feat(tools): map kv library variants to kb ids

pv-agent/tests/test_kv_variant_map.py

@@ -0,0 +1,76 @@
+"""Tests für die KV-Varianten-Map (Odoo library_variant_id ↔ KB kv-kvt-IDs)."""
+import json
+from pathlib import Path
+
+ROOT = Path(__file__).resolve().parents[1]
+
+
+def load_map() -> dict:
+    return json.loads(
+        (ROOT / "tools" / "catalogs" / "kv_variant_map.json").read_text(encoding="utf-8")
+    )
+
+
+def load_kb_catalog() -> dict:
+    return json.loads(
+        (ROOT / "tools" / "catalogs" / "kv_catalog.json").read_text(encoding="utf-8")
+    )
+
+
+def test_map_structure_and_kb_id_integrity():
+    kb = load_kb_catalog()
+    kb_ids = {e["id"] for e in kb["entries"].values()}
+    variant_map = load_map()
+
+    variants = variant_map["variants"]
+    assert variants, "Mapping muss Varianten enthalten"
+    referenced = set()
+    for vid, variant in variants.items():
+        assert variant["variant_id"] == vid
+        assert variant["docs"], f"leere Variante: {vid}"
+        assert variant["confidence"] in ("matched", "low")
+        for doc in variant["docs"]:
+            assert {"kv_kvt_id", "slug", "doctype"} <= set(doc)
+            referenced.add(doc["kv_kvt_id"])
+    assert referenced <= kb_ids, "Mapping darf nur existierende KB-IDs führen"
+
+    # Unmatched-Einträge sind ebenfalls echte KB-IDs und decken den Rest ab
+    unmatched = variant_map["unmatched_kb_entries"]
+    assert set(unmatched) <= kb_ids
+    stats = variant_map["stats"]
+    assert stats["n_kb_entries_covered"] + stats["n_kb_entries_unmatched"] == (
+        stats["n_kb_entries_total"]
+    )
+    assert stats["n_variants"] == len(variants)
+
+
+def test_odoo_seed_variants_are_covered():
+    """Odoo-Seed-KVs (kv_seed.xml): SI-2203 Handwerk und Gewerbe,
+    SI-2748 Metallgewerbe müssen KV- und Tabellendokumente führen."""
+    variants = load_map()["variants"]
+    seeds = {vid: variants.get(vid) for vid in ("SI-2203_de", "SI-2748_de")}
+    assert seeds["SI-2203_de"] and seeds["SI-2203_de"]["confidence"] == "matched"
+    assert seeds["SI-2748_de"] and seeds["SI-2748_de"]["confidence"] == "matched"
+
+    slugs_2203 = {d["slug"] for d in seeds["SI-2203_de"]["docs"]}
+    slugs_2748 = {d["slug"] for d in seeds["SI-2748_de"]["docs"]}
+    assert "kollektivvertrag-gewerbe-handwerk-und-dienstleistung-2026" in slugs_2203
+    assert "gehaltsordnung-gewerbe-handwerk-dienstleistung-2026" in slugs_2203
+    assert "kollektivvertrag-abschluss-angestellte-metallgewerbe-2026" in slugs_2748
+    assert "lohnordnung-metallgewerbe-arbeiter-2026" in slugs_2748
+
+
+def test_doc_pairs_keep_slug_and_id_correspondence():
+    kb = load_kb_catalog()
+    by_id = {e["id"]: e for e in kb["entries"].values()}
+    variant_map = load_map()
+    checked = 0
+    for variant in variant_map["variants"].values():
+        for doc in variant["docs"]:
+            entry = by_id[doc["kv_kvt_id"]]
+            url_slug = entry["url"].rstrip("/").rsplit("/", 1)[-1]
+            assert url_slug == doc["slug"], (
+                f"{doc['kv_kvt_id']}: {url_slug} != {doc['slug']}"
+            )
+            checked += 1
+    assert checked >= 400