V10.391 handle catalog variant listings

services/competitor_price_feeder.py

@@ -55,7 +55,6 @@ BROWSE_SH_MAX_EXECUTIONS_PER_RUN = int(os.getenv("PCHOME_FEEDER_BROWSE_SH_MAX_PE
 BROWSE_SH_OUTPUT_PREVIEW_CHARS = int(os.getenv("PCHOME_FEEDER_BROWSE_SH_OUTPUT_PREVIEW_CHARS", "1200"))
 RECOVERABLE_DIAGNOSTIC_REASONS = {
     "strong_product_line_match",
-    "strong_exact_spec_match",
     "shared_identity_anchor",
     "shared_identity_anchor_no_spec",
     "shared_identity_anchor_packaging_variant",
@@ -93,6 +92,15 @@ def _has_recoverable_identity_signal(diagnostics) -> bool:
     )
 
 
+def _is_multi_variant_listing_name(name: str) -> bool:
+    return bool(
+        re.search(
+            r"(多款任選|多款可選|多色任選|多色可選|多種香味|多種香氣|香味任選|香味可選|味道可選)",
+            name or "",
+        )
+    )
+
+
 def _classify_low_score_attempt(score: float, diagnostics) -> str:
     if getattr(diagnostics, "hard_veto", False):
         return "identity_veto"
@@ -117,10 +125,11 @@ def _has_variant_selection_gap(
 
     source_identity = parse_product_identity(momo_name)
     source_options = set(_explicit_variant_option_tokens(source_identity))
+    source_multi_listing = _is_multi_variant_listing_name(momo_name)
     if re.search(r"任選\s*[一二兩三四五六七八九十0-9]+\s*款", momo_name):
         source_options -= {str(value) for value in range(1, 11)}
         source_options -= {f"{value:02d}" for value in range(1, 11)}
-    if source_options:
+    if source_options and not source_multi_listing:
         return False
 
     threshold = max(best_score - 0.02, RECOVERABLE_LOW_SCORE_FLOOR)
@@ -128,7 +137,10 @@ def _has_variant_selection_gap(
     for product, score, diagnostics in ranked_matches[:5]:
         if getattr(diagnostics, "hard_veto", False) or score < threshold:
             continue
-        candidate_identity = parse_product_identity(getattr(product, "name", "") or "")
+        candidate_name = getattr(product, "name", "") or ""
+        if source_multi_listing != _is_multi_variant_listing_name(candidate_name):
+            return True
+        candidate_identity = parse_product_identity(candidate_name)
         options = _explicit_variant_option_tokens(candidate_identity)
         if len(options) >= 2:
             return True

services/marketplace_product_matcher.py

@@ -454,6 +454,9 @@ VARIANT_SENSITIVE_KEYWORDS = {
     "車用香氛",
     "香味",
     "私密清潔慕斯",
+    "私密清潔凝露",
+    "私密潔淨凝露",
+    "私密淨白清潔凝露",
     "定妝噴霧",
     "妝前防護乳",
     "妝前乳",
@@ -492,6 +495,8 @@ VARIANT_OPTION_COLOR_WORDS = {
     "金縷梅",
     "柔焦霧面",
     "水光亮面",
+    "菸鹼醯胺",
+    "胺基酸",
     "黑色",
     "棕色",
     "咖啡色",
@@ -534,6 +539,20 @@ VARIANT_DESCRIPTOR_NOISE_KEYWORDS = {
     "盒組",
 }
 
+MULTI_VARIANT_LISTING_PHRASES = (
+    "多款任選",
+    "多款可選",
+    "多色任選",
+    "多色可選",
+    "多種香味",
+    "多種香氣",
+    "香味任選",
+    "香味可選",
+    "味道可選",
+    "任選",
+    "可選",
+)
+
 SEARCH_AMBIGUOUS_PRODUCT_TERMS = {
     "保護膜",
     "保護貼",
@@ -2009,6 +2028,18 @@ def score_marketplace_match(
     ):
         score += 0.04
         reasons.append("shared_model_token_brush_baby_wildones")
+    if (
+        _has_catalog_variant_listing_alignment(left, right)
+        and brand_score >= 0.95
+        and not hard_veto
+        and price_penalty == 0
+        and spec_score >= 0.85
+        and type_score >= 0.95
+        and sequence_score >= 0.50
+        and not variant_descriptor_conflict
+    ):
+        score += 0.06
+        reasons.append("catalog_variant_listing_alignment")
     if (
         shared_anchor
         and len(shared_anchor.replace(" ", "")) >= 5
@@ -2390,6 +2421,37 @@ def _has_brush_baby_wildones_toothbrush_alignment(left: ProductIdentity, right:
     )
 
 
+def _is_relove_private_cleanser_line(left: ProductIdentity, right: ProductIdentity) -> bool:
+    return (
+        "relove" in (left.brand_tokens | right.brand_tokens)
+        and "私密" in left.searchable_name
+        and "私密" in right.searchable_name
+        and "凝露" in left.searchable_name
+        and "凝露" in right.searchable_name
+    )
+
+
+def _is_multi_variant_catalog_listing(identity: ProductIdentity) -> bool:
+    text = identity.normalized_name
+    return any(phrase in text for phrase in MULTI_VARIANT_LISTING_PHRASES)
+
+
+def _has_catalog_variant_listing_alignment(left: ProductIdentity, right: ProductIdentity) -> bool:
+    if not (_is_multi_variant_catalog_listing(left) and _is_multi_variant_catalog_listing(right)):
+        return False
+    if left.product_type != right.product_type or left.product_type not in {"精油", "護唇膏"}:
+        return False
+    if not _has_overlapping_base_spec(left, right):
+        return False
+    shared_core = left.core_tokens & right.core_tokens
+    if shared_core:
+        return True
+    left_text = left.searchable_name
+    right_text = right.searchable_name
+    catalog_terms = ("香氛擴香罐", "香氛蠟燭", "蠟燭", "擴香罐", "修護唇膏")
+    return any(term in left_text and term in right_text for term in catalog_terms)
+
+
 def _is_variant_sensitive_identity(
     left: ProductIdentity,
     right: ProductIdentity,
@@ -2412,6 +2474,8 @@ def _has_variant_descriptor_conflict(left: ProductIdentity, right: ProductIdenti
         return False
     if _has_brush_baby_wildones_toothbrush_alignment(left, right):
         return False
+    if _is_relove_private_cleanser_line(left, right):
+        return False
     if (
         shared_anchor
         and shared_anchor not in SEARCH_BROAD_ANCHORS