digicraft-be/scripts/legacy/reconstruct_metadata.ts

import { PrismaClient, Prisma } from '@prisma/client';
import { GoogleGenAI, Type, Part } from "@google/genai";
import fs from 'fs';
import path from 'path';
import dotenv from 'dotenv';
import sharp from 'sharp'; // Correct top-level import

// Load environment variables manually if helpful, but usually preloaded
dotenv.config();

const prisma = new PrismaClient();
// Adjust API Key retrieval if needed
const API_KEY = process.env.GEMINI_API_KEY || process.env.API_KEY || "";
const ai = new GoogleGenAI({ apiKey: API_KEY });

const STORAGE_ROOT = path.join(process.cwd(), '..', 'storage');

async function reconstruct() {
    console.log("🚑 STARTING AI METADATA RECONSTRUCTION (v2)...");

    // We target "Recovered Project" in particular
    // Note: Project model does not have strategy column, so we rely on Niche check
    const recoveredProjects = await prisma.project.findMany({
        where: {
            OR: [
                { niche: { contains: "Recovered Project" } },
                { seoData: { is: null } }
            ]
        },
        include: {
            assets: true,
            seoData: true
        }
    });

    console.log(`🔍 Found ${recoveredProjects.length} projects needing reconstruction.`);

    let successCount = 0;

    for (const project of recoveredProjects) {
        try {
            // Find Master Asset
            const masterAsset = project.assets.find(a => a.type === 'master' || a.type === 'MASTER');
            if (!masterAsset) {
                // console.log(`⚠️ Skiping ${project.id}: No master asset.`);
                continue;
            }

            const imagePath = path.join(STORAGE_ROOT, masterAsset.path.replace('projects/', 'projects/'));
            // Correction: masterAsset.path is relative to storage root?
            // masterAsset.path usually is "projects/UUID/master/file.png"
            // STORAGE_ROOT is ".../storage"
            // Join safely.
            const fullPath = path.join(STORAGE_ROOT, ...masterAsset.path.split('/').slice(1));
            // Wait, if path is "projects/..." and STORAGE_ROOT has "storage", we need to check real fs path.
            // Let's try direct join first.
            let realPath = path.join(process.cwd(), '../storage', masterAsset.path);

            if (!fs.existsSync(realPath)) {
                // Try alternative path (sometimes assets stored with leading slash?)
                realPath = path.join(process.cwd(), '../storage', masterAsset.path.replace(/^projects\//, ''));
            }

            if (!fs.existsSync(realPath)) {
                console.log(`❌ Image not found on disk: ${realPath}`);
                continue;
            }

            // Read Image
            const rawBuffer = fs.readFileSync(realPath);

            // Resize if too large (Optimization)
            const resizedBuffer = await sharp(rawBuffer)
                .resize({ width: 1024, height: 1024, fit: 'inside' })
                .toBuffer();

            const base64Image = resizedBuffer.toString('base64');

            // CALL GEMINI
            console.log(`🤖 Analyzing ${project.id.substring(0, 8)}...`);

            // Construct Prompt Parts
            const parts: Part[] = [
                {
                    text: `
                    You are a Professional Market Analyst and SEO Specialist.
                    Analyze this image (which is a digital product for Etsy) to reconstruct its lost metadata.
                    
                    IMPORTANT:
                    - Keep "description" SHORT and CONCISE (max 50 words).
                    - Return exactly 13 "keywords".
                    - Do not hallucinate long text.
                    
                    Return ONLY a JSON object:
                    {
                        "niche": "specific niche string",
                        "productType": "Wall Art",
                        "title": "SEO Title",
                        "description": "Short Marketing Description",
                        "keywords": ["tag1", "tag2", "tag3", "tag4", "tag5", "tag6", "tag7", "tag8", "tag9", "tag10", "tag11", "tag12", "tag13"],
                        "creativity": "Balanced",
                        "jsonLd": { ...valid schema... }
                    }
                ` },
                { inlineData: { data: base64Image, mimeType: "image/png" } }
            ];

            const result = await ai.models.generateContent({
                model: "gemini-flash-latest", // 1.5 Flash (Confirmed available)
                contents: { parts },
                config: {
                    responseMimeType: "application/json",
                    responseSchema: {
                        type: Type.OBJECT,
                        properties: {
                            niche: { type: Type.STRING },
                            productType: { type: Type.STRING },
                            title: { type: Type.STRING },
                            description: { type: Type.STRING },
                            keywords: { type: Type.ARRAY, items: { type: Type.STRING } },
                            creativity: { type: Type.STRING },
                            jsonLd: {
                                type: Type.OBJECT,
                                properties: {
                                    "@context": { type: Type.STRING },
                                    "@type": { type: Type.STRING },
                                    name: { type: Type.STRING },
                                    description: { type: Type.STRING },
                                    image: { type: Type.STRING },
                                    brand: { type: Type.STRING }
                                }
                            }
                        }
                    }
                }
            });

            let responseText = result.text;
            if (!responseText) throw new Error("Empty AI Response");

            // Clean markdown fences
            responseText = responseText.replace(/```json/g, '').replace(/```/g, '').trim();

            let metadata;
            try {
                metadata = JSON.parse(responseText);
            } catch (jsonErr) {
                throw new Error("Failed to parse JSON: " + responseText.substring(0, 100));
            }

            // Defaults if missing
            const finalTitle = metadata.title || `Recovered Project ${project.niche.substring(0, 10)}`;
            const finalDescription = metadata.description || "Project metadata recovered from image analysis.";
            const finalKeywords = Array.isArray(metadata.keywords) ? metadata.keywords : ["Recovered"];

            // UDPATE DATABASE
            await prisma.project.update({
                where: { id: project.id },
                data: {
                    niche: metadata.niche || "Recovered Art",
                    productType: metadata.productType || "Wall Art"
                }
            });

            await prisma.seoData.upsert({
                where: { projectId: project.id },
                create: {
                    projectId: project.id,
                    title: finalTitle,
                    description: finalDescription,
                    keywords: JSON.stringify(finalKeywords),
                    jsonLd: JSON.stringify(metadata.jsonLd || {}),
                    printingGuide: "Standard",
                    suggestedPrice: "5.00"
                },
                update: {
                    title: finalTitle,
                    description: finalDescription,
                    keywords: JSON.stringify(finalKeywords),
                    jsonLd: JSON.stringify(metadata.jsonLd || {})
                }
            });

            process.stdout.write('+');
            successCount++;

        } catch (err: any) {
            console.error(`❌ Error ${project.id}:`, err.message);
            // Log preview if it was a parse error
            if (err.message.includes('JSON')) {
                console.log("Response Preview:", err.stack); // Stack is too noisy, but okay for now.
            }
        }
    }

    console.log(`\n🎉 RECONSTRUCTION COMPLETE. Fixed ${successCount} projects.`);
    await prisma.$disconnect();
}

reconstruct();