Sanitize version num from string to int by removing non-numeric characters (#58)

* Sanitize version string by removing non-numeric characters

* test versions are extracted correctly

* test borealis compliance

---------

Co-authored-by: Jan Range <30547301+JR-1991@users.noreply.github.com>

Author: kenlhlui

easyDataverse/dataverse.py

@@ -226,7 +226,8 @@ def _version_is_compliant(self) -> bool:
                 f"URL '{self.server_url}' is not a valid Dataverse installation. Couldn't find version info."
             )
 
-        major, minor, *_ = response.json()["data"]["version"].split(".")
+        version = response.json()["data"]["version"]
+        major, minor = self._extract_major_minor(version)
 
         if int(major) >= 6:
             return True
@@ -235,6 +236,26 @@ def _version_is_compliant(self) -> bool:
 
         return False
 
+    @staticmethod
+    def _extract_major_minor(version: str) -> Tuple[int, int]:
+        """Extracts the major and minor version numbers from a Dataverse version string."""
+        try:
+            major, minor, *_ = version.split(".")
+            major = "".join(filter(str.isdigit, major))
+            minor = "".join(filter(str.isdigit, minor))
+            return int(major), int(minor)
+        except ValueError:
+            raise ValueError(f"Version '{version}' is not a valid Dataverse version.")
+
+    @staticmethod
+    def _check_version(major: int, minor: int) -> bool:
+        """Checks if the version is compliant."""
+        if int(major) >= 6:
+            return True
+        elif int(major) >= 5 and int(minor) >= 13:
+            return True
+        return False
+
     def _fetch_licenses(self) -> Dict[str, License]:
         """Fetches the licenses from the Dataverse installation."""
         response = httpx.get(parse.urljoin(str(self.server_url), "/api/licenses"))

tests/integration/test_connection.py

@@ -77,3 +77,8 @@ def test_numeric_namespace(self):
             Dataverse(server_url="https://dataverse.harvard.edu")
         except ValueError as e:
             AssertionError("Failed to parse numeric namespace: " + str(e))
+
+    @pytest.mark.integration
+    def test_version_borealis(self):
+        """Tests compatibility with BorealisData, which uses a different versioning scheme."""
+        Dataverse("https://borealisdata.ca/")

tests/unit/test_dataverse.py

@@ -21,3 +21,40 @@ def test_invalid_api_token(self):
                 server_url="http://localhost:8080",
                 api_token="not a uuid",
             )
+
+    @pytest.mark.unit
+    def test_valid_versions(self):
+        """Test that the version is compliant"""
+        cases = [
+            "6.0",
+            "6.1",
+            "5.13",
+            "5.14",
+            "v6.0",
+            "v5.14",
+            "v5.13-beta",
+            "6.1.0",
+        ]
+
+        for version in cases:
+            major, minor = Dataverse._extract_major_minor(version)
+            assert Dataverse._check_version(major, minor)
+
+    @pytest.mark.unit
+    def test_invalid_version(self):
+        """Test that an invalid version raises a ValueError"""
+        with pytest.raises(ValueError):
+            Dataverse._extract_major_minor("not a version")
+
+    @pytest.mark.unit
+    def test_unsupported_version(self):
+        """Test that the version is compliant"""
+        cases = [
+            "4.0",
+            "4.1",
+            "5.12",
+            "5.11",
+        ]
+        for version in cases:
+            major, minor = Dataverse._extract_major_minor(version)
+            assert not Dataverse._check_version(major, minor)