fix: arxiv year

2026-04-17 17:03:59 -04:00
parent 432010f431
commit 174801242d
4 changed files with 158 additions and 8 deletions
@@ -0,0 +1,139 @@
+"""Test for arXiv year extraction bug fix."""
+
+import shutil
+from pathlib import Path
+
+import pytest
+
+from paperlib.config import LibraryPaths
+from paperlib.models import SourceType
+from paperlib.storage import PaperStorageManager
+
+
+class TestArxivYearFix:
+    """Test the arXiv year extraction fix."""
+
+    @pytest.fixture
+    def temp_library(self):
+        """Create a temporary library for testing."""
+        temp_dir = Path("./.tmp") / f"test_arxiv_year_{hash(self)}"
+        temp_dir.mkdir(parents=True, exist_ok=True)
+        library_paths = LibraryPaths.from_root(temp_dir)
+        library_paths.create_directories()
+
+        yield library_paths
+
+        # Cleanup
+        if temp_dir.exists():
+            shutil.rmtree(temp_dir)
+
+    @pytest.fixture
+    def storage_manager(self, temp_library):
+        """Create a storage manager for testing."""
+        return PaperStorageManager(temp_library)
+
+    def test_arxiv_year_extraction_2022(self, storage_manager):
+        """Test year extraction for 2022 paper (2212.06340)."""
+        paper_dir = storage_manager.get_paper_directory(
+            "arxiv-2212_06340", SourceType.ARXIV
+        )
+
+        # Should extract year 2022 from 2212.06340
+        expected = (
+            storage_manager.library_paths.papers_dir
+            / "arxiv"
+            / "2022"
+            / "arxiv-2212_06340"
+        )
+        assert paper_dir == expected
+
+    def test_arxiv_year_extraction_2023(self, storage_manager):
+        """Test year extraction for 2023 paper (2301.12345)."""
+        paper_dir = storage_manager.get_paper_directory(
+            "arxiv-2301_12345", SourceType.ARXIV
+        )
+
+        # Should extract year 2023 from 2301.12345
+        expected = (
+            storage_manager.library_paths.papers_dir
+            / "arxiv"
+            / "2023"
+            / "arxiv-2301_12345"
+        )
+        assert paper_dir == expected
+
+    def test_arxiv_year_extraction_2020(self, storage_manager):
+        """Test year extraction for 2020 paper (2005.67890)."""
+        paper_dir = storage_manager.get_paper_directory(
+            "arxiv-2005_67890", SourceType.ARXIV
+        )
+
+        # Should extract year 2020 from 2005.67890
+        expected = (
+            storage_manager.library_paths.papers_dir
+            / "arxiv"
+            / "2020"
+            / "arxiv-2005_67890"
+        )
+        assert paper_dir == expected
+
+    def test_arxiv_year_extraction_1999(self, storage_manager):
+        """Test year extraction for 1999 paper (9912.12345)."""
+        paper_dir = storage_manager.get_paper_directory(
+            "arxiv-9912_12345", SourceType.ARXIV
+        )
+
+        # Should extract year 1999 from 9912.12345 (99 -> 1999)
+        expected = (
+            storage_manager.library_paths.papers_dir
+            / "arxiv"
+            / "1999"
+            / "arxiv-9912_12345"
+        )
+        assert paper_dir == expected
+
+    def test_arxiv_year_extraction_2000(self, storage_manager):
+        """Test year extraction for 2000 paper (0001.12345)."""
+        paper_dir = storage_manager.get_paper_directory(
+            "arxiv-0001_12345", SourceType.ARXIV
+        )
+
+        # Should extract year 2000 from 0001.12345 (00 -> 2000)
+        expected = (
+            storage_manager.library_paths.papers_dir
+            / "arxiv"
+            / "2000"
+            / "arxiv-0001_12345"
+        )
+        assert paper_dir == expected
+
+    def test_arxiv_id_with_version(self, storage_manager):
+        """Test year extraction with version number."""
+        paper_dir = storage_manager.get_paper_directory(
+            "arxiv-2212_06340v1", SourceType.ARXIV
+        )
+
+        # Should extract year 2022 from 2212.06340v1
+        expected = (
+            storage_manager.library_paths.papers_dir
+            / "arxiv"
+            / "2022"
+            / "arxiv-2212_06340v1"
+        )
+        assert paper_dir == expected
+
+    def test_existing_storage_test_still_passes(self, storage_manager):
+        """Ensure we didn't break the existing test case."""
+        # This matches the test case in test_storage.py
+        paper_dir = storage_manager.get_paper_directory(
+            "arxiv-2212_06340", SourceType.ARXIV
+        )
+
+        # The old test expected papers/arxiv/2212/ but should now be papers/arxiv/2022/
+        expected = (
+            storage_manager.library_paths.papers_dir
+            / "arxiv"
+            / "2022"
+            / "arxiv-2212_06340"
+        )
+        assert paper_dir == expected