import json import shutil from pathlib import Path from sqlalchemy import select from db.kitchen_cabinets_taxonomy_import import ( build_seed_payload_from_taxonomy_collection, expand_taxonomy_collections_with_brochure_twins, seed_kitchen_cabinets_taxonomy, ) from db.models import BrochureTaxonomyGroup, MasterCollectionRegistry, MasterTaxonomyNode def test_expand_taxonomy_collections_with_brochure_twins_copies_declared_sibling(): taxonomy_payload = { "collections": { "johnson-sandstone": { "display_name": "Johnson Sandstone", "sku_count": 1, "assignments": [ { "sku": "GD-W1530", "l2_category": "Accessories", "l3_category": "Loose Glass Door", } ], } } } brochure_payloads = { "johnson-sandstone": { "collection": {"canonical_code": "johnson-sandstone", "display_name": "Johnson Sandstone"}, "source_document": {"notes": []}, }, "johnson-white": { "collection": {"canonical_code": "johnson-white", "display_name": "Johnson White"}, "source_document": { "notes": ["Brand/line relationship inherited from Johnson Sandstone brochure sibling."] }, }, } expanded, summary = expand_taxonomy_collections_with_brochure_twins(taxonomy_payload, brochure_payloads) assert expanded["collections"]["johnson-white"]["display_name"] == "Johnson White" assert expanded["collections"]["johnson-white"]["assignments"][0]["sku"] == "GD-W1530" assert expanded["collections"]["johnson-white"]["copied_from_collection"] == "johnson-sandstone" assert summary["created"][0]["canonical_code"] == "johnson-white" def test_seed_kitchen_cabinets_taxonomy_seeds_twin_collection_groups(catalog_intent_session): scratch = Path("E:/workroom/2026/PlytixMage/.tmp_test_kitchen_taxonomy") if scratch.exists(): shutil.rmtree(scratch) try: brochure_dir = scratch / "brochures" brochure_dir.mkdir(parents=True) taxonomy_path = scratch / "kitchen_cabinets_taxonomy.json" taxonomy_path.write_text( json.dumps( { "master_taxonomy_reference": { "confirmed_hierarchy": {"Accessories": ["Loose Glass Door"]}, "proposed_new_l3_categories": {}, }, "collections": { "johnson-sandstone": { "display_name": "Johnson Sandstone", "sku_count": 1, "assignments": [ { "sku": "GD-W1530", "l2_category": "Accessories", "l3_category": "Loose Glass Door", "source_family": "accessories", "source_sku_type": "glass_door", "confidence": "high", } ], } }, } ), encoding="utf-8", ) (brochure_dir / "JohnsonSandstone_0226.json").write_text( json.dumps( { "source_document": {"collection_code": "JS", "file_name": "JohnsonSandstone_0226.pdf", "notes": []}, "collection": { "canonical_code": "johnson-sandstone", "display_name": "Johnson Sandstone", "brand_label": "Legion Manor", "line_name": "Johnson", "color_label": "Sandstone", "category_l1": "Kitchen Cabinets", "product_family": "Cabinets", }, "availability": {"modes": []}, "matching_styles": [], "sample_products": [], "specifications": {}, "kitchen_packages": [], } ), encoding="utf-8", ) (brochure_dir / "JohnsonWhite_site.json").write_text( json.dumps( { "source_document": { "collection_code": "JW", "notes": ["Brand/line relationship inherited from Johnson Sandstone brochure sibling."], }, "collection": { "canonical_code": "johnson-white", "display_name": "Johnson White", "brand_label": "Legion Manor", "line_name": "Johnson", "color_label": "White", "category_l1": "Kitchen Cabinets", "product_family": "Cabinets", }, "availability": {"modes": []}, "matching_styles": [], "sample_products": [], "specifications": {}, "kitchen_packages": [], } ), encoding="utf-8", ) result = seed_kitchen_cabinets_taxonomy( catalog_intent_session, taxonomy_path=taxonomy_path, brochure_folder=brochure_dir, ) registry = catalog_intent_session.scalar( select(MasterCollectionRegistry).where(MasterCollectionRegistry.code == "JW") ) assert registry is not None assert registry.name == "Johnson White" assert any(item["canonical_code"] == "johnson-white" for item in result["twin_expansion"]["created"]) groups = list( catalog_intent_session.scalars( select(BrochureTaxonomyGroup).where(BrochureTaxonomyGroup.collection_registry_id == registry.id) ).all() ) assert len(groups) == 1 assert groups[0].canonical_taxonomy_path_label == "Kitchen Cabinets / Accessories / Loose Glass Door" node = catalog_intent_session.scalar( select(MasterTaxonomyNode).where( MasterTaxonomyNode.path_slug == "kitchen-cabinets/accessories/loose-glass-door" ) ) assert node is not None finally: if scratch.exists(): shutil.rmtree(scratch) def test_build_seed_payload_prefers_explicit_canonical_taxonomy_path_slug(): collection_entry = { "display_name": "Anna Caramel Harvest", "assignments": [ { "sku": "BEC1224L", "l2_category": "Base Cabinets", "l3_category": "Angle End", "canonical_taxonomy_path_slug": "kitchen-cabinets/base-cabinets/angle-end-base-cabinet", } ], } brochure_payload = { "collection": { "canonical_code": "anna-caramel-harvest", "display_name": "Anna Caramel Harvest", "category_l1": "Kitchen Cabinets", } } payload = build_seed_payload_from_taxonomy_collection(collection_entry, brochure_payload) assert payload["taxonomy_groups"][0]["brochure_l1_label"] == "Base Cabinets" assert payload["taxonomy_groups"][0]["brochure_l2_label"] == "Angle End" assert payload["taxonomy_groups"][0]["canonical_taxonomy_path_slug"] == ( "kitchen-cabinets/base-cabinets/angle-end-base-cabinet" ) assert payload["taxonomy_groups"][0]["canonical_taxonomy_path_label"] == ( "Kitchen Cabinets / Base Cabinets / Angle End Base Cabinet" ) def test_sanitize_strips_polluted_clearance_kit_collection_paths(): from db.kitchen_cabinets_taxonomy_import import ( sanitize_brochure_collection_payload, sanitize_kitchen_cabinets_taxonomy_payload, ) taxonomy = { "master_taxonomy_reference": { "confirmed_hierarchy": {"Accessories": ["Clearance Kit"]}, "cross_l2_notes": [], }, "collections": { "anna-snow-white": { "display_name": "Anna Snow White", "path_slug": "kitchen-cabinets/anna-snow-white", "assignments": [], }, "bad-clearance": { "display_name": "Clearance Kit Base Cabinets", "path_slug": "kitchen-cabinets/accessories/clearance-kit/base-cabinets", "assignments": [], }, }, } cleaned = sanitize_kitchen_cabinets_taxonomy_payload(taxonomy) assert "bad-clearance" not in cleaned["payload"]["collections"] assert "anna-snow-white" in cleaned["payload"]["collections"] assert any("never finish collections" in n.lower() for n in cleaned["payload"]["master_taxonomy_reference"]["cross_l2_notes"]) brochure = { "collection": {"canonical_code": "ACH", "display_name": "Anna"}, "taxonomy_groups": [ { "canonical_taxonomy_path_slug": "kitchen-cabinets/accessories/clearance-kit", "skus": ["CK1"], }, { "canonical_taxonomy_path_slug": "kitchen-cabinets/accessories/clearance-kit/base-cabinets/double-door-base-cabinet", "skus": ["B12"], }, ], "start_shopping_config": { "items": [ {"path": "kitchen-cabinets/accessories/clearance-kit/base-cabinets"}, {"path": "kitchen-cabinets/anna-snow-white/base-cabinets"}, ] }, } brochure_cleaned = sanitize_brochure_collection_payload(brochure) slugs = [ g["canonical_taxonomy_path_slug"] for g in brochure_cleaned["payload"]["taxonomy_groups"] ] assert "kitchen-cabinets/accessories/clearance-kit" in slugs assert all("/clearance-kit/base-cabinets" not in s for s in slugs)