|
| 1 | +import { anthropic } from "@ai-sdk/anthropic"; |
| 2 | +import { prisma } from "@dub/prisma"; |
| 3 | +import { Category } from "@dub/prisma/client"; |
| 4 | +import FireCrawlApp from "@mendable/firecrawl-js"; |
| 5 | +import { generateObject } from "ai"; |
| 6 | +import "dotenv-flow/config"; |
| 7 | +import { z } from "zod"; |
| 8 | + |
| 9 | +const CategoryEnum = z.nativeEnum(Category); |
| 10 | + |
| 11 | +// AI response schema |
| 12 | +const categorizationSchema = z.object({ |
| 13 | + categories: z.array(CategoryEnum).min(1).max(3), |
| 14 | + reasoning: z.string(), |
| 15 | +}); |
| 16 | + |
| 17 | +// Result interface |
| 18 | +interface ProgramResult { |
| 19 | + programId: string; |
| 20 | + programName: string; |
| 21 | + categories: Category[]; |
| 22 | + url?: string; |
| 23 | + error?: string; |
| 24 | +} |
| 25 | + |
| 26 | +if (!process.env.FIRECRAWL_API_KEY) |
| 27 | + throw new Error("FIRECRAWL_API_KEY is not set"); |
| 28 | + |
| 29 | +// Initialize FireCrawl |
| 30 | +const firecrawl = new FireCrawlApp({ |
| 31 | + apiKey: process.env.FIRECRAWL_API_KEY, |
| 32 | +}); |
| 33 | + |
| 34 | +// Function to scrape website content |
| 35 | +async function scrapeWebsite(url: string) { |
| 36 | + try { |
| 37 | + const scrapeResult = await firecrawl.scrapeUrl(url, { |
| 38 | + formats: ["markdown"], |
| 39 | + onlyMainContent: true, |
| 40 | + parsePDF: false, |
| 41 | + maxAge: 14400000, // 4 hours cache |
| 42 | + excludeTags: ["img"], |
| 43 | + }); |
| 44 | + |
| 45 | + if (!scrapeResult.success) { |
| 46 | + throw new Error(scrapeResult.error || "Failed to scrape"); |
| 47 | + } |
| 48 | + |
| 49 | + return { |
| 50 | + content: scrapeResult.markdown || "", |
| 51 | + title: scrapeResult.metadata?.title || "", |
| 52 | + description: scrapeResult.metadata?.description || "", |
| 53 | + }; |
| 54 | + } catch (error) { |
| 55 | + console.error(`Error scraping ${url}:`, error); |
| 56 | + return null; |
| 57 | + } |
| 58 | +} |
| 59 | + |
| 60 | +// Function to categorize content using AI |
| 61 | +async function categorizeProgram( |
| 62 | + programName: string, |
| 63 | + url: string, |
| 64 | + content: string, |
| 65 | + title: string, |
| 66 | + description: string, |
| 67 | +) { |
| 68 | + try { |
| 69 | + const prompt = `Analyze this website and categorize it into 1-3 most relevant categories. |
| 70 | +
|
| 71 | +IMPORTANT: You must select categories from this EXACT list (case-sensitive): |
| 72 | +- Artificial_Intelligence |
| 73 | +- Development |
| 74 | +- Design |
| 75 | +- Productivity |
| 76 | +- Finance |
| 77 | +- Marketing |
| 78 | +- Ecommerce |
| 79 | +- Security |
| 80 | +- Education |
| 81 | +- Health |
| 82 | +- Consumer |
| 83 | +
|
| 84 | +Category descriptions: |
| 85 | +- Artificial_Intelligence: AI/ML tools, chatbots, automation, machine learning platforms |
| 86 | +- Development: Code tools, APIs, developer platforms, programming resources |
| 87 | +- Design: Design tools, UI/UX, creative software, graphics |
| 88 | +- Productivity: Task management, collaboration, workflow tools, organization |
| 89 | +- Finance: Financial services, payments, accounting, investment, banking |
| 90 | +- Marketing: Marketing tools, analytics, advertising, social media, SEO |
| 91 | +- Ecommerce: Online stores, commerce platforms, marketplaces, retail |
| 92 | +- Security: Cybersecurity, privacy, protection tools, data security |
| 93 | +- Education: Learning platforms, courses, educational content, training |
| 94 | +- Health: Healthcare, fitness, wellness apps, medical services |
| 95 | +- Consumer: General consumer products/services that don't fit other categories |
| 96 | +
|
| 97 | +CRITICAL: Only use the exact category names listed above. DO NOT create new categories or modify existing ones. Do not select "Entrepreneurship" or "Business" as a category. |
| 98 | +
|
| 99 | +Website information: |
| 100 | +Name: ${programName} |
| 101 | +Website URL: ${url} |
| 102 | +Page Title: ${title} |
| 103 | +Meta Description: ${description} |
| 104 | +Website Content Preview: ${content.slice(0, 300)}...`; |
| 105 | + |
| 106 | + const { object } = await generateObject({ |
| 107 | + model: anthropic("claude-sonnet-4-20250514"), |
| 108 | + schema: categorizationSchema, |
| 109 | + prompt, |
| 110 | + }); |
| 111 | + |
| 112 | + return object.categories; |
| 113 | + } catch (error) { |
| 114 | + // console.error(`Error categorizing ${programName}:`, error); |
| 115 | + |
| 116 | + // If it's a validation error (invalid enum values), return empty array |
| 117 | + if ( |
| 118 | + error?.name === "AI_NoObjectGeneratedError" || |
| 119 | + error?.cause?.name === "AI_TypeValidationError" |
| 120 | + ) { |
| 121 | + console.log( |
| 122 | + ` Invalid categories returned for ${programName}, skipping categorization`, |
| 123 | + ); |
| 124 | + return []; // Return empty array for invalid categories |
| 125 | + } |
| 126 | + |
| 127 | + return []; // Return empty array for any other errors too |
| 128 | + } |
| 129 | +} |
| 130 | + |
| 131 | +// Main processing function |
| 132 | +async function main() { |
| 133 | + const program = await prisma.program.findUniqueOrThrow({ |
| 134 | + where: { |
| 135 | + slug: "", |
| 136 | + url: { |
| 137 | + not: null, |
| 138 | + }, |
| 139 | + }, |
| 140 | + select: { |
| 141 | + id: true, |
| 142 | + name: true, |
| 143 | + url: true, |
| 144 | + }, |
| 145 | + }); |
| 146 | + |
| 147 | + // Scrape website |
| 148 | + console.log(`Scraping: ${program.url}...`); |
| 149 | + |
| 150 | + const scraped = await scrapeWebsite(program.url!); // already filtered above |
| 151 | + |
| 152 | + if (!scraped) { |
| 153 | + throw new Error("Failed to scrape website"); |
| 154 | + } |
| 155 | + |
| 156 | + console.log(`Description: ${scraped.description}`); |
| 157 | + |
| 158 | + // Categorize with AI |
| 159 | + console.log(`Analyzing content...`); |
| 160 | + const categories = await categorizeProgram( |
| 161 | + program.name, |
| 162 | + program.url!, // already filtered above |
| 163 | + scraped.content, |
| 164 | + scraped.title, |
| 165 | + scraped.description, |
| 166 | + ); |
| 167 | + |
| 168 | + console.log( |
| 169 | + `Categories: ${categories.length > 0 ? categories.join(", ") : "None (invalid/failed categorization)"}`, |
| 170 | + ); |
| 171 | + |
| 172 | + const res = await prisma.program.update({ |
| 173 | + where: { id: program.id }, |
| 174 | + data: { |
| 175 | + addedToMarketplaceAt: new Date(), |
| 176 | + description: scraped.description, |
| 177 | + categories: { |
| 178 | + deleteMany: {}, |
| 179 | + create: categories.map((category) => ({ category })), |
| 180 | + }, |
| 181 | + }, |
| 182 | + }); |
| 183 | + |
| 184 | + console.log( |
| 185 | + `Added ${res.name} to the marketplace with description: ${res.description}`, |
| 186 | + ); |
| 187 | +} |
| 188 | + |
| 189 | +main() |
| 190 | + .catch((error) => { |
| 191 | + console.error("Script failed:", error); |
| 192 | + process.exit(1); |
| 193 | + }) |
| 194 | + .finally(async () => { |
| 195 | + await prisma.$disconnect(); |
| 196 | + }); |
0 commit comments