hp-prod-tracker/scripts/backfill-embeddings.ts

/**
 * Backfill Embeddings Script
 *
 * One-time script to generate embeddings for all existing projects and deliverables.
 * Run with: npx tsx scripts/backfill-embeddings.ts
 *
 * Prerequisites:
 * - PostgreSQL with pgvector extension enabled
 * - Ollama running with nomic-embed-text model pulled
 * - DATABASE_URL set in environment
 */

import "dotenv/config";
import { PrismaPg } from "@prisma/adapter-pg";
import { PrismaClient } from "../src/generated/prisma/client";
import {
  buildProjectText,
  buildDeliverableText,
  generateEmbedding,
  checkOllamaHealth,
} from "../src/lib/services/embedding-service";

const connectionString = process.env.DATABASE_URL!;
const adapter = new PrismaPg({ connectionString });
const prisma = new PrismaClient({ adapter });

async function backfillProjects() {
  const projects = await prisma.project.findMany({
    include: {
      deliverables: {
        select: { name: true, status: true, priority: true },
      },
    },
  });

  console.log(`\n📦 Processing ${projects.length} projects...`);
  let success = 0;
  let failed = 0;

  for (const project of projects) {
    try {
      const text = buildProjectText(project);
      const embedding = await generateEmbedding(text);

      if (embedding) {
        const vectorStr = `[${embedding.join(",")}]`;
        await prisma.$executeRawUnsafe(
          `UPDATE "projects" SET "embedding" = $1::vector WHERE "id" = $2`,
          vectorStr,
          project.id
        );
        success++;
        process.stdout.write(
          `\r  ✅ Projects: ${success} done, ${failed} failed (${success + failed}/${projects.length})`
        );
      } else {
        failed++;
        console.warn(`\n  ⚠️  Failed to generate embedding for project: ${project.name}`);
      }
    } catch (error) {
      failed++;
      console.error(
        `\n  ❌ Error processing project "${project.name}":`,
        error instanceof Error ? error.message : error
      );
    }
  }

  console.log(`\n  Projects complete: ${success} succeeded, ${failed} failed`);
}

async function backfillDeliverables() {
  const deliverables = await prisma.deliverable.findMany({
    include: {
      project: { select: { name: true, projectCode: true } },
      stages: {
        include: { template: { select: { name: true } } },
      },
    },
  });

  console.log(`\n📋 Processing ${deliverables.length} deliverables...`);
  let success = 0;
  let failed = 0;

  for (const deliverable of deliverables) {
    try {
      const text = buildDeliverableText({
        ...deliverable,
        stages: deliverable.stages.map((s) => ({
          template: s.template,
          status: s.status,
        })),
      });
      const embedding = await generateEmbedding(text);

      if (embedding) {
        const vectorStr = `[${embedding.join(",")}]`;
        await prisma.$executeRawUnsafe(
          `UPDATE "deliverables" SET "embedding" = $1::vector WHERE "id" = $2`,
          vectorStr,
          deliverable.id
        );
        success++;
        process.stdout.write(
          `\r  ✅ Deliverables: ${success} done, ${failed} failed (${success + failed}/${deliverables.length})`
        );
      } else {
        failed++;
        console.warn(
          `\n  ⚠️  Failed to generate embedding for deliverable: ${deliverable.name}`
        );
      }
    } catch (error) {
      failed++;
      console.error(
        `\n  ❌ Error processing deliverable "${deliverable.name}":`,
        error instanceof Error ? error.message : error
      );
    }
  }

  console.log(
    `\n  Deliverables complete: ${success} succeeded, ${failed} failed`
  );
}

async function main() {
  console.log("🔍 Backfill Embeddings — HP CG Production Tracker");
  console.log("═".repeat(50));

  // Check Ollama health first
  console.log("\n🏥 Checking Ollama health...");
  const health = await checkOllamaHealth();

  if (!health.available) {
    console.error(
      "❌ Ollama is not available. Make sure it's running at",
      process.env.OLLAMA_HOST || "http://localhost:11434"
    );
    console.error(
      "   Start it with: ollama serve (or docker run -p 11434:11434 ollama/ollama)"
    );
    process.exit(1);
  }

  console.log(`✅ Ollama is running. Available models: ${health.models.join(", ")}`);

  const embedModel = process.env.OLLAMA_EMBED_MODEL || "nomic-embed-text";
  const hasModel = health.models.some((m) => m.includes(embedModel.split(":")[0]));

  if (!hasModel) {
    console.error(
      `❌ Embedding model "${embedModel}" not found. Pull it with: ollama pull ${embedModel}`
    );
    process.exit(1);
  }

  console.log(`✅ Using embedding model: ${embedModel}`);

  const startTime = Date.now();

  await backfillProjects();
  await backfillDeliverables();

  const elapsed = ((Date.now() - startTime) / 1000).toFixed(1);
  console.log(`\n${"═".repeat(50)}`);
  console.log(`✨ Backfill complete in ${elapsed}s`);

  await prisma.$disconnect();
}

main().catch((error) => {
  console.error("\n💥 Fatal error:", error);
  prisma.$disconnect();
  process.exit(1);
});