修正 PChome rescore 重複入隊判斷
This commit is contained in:
OoO
@@ -4,6 +4,7 @@
|
||||
================================================================================
|
||||
|
||||
【已完成】
|
||||
- V10.466 修正 rescore audit duplicate 判斷：只在「最新 attempt 已是同候選 `rescore_accepted_current`」時跳過；若歷史曾 accepted、但後續 crawler 又追加低信心列，允許重新 materialize，避免 Dashboard latest-state 仍停在 `true_low_confidence`。Production pilot 已將 SKU `14756069`、`11159042`、`13842560`、`8394210`、`15192547`、`10509765`、`10603780` 送入人工覆核隊列；只寫 `competitor_match_attempts`，`competitor_prices` / `competitor_price_history` 未變。
|
||||
- V10.465 修正 embedding fallback-disabled 控制流：`allow_111_fallback=False` 時若 resolver 回 111，不再直接退出或只試單台 GCP-B，會強制改試尚未嘗試的 GCP-A/GCP-B；背景 embedding 仍不落 111。
|
||||
- V10.464 補 rescore audit 精準 SKU pilot：`audit_competitor_match_attempt_rescore.py --sku` 可只掃指定 SKU，再搭配 `--apply-accepted` 只把通過新版 matcher 的目標 SKU 追加到 `rescore_accepted_current` 人工覆核隊列，不寫正式價格表。
|
||||
- V10.463 補 DR.WU / 達爾膚品牌 alias：同規格 `DR.WU 達爾膚` 與 `DR.WU` 候選不再被當成 brandless identity review，會以既有 exact_identity / total_price / price_alert_exact 閘門處理；未調整 `MIN_MATCH_SCORE`，保留 variant / hard veto 保護。
|
||||
|
||||
@@ -325,7 +325,7 @@ YOUTUBE_API_KEY = os.getenv('YOUTUBE_API_KEY', '')
|
||||
# ==========================================
|
||||
# 系統版本與路徑
|
||||
# ==========================================
|
||||
SYSTEM_VERSION = "V10.465"
|
||||
SYSTEM_VERSION = "V10.466"
|
||||
LOG_FILE_PATH = os.path.join(BASE_DIR, 'logs/system.log')
|
||||
public_url = PUBLIC_URL # 用於模板顯示
|
||||
|
||||
|
||||
@@ -1,8 +1,8 @@
|
||||
# MOMO PRO — AI 競價情報模組 Single Source of Truth
|
||||
|
||||
> **最後更新**: 2026-05-24 (台北時間)
|
||||
> **最後更新**: 2026-05-25 (台北時間)
|
||||
> **狀態**: 🟢 四 AI Agent 自動化閉環已落地；LLM 路由紅線升級為 Ollama-first 三主機級聯，Gemini 備援預設關閉
|
||||
> **適用版本**: V10.465
|
||||
> **適用版本**: V10.466
|
||||
|
||||
---
|
||||
|
||||
|
||||
@@ -53,6 +53,7 @@
|
||||
- 2026-05-25 08:30 CST 起，rescore audit 支援 `--sku` repeatable 精準篩選；production pilot 可只指定 3-10 個 SKU 執行 read-only audit 或 `--apply-accepted`，避免寬範圍掃描誤把不同 cohort 混在同一次驗證。
|
||||
- 2026-05-25 08:25 CST 起，`DR.WU / DR WU / DRWU / 達爾膚` 視為同一品牌 alias；正式樣本中的 DR.WU 玻尿酸保濕精華乳 50ML、2入組與杏仁酸亮白煥膚精華 18% 30ML 2入組，在不調整全域門檻下可由 brandless identity review 回到 exact total-price lane。
|
||||
- 2026-05-25 08:36 CST production pilot：SKU `10362820`、`10653216`、`10653329` 已從 `true_low_confidence` materialize 為 `rescore_accepted_current`，只進人工覆核隊列，不寫 `competitor_prices`。
|
||||
- 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。
|
||||
|
||||
## 3. 12 Agent 決策信封整合
|
||||
|
||||
|
||||
@@ -13,6 +13,7 @@
|
||||
## 📅 詳細更新日誌 (考古存檔)
|
||||
|
||||
### 2026-05-24：PChome 近門檻身份回收第二輪
|
||||
- **V10.466 Rescore latest-state duplicate 修正與 7 SKU pilot**: `materialize_rescore_accept_reviews()` 的 duplicate 判斷改看最新 attempt，而不是歷史任一 accepted；若後續 crawler 又把同 SKU/候選覆蓋成 `true_low_confidence`，可重新追加 `rescore_accepted_current` 讓 Dashboard latest-state 正確進人工覆核。Production pilot 已將 SKU `14756069`、`11159042`、`13842560`、`8394210`、`15192547`、`10509765`、`10603780` materialize 到人工覆核隊列；`competitor_prices` 目標計數維持 7、`competitor_price_history` 目標計數維持 210，未寫正式價差表。
|
||||
- **V10.465 Embedding GCP fallback 修正**: `OllamaService.generate_embedding(..., allow_111_fallback=False)` 若 resolver 因 unhealthy cache 回 111，會強制改試尚未嘗試的 GCP-A/GCP-B，不再直接 `break` 造成 `tried=[]` 或只試單台 GCP-B；背景 embedding 仍不允許落 111。
|
||||
- **V10.464 Rescore SKU pilot 篩選**: `audit_competitor_match_attempt_rescore.py` 與 `fetch_match_attempt_rescore_rows()` 增加 `--sku` / `skus` 篩選，可針對 DR.WU 這類明確 cohort 做 3-10 筆精準 materialize，不必為了 pilot 掃整批 `true_low_confidence`。
|
||||
- **V10.463 DR.WU / 達爾膚品牌 alias**: `marketplace_product_matcher` 補 `DR.WU / DR WU / DRWU / 達爾膚` 正規化，讓正式樣本中同規格玻尿酸保濕精華乳、杏仁酸亮白煥膚精華不再因品牌 token 不同被降成 brandless identity review；測試鎖住 exact / total_price / price_alert_exact。
|
||||
|
||||
@@ -231,20 +231,22 @@ def _ensure_attempt_table(conn) -> None:
|
||||
|
||||
def _already_materialized(conn, *, source: str, sku: str, candidate_id: str) -> bool:
|
||||
row = conn.execute(text("""
|
||||
SELECT 1
|
||||
SELECT attempt_status, COALESCE(best_competitor_product_id, '') AS candidate_id
|
||||
FROM competitor_match_attempts
|
||||
WHERE sku = :sku
|
||||
AND source = :source
|
||||
AND attempt_status = :attempt_status
|
||||
AND COALESCE(best_competitor_product_id, '') = :candidate_id
|
||||
ORDER BY attempted_at DESC, id DESC
|
||||
LIMIT 1
|
||||
"""), {
|
||||
"sku": sku,
|
||||
"source": source,
|
||||
"attempt_status": RESCORE_ACCEPTED_CURRENT_STATUS,
|
||||
"candidate_id": candidate_id,
|
||||
}).first()
|
||||
return row is not None
|
||||
if row is None:
|
||||
return False
|
||||
return (
|
||||
row.attempt_status == RESCORE_ACCEPTED_CURRENT_STATUS
|
||||
and row.candidate_id == candidate_id
|
||||
)
|
||||
|
||||
|
||||
def materialize_rescore_accept_reviews(
|
||||
|
||||
@@ -210,6 +210,55 @@ def test_match_attempt_rescore_materializes_accepted_current_for_manual_review()
|
||||
assert "matcher_rescore=accepted_current" in stored[0]["error_message"]
|
||||
|
||||
|
||||
def test_match_attempt_rescore_materialize_allows_requeue_when_latest_is_low_confidence():
|
||||
from services.competitor_match_attempt_rescore_audit import materialize_rescore_accept_reviews
|
||||
|
||||
engine = create_engine("sqlite:///:memory:")
|
||||
rows = [{
|
||||
"sku": "10509765",
|
||||
"attempt_status": "true_low_confidence",
|
||||
"momo_product_id": 10509765,
|
||||
"momo_product_name": "【悠斯晶】經典乳霜120g(2入組)",
|
||||
"momo_price": 599,
|
||||
"candidate_count": 1,
|
||||
"best_competitor_product_id": "YUSKIN-120G-2",
|
||||
"best_competitor_product_name": "【Yuskin悠斯晶】經典乳霜 2盒組(120g/盒)",
|
||||
"best_competitor_price": 540,
|
||||
"best_match_score": 0.779,
|
||||
}]
|
||||
|
||||
with engine.begin() as conn:
|
||||
initial_stats = materialize_rescore_accept_reviews(conn, rows)
|
||||
duplicate_stats = materialize_rescore_accept_reviews(conn, rows)
|
||||
conn.execute(text("""
|
||||
INSERT INTO competitor_match_attempts
|
||||
(sku, source, attempt_status, momo_product_id, momo_product_name,
|
||||
momo_price, candidate_count, best_competitor_product_id,
|
||||
best_competitor_product_name, best_competitor_price,
|
||||
best_match_score, diagnostic_codes, error_message, attempted_at)
|
||||
VALUES
|
||||
('10509765', 'pchome', 'true_low_confidence', 10509765,
|
||||
'【悠斯晶】經典乳霜120g(2入組)', 599, 1, 'YUSKIN-120G-2',
|
||||
'【Yuskin悠斯晶】經典乳霜 2盒組(120g/盒)', 540, 0.779,
|
||||
'["strong_exact_spec_match"]', 'later crawler low-confidence row',
|
||||
CURRENT_TIMESTAMP)
|
||||
"""))
|
||||
requeue_stats = materialize_rescore_accept_reviews(conn, rows)
|
||||
latest_status = conn.execute(text("""
|
||||
SELECT attempt_status
|
||||
FROM competitor_match_attempts
|
||||
WHERE sku = '10509765'
|
||||
ORDER BY attempted_at DESC, id DESC
|
||||
LIMIT 1
|
||||
""")).scalar_one()
|
||||
|
||||
assert initial_stats["materialized"] == 1
|
||||
assert duplicate_stats["skipped_duplicate"] == 1
|
||||
assert requeue_stats["materialized"] == 1
|
||||
assert requeue_stats["skipped_duplicate"] == 0
|
||||
assert latest_status == "rescore_accepted_current"
|
||||
|
||||
|
||||
def test_match_attempt_rescore_retracts_variant_review_from_accepted_queue():
|
||||
from services.competitor_match_attempt_rescore_audit import (
|
||||
fetch_variant_rescore_accept_review_rows,
|
||||
|
||||
Reference in New Issue
Block a user