{}]: [{}]",
-                self.guid,
-                len(key_values),
+                f"Unexpected number of parts in key_value [{key_values[-1]}]: [{len(key_value)}]"
             )
 
-        return result
+        return key_value[index]
+

ckanext/fairdatapoint/tests/test_identifier.py

@@ -33,11 +33,17 @@ def test_get_part_zero_index(self):
         assert actual == expected
 
     def test_get_part_no_separator(self):
-        with pytest.raises(IndexError):
+        with pytest.raises(IdentifierException):
             identifier = Identifier("some_id_no_separator")
             part = identifier.get_part(index=0)
 
     def test_get_part_raises(self):
         with pytest.raises(IdentifierException):
             identifier = Identifier("too_many;id_separators;in_an_id")
             part = identifier.get_part(index=1)
+
+    def test_get_id_type_and_value(self):
+        guid = "dataset=http://example.com/resource?id=123"
+        identifier = Identifier(guid)
+        assert identifier.get_id_type() == "dataset"
+        assert identifier.get_id_value() == "http://example.com/resource?id=123"