busi488energy/prisma/seed.ts

import { PrismaPg } from '@prisma/adapter-pg';
import { randomUUID } from 'crypto';
import { readFileSync } from 'fs';
import { resolve } from 'path';
import { z } from 'zod/v4';
import { PrismaClient } from '../src/generated/prisma/client.js';

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

const PointGeometrySchema = z.object({
  type: z.literal('Point'),
  coordinates: z.tuple([z.number(), z.number()]),
});

const MultiPolygonGeometrySchema = z.object({
  type: z.literal('MultiPolygon'),
  coordinates: z.array(z.array(z.array(z.tuple([z.number(), z.number()])))),
});

const RegionPropertiesSchema = z.object({
  name: z.string(),
  code: z.string(),
  iso: z.string(),
});

const DatacenterPropertiesSchema = z.object({
  name: z.string(),
  operator: z.string(),
  capacity_mw: z.number(),
  status: z.string(),
  year_opened: z.number(),
  region: z.string(),
});

const RegionFeatureSchema = z.object({
  type: z.literal('Feature'),
  geometry: MultiPolygonGeometrySchema,
  properties: RegionPropertiesSchema,
});

const DatacenterFeatureSchema = z.object({
  type: z.literal('Feature'),
  geometry: PointGeometrySchema,
  properties: DatacenterPropertiesSchema,
});

const RegionCollectionSchema = z.object({
  type: z.literal('FeatureCollection'),
  features: z.array(RegionFeatureSchema),
});

const DatacenterCollectionSchema = z.object({
  type: z.literal('FeatureCollection'),
  features: z.array(DatacenterFeatureSchema),
});

const PowerPlantPropertiesSchema = z.object({
  Plant_Name: z.string(),
  Plant_Code: z.number(),
  Utility_Na: z.string(),
  State: z.string(),
  PrimSource: z.string(),
  Total_MW: z.number(),
});

const PowerPlantFeatureSchema = z.object({
  type: z.literal('Feature'),
  geometry: PointGeometrySchema,
  properties: PowerPlantPropertiesSchema,
});

const PowerPlantCollectionSchema = z.object({
  type: z.literal('FeatureCollection'),
  features: z.array(PowerPlantFeatureSchema),
});

const AIMilestoneSchema = z.object({
  date: z.string(),
  title: z.string(),
  description: z.string(),
  category: z.string(),
});

function readAndParse<T>(relativePath: string, schema: z.ZodType<T>): T {
  const fullPath = resolve(import.meta.dirname, '..', relativePath);
  const raw: unknown = JSON.parse(readFileSync(fullPath, 'utf-8'));
  return schema.parse(raw);
}

async function seedGridRegions() {
  console.log('Seeding grid regions...');

  const geojson = readAndParse('data/grid-regions.geojson', RegionCollectionSchema);

  // Delete only static seed data that doesn't have time-series FK references
  await prisma.$executeRawUnsafe('DELETE FROM datacenters');

  // Upsert grid_regions by code to preserve FK references from time-series tables
  for (const feature of geojson.features) {
    const id = randomUUID();
    const geojsonStr = JSON.stringify(feature.geometry);

    await prisma.$executeRawUnsafe(
      `INSERT INTO grid_regions (id, name, code, iso, boundary, created_at)
       VALUES ($1::uuid, $2, $3, $4, ST_GeomFromGeoJSON($5)::geography, NOW())
       ON CONFLICT (code) DO UPDATE SET
         name = EXCLUDED.name,
         iso = EXCLUDED.iso,
         boundary = EXCLUDED.boundary`,
      id,
      feature.properties.name,
      feature.properties.code,
      feature.properties.iso,
      geojsonStr,
    );

    console.log(`  Upserted region: ${feature.properties.code}`);
  }
}

async function seedDatacenters() {
  console.log('Seeding datacenters...');

  const geojson = readAndParse('data/datacenters.geojson', DatacenterCollectionSchema);

  // Get all region codes from DB
  const regions = await prisma.$queryRawUnsafe<Array<{ id: string; code: string }>>(
    'SELECT id, code FROM grid_regions',
  );
  const regionMap = new Map(regions.map(r => [r.code, r.id]));

  let inserted = 0;
  let skipped = 0;

  for (const feature of geojson.features) {
    const props = feature.properties;
    const regionId = regionMap.get(props.region);

    if (!regionId) {
      // Try to find the region by spatial containment
      const [lng, lat] = feature.geometry.coordinates;
      const spatialMatch = await prisma.$queryRawUnsafe<Array<{ id: string; code: string }>>(
        `SELECT id, code FROM grid_regions
         WHERE ST_Contains(boundary::geometry, ST_SetSRID(ST_MakePoint($1, $2), 4326))
         LIMIT 1`,
        lng,
        lat,
      );

      if (spatialMatch.length > 0) {
        const match = spatialMatch[0]!;
        const id = randomUUID();
        const geojsonStr = JSON.stringify(feature.geometry);

        await prisma.$executeRawUnsafe(
          `INSERT INTO datacenters (id, name, operator, location, capacity_mw, status, year_opened, region_id, created_at)
           VALUES ($1::uuid, $2, $3, ST_GeomFromGeoJSON($4)::geography, $5, $6, $7, $8::uuid, NOW())`,
          id,
          props.name,
          props.operator,
          geojsonStr,
          props.capacity_mw,
          props.status,
          props.year_opened,
          match.id,
        );
        console.log(`  Inserted DC: ${props.name} (spatial match -> ${match.code})`);
        inserted++;
      } else {
        console.log(`  Skipped DC: ${props.name} (no matching region for "${props.region}")`);
        skipped++;
      }
      continue;
    }

    const id = randomUUID();
    const geojsonStr = JSON.stringify(feature.geometry);

    await prisma.$executeRawUnsafe(
      `INSERT INTO datacenters (id, name, operator, location, capacity_mw, status, year_opened, region_id, created_at)
       VALUES ($1::uuid, $2, $3, ST_GeomFromGeoJSON($4)::geography, $5, $6, $7, $8::uuid, NOW())`,
      id,
      props.name,
      props.operator,
      geojsonStr,
      props.capacity_mw,
      props.status,
      props.year_opened,
      regionId,
    );
    console.log(`  Inserted DC: ${props.name} (${props.region})`);
    inserted++;
  }

  console.log(`  Total: ${inserted.toString()} inserted, ${skipped.toString()} skipped`);
}

/** Normalize ArcGIS PrimSource strings to consistent capitalized fuel types. */
function normalizeFuelType(primSource: string): string {
  const lower = primSource.toLowerCase().trim();
  if (lower === 'natural gas') return 'Natural Gas';
  if (lower === 'coal') return 'Coal';
  if (lower === 'nuclear') return 'Nuclear';
  if (lower.includes('hydro') || lower === 'hydroelectric conventional' || lower === 'pumped storage')
    return 'Hydroelectric';
  if (lower === 'wind') return 'Wind';
  if (lower === 'solar') return 'Solar';
  if (lower.includes('petroleum') || lower === 'petroleum') return 'Petroleum';
  if (lower.includes('biomass') || lower === 'wood' || lower === 'wood and wood derived fuels') return 'Biomass';
  if (lower === 'geothermal') return 'Geothermal';
  return 'Other';
}

async function seedPowerPlants() {
  console.log('Seeding power plants...');

  const geojson = readAndParse('data/power-plants.geojson', PowerPlantCollectionSchema);

  let upserted = 0;
  let skipped = 0;

  for (const feature of geojson.features) {
    const props = feature.properties;
    const [lng, lat] = feature.geometry.coordinates;

    // Skip features with invalid coordinates
    if (!lng || !lat || lat === 0 || lng === 0) {
      skipped++;
      continue;
    }

    const id = randomUUID();
    const fuelType = normalizeFuelType(props.PrimSource);

    await prisma.$executeRawUnsafe(
      `INSERT INTO power_plants (id, plant_code, name, operator, location, capacity_mw, fuel_type, state, created_at)
       VALUES ($1::uuid, $2, $3, $4, ST_SetSRID(ST_MakePoint($5, $6), 4326)::geography, $7, $8, $9, NOW())
       ON CONFLICT (plant_code) DO UPDATE SET
         name = EXCLUDED.name,
         operator = EXCLUDED.operator,
         location = EXCLUDED.location,
         capacity_mw = EXCLUDED.capacity_mw,
         fuel_type = EXCLUDED.fuel_type,
         state = EXCLUDED.state`,
      id,
      props.Plant_Code,
      props.Plant_Name,
      props.Utility_Na,
      lng,
      lat,
      props.Total_MW,
      fuelType,
      props.State,
    );
    upserted++;
  }

  console.log(`  Total: ${upserted.toString()} upserted, ${skipped.toString()} skipped`);
}

function validateAIMilestones() {
  console.log('Validating AI milestones...');
  const milestones = readAndParse('data/ai-milestones.json', z.array(AIMilestoneSchema));
  console.log(`  Valid JSON with ${milestones.length.toString()} milestones`);

  for (const m of milestones) {
    if (isNaN(Date.parse(m.date))) {
      throw new Error(`Invalid date in milestone: ${m.date}`);
    }
  }
  console.log('  All milestones valid');
}

async function main() {
  console.log('Starting seed...\n');

  await seedGridRegions();
  console.log('');

  await seedDatacenters();
  console.log('');

  await seedPowerPlants();
  console.log('');

  validateAIMilestones();

  // Print summary
  console.log('\n--- Seed Summary ---');
  const regionCount = await prisma.$queryRawUnsafe<Array<{ count: bigint }>>(
    'SELECT count(*) as count FROM grid_regions',
  );
  const dcCount = await prisma.$queryRawUnsafe<Array<{ count: bigint }>>('SELECT count(*) as count FROM datacenters');
  const ppCount = await prisma.$queryRawUnsafe<Array<{ count: bigint }>>('SELECT count(*) as count FROM power_plants');
  console.log(`Grid regions: ${regionCount[0]!.count.toString()}`);
  console.log(`Datacenters: ${dcCount[0]!.count.toString()}`);
  console.log(`Power plants: ${ppCount[0]!.count.toString()}`);

  // Show sample spatial data
  const sample = await prisma.$queryRawUnsafe<Array<{ name: string; location_text: string }>>(
    'SELECT name, ST_AsText(location) as location_text FROM datacenters LIMIT 3',
  );
  console.log('\nSample datacenter locations:');
  for (const s of sample) {
    console.log(`  ${s.name}: ${s.location_text}`);
  }

  console.log('\nSeed complete!');
}

main()
  .catch((e: unknown) => {
    console.error('Seed failed:', e);
    process.exit(1);
  })
  .finally(() => {
    void prisma.$disconnect();
  });