Add versioned glossary YAML import and export

2026-09-12 11:23:18 +02:00
parent 9840953687
commit e2a0434941
7 changed files with 396 additions and 0 deletions
@@ -0,0 +1,147 @@
+from pathlib import Path
+
+import pytest
+
+from mka.application.glossary import GlossaryRepository
+from mka.application.glossary_yaml import (
+    GlossaryYamlError,
+    export_glossary_yaml,
+    import_glossary_yaml,
+)
+
+
+def repository(tmp_path: Path) -> GlossaryRepository:
+    result = GlossaryRepository(tmp_path / "glossary.sqlite3")
+    result.initialize()
+    return result
+
+
+def test_round_trip_preserves_ids_aliases_and_active_state(tmp_path: Path) -> None:
+    source = repository(tmp_path / "source")
+    active = source.create(
+        "Größenmaß",
+        "technical_term",
+        aliases=("Groessenmass", "Größen-Maß"),
+        description="Canonical UTF-8 spelling",
+    )
+    inactive = source.create("Legacy Product", "product", is_active=False)
+
+    encoded = export_glossary_yaml(source.list()).encode("utf-8")
+    target = repository(tmp_path / "target")
+    target.create("Will be replaced", "other")
+    target.replace_all(import_glossary_yaml(encoded))
+
+    entries = target.list()
+    assert [entry.id for entry in entries] == [active.id, inactive.id]
+    assert entries[0].canonical_term == "Größenmaß"
+    assert entries[0].aliases == ("Groessenmass", "Größen-Maß")
+    assert entries[0].description == "Canonical UTF-8 spelling"
+    assert entries[0].is_active is True
+    assert entries[1].is_active is False
+
+
+def test_malformed_yaml_does_not_replace_existing_glossary(tmp_path: Path) -> None:
+    glossary = repository(tmp_path)
+    existing = glossary.create("PBAT", "acronym")
+
+    with pytest.raises(GlossaryYamlError, match="Malformed glossary YAML"):
+        import_glossary_yaml(b"version: 1\nglossary: [")
+
+    assert glossary.list() == [existing]
+
+
+@pytest.mark.parametrize(
+    "entries, message",
+    [
+        (
+            [
+                {
+                    "id": 1,
+                    "canonical_term": "PBAT",
+                    "aliases": [],
+                    "category": "acronym",
+                    "description": None,
+                    "active": True,
+                },
+                {
+                    "id": 2,
+                    "canonical_term": "pbat",
+                    "aliases": [],
+                    "category": "material",
+                    "description": None,
+                    "active": True,
+                },
+            ],
+            "duplicate canonical term or alias",
+        ),
+        (
+            [
+                {
+                    "id": 1,
+                    "canonical_term": "Secugrid",
+                    "aliases": ["Sikirgut", "SIKIRGUT"],
+                    "category": "product",
+                    "description": None,
+                    "active": True,
+                },
+            ],
+            "aliases must be unique",
+        ),
+        (
+            [
+                {
+                    "id": 1,
+                    "canonical_term": "Secugrid",
+                    "aliases": ["PBAT"],
+                    "category": "product",
+                    "description": None,
+                    "active": True,
+                },
+                {
+                    "id": 2,
+                    "canonical_term": "PBAT",
+                    "aliases": [],
+                    "category": "acronym",
+                    "description": None,
+                    "active": False,
+                },
+            ],
+            "duplicate canonical term or alias",
+        ),
+    ],
+)
+def test_duplicate_terms_and_aliases_are_rejected(
+    entries: list[dict[str, object]], message: str
+) -> None:
+    import yaml
+
+    content = yaml.safe_dump({"version": 1, "glossary": entries})
+
+    with pytest.raises(GlossaryYamlError, match=message):
+        import_glossary_yaml(content)
+
+
+def test_validation_failure_leaves_database_unchanged(tmp_path: Path) -> None:
+    glossary = repository(tmp_path)
+    existing = glossary.create("Existing", "other", aliases=("Existing alias",))
+    invalid = b"""version: 1
+glossary:
+  - id: 10
+    canonical_term: Duplicate
+    aliases: []
+    category: other
+    description: null
+    active: true
+  - id: 11
+    canonical_term: duplicate
+    aliases: []
+    category: other
+    description: null
+    active: false
+"""
+
+    with pytest.raises(GlossaryYamlError):
+        imported = import_glossary_yaml(invalid)
+        glossary.replace_all(imported)
+
+    assert glossary.list() == [existing]