Add robust Dataverse version parsing and tests

Allow parsing of Dataverse version strings with trailing descriptors by relaxing VERSION_PATTERN and adding _extract_major_version(version_string). The helper extracts the major version or raises a ValueError for invalid prefixes; Dataverse now uses this helper to enforce MINIMUM_MAJOR_VERSION. Added unit tests (tests/dataverse/test_version_parsing.py) covering accepted formats (e.g., "6.10", "v6.10.1", "6.10 bugfixes") and invalid inputs.

Author: JR-1991

pyDataverse/dataverse/dataverse.py

@@ -38,9 +38,37 @@
     from .collection import Collection
     from .dataset import Dataset
 
-VERSION_PATTERN = re.compile(r"^v?(\d+)(?:\.\d+)*(?:-.*)?$")
+VERSION_PATTERN = re.compile(r"^v?(\d+)(?:\.\d+)*")
 MINIMUM_MAJOR_VERSION = 6
 
+
+def _extract_major_version(version_string: str) -> int:
+    """
+    Extract the major version from a Dataverse version string.
+
+    Supports canonical versions (e.g., ``6.10``, ``v6.10.1``, ``6.10.1-SNAPSHOT``)
+    and versions with trailing descriptors returned by some installations
+    (e.g., ``6.10 bugfixes``).
+
+    Args:
+        version_string: Raw version string returned by the Dataverse API.
+
+    Returns:
+        int: Parsed major version.
+
+    Raises:
+        ValueError: If the version string does not start with a valid version number.
+    """
+    match = VERSION_PATTERN.match(version_string.strip())
+
+    if match is None:
+        raise ValueError(
+            f"Unable to parse version string: {version_string}. "
+            "Expected to start with: vX.Y.Z, X.Y, X.Y.Z-SUFFIX, or X.Y <description>"
+        )
+
+    return int(match.group(1))
+
 # Used for autocomplete in the create_dataset method
 Subject = Literal[
     "Agricultural Sciences",
@@ -183,17 +211,9 @@ def _check_api_version(self) -> None:
             ValueError: If the Dataverse instance version is below the minimum required version.
         """
         version = self.native_api.get_info_version()
-        match = VERSION_PATTERN.match(version.version)
-
-        if match is None:
-            raise ValueError(
-                f"Unable to parse version string: {version.version}. "
-                f"Expected format: vX.Y.Z or X.Y or X.Y.Z-SUFFIX"
-            )
-
-        major_version_str = match.group(1)
+        major_version = _extract_major_version(version.version)
 
-        if int(major_version_str) < MINIMUM_MAJOR_VERSION:
+        if major_version < MINIMUM_MAJOR_VERSION:
             raise ValueError(
                 f"Dataverse version {version.version} is below the minimum required "
                 f"version {MINIMUM_MAJOR_VERSION}. Please upgrade your Dataverse instance."

tests/dataverse/test_version_parsing.py

@@ -0,0 +1,29 @@
+"""Unit tests for Dataverse version parsing."""
+
+import pytest
+
+from pyDataverse.dataverse.dataverse import _extract_major_version
+
+
+@pytest.mark.parametrize(
+    "version_string, expected_major",
+    [
+        ("6.10", 6),
+        ("v6.10.1", 6),
+        ("6.10.1-SNAPSHOT", 6),
+        ("6.10 bugfixes", 6),
+        ("  7.2 custom-build  ", 7),
+    ],
+)
+def test_extract_major_version_accepts_known_formats(
+    version_string: str, expected_major: int
+) -> None:
+    """It parses major versions from canonical and descriptor-suffixed strings."""
+    assert _extract_major_version(version_string) == expected_major
+
+
+@pytest.mark.parametrize("version_string", ["", "bugfixes 6.10", "release-v6.10"])
+def test_extract_major_version_rejects_invalid_prefix(version_string: str) -> None:
+    """It fails if the version string does not start with a version number."""
+    with pytest.raises(ValueError, match="Unable to parse version string"):
+        _extract_major_version(version_string)