優化 PChome 近門檻商品比對回收

TODO_NEXT_STEPS.txt

@@ -4,6 +4,7 @@
 ================================================================================
 
 【已完成】
+   - V10.474 補 PChome near-threshold matcher / feeder 下一階段：新增 HOOOME 白色經典香氛暖燈與 Gdesign Aroma Lava 2.0 的窄範圍 total-price exact 回收；Recipe Box 可撕式水性兒童指甲油只進 identity_review，不自動寫正式價差；Pavaruni 蠟燭 vs 精油、DASHING DIVA 不同款式仍維持 veto/低信心。known-id refresh 現在會對 hard-veto 舊候選執行 fresh search recovery，missing known-id 若 fresh search 只找到低分候選也會保留 best candidate + diagnostics，而非落成 `refresh_no_result`；正式覆寫保護新增 stronger existing guard，避免較弱新候選以高分覆蓋既有強正式配對。測試：`tests/test_marketplace_product_matcher.py`、`tests/test_competitor_match_attempts_persistence.py`、`tests/test_competitor_match_attempt_rescore_audit.py` 通過。
    - V10.473 補背景 embedding host_health skip：`allow_111_fallback=false` 會讀最近 `host_health_probes`，跳過 runtime unhealthy 的 GCP 節點（預設 20 分鐘，DB fail-open），避免每筆任務都等待已知壞節點 timeout；路由安全不變，不把背景 embedding 落 111。
    - V10.472 補 GCP Ollama failover rootless 診斷：新增 `scripts/ops/diagnose_ollama_gcp_failover.sh`，可一鍵檢查 GCP-A direct、GCP-B direct、111、110:11435、110:11436 與 GCP-B `bge-m3` runtime；目前輸出確認 GCP-A `11434` refused、GCP-B direct/embed OK、110:11435 502、110:11436 OK。110 無免密 sudo、GCP-A 22 refused、GCP-B SSH key denied，因此 primary 修復需 GCP/SSH 或 110 root 權限；應用層維持 GCP-A → GCP-B → 111，不把背景 embedding 落 111。
    - V10.471 依 GCP-B `bge-m3` 實測 latency 調整 embedding timeout，已部署正式環境並確認 `/health=V10.471`：GCP-B `/api/embed` 三次實測約 6.4s / 7.3s / 23.5s，原本 `OLLAMA_EMBED_MAX_TIMEOUT=15` 與 host health `8s` 會誤殺慢但成功的 embedding；已將背景 embedding cap 與 host health model probe timeout 預設調為 30s。正式 smoke 顯示容器內 embedding 回 1024 維、耗時 10.07s；手動 host health probe 後最新狀態為 GCP-A unhealthy、GCP-B healthy、111 healthy。路由安全不變：背景 embedding 仍只跑 GCP-A/GCP-B，不落 111。

config.py

@@ -350,7 +350,7 @@ YOUTUBE_API_KEY = os.getenv('YOUTUBE_API_KEY', '')
 # ==========================================
 # 系統版本與路徑
 # ==========================================
-SYSTEM_VERSION = "V10.473"
+SYSTEM_VERSION = "V10.474"
 LOG_FILE_PATH = os.path.join(BASE_DIR, 'logs/system.log')
 public_url = PUBLIC_URL  # 用於模板顯示
 

docs/memory/current_execution_queue_20260524.md

@@ -27,6 +27,7 @@
 - 2026-05-25 13:38 CST 狀態：已部署 `V10.471` 到 188，正式 `/health` 為 `V10.471`。本輪 recreate `momo-app`、`scheduler`、`telegram-bot`；未使用 `--remove-orphans`，未碰 `momo-db`。Smoke 通過：三容器 healthy、首頁 / daily / growth / host_health / ppt_audit_history / PChome review queue HTTP 200。GCP-B `bge-m3` `/api/embed` 直接實測約 6.4s、7.3s、23.5s，原 `OLLAMA_EMBED_MAX_TIMEOUT=15` 與 host health `OLLAMA_HOST_HEALTH_EMBED_TIMEOUT=8` 會誤殺慢但成功的 embedding；預設改為 30s。正式容器內 embedding smoke 回 1024 維、耗時 10.07s；手動 host health probe 後最新狀態為 GCP-A unhealthy、GCP-B healthy、111 healthy。背景 embedding 路由安全不變：GCP-A → GCP-B，不落 111。
 - 2026-05-25 14:10 CST 起，`V10.472` 補 rootless GCP Ollama failover 診斷腳本與 DevOps SOP：`scripts/ops/diagnose_ollama_gcp_failover.sh` 會檢查 direct GCP-A/GCP-B/111、110 proxy `11435/11436` 與 GCP-B `bge-m3` runtime。現況輸出：GCP-A direct `/api/version` failed/refused、GCP-B direct OK、111 OK、110:11435 502、110:11436 OK、GCP-B embed OK；110 無免密 sudo，`ssh gcp-a` 22 refused、`ssh gcp-b` publickey denied，因此 primary 修復需 GCP/SSH 或 110 root 權限。
 - 2026-05-25 14:12 CST 起，`V10.473` 進行背景 embedding host_health skip：`allow_111_fallback=false` 的背景 embedding 會讀最近 `host_health_probes`，若 GCP-A/GCP-B runtime 已被標 unhealthy，直接跳過該節點並開 GCP circuit，不等待 30 秒 timeout、不落 111；DB 讀取失敗 fail-open。
+- 2026-05-25 14:45 CST 起，`V10.474` 補 PChome near-threshold matcher / feeder 下一階段：HOOOME 白色經典香氛暖燈、Gdesign Aroma Lava 2.0 進 total-price exact；Recipe Box 可撕式水性兒童指甲油保留 identity_review，不自動寫正式價差；Pavaruni 蠟燭 vs 精油與 DASHING DIVA 不同款式仍不放行。known-id refresh 會對 hard-veto 舊候選跑 fresh search recovery；missing known-id 若 fresh search 只有低分候選，也保留 best candidate + diagnostics，不再只記 `refresh_no_result`；正式覆寫保護新增 stronger existing guard。
 - 2026-05-25 12:05 CST 狀態：`main` 已部署到 188，正式 `/health` 為 `V10.467`，待推 Gitea。兩段變更已合併驗證：V10.466 rescore duplicate 改看 latest-state，7 筆 SKU 最新 attempt 全為 `rescore_accepted_current`，`competitor_prices` / `competitor_price_history` 目標計數未變；V10.467 focused exact matcher 在容器內回 `exact / total_price / price_alert_exact`。本輪 recreate `momo-app`、`scheduler`、`telegram-bot`；未使用 `--remove-orphans`，未碰 `momo-db`。Smoke 通過：三容器 healthy、PChome rescore queue API HTTP 200、Gemini 24 小時無 provider 紀錄、Ollama env 順序維持 GCP-A → GCP-B → 111、3 分鐘三容器 log 未見 Traceback / ERROR / CRITICAL / IntegrityError。
 
 ## 1. MOMO / PChome 核心比價準確率
@@ -64,6 +65,7 @@
 - 2026-05-25 11:55 CST 起，rescore audit duplicate 判斷只看最新 attempt；若歷史已有 accepted 但後續 crawler 又追加低信心列，可重新 materialize 成最新 `rescore_accepted_current`。Production pilot 已將 SKU `14756069`、`11159042`、`13842560`、`8394210`、`15192547`、`10509765`、`10603780` 入人工覆核隊列；正式 `competitor_prices` / `competitor_price_history` 未寫入或改變。
 - 2026-05-25 12:20 CST 起，matcher 新增 `focused_exact_total_price_safe` 窄範圍通道；目前只覆蓋 3W CLINIC 粉底液 2入、花美水凝膠 3支、The Ordinary 咖啡因 EGCG 30ml、KUSSEN 屁屁膏 3入、Bone 擴香禮盒、1990 融燭燈白色款與 CANMAKE 淚袋盤等已確認同款樣本。這讓高信心 `exact/manual_review` 能轉為 `exact/total_price` 供 rescore pilot 入人工覆核；DASHING DIVA、唇彩、香味、色號/款式敏感商品仍不放行。
 - 2026-05-25 12:25 CST production pilot：SKU `6101639`、`10074951`、`7760902`、`TP00074980000005`、`14774766`、`10142589`、`10262470`、`10262471`、`11308520` 已從 `true_low_confidence` materialize 為 `rescore_accepted_current`，全數 `exact/total_price/price_alert_exact` 且理由含 `focused_exact_total_price_safe`。SKU `6101784` 因「即期品」商業條件不同，刻意保留在 `true_low_confidence`，不納入本輪自動入隊。
+- 2026-05-25 14:45 CST 起，matcher 擴充至香氛/精油近門檻安全 cohort：HOOOME 白色經典香氛暖燈與 Gdesign Aroma Lava 2.0 可進 `exact/total_price/price_alert_exact`；Recipe Box 可撕式水性兒童指甲油只進 `identity_review`，因兒童指甲油仍可能藏色款/款式。DASHING DIVA 與 Pavaruni cross-type 負例已補測試，避免跨款式、跨劑型誤配。
 
 ## 3. 12 Agent 決策信封整合
 

services/competitor_price_feeder.py

@@ -1371,6 +1371,13 @@ class CompetitorPriceFeeder:
                     "replace_same_identity_better_score="
                     f"{existing_score:.3f}->{match_score:.3f}",
                 )
+        if existing_score >= match_score:
+            return (
+                False,
+                f"existing_match_conflict;stronger_existing;existing_id={existing_id};"
+                f"incoming_id={incoming_id};existing_score={existing_score:.3f};"
+                f"incoming_score={match_score:.3f}",
+            )
         if match_score >= REPLACE_DIFFERENT_PRODUCT_SCORE:
             return True, f"replace_high_confidence_score={match_score:.3f}"
         return (
@@ -1937,6 +1944,35 @@ class CompetitorPriceFeeder:
                             attempts_written += 1
                             continue
 
+                        attempt_status = _classify_low_score_attempt(score, diagnostics)
+                        if (
+                            attempt_status == "recoverable_low_score"
+                            and _has_variant_selection_gap(momo_name, [(best_product, score, diagnostics)], score)
+                        ):
+                            attempt_status = "true_low_confidence"
+                        attempt_terms = search_terms + [term for term in recovery_terms if term not in search_terms]
+                        self._record_match_attempt(
+                            sku,
+                            momo_name,
+                            momo_product_id=momo_product_id,
+                            momo_price=momo_price,
+                            search_terms=attempt_terms,
+                            candidate_count=max(1, recovery_candidate_count),
+                            attempt_status=attempt_status,
+                            best_product=best_product,
+                            best_score=score,
+                            diagnostics=diagnostics,
+                            error_message=(
+                                "missing_known_product_id_fresh_search_low_confidence; "
+                                f"PChome product_id not returned: {competitor_product_id}; "
+                                f"{_format_match_diagnostics(diagnostics)}"
+                            ),
+                            source=source,
+                        )
+                        skipped_low += 1
+                        attempts_written += 1
+                        continue
+
                     self._record_match_attempt(
                         sku,
                         momo_name,
@@ -1991,17 +2027,20 @@ class CompetitorPriceFeeder:
                 if score < MIN_MATCH_SCORE:
                     recovery_terms: list[str] = []
                     recovery_candidate_count = 0
-                    if not getattr(diagnostics, "hard_veto", False):
-                        recovered, recovery_terms, recovery_candidate_count = _recover_low_score_with_fresh_search(
-                            crawler,
-                            momo_name,
-                            momo_price=momo_price,
-                            existing_product_id=competitor_product_id,
-                        )
-                        if recovered:
-                            recovered_product, recovered_score, recovered_diagnostics = recovered
-                            if recovered_score > score:
-                                best_product, score, diagnostics = recovered_product, recovered_score, recovered_diagnostics
+                    recovered, recovery_terms, recovery_candidate_count = _recover_low_score_with_fresh_search(
+                        crawler,
+                        momo_name,
+                        momo_price=momo_price,
+                        existing_product_id=competitor_product_id,
+                    )
+                    if recovered:
+                        recovered_product, recovered_score, recovered_diagnostics = recovered
+                        if (
+                            recovered_score > score
+                            or getattr(diagnostics, "hard_veto", False)
+                            and not getattr(recovered_diagnostics, "hard_veto", False)
+                        ):
+                            best_product, score, diagnostics = recovered_product, recovered_score, recovered_diagnostics
 
                     if score >= MIN_MATCH_SCORE:
                         extras = ["refresh_known_identity"]

services/marketplace_product_matcher.py

@@ -365,6 +365,9 @@ SEARCH_IDENTITY_ANCHORS = (
     "天然植物香氛精油",
     "爆水擦澡濕巾",
     "嬰兒潤膚乳",
+    "可撕式水性兒童指甲油",
+    "aroma lava 解憂放鬆緩緩燈",
+    "經典款香氛蠟燭暖燈",
     "我愛超磁妝定妝噴霧",
     "全天候超完美定妝噴霧",
     "怪獸級持色唇膏",
@@ -475,6 +478,7 @@ FOCUSED_IDENTITY_REVIEW_ONLY_REASONS = {
     "kate_monster_lipstick_catalog",
     "opi_gel_polish_series_catalog",
     "romand_juicy_lip_tint_2_catalog",
+    "recipe_box_peelable_child_polish_catalog",
     "solone_longlasting_eyeliner",
     "shu_auto_hard_formula_refill_catalog",
     "summer_eve_full_skin_wash_2pack",
@@ -497,6 +501,8 @@ FOCUSED_IDENTITY_TOTAL_PRICE_REASONS = {
     "selection1990_half_dome_wax_lamp_white",
     "selection1990_bendable_wax_lamp_white",
     "canmake_tear_bag_palette",
+    "gdesign_aroma_lava_lamp_2",
+    "hooome_classic_white_wax_lamp_bulbs_giftbox",
 }
 
 SEARCH_BROAD_ANCHORS = {
@@ -3512,6 +3518,38 @@ def _has_focused_low_score_exact_identity_line(left: ProductIdentity, right: Pro
         and "淚袋眼影盤" in right_text
     ):
         return "canmake_tear_bag_palette"
+    if (
+        {"recipe", "box"} <= brand_tokens
+        and "可撕式水性兒童指甲油" in left_text
+        and "可撕式水性兒童指甲油" in right_text
+    ):
+        return "recipe_box_peelable_child_polish_catalog"
+    if (
+        "gdesign" in (left.brand_tokens & right.brand_tokens)
+        and "aroma" in left_text
+        and "aroma" in right_text
+        and "lava" in left_text
+        and "lava" in right_text
+        and "解憂放鬆緩緩燈2.0" in left_text
+        and "解憂放鬆緩緩燈2.0" in right_text
+        and "熔岩燈" in left_text
+        and "熔岩燈" in right_text
+        and "精油擴香" in left_text
+        and "精油擴香" in right_text
+    ):
+        return "gdesign_aroma_lava_lamp_2"
+    if (
+        "hooome" in (left.brand_tokens & right.brand_tokens)
+        and "白色" in left_text
+        and "白色" in right_text
+        and "香氛蠟燭暖燈" in left_text
+        and "香氛蠟燭暖燈" in right_text
+        and "兩顆燈泡" in left_text
+        and "兩顆燈泡" in right_text
+        and "禮盒" in left_text
+        and "禮盒" in right_text
+    ):
+        return "hooome_classic_white_wax_lamp_bulbs_giftbox"
     return ""
 
 

tests/test_competitor_match_attempts_persistence.py

@@ -1658,6 +1658,251 @@ def test_competitor_feeder_refresh_recovers_when_known_id_missing(monkeypatch):
     assert attempts[0]["attempt_status"] == "matched"
 
 
+def test_competitor_feeder_records_missing_known_id_low_score_candidate(monkeypatch):
+    from services.competitor_price_feeder import CompetitorPriceFeeder
+    from services.pchome_crawler import PChomeProduct
+
+    candidate = PChomeProduct(
+        product_id="DDAB01-LOW",
+        name="Recipe Box 韓國 recipebox 可撕式水性兒童指甲油",
+        price=299,
+        original_price=350,
+        discount=15,
+        image_url="",
+        product_url="https://24h.pchome.com.tw/prod/DDAB01-LOW",
+        stock=20,
+        store="24h",
+        rating=4.8,
+        review_count=8,
+        is_on_sale=True,
+        crawled_at=datetime.now(),
+    )
+
+    class FakeCrawler:
+        def __init__(self, *_args, **_kwargs):
+            pass
+
+        def fetch_product_details(self, product_ids, batch_size=20):
+            assert product_ids == ["DDAB01-MISSING"]
+            return True, "ok", []
+
+        def search_products(self, *_args, **_kwargs):
+            return True, "ok", [candidate]
+
+    def fake_score(*_args, **_kwargs):
+        return SimpleNamespace(
+            score=0.742,
+            brand_score=1.0,
+            token_score=0.66,
+            spec_score=0.55,
+            sequence_score=0.60,
+            type_score=0.55,
+            price_penalty=0.0,
+            hard_veto=False,
+            reasons=("shared_identity_anchor_variant_safe",),
+            comparison_mode="exact_identity",
+            match_type="no_match",
+            price_basis="none",
+            alert_tier="suppress",
+            tags=["identity_v2", "comparison_exact_identity", "brand_match"],
+        )
+
+    monkeypatch.setattr("services.pchome_crawler.PChomeCrawler", FakeCrawler)
+    monkeypatch.setattr("services.marketplace_product_matcher.score_marketplace_match", fake_score)
+    feeder = CompetitorPriceFeeder(engine=object())
+    attempts = []
+    monkeypatch.setattr(
+        feeder,
+        "_record_match_attempt",
+        lambda *args, **kwargs: attempts.append(kwargs),
+    )
+
+    result = feeder._run_known_identity_refresh_items([{
+        "sku": "TP00018610000157",
+        "name": "韓國 recipebox 可撕式水性兒童指甲油(兒童水性指甲油 可撕式指甲油 韓兔指甲油)",
+        "product_id": 1,
+        "momo_price": 299,
+        "competitor_product_id": "DDAB01-MISSING",
+    }])
+
+    assert result.matched == 0
+    assert result.skipped_low_score == 1
+    assert attempts[0]["attempt_status"] == "recoverable_low_score"
+    assert attempts[0]["best_product"].product_id == "DDAB01-LOW"
+    assert "missing_known_product_id_fresh_search_low_confidence" in attempts[0]["error_message"]
+
+
+def test_competitor_feeder_refresh_recovers_when_known_id_is_hard_veto(monkeypatch):
+    from services.competitor_price_feeder import CompetitorPriceFeeder
+    from services.pchome_crawler import PChomeProduct
+
+    stale = PChomeProduct(
+        product_id="DDAB01-STALE",
+        name="Pavaruni 香氛蠟燭500g",
+        price=980,
+        original_price=1200,
+        discount=18,
+        image_url="",
+        product_url="https://24h.pchome.com.tw/prod/DDAB01-STALE",
+        stock=20,
+        store="24h",
+        rating=4.7,
+        review_count=8,
+        is_on_sale=True,
+        crawled_at=datetime.now(),
+    )
+    recovered = PChomeProduct(
+        product_id="DDAB01-RECOVERED",
+        name="Pavaruni 天然植物香氛精油40種香味10ml",
+        price=399,
+        original_price=499,
+        discount=20,
+        image_url="",
+        product_url="https://24h.pchome.com.tw/prod/DDAB01-RECOVERED",
+        stock=20,
+        store="24h",
+        rating=4.8,
+        review_count=8,
+        is_on_sale=True,
+        crawled_at=datetime.now(),
+    )
+
+    class FakeCrawler:
+        def __init__(self, *_args, **_kwargs):
+            pass
+
+        def fetch_product_details(self, product_ids, batch_size=20):
+            assert product_ids == ["DDAB01-STALE"]
+            return True, "ok", [stale]
+
+        def search_products(self, *_args, **_kwargs):
+            return True, "ok", [stale, recovered]
+
+    def fake_score(_momo_name, competitor_name, **_kwargs):
+        if "RECOVERED" in competitor_name or "天然植物香氛精油" in competitor_name:
+            return SimpleNamespace(
+                score=0.84,
+                brand_score=1.0,
+                token_score=0.78,
+                spec_score=1.0,
+                sequence_score=0.70,
+                type_score=1.0,
+                price_penalty=0.0,
+                hard_veto=False,
+                reasons=("shared_identity_anchor",),
+                comparison_mode="exact_identity",
+                match_type="exact",
+                price_basis="total_price",
+                alert_tier="price_alert_exact",
+                tags=["identity_v2", "comparison_exact_identity", "brand_match"],
+            )
+        return SimpleNamespace(
+            score=0.32,
+            brand_score=1.0,
+            token_score=0.20,
+            spec_score=0.0,
+            sequence_score=0.20,
+            type_score=0.0,
+            price_penalty=0.0,
+            hard_veto=True,
+            reasons=("type_conflict",),
+            comparison_mode="not_comparable",
+            match_type="no_match",
+            price_basis="none",
+            alert_tier="suppress",
+            tags=["identity_v2", "identity_veto"],
+        )
+
+    monkeypatch.setattr("services.pchome_crawler.PChomeCrawler", FakeCrawler)
+    monkeypatch.setattr("services.marketplace_product_matcher.score_marketplace_match", fake_score)
+    feeder = CompetitorPriceFeeder(engine=object())
+    attempts = []
+    writes = []
+    monkeypatch.setattr(
+        feeder,
+        "_should_upsert_competitor_price",
+        lambda *_args, **_kwargs: (True, "same_or_empty_existing"),
+    )
+    monkeypatch.setattr(
+        feeder,
+        "_upsert_competitor_price",
+        lambda sku, product, score, tags, **kwargs: writes.append({
+            "sku": sku,
+            "product_id": product.product_id,
+            "score": score,
+            "tags": tags,
+            **kwargs,
+        }),
+    )
+    monkeypatch.setattr(
+        feeder,
+        "_record_match_attempt",
+        lambda *args, **kwargs: attempts.append(kwargs),
+    )
+
+    result = feeder._run_known_identity_refresh_items([{
+        "sku": "PAVARUNI-OIL",
+        "name": "【Pavaruni】天然植物香氛精油40種香味10ml",
+        "product_id": 1,
+        "momo_price": 399,
+        "competitor_product_id": "DDAB01-STALE",
+    }])
+
+    assert result.matched == 1
+    assert writes[0]["product_id"] == "DDAB01-RECOVERED"
+    assert "fresh_search_recovery" in writes[0]["tags"]
+    assert attempts[0]["attempt_status"] == "matched"
+
+
+def test_should_upsert_protects_stronger_existing_identity_candidate():
+    from sqlalchemy import create_engine, text
+
+    from services.competitor_price_feeder import CompetitorPriceFeeder
+
+    engine = create_engine("sqlite:///:memory:")
+    with engine.begin() as conn:
+        conn.execute(text("""
+            CREATE TABLE competitor_prices (
+                sku TEXT,
+                source TEXT,
+                competitor_product_id TEXT,
+                competitor_product_name TEXT,
+                match_score REAL,
+                tags TEXT
+            )
+        """))
+        conn.execute(text("""
+            INSERT INTO competitor_prices (
+                sku, source, competitor_product_id, competitor_product_name, match_score, tags
+            ) VALUES (
+                '14133077',
+                'pchome',
+                'DDAB01-STRONG',
+                'PONY EFFECT 絕對持久定妝噴霧 100ml',
+                0.950,
+                '["identity_v2","comparison_exact_identity","brand_match"]'
+            )
+        """))
+
+    feeder = CompetitorPriceFeeder(engine=engine)
+    product = SimpleNamespace(
+        product_id="DDAB01-WEAKER",
+        name="PONY EFFECT 絕對持久定妝噴霧",
+    )
+
+    should_write, reason = feeder._should_upsert_competitor_price(
+        "14133077",
+        product,
+        0.850,
+        source="pchome",
+    )
+
+    assert should_write is False
+    assert "stronger_existing" in reason
+    assert "existing_score=0.950" in reason
+    assert "incoming_score=0.850" in reason
+
+
 def test_competitor_feeder_records_unit_comparable_without_price_upsert(monkeypatch):
     from services.competitor_price_feeder import CompetitorPriceFeeder
     from services.pchome_crawler import PChomeProduct

tests/test_marketplace_product_matcher.py

@@ -2125,6 +2125,68 @@ def test_marketplace_matcher_promotes_lactacyd_private_wash_exact_line():
     assert "shared_identity_anchor" in diagnostics.reasons
 
 
+def test_marketplace_matcher_promotes_safe_aroma_lamps_to_total_price():
+    from services.marketplace_product_matcher import score_marketplace_match
+
+    hooome = score_marketplace_match(
+        "【HOOOME】白色 經典款香氛蠟燭暖燈 薰香燈 蠟燭燈 香氛燈 燭燈 融蠟燈(可調光+附兩顆燈泡+精美禮盒包裝 /",
+        "【HOOOME】白色 經典款香氛蠟燭暖燈 蠟燭燈 香氛燈 融燭燈 融蠟燈 (可調光+附兩顆燈泡+禮盒包裝)",
+        momo_price=1290,
+        competitor_price=1290,
+    )
+    gdesign = score_marketplace_match(
+        "【Gdesign】全新升級 Aroma Lava 解憂放鬆緩緩燈2.0｜熔岩燈、精油擴香 2 in 1(熔岩燈/精油/氛圍燈)",
+        "全新升級 Aroma Lava 解憂放鬆緩緩燈2.0｜熔岩燈、精油擴香 2 in 1(熔岩燈/精油/氛圍燈)",
+        momo_price=1980,
+        competitor_price=1980,
+    )
+
+    for diagnostics in (hooome, gdesign):
+        assert diagnostics.score >= 0.76
+        assert diagnostics.hard_veto is False
+        assert diagnostics.match_type == "exact"
+        assert diagnostics.price_basis == "total_price"
+        assert diagnostics.alert_tier == "price_alert_exact"
+
+    assert "focused_exact_identity_hooome_classic_white_wax_lamp_bulbs_giftbox" in hooome.reasons
+
+
+def test_marketplace_matcher_keeps_recipe_box_child_polish_catalog_in_review():
+    from services.marketplace_product_matcher import score_marketplace_match
+
+    diagnostics = score_marketplace_match(
+        "韓國 recipebox 可撕式水性兒童指甲油(兒童水性指甲油 可撕式指甲油 韓兔指甲油)",
+        "【Recipe Box】韓國 recipebox 可撕式水性兒童指甲油",
+        momo_price=299,
+        competitor_price=299,
+    )
+
+    assert diagnostics.score >= 0.76
+    assert diagnostics.hard_veto is False
+    assert diagnostics.price_basis == "manual_review"
+    assert diagnostics.alert_tier == "identity_review"
+    assert "variant_selection_review" in diagnostics.reasons
+    assert "focused_exact_identity_recipe_box_peelable_child_polish_catalog" in diagnostics.reasons
+
+
+def test_marketplace_matcher_keeps_aroma_and_nail_variant_gaps_blocked():
+    from services.marketplace_product_matcher import score_marketplace_match
+
+    pavaruni_cross_type = score_marketplace_match(
+        "【Pavaruni】美國香氛蠟燭20種香味500g(大豆蠟燭禮盒/天然植物精油/花香/木質香/果香)",
+        "【美國Pavaruni】天然植物香氛精油40種香味10ml 多款任選",
+    )
+    dashing_cross_style = score_marketplace_match(
+        "【DASHING DIVA】MAGICPRESS 時尚潮流美甲片-輕躍裸粉(貓眼 漸層)",
+        "Dashing Diva/F 時尚潮流美甲片-流光裸色 30片/盒 MDF5F015CG",
+    )
+
+    assert pavaruni_cross_type.hard_veto is True
+    assert pavaruni_cross_type.score < 0.76
+    assert dashing_cross_style.score < 0.76
+    assert "variant_descriptor_conflict" in dashing_cross_style.reasons
+
+
 def test_marketplace_matcher_promotes_eaoron_classic_tone_up_cream_exact_line():
     from services.marketplace_product_matcher import score_marketplace_match