tt-help-cli-ycl 1.4.17 → 1.4.19

package/src/watch/tag-service.js

@@ -150,6 +150,7 @@ export function buildDiscoverPrompt(
   hardRules = [],
   existingCount = 0,
   forceNicheSwitch = null,
+  lastRoundFeedback = null,
 ) {
   const lang = getLang(country);
   const langNames = {
@@ -167,8 +168,285 @@ export function buildDiscoverPrompt(
   };
   const langName = langNames[lang] || lang;
 
-  // 正样本：该国高分 tag（带分数，让 LLM 知道哪些方向最有效）
-  const productive = history.productive || [];
+  // ====== 上轮反馈：告知 LLM 上一轮生成了什么、哪些被去重 ======
+  let lastRoundRejectionHint = "";
+  if (lastRoundFeedback) {
+    if (lastRoundFeedback.allRejected) {
+      lastRoundRejectionHint = `\n=== ⛔ CRITICAL: Last round ALL tags were DUPLICATES ===\nYou generated: ${lastRoundFeedback.raw.join(", ")}\nALL of these already exist in our database — NONE were added.\nYou MUST generate COMPLETELY DIFFERENT tags this round. Do NOT repeat any of the above.\nTry a totally different product category, use-case, or selling context.`;
+    } else if (lastRoundFeedback.rejected.length > 0) {
+      lastRoundRejectionHint = `\n=== ⚠️ Last round feedback ===\nYou generated: ${lastRoundFeedback.raw.join(", ")}\nRejected (already exist): ${lastRoundFeedback.rejected.join(", ")}\nAdded: ${lastRoundFeedback.survived.join(", ") || "(none)"}\nAvoid the rejected tags above — they already exist in our database.`;
+    }
+  }
+
+  // 正样本：该国高分 tag。当 forceNicheSwitch 激活时，仅展示目标 niche 的 productive tag
+  let productive = history.productive || [];
+  if (forceNicheSwitch && productive.length > 0) {
+    const targetNiche = forceNicheSwitch.to;
+    const NICHE_KEYWORDS = {
+      fashion: [
+        "clothes",
+        "kleidung",
+        "kleding",
+        "vestiti",
+        "ropa",
+        "outfit",
+        "mode",
+        "shoes",
+        "sneakers",
+        "zapatos",
+        "chaussures",
+        "vintage",
+        "secondhand",
+        "thrift",
+        "moda",
+        "scarpe",
+        "schuhe",
+        "schoenen",
+        "cipő",
+        "boty",
+        "tenisky",
+        "sapatos",
+        "calzado",
+      ],
+      food: [
+        "food",
+        "bakery",
+        "brot",
+        "brood",
+        "pane",
+        "comida",
+        "restaurant",
+        "cafe",
+        "kitchen",
+        "konyha",
+        "patisserie",
+        "boulangerie",
+        "wypieki",
+        "etterm",
+        "essen",
+        "eten",
+        "jedzenie",
+        "étel",
+        "cibo",
+        "chleb",
+        "bread",
+        "pekar",
+        "bageri",
+        "forno",
+        "padaria",
+      ],
+      handmade_crafts: [
+        "handmade",
+        "handwerk",
+        "handgemaakt",
+        "artigianato",
+        "hechoamano",
+        "kézműves",
+        "kreativ",
+        "craft",
+        "diy",
+        "knutselen",
+        "ceramic",
+        "keramiek",
+        "kerzen",
+        "gyertya",
+        "resin",
+        "macrame",
+        "pottery",
+        "weaving",
+        "fattoamano",
+        "handgemacht",
+        "rękodzieło",
+        "ručně",
+      ],
+      home_living: [
+        "home",
+        "wohnung",
+        "huis",
+        "casa",
+        "lakberendezes",
+        "interior",
+        "deco",
+        "decoration",
+        "einrichten",
+        "furniture",
+        "möbel",
+        "meubilair",
+        "epipla",
+        "butor",
+        "nabytek",
+        "tapic",
+        "reinigung",
+        "cleaning",
+        "muebles",
+        "mobili",
+        "meble",
+        "bútor",
+        "dom",
+        "otthon",
+        "thuis",
+        "zuhause",
+      ],
+      electronics_tech: [
+        "tech",
+        "elektronik",
+        "electronica",
+        "gadget",
+        "phone",
+        "smartphone",
+        "laptop",
+        "computer",
+        "audio",
+        "headphones",
+        "charger",
+        "usb",
+        "playstation",
+        "gaming",
+        "telefon",
+        "handy",
+        "tablet",
+        "pc",
+        "console",
+        "smartwatch",
+        "cuffie",
+        "auricolari",
+        "caricatore",
+      ],
+      pets_animals: [
+        "pet",
+        "hund",
+        "dog",
+        "cat",
+        "kat",
+        "kuty",
+        "konia",
+        "pets",
+        "tier",
+        "dieren",
+        "animale",
+        "animal",
+        "mascota",
+        "animais",
+        "zwierzę",
+        "állat",
+        "cane",
+        "gatto",
+        "perro",
+        "chien",
+        "chat",
+      ],
+      beauty_health: [
+        "beauty",
+        "kosmetik",
+        "cosmetica",
+        "beaute",
+        "belleza",
+        "szepseg",
+        "hair",
+        "haj",
+        "skincare",
+        "makeup",
+        "parfum",
+        "perfume",
+        "wellness",
+        "gesundheit",
+        "salud",
+        "sante",
+        "szépség",
+        "krása",
+      ],
+      books_collectibles: [
+        "book",
+        "buch",
+        "boek",
+        "libro",
+        "knihy",
+        "konyv",
+        "collectible",
+        "sammlung",
+        "verzameling",
+        "collezionismo",
+        "coleccion",
+        "antique",
+        "antiquariat",
+        "vintage",
+        "vinyl",
+        "schallplatten",
+        "pokemon",
+        "kaarten",
+        "moedas",
+        "mince",
+        "livre",
+        "książka",
+        "carte",
+      ],
+      services_professional: [
+        "service",
+        "dienst",
+        "servizio",
+        "servicio",
+        "szolgaltatas",
+        "handwerker",
+        "craftsman",
+        "repair",
+        "reparatur",
+        "reparatie",
+        "riparazione",
+        "reparacion",
+        "cleaning",
+        "reinigung",
+        "moving",
+        "umzug",
+        "verhuur",
+        "mudanca",
+        "florist",
+        "photography",
+        "fotograf",
+        "usługa",
+        "szolgáltatás",
+        "serviço",
+      ],
+      marketplace_local: [
+        "market",
+        "markt",
+        "mercato",
+        "mercado",
+        "piac",
+        "trh",
+        "flohmarkt",
+        "rommelmarkt",
+        "brocante",
+        "bazar",
+        "mercadillo",
+        "feira",
+        "local",
+        "lokal",
+        "locale",
+        "winkel",
+        "boutique",
+        "shop",
+        "tienda",
+        "negozio",
+        "bol",
+        "webshop",
+        "eshop",
+        "sklep",
+        "obchod",
+        "bolt",
+        "butik",
+      ],
+    };
+    const keywords = NICHE_KEYWORDS[targetNiche] || [];
+    const nicheFiltered = productive.filter((t) =>
+      keywords.some((kw) => t.tag.toLowerCase().includes(kw.toLowerCase())),
+    );
+    if (nicheFiltered.length > 0) {
+      console.error(
+        `[tag-service] 🎯 niche-filtered productive: ${productive.length} → ${nicheFiltered.length} (target: ${targetNiche})`,
+      );
+      productive = nicheFiltered;
+    }
+  }
   const productiveHint =
     productive.length > 0
       ? `\nTags that already worked well for ${country} (with scores): ${productive
@@ -335,7 +613,7 @@ ${hardRules.map((r) => `  ${r.ruleType === "hard" ? "⛔" : "💡"} ${r.ruleText
     : ""
 }
 
-${productiveHint}${deadHint}${repeatedDeadHint}${langPatternHint}${errorHint}${userHint}${strategyReview}${healthHint}${nicheHint}${forceNicheSwitchHint}
+${lastRoundRejectionHint}${productiveHint}${deadHint}${repeatedDeadHint}${langPatternHint}${errorHint}${userHint}${strategyReview}${healthHint}${nicheHint}${forceNicheSwitchHint}
 
 ## Quality check before responding
 
@@ -517,17 +795,44 @@ export async function discoverTagsForCountry(
   // 读取历史打分记录
   const productive = store.getTagsByCountry(country, 50);
   const dead = store.getDeadTags(country);
-  // 全局去重：获取所有已存在的 tag 名（不限国家、不限状态），避免跨国家重复入库
-  const { getAllExistingTagNames } = await import("./db-tags.js");
-  const allExisting = getAllExistingTagNames();
+
+  // ====== P0修复: 清理 truly dead + 分类未分类 dead，释放去重空间 ======
+  const { classifyDeadTags, cleanupTrulyDeadTags: cleanupTrulyDead } =
+    await import("./tag-selflearn.js");
+  const classifyResult = classifyDeadTags();
+  if (classifyResult.classified > 0) {
+    console.error(
+      `[tag-service] 🧹 ${country} 分类了 ${classifyResult.classified} 个 dead tag (not_exist=${classifyResult.not_exist}, no_match=${classifyResult.no_match}, wrong_country=${classifyResult.wrong_country})`,
+    );
+  }
+  const cleanupResult = cleanupTrulyDead(2);
+  if (cleanupResult.deleted > 0) {
+    console.error(
+      `[tag-service] 🗑️ ${country} 清理了 ${cleanupResult.deleted} 个确认不存在的 dead tag，释放去重空间`,
+    );
+  }
+
+  // 全局去重：仅用 productive + scored 的活跃 tag 名做去重（排除 dead）
+  // dead tag 同名重新生成时 insertTag 会将其重置为 new
+  const { getActiveTagNames } = await import("./db-tags.js");
+  const allExisting = getActiveTagNames();
   console.error(
-    `[tag-service] ${country} 全局去重集合: ${allExisting.size} 个已存在 tag`,
+    `[tag-service] ${country} 活跃去重集合: ${allExisting.size} 个 (productive+scored, 排除dead)`,
   );
 
   // 获取最近生成的标签（用于检测重复生成）
-  const { getRecentGeneratedTags } = await import("./db-tags.js");
+  const { getRecentGeneratedTags, getLastRoundAllGeneratedTags } =
+    await import("./db-tags.js");
   const recentGenerated = getRecentGeneratedTags(country, 10);
 
+  // ====== 上轮反馈：告知 LLM 上一轮全部被去重的情况 ======
+  const lastRoundFeedback = getLastRoundAllGeneratedTags(country);
+  if (lastRoundFeedback?.allRejected) {
+    console.error(
+      `[tag-service] 🔁 ${country} 上轮全部被去重: ${lastRoundFeedback.raw.join(", ")}`,
+    );
+  }
+
   const history = { productive, dead, recentGenerated };
 
   // 读取历史策略效果（自学习）
@@ -542,7 +847,7 @@ export async function discoverTagsForCountry(
     );
   }
 
-  // 组装 prompt（加入健康指标、niche 推荐、硬规则）
+  // 组装 prompt（加入健康指标、niche 推荐、硬规则、上轮反馈）
   const prompt = buildDiscoverPrompt(
     country,
     count,
@@ -554,6 +859,7 @@ export async function discoverTagsForCountry(
     hardRules,
     allExisting.size,
     forceNicheSwitch,
+    lastRoundFeedback,
   );
   console.error(
     `[tag-service] LLM discover ${country} round=${round} (lang=${getLang(country)}, count=${count}, temp=${temperature}, health=${health?.healthStatus || "unknown"})`,
@@ -591,12 +897,13 @@ export async function discoverTagsForCountry(
     if (result.inserted) inserted.push(result.tag);
   }
 
-  // 记录 discover log（策略 + 结果 + 健康状态 + niche）
+  // 记录 discover log（策略 + 结果 + 健康状态 + niche + 原始 tag）
   const { insertDiscoverLog } = await import("./db-tags.js");
   insertDiscoverLog({
     country,
     round,
     strategy,
+    tagsRaw: deduped,
     tagsGenerated: unique,
     tagsAdded: inserted.length,
     productiveSample: productive.slice(0, 20).map((t) => t.tag),