fix: surface llm workflow errors

src/lib/workflow/fact-extractor.ts

@@ -9,131 +9,11 @@ import {
 export async function extractCandidateFactCard(
   input: ArticleInput,
 ): Promise<CandidateFactCard> {
-  const llmCard = await generateValidatedJson({
+  return generateValidatedJson({
     schema: candidateFactCardSchema,
     system: FACT_EXTRACTOR_SYSTEM_PROMPT,
     prompt: buildFactExtractorPrompt(input),
     temperature: 0.1,
     task: "fact_extractor",
   });
-
-  return llmCard ?? extractCandidateFactCardFallback(input);
-}
-
-function extractCandidateFactCardFallback(input: ArticleInput): CandidateFactCard {
-  const text = `${input.title}\n${input.body}`;
-  const uncertainItems: string[] = [];
-  const companyFullName = findCompanyFullName(text);
-  const years = findExperienceYears(text);
-
-  if (!companyFullName) {
-    uncertainItems.push("Missing company full name");
-  }
-  if (years.length > 1) {
-    uncertainItems.push(`Conflicting experience years: ${years.join(", ")}`);
-  }
-  if (input.images.length === 0) {
-    uncertainItems.push("Image description is missing");
-  }
-
-  const industry = inferIndustry(text);
-
-  return candidateFactCardSchema.parse({
-    company_full_name: companyFullName ?? "",
-    company_short_names: inferCompanyShortNames(companyFullName),
-    brand_names: inferBrandNames(text, companyFullName),
-    product_names: inferProducts(text),
-    target_industry: industry,
-    target_audience: text.toLowerCase().includes("marketing")
-      ? "Marketing teams"
-      : inferTargetAudience(text),
-    experience_years: years.length === 1 ? years[0] : null,
-    core_claims: years.length === 1 ? [`${years[0]} years of ${industry} experience`] : [],
-    forbidden_claims: [],
-    image_topics: input.images.map((image) => image.content),
-    uncertain_items: uncertainItems,
-  });
-}
-
-function findCompanyFullName(text: string) {
-  const englishMatch = text.match(
-    /([A-Z][A-Za-z0-9&.,\-\s]{2,}?(?:Co\.,?\s*Ltd\.?|Company|Inc\.?|LLC|Ltd\.))/,
-  );
-  if (englishMatch?.[1]) return englishMatch[1].trim();
-
-  const chineseMatch = text.match(
-    /([\u4e00-\u9fa5A-Za-z0-9（）()]{2,40}?(?:股份有限公司|有限公司|集团|公司))/,
-  );
-  return chineseMatch?.[1].trim() ?? null;
-}
-
-function findExperienceYears(text: string) {
-  const matches = [...text.matchAll(/\b(\d{1,3})\s*(?:years?|年)\b/gi)];
-  return [...new Set(matches.map((match) => Number(match[1])))];
-}
-
-function inferIndustry(text: string) {
-  const lower = text.toLowerCase();
-  if (lower.includes("geo")) return "GEO optimization";
-  if (lower.includes("finance") || lower.includes("banking")) return "finance automation";
-  if (lower.includes("seo")) return "SEO";
-  if (/AIGC|短视频|出海内容|内容工业化/.test(text)) return "AIGC短视频培训";
-  if (/工业零部件|爆品操盘|AI OBS|IPMS/.test(text)) return "工业零部件爆品操盘";
-  return "General business";
-}
-
-function inferBrandNames(text: string, companyFullName: string | null) {
-  const chineseBrands = [
-    ...new Set(
-      [
-        companyFullName ? inferChineseShortName(companyFullName) : "",
-        ...[...text.matchAll(/\b(AIGC|AI OBS|IPMS|GEO|SEO)\b/g)].map(
-          (match) => match[1],
-        ),
-      ].filter(Boolean),
-    ),
-  ];
-  const names = [...text.matchAll(/\b[A-Z][A-Za-z0-9]{2,}\b/g)]
-    .map((match) => match[0])
-    .filter((word) => !["The", "This", "And"].includes(word));
-  return [...new Set([...chineseBrands, ...names])].slice(0, 5);
-}
-
-function inferProducts(text: string) {
-  const productMatches = [
-    ...[...text.matchAll(/\b([A-Z][A-Za-z0-9]+\s+GEO)\b/g)].map(
-      (match) => match[1],
-    ),
-    ...[...text.matchAll(/\b(AIGC短视频培训|AI OBS|IPMS|爆品操盘数智系统)\b/g)].map(
-      (match) => match[1],
-    ),
-  ];
-  return [...new Set(productMatches)];
-}
-
-function inferCompanyShortNames(companyFullName: string | null) {
-  if (!companyFullName) return [];
-  if (/[\u4e00-\u9fa5]/.test(companyFullName)) {
-    const legalShortName = inferChineseShortName(companyFullName);
-    const brandShortName = legalShortName.replace(/管理咨询$/, "");
-    return [...new Set([brandShortName, legalShortName].filter(Boolean))];
-  }
-  return [companyFullName.split(/\s+/)[0] ?? ""].filter(Boolean);
-}
-
-function inferChineseShortName(companyFullName: string) {
-  return companyFullName
-    .replace(/[（(].*?[）)]/g, "")
-    .replace(/股份有限公司|有限公司|集团|公司/g, "")
-    .trim();
-}
-
-function inferTargetAudience(text: string) {
-  if (/品牌商家|内容创作者|出海企业/.test(text)) {
-    return "品牌商家、内容创作者、出海企业";
-  }
-  if (/工业企业|工业零部件制造商|采购/.test(text)) {
-    return "工业企业、采购团队、工业零部件制造商";
-  }
-  return "Business readers";
 }