haixunMaster/lib/topic-anchor.ts

/** 主題錨點：標籤／帳號／海巡結果是否扣緊核心主題 */

const CONCEPT_PARTS = [
  "寵物洗毛精",
  "寵物沐浴",
  "寵物美容",
  "寵物用品",
  "貓咪洗澡",
  "狗狗洗澡",
  "洗毛精",
  "沐浴乳",
  "沐浴精",
  "洗髮精",
  "洗碗精",
  "寵物",
  "毛孩",
  "貓咪",
  "狗狗",
  "貓",
  "狗",
  "犬",
];

const CONCEPT_ALIASES: Record<string, string[]> = {
  寵物: ["寵物", "毛孩", "毛小孩", "毛寶", "貓咪", "狗狗", "貓", "狗", "犬"],
  洗毛精: ["洗毛精", "沐浴乳", "沐浴精", "洗澡", "沐浴", "梳毛"],
  沐浴: ["沐浴", "沐浴乳", "洗澡", "洗毛精"],
  洗澡: ["洗澡", "沐浴", "洗毛精"],
  貓咪: ["貓咪", "貓", "毛孩"],
  狗狗: ["狗狗", "狗", "犬", "毛孩"],
};

const COMPOUND_CONCEPTS: Record<string, string[]> = {
  寵物洗毛精: ["寵物", "洗毛精"],
  寵物沐浴: ["寵物", "沐浴"],
  寵物洗澡: ["寵物", "洗澡"],
  貓咪洗澡: ["貓咪", "洗澡"],
  狗狗洗澡: ["狗狗", "洗澡"],
};

/** 太寬泛、容易跑題的單字標籤 */
export const BROAD_TAG_BLOCKLIST = new Set([
  "寵物",
  "洗毛精",
  "沐浴",
  "貓",
  "狗",
  "犬",
  "貓咪",
  "狗狗",
  "毛孩",
  "洗碗精",
  "掉髮",
  "落髮",
  "語錄",
  "心情",
  "日常",
  "分享",
  "推薦",
  "求助",
  "請益",
]);

export interface TopicAnchorInput {
  label: string;
  query: string;
  brief?: string | null;
  pillars?: string[];
  suggestedTags?: string[];
  exclusions?: string[];
}

export interface TopicAnchor {
  corePhrase: string;
  requiredConcepts: string[];
  rejects: Array<{ test: (text: string) => boolean }>;
}

export function extractTopicConcepts(phrase: string): string[] {
  const normalized = phrase.replace(/\s+/g, "").trim();
  if (!normalized) return [];

  if (COMPOUND_CONCEPTS[normalized]) {
    return COMPOUND_CONCEPTS[normalized];
  }

  const found: string[] = [];
  let scratch = normalized;
  const parts = [...CONCEPT_PARTS].sort((a, b) => b.length - a.length);

  for (const part of parts) {
    if (scratch.includes(part)) {
      found.push(part);
      scratch = scratch.replace(part, "");
    }
  }

  const pruned = found.filter(
    (part) => !found.some((other) => other !== part && other.includes(part))
  );

  if (pruned.length === 1 && pruned[0].length >= 4) {
    const atoms = ["寵物", "毛孩", "貓咪", "狗狗", "貓", "狗", "洗毛精", "沐浴", "洗澡", "梳毛"];
    const inner = atoms.filter((a) => pruned[0].includes(a));
    if (inner.length >= 2) return [...new Set(inner)];
  }

  if (pruned.length >= 2) return [...new Set(pruned)];
  return [normalized];
}

function conceptMatched(text: string, concept: string): boolean {
  const aliases = CONCEPT_ALIASES[concept] ?? [concept];
  return aliases.some((alias) => text.includes(alias));
}

function buildRejectChecks(corePhrase: string, exclusions?: string[]): TopicAnchor["rejects"] {
  const rejects: TopicAnchor["rejects"] = [];

  if (/洗毛精|寵物沐浴|寵物洗澡/.test(corePhrase) && !/洗碗/.test(corePhrase)) {
    rejects.push({ test: (t) => /洗碗精|洗碗乳|碗盤清潔|餐具清潔|廚房清潔/.test(t) });
  }

  if (/寵物|毛孩|貓|狗|犬/.test(corePhrase)) {
    rejects.push({
      test: (t) =>
        /掉髮|落髮|植髮|生髮|禿頭|髮量減少|頭皮療程|人髮/.test(t) &&
        !/寵物|毛孩|貓|狗|犬|毛小孩/.test(t),
    });
  }

  if (/寵物/.test(corePhrase) && /洗毛精|沐浴|洗澡|梳毛/.test(corePhrase)) {
    rejects.push({
      test: (t) =>
        /洗髮精|護髮|頭皮屑|控油洗髮/.test(t) &&
        !/寵物|毛孩|貓|狗|犬|毛小孩|洗毛精/.test(t),
    });
  }

  for (const ex of exclusions ?? []) {
    const term = ex.trim();
    if (term.length >= 2) {
      rejects.push({ test: (t) => t.includes(term) });
    }
  }

  return rejects;
}

export function buildTopicAnchor(input: TopicAnchorInput): TopicAnchor {
  const corePhrase = (input.query || input.label).trim();
  return {
    corePhrase,
    requiredConcepts: extractTopicConcepts(corePhrase),
    rejects: buildRejectChecks(corePhrase, input.exclusions),
  };
}

/** 分數 ≥3 視為扣緊主題；-100 為明確跑題 */
export function scoreTopicRelevance(text: string, anchor: TopicAnchor): number {
  const t = text.trim();
  if (!t) return -10;

  for (const reject of anchor.rejects) {
    if (reject.test(t)) return -100;
  }

  const compact = t.replace(/\s+/g, "");
  const coreCompact = anchor.corePhrase.replace(/\s+/g, "");

  if (coreCompact.length >= 3 && compact.includes(coreCompact)) return 12;
  if (t.includes(anchor.corePhrase)) return 10;

  if (anchor.requiredConcepts.length >= 2) {
    const matched = anchor.requiredConcepts.filter((c) => conceptMatched(t, c));
    if (matched.length >= anchor.requiredConcepts.length) return 8;
    if (matched.length === 1) return -8;
    return -12;
  }

  if (anchor.corePhrase.length >= 3 && t.includes(anchor.corePhrase)) return 6;
  return -5;
}

export function isOnTopicText(
  text: string,
  input: TopicAnchorInput,
  minScore = 3
): boolean {
  return scoreTopicRelevance(text, buildTopicAnchor(input)) >= minScore;
}

export function isOnTopicTag(
  tag: string,
  input: TopicAnchorInput,
  options?: { allowBroad?: boolean }
): boolean {
  const clean = tag.replace(/^@/, "").trim();
  if (!clean) return false;
  if (!options?.allowBroad && BROAD_TAG_BLOCKLIST.has(clean)) return false;
  return isOnTopicText(clean, input, 3);
}

export function buildTopicAnchorPromptBlock(input: TopicAnchorInput): string {
  const anchor = buildTopicAnchor(input);
  return `【主題錨點】核心：${anchor.corePhrase}
${
  anchor.requiredConcepts.length >= 2
    ? `必備概念（需同時符合，不可只命中其一）：${anchor.requiredConcepts.join(" + ")}`
    : ""
}
常見跑題（務必排除）：只沾邊單字、易混淆產品（如洗碗精≠洗毛精）、人類掉髮≠寵物洗澡、泛用詞（心情/日常/分享）`;
}

/** 置入模式：貼文是否像「求推薦／求助／有痛點」 */
export function looksLikeRecommendationPost(text: string): boolean {
  return /推薦|求助|請益|請問|哪裡買|有沒有|求分享|困擾|煩惱|怎麼辦|怎麼選|不知道|有推|拜託|求救|卡關/.test(
    text
  );
}

/** 置入模式：是否有產品需求或困擾訊號（含未明說「求推薦」的痛點描述） */
export function hasPlacementIntent(text: string): boolean {
  if (looksLikeRecommendationPost(text)) return true;
  return /用什麼洗|哪款|哪牌|哪一牌|洗什麼|買什麼|在家洗|自己洗|洗澡怕|洗不乾淨|味道重|皮膚癢|皮膚紅|一直抓|掉毛多|敏感肌|過敏|紅腫|抓癢|異味|不敢洗|第一次洗|洗完還是|越洗越/.test(
    text
  );
}

/** 純閒聊／晒照，沒有可置入的需求訊號 */
export function looksLikeCasualChat(text: string): boolean {
  if (hasPlacementIntent(text)) return false;
  return /好可愛|太萌|晒照|日常分享|隨便發|廢文|路過|笑死|哈哈哈|哈囉|早安|晚安|按讚|追蹤我|純分享|沒有要問/.test(
    text
  );
}