V10.398 recover true low confidence matches

services/marketplace_product_matcher.py

@@ -352,6 +352,16 @@ SEARCH_IDENTITY_ANCHORS = (
     "麗駐蘭修復舒緩面膜",
     "濕度數顯智能加濕器",
     "新艷陽夏水離子熱防禦隔離露",
+    "3d極細防水眼線膠筆",
+    "恆久完美透膚煙染腮紅",
+    "私密植萃美白緊緻凝露",
+    "學習刷牙漱口水",
+    "羅馬柱智慧居家車用香氛機",
+    "經典菲菲染唇液",
+    "染唇液",
+    "晨霧純精油擴香儀",
+    "天然植物香氛精油",
+    "爆水擦澡濕巾",
     "3d立體持色眉彩盤",
     "細芯睛彩雙頭眉筆",
     "雙頭旋轉極細眉筆",
@@ -598,6 +608,7 @@ BRAND_ALIAS_OVERRIDES = {
     "jo malone": ("jo malone",),
     "prada": ("prada", "普拉達"),
     "za": ("za",),
+    "hh": ("hh", "草本新淨界"),
     "小浪": ("小浪",),
     "xiaomi": ("小米有品", "小米", "xiaomi"),
     "mac": ("m.a.c", "mac", "m a c"),
@@ -1649,6 +1660,7 @@ def _classify_match_quality(
     shared_anchor: str,
     shared_models: set[str],
     catalog_count_omission: bool,
+    multi_component_pair: bool,
 ) -> tuple[str, str, str]:
     """Map raw matcher scores into operator-facing price comparison lanes."""
     reason_set = set(reasons)
@@ -1670,6 +1682,8 @@ def _classify_match_quality(
         and (direct_spec_evidence or (shared_anchor and token_score >= 0.62 and sequence_score >= 0.58))
     )
     if strong_identity_evidence and not catalog_count_omission:
+        if multi_component_pair:
+            return "exact", "manual_review", "identity_review"
         return "exact", "total_price", "price_alert_exact"
 
     if score >= 0.76:
@@ -2288,6 +2302,7 @@ def score_marketplace_match(
         shared_anchor=shared_anchor,
         shared_models=shared_models,
         catalog_count_omission=catalog_count_omission,
+        multi_component_pair=_has_multi_component(left) and _has_multi_component(right),
     )
     evidence_flags = _build_evidence_flags(
         brand_score=brand_score,
@@ -2571,6 +2586,20 @@ def _has_baan_baby_lip_catalog_alignment(left: ProductIdentity, right: ProductId
     )
 
 
+def _has_pavaruni_40_scent_oil_alignment(left: ProductIdentity, right: ProductIdentity) -> bool:
+    left_text = left.searchable_name
+    right_text = right.searchable_name
+    return (
+        "pavaruni" in (left.brand_tokens & right.brand_tokens)
+        and "天然植物" in f"{left_text} {right_text}"
+        and "精油" in left_text
+        and "精油" in right_text
+        and _has_shared_volume(left, right, 10)
+        and ("40香味" in left_text or "40種香味" in left_text)
+        and ("40香味" in right_text or "40種香味" in right_text)
+    )
+
+
 def _has_shared_count(left: ProductIdentity, right: ProductIdentity, count: int, unit: str) -> bool:
     return (count, unit) in set(left.counts) and (count, unit) in set(right.counts)
 
@@ -2677,6 +2706,63 @@ def _has_focused_low_score_exact_identity_line(left: ProductIdentity, right: Pro
         and "隔離露" in right_text
     ):
         return "shiseido_blue_sunscreen"
+    if (
+        {"baan", "貝恩"} & (left.brand_tokens & right.brand_tokens)
+        and "嬰兒修護唇膏" in left_text
+        and "嬰兒修護唇膏" in right_text
+        and left.product_type == right.product_type == "護唇膏"
+    ):
+        return "baan_baby_lip_base_catalog"
+    if (
+        {"shu uemura", "植村秀"} & (left.brand_tokens & right.brand_tokens)
+        and "3d極細防水眼線膠筆" in left_text
+        and "3d極細防水眼線膠筆" in right_text
+    ):
+        return "shu_3d_eyeliner"
+    if (
+        {"ysl", "聖羅蘭"} & (left.brand_tokens & right.brand_tokens)
+        and "恆久完美透膚煙染腮紅" in left_text
+        and "恆久完美透膚煙染腮紅" in right_text
+    ):
+        return "ysl_blush_catalog"
+    if (
+        {"hh", "草本新淨界"} & (left.brand_tokens & right.brand_tokens)
+        and "私密植萃美白緊緻凝露" in left_text
+        and "私密植萃美白緊緻凝露" in right_text
+        and _has_shared_volume(left, right, 30)
+    ):
+        return "hh_private_gel"
+    if (
+        {"lab52", "齒妍堂"} & (left.brand_tokens & right.brand_tokens)
+        and "學習刷牙漱口水" in left_text
+        and "學習刷牙漱口水" in right_text
+        and _has_overlapping_base_spec(left, right)
+    ):
+        return "lab52_mouthwash"
+    if (
+        "benefit" in (left.brand_tokens & right.brand_tokens)
+        and "染唇液" in left_text
+        and "染唇液" in right_text
+        and "唇頰兩用" in pair_text
+    ):
+        return "benefit_lip_tint"
+    if (
+        {"herb24", "草本"} & (left.brand_tokens & right.brand_tokens)
+        and "晨霧純精油擴香儀ii" in left_text
+        and "晨霧純精油擴香儀ii" in right_text
+        and (("霧黑" in left_text and "黑色" in right_text) or ("霧黑" in right_text and "黑色" in left_text))
+    ):
+        return "herb24_mist_diffuser_black"
+    if _has_pavaruni_40_scent_oil_alignment(left, right):
+        return "pavaruni_40_scent_oil"
+    if (
+        "gatsby" in (left.brand_tokens & right.brand_tokens)
+        and "爆水擦澡濕巾" in left_text
+        and "爆水擦澡濕巾" in right_text
+        and "24張入" in left_text
+        and "24張入" in right_text
+    ):
+        return "gatsby_body_wipes_24"
     return ""
 
 
@@ -2735,6 +2821,8 @@ def _has_variant_descriptor_conflict(left: ProductIdentity, right: ProductIdenti
         return False
     if _has_baan_baby_lip_catalog_alignment(left, right):
         return False
+    if _has_pavaruni_40_scent_oil_alignment(left, right):
+        return False
     if _is_relove_private_cleanser_line(left, right):
         return False
     if (