V10.584 強化 PChome Nick 比對窄門

services/marketplace_product_matcher.py

@@ -526,6 +526,7 @@ FOCUSED_IDENTITY_VARIANT_REVIEW_BYPASS_REASONS = {
     "ts6_lubricant_100g_3pack",
     "ts6_peach_bright_gel_45g_3pack",
     "ts6_white_wash_peach_gel_kit",
+    "ts6_cooling_clean_mousse_100g",
     "vaseline_baby_jelly_368g_3pack",
 }
 
@@ -601,7 +602,10 @@ FOCUSED_IDENTITY_TOTAL_PRICE_REASONS = {
     "ts6_lubricant_100g_3pack",
     "ts6_peach_bright_gel_45g_3pack",
     "ts6_white_wash_peach_gel_kit",
+    "ts6_cooling_clean_mousse_100g",
     "vaseline_baby_jelly_368g_3pack",
+    "nivea_creme_100ml",
+    "schick_womens_sensitive_blade_3pack",
 }
 
 SEARCH_BROAD_ANCHORS = {
@@ -1665,6 +1669,15 @@ def _has_paulas_choice_body_lotion_210ml_2pack_alignment(
     return has_two_pack(left) and has_two_pack(right)
 
 
+def _has_nivea_creme_100ml_alignment(left: ProductIdentity, right: ProductIdentity) -> bool:
+    brand_tokens = left.brand_tokens | right.brand_tokens
+    if not ({"nivea", "妮維雅"} & brand_tokens):
+        return False
+    if not _has_shared_volume(left, right, 100):
+        return False
+    return all("妮維雅霜" in item.searchable_name and "隨身版" in item.searchable_name for item in (left, right))
+
+
 def _has_refill_pack(identity: ProductIdentity) -> bool:
     text = identity.normalized_name
     return bool(
@@ -2332,8 +2345,11 @@ def score_marketplace_match(
     if brand_reason:
         reasons.append(brand_reason)
     reasons.extend(spec_reasons)
-    if left.product_type and right.product_type and left.product_type != right.product_type:
+    nivea_creme_100ml_alignment = _has_nivea_creme_100ml_alignment(left, right)
+    if left.product_type and right.product_type and left.product_type != right.product_type and not nivea_creme_100ml_alignment:
         reasons.append("type_conflict")
+    if nivea_creme_100ml_alignment:
+        reasons.append("nivea_creme_100ml_type_alignment")
     model_line_conflict = _has_model_line_conflict(left, right)
     if model_line_conflict:
         reasons.append("model_line_conflict")
@@ -2548,7 +2564,7 @@ def score_marketplace_match(
         hard_veto = True
     if chinese_name_score < 0.16 and token_score < 0.72:
         hard_veto = True
-    if left.product_type and right.product_type and left.product_type != right.product_type:
+    if left.product_type and right.product_type and left.product_type != right.product_type and not nivea_creme_100ml_alignment:
         hard_veto = True
     if sun_protection_line_conflict:
         hard_veto = True
@@ -4562,6 +4578,14 @@ def _has_focused_low_score_exact_identity_line(left: ProductIdentity, right: Pro
         and _has_exact_count_alignment(left, right)
     ):
         return "ts6_white_mousse_180g_3pack"
+    if (
+        {"ts6", "護一生"} & (left.brand_tokens & right.brand_tokens)
+        and "沁涼潔淨慕斯" in left_text
+        and "沁涼潔淨慕斯" in right_text
+        and _has_shared_weight(left, right, 100)
+        and _has_exact_count_alignment(left, right)
+    ):
+        return "ts6_cooling_clean_mousse_100g"
     if (
         {"ts6", "護一生"} & (left.brand_tokens & right.brand_tokens)
         and "蜜愛潤滑液" in left_text
@@ -4659,6 +4683,15 @@ def _has_focused_low_score_exact_identity_line(left: ProductIdentity, right: Pro
         and _has_shared_volume(left, right, 473)
     ):
         return "cetaphil_long_lotion_473ml"
+    if (
+        {"nivea", "妮維雅"} & (left.brand_tokens & right.brand_tokens)
+        and "妮維雅霜" in left_text
+        and "妮維雅霜" in right_text
+        and "隨身版" in left_text
+        and "隨身版" in right_text
+        and _has_shared_volume(left, right, 100)
+    ):
+        return "nivea_creme_100ml"
     if (
         "nailmatic" in (left.brand_tokens & right.brand_tokens)
         and "小精靈" in left_text
@@ -4801,6 +4834,17 @@ def _has_focused_low_score_exact_identity_line(left: ProductIdentity, right: Pro
         and "唇頰兩用" in pair_text
     ):
         return "benefit_lip_tint"
+    if (
+        {"schick", "舒適牌"} & (left.brand_tokens & right.brand_tokens)
+        and "舒綺" in left_text
+        and "舒綺" in right_text
+        and "除毛刀片" in left_text
+        and "除毛刀片" in right_text
+        and "敏感肌" in left_text
+        and "敏感肌" in right_text
+        and _has_shared_count(left, right, 3, "入")
+    ):
+        return "schick_womens_sensitive_blade_3pack"
     if (
         {"herb24", "草本"} & (left.brand_tokens & right.brand_tokens)
         and "晨霧純精油擴香儀ii" in left_text

services/pchome_crawler.py

@@ -11,6 +11,7 @@ API 參考:
 - 圖片 URL: https://img.pchome.com.tw/cs{Pic.B}
 """
 
+import html
 import re
 import time
 import logging
@@ -49,14 +50,34 @@ class PChomeProduct:
         return data
 
 
+def _clean_subtitle_text(value: str) -> str:
+    """Normalize PChome Nick text before using it as identity evidence."""
+    text = html.unescape(str(value or ""))
+    text = re.sub(r"<\s*br\s*/?\s*>", " ", text, flags=re.I)
+    text = re.sub(r"<[^>]+>", " ", text)
+    text = re.sub(r"[★☆][^★☆]{0,80}[★☆]", " ", text)
+    return re.sub(r"\s+", " ", text).strip()
+
+
+def _compact_identity_text(value: str) -> str:
+    return re.sub(r"[^0-9a-zA-Z\u4e00-\u9fff]+", "", str(value or "").lower())
+
+
 def _build_match_name(name: str, subtitle: str) -> str:
     """Build an identity-rich title without duplicating the PChome display name."""
     display_name = str(name or '').strip()
-    nick = str(subtitle or '').strip()
+    nick = _clean_subtitle_text(subtitle)
     if not nick or nick == display_name:
         return display_name
-    if display_name and nick.startswith(display_name):
+    display_compact = _compact_identity_text(display_name)
+    nick_compact = _compact_identity_text(nick)
+    if display_name and (
+        nick.startswith(display_name)
+        or (display_compact and nick_compact.startswith(display_compact))
+    ):
         return nick
+    if display_compact and display_compact == nick_compact:
+        return display_name
     return f"{display_name} {nick}".strip()
 
 
@@ -348,7 +369,7 @@ class PChomeCrawler:
             image_url = f"{self.IMAGE_BASE_URL}{pic_path}" if pic_path else ''
 
             name = data.get('Name', '') or ''
-            subtitle = data.get('Nick', '') or ''
+            subtitle = _clean_subtitle_text(data.get('Nick', '') or '')
 
             return PChomeProduct(
                 product_id=product_id,