goodgo-platform/scripts/sync-osm-industrial-parks.ts

/**
 * OSM → goodgo Industrial Park bulk sync (PR 2/4 of OSM-sync project).
 *
 *   Usage:
 *     NODE_OPTIONS="-r dotenv/config" DOTENV_CONFIG_PATH=.env \
 *       pnpm tsx scripts/sync-osm-industrial-parks.ts [--dry-run] [--chunk=north]
 *
 *   Flags:
 *     --dry-run        Fetch from Overpass + show counts, don't write to DB.
 *     --chunk=NAME     Process only one chunk (north|northCentral|southCentral|south).
 *                      Default: process all four.
 *
 * Strategy:
 *   • Vietnam is split into 4 horizontal bbox slices (Overpass times out
 *     on the whole country at once).
 *   • For each chunk we ask Overpass for `landuse=industrial` ways +
 *     relations + nodes within the bbox, with inline geometry.
 *   • osmtogeojson normalises the OSM JSON into a GeoJSON FeatureCollection.
 *   • For each Feature we upsert a row keyed on `osmId`. The schema
 *     guarantees uniqueness via the constraint added in PR 1.
 *   • PostGIS columns (`location` Point + `boundary` MultiPolygon) are
 *     written via raw SQL because Prisma cannot manage Geometry types.
 *   • Nodes get `location` only — no boundary.
 *   • Conflict resolution: respect `osmLocked` and `lockedFields` from PR 1.
 */
import 'dotenv/config';
import { createId } from '@paralleldrive/cuid2';
import { PrismaPg } from '@prisma/adapter-pg';
import { type Prisma, PrismaClient } from '@prisma/client';
import area from '@turf/area';
import centroid from '@turf/centroid';
import type { Feature, MultiPolygon, Polygon, Point } from 'geojson';
import osmtogeojson from 'osmtogeojson';
import pg from 'pg';
import { nearestProvince } from './data/vn-province-centroids';

const generateCuid = (): Promise<string> => Promise.resolve(createId());

// ─── Setup ────────────────────────────────────────────────────────────────
const pool = new pg.Pool({ connectionString: process.env['DATABASE_URL'] });
const adapter = new PrismaPg(pool);
const prisma = new PrismaClient({ adapter });

const OVERPASS_URL =
  process.env['OVERPASS_URL'] ?? 'https://overpass-api.de/api/interpreter';

interface BBox {
  south: number;
  west: number;
  north: number;
  east: number;
}

/** Vietnam split horizontally — Overpass timeouts at country scale. */
const CHUNKS: Record<string, BBox> = {
  north: { south: 19.0, west: 102.0, north: 23.5, east: 110.0 },
  northCentral: { south: 16.5, west: 102.0, north: 19.0, east: 110.0 },
  southCentral: { south: 13.0, west: 102.0, north: 16.5, east: 110.0 },
  south: { south: 8.0, west: 102.0, north: 13.0, east: 110.0 },
};

// ─── CLI flags ────────────────────────────────────────────────────────────
const argv = process.argv.slice(2);
const dryRun = argv.includes('--dry-run');
const chunkArg = argv.find((a) => a.startsWith('--chunk='))?.slice('--chunk='.length);

// ─── Province / region heuristic from centroid ────────────────────────────

/** Approximate centroid → region using latitude bands. Good enough as a
 *  fallback when the OSM tags lack `addr:state`; admin can correct later. */
function guessRegion(lat: number): 'NORTH' | 'CENTRAL' | 'SOUTH' {
  if (lat >= 19) return 'NORTH';
  if (lat >= 13) return 'CENTRAL';
  return 'SOUTH';
}

// ─── Helpers ──────────────────────────────────────────────────────────────

function slugify(name: string, osmId: string): string {
  const base = name
    .toLowerCase()
    .replace(/đ/g, 'd')
    // strip Vietnamese diacritics
    .normalize('NFD')
    .replace(/[̀-ͯ]/g, '')
    .replace(/[^a-z0-9]+/g, '-')
    .replace(/^-+|-+$/g, '')
    .slice(0, 60);
  return `${base || 'kcn'}-osm-${osmId}`;
}

interface OverpassResult {
  elements: unknown[];
}

async function fetchChunk(name: string, bbox: BBox): Promise<OverpassResult> {
  const query = `
    [out:json][timeout:180];
    (
      way["landuse"="industrial"](${bbox.south},${bbox.west},${bbox.north},${bbox.east});
      relation["landuse"="industrial"](${bbox.south},${bbox.west},${bbox.north},${bbox.east});
      node["landuse"="industrial"](${bbox.south},${bbox.west},${bbox.north},${bbox.east});
    );
    out body geom;
  `;
  console.log(`  → fetching chunk "${name}" from Overpass…`);
  const start = Date.now();
  // Overpass expects the query in a `data=` body field with form-encoding.
  const res = await fetch(OVERPASS_URL, {
    method: 'POST',
    headers: {
      'Content-Type': 'application/x-www-form-urlencoded',
      'User-Agent': 'goodgo-osm-sync/1.0 (https://goodgo.vn)',
    },
    body: 'data=' + encodeURIComponent(query),
  });
  if (!res.ok) {
    const body = await res.text();
    throw new Error(`Overpass returned ${res.status}: ${body.slice(0, 200)}`);
  }
  const json = (await res.json()) as OverpassResult;
  console.log(
    `  ← ${name}: ${json.elements?.length ?? 0} elements in ${(
      (Date.now() - start) /
      1000
    ).toFixed(1)}s`,
  );
  return json;
}

// ─── Per-feature upsert ──────────────────────────────────────────────────

interface ParsedFeature {
  osmId: bigint;
  osmType: 'NODE' | 'WAY' | 'RELATION';
  osmVersion: number | null;
  name: string;
  nameEn: string | null;
  developer: string;
  operator: string | null;
  tags: Record<string, string>;
  centroid: { lng: number; lat: number };
  /** GeoJSON Polygon|MultiPolygon as an unparsed JSON string for ST_GeomFromGeoJSON. */
  boundaryGeoJson: string | null;
  totalAreaHa: number;
  province: string;
  district: string;
  address: string;
}

/** OSM tag keys whose Vietnamese-province values we recognise. */
const VN_PROVINCE_HINTS = ['addr:province', 'addr:state', 'addr:region', 'is_in:province'];

function parseFeature(
  feat: Feature<Polygon | MultiPolygon | Point>,
): ParsedFeature | null {
  const propsRaw = feat.properties as Record<string, unknown> | null;
  if (!propsRaw) return null;
  const idStr = String(propsRaw['id'] ?? '');
  // osmtogeojson encodes id as "way/123", "relation/456", "node/789"
  const slashIdx = idStr.indexOf('/');
  if (slashIdx < 0) return null;
  const typeRaw = idStr.slice(0, slashIdx).toUpperCase();
  if (typeRaw !== 'NODE' && typeRaw !== 'WAY' && typeRaw !== 'RELATION') {
    return null;
  }
  const osmType = typeRaw as 'NODE' | 'WAY' | 'RELATION';
  const osmId = BigInt(idStr.slice(slashIdx + 1));
  const tagsRaw = propsRaw['tags'];
  const tags: Record<string, string> =
    tagsRaw && typeof tagsRaw === 'object'
      ? (tagsRaw as Record<string, string>)
      : (propsRaw as Record<string, string>);

  const name = tags['name:vi'] ?? tags['name'] ?? null;
  // Skip purely unnamed industrial polygons — too noisy for our catalog.
  if (!name) return null;
  // Skip rows whose names contain zero Latin / Vietnamese letters. This
  // catches polygons that bleed across the northern border (Quảng Ninh /
  // Lạng Sơn bbox edges) and have only CJK names — those are Chinese
  // industrial sites, not VN KCN.
  if (!/[A-Za-zÀ-ỹ]/.test(name)) return null;

  const operator = tags['operator'] ?? null;
  const developer = operator ?? tags['operator:wikidata'] ?? 'Chưa xác định';

  // Derive centroid + (optional) boundary GeoJSON.
  let cLng: number;
  let cLat: number;
  let boundaryGeoJson: string | null = null;
  let totalAreaHa = 0;

  if (feat.geometry.type === 'Point') {
    [cLng, cLat] = feat.geometry.coordinates;
  } else {
    // Polygon or MultiPolygon
    const c = centroid(feat as Feature);
    [cLng, cLat] = c.geometry.coordinates;
    // Normalise to MultiPolygon for storage in the boundary column.
    const geom: MultiPolygon =
      feat.geometry.type === 'Polygon'
        ? { type: 'MultiPolygon', coordinates: [feat.geometry.coordinates] }
        : feat.geometry;
    boundaryGeoJson = JSON.stringify(geom);
    // turf/area returns m² — convert to ha (1 ha = 10 000 m²).
    totalAreaHa = Math.round((area(feat as Feature) / 10000) * 100) / 100;
  }

  // Province resolution: prefer explicit OSM tags, then fall back to a
  // nearest-centroid lookup against our 63-province table. The fallback
  // catches the (very common) case where Vietnamese landuse polygons have
  // no addr:* tags at all.
  const province =
    VN_PROVINCE_HINTS.map((k) => tags[k]).find(Boolean) ??
    tags['addr:city'] ??
    nearestProvince(cLat, cLng);
  const district = tags['addr:district'] ?? tags['addr:suburb'] ?? '';
  const address =
    tags['addr:full'] ??
    [tags['addr:housenumber'], tags['addr:street']].filter(Boolean).join(' ') ??
    '';

  const versionRaw = propsRaw['version'];
  const osmVersion = typeof versionRaw === 'number' ? versionRaw : null;

  return {
    osmId,
    osmType,
    osmVersion,
    name,
    nameEn: tags['name:en'] ?? null,
    developer,
    operator,
    tags,
    centroid: { lng: cLng, lat: cLat },
    boundaryGeoJson,
    totalAreaHa,
    province,
    district,
    address,
  };
}

interface UpsertStats {
  inserted: number;
  updated: number;
  skipped: number;
  locked: number;
}

async function upsertFeature(
  parsed: ParsedFeature,
  stats: UpsertStats,
): Promise<void> {
  // Cheap pre-flight: if a row already exists with osmLocked=true, skip.
  const existing = await prisma.industrialPark.findUnique({
    where: { osmId: parsed.osmId },
    select: { id: true, osmLocked: true, lockedFields: true, dataSource: true },
  });

  if (existing?.osmLocked) {
    stats.locked += 1;
    return;
  }

  const region = guessRegion(parsed.centroid.lat);
  const slug = slugify(parsed.name, parsed.osmId.toString());

  if (!existing) {
    // INSERT path via raw SQL — Prisma's create() can't satisfy the
    // `location` PostGIS Geometry NOT NULL column (it's `Unsupported`).
    const cuid = await generateCuid();
    const tagsJson = JSON.stringify(parsed.tags);
    const boundarySql = parsed.boundaryGeoJson
      ? `ST_Multi(ST_GeomFromGeoJSON('${parsed.boundaryGeoJson.replace(/'/g, "''")}'))`
      : 'NULL';
    // Use a parameterised statement for safety + ST_* exprs inlined.
    await prisma.$executeRawUnsafe(
      `
      INSERT INTO "IndustrialPark" (
        id, name, "nameEn", slug, developer, operator, status,
        location, address, district, province, region,
        "totalAreaHa", "leasableAreaHa", "occupancyRate", "remainingAreaHa",
        "tenantCount", "targetIndustries",
        "osmId", "osmType", "osmVersion", "osmTags",
        "dataSource", "isPublic", "lastSyncedAt", "createdAt", "updatedAt",
        boundary
      ) VALUES (
        $1, $2, $3, $4, $5, $6, $7::"IndustrialParkStatus",
        ST_SetSRID(ST_MakePoint($8, $9), 4326), $10, $11, $12, $13::"VietnamRegion",
        $14, $15, $16, $17,
        $18, $19::text[],
        $20::bigint, $21::"IndustrialParkOsmType", $22, $23::jsonb,
        $24::"IndustrialParkDataSource", $25, $26, NOW(), NOW(),
        ${boundarySql}
      )
      `,
      cuid,
      parsed.name,
      parsed.nameEn,
      slug,
      parsed.developer,
      parsed.operator,
      'OPERATIONAL',
      parsed.centroid.lng,
      parsed.centroid.lat,
      parsed.address,
      parsed.district,
      parsed.province,
      region,
      parsed.totalAreaHa,
      parsed.totalAreaHa,
      0,
      parsed.totalAreaHa,
      0,
      [],
      parsed.osmId.toString(),
      parsed.osmType,
      parsed.osmVersion,
      tagsJson,
      'OSM',
      false,
      new Date(),
    );
    stats.inserted += 1;
  } else {
    // UPDATE path. Honour `lockedFields` — preserve listed columns.
    const locked = new Set(existing.lockedFields);
    const data: Prisma.IndustrialParkUpdateInput = {
      osmTags: parsed.tags as Prisma.InputJsonValue,
      osmVersion: parsed.osmVersion,
      lastSyncedAt: new Date(),
    };
    if (!locked.has('name')) data.name = parsed.name;
    if (!locked.has('nameEn')) data.nameEn = parsed.nameEn;
    if (!locked.has('operator')) data.operator = parsed.operator;
    if (!locked.has('developer')) data.developer = parsed.developer;
    if (!locked.has('address')) data.address = parsed.address;
    if (!locked.has('district')) data.district = parsed.district;
    if (!locked.has('province')) data.province = parsed.province;
    if (!locked.has('region')) data.region = region;
    if (!locked.has('totalAreaHa') && parsed.totalAreaHa > 0) {
      data.totalAreaHa = parsed.totalAreaHa;
    }
    await prisma.industrialPark.update({
      where: { osmId: parsed.osmId },
      data,
    });
    stats.updated += 1;
  }

  // For UPDATE path, also refresh the geometry columns (INSERT already
  // wrote them inline). Skip for fresh INSERTs we just wrote.
  if (existing) {
    if (parsed.boundaryGeoJson != null) {
      await prisma.$executeRaw`
        UPDATE "IndustrialPark"
        SET
          location = ST_SetSRID(ST_MakePoint(${parsed.centroid.lng}, ${parsed.centroid.lat}), 4326),
          boundary = ST_Multi(ST_GeomFromGeoJSON(${parsed.boundaryGeoJson}))
        WHERE "osmId" = ${parsed.osmId}
      `;
    } else {
      await prisma.$executeRaw`
        UPDATE "IndustrialPark"
        SET location = ST_SetSRID(ST_MakePoint(${parsed.centroid.lng}, ${parsed.centroid.lat}), 4326)
        WHERE "osmId" = ${parsed.osmId}
      `;
    }
  }
}

// ─── Main ─────────────────────────────────────────────────────────────────

async function main() {
  const chunkNames = chunkArg
    ? [chunkArg]
    : (Object.keys(CHUNKS) as Array<keyof typeof CHUNKS>);
  const totalStats: UpsertStats = { inserted: 0, updated: 0, skipped: 0, locked: 0 };
  let parseSkipped = 0;

  console.log(
    `🌍 OSM industrial-park sync — chunks: ${chunkNames.join(', ')} ` +
      `(dry-run: ${dryRun})\n`,
  );

  for (const chunkName of chunkNames) {
    const bbox = CHUNKS[chunkName as keyof typeof CHUNKS];
    if (!bbox) {
      console.warn(`  ⚠️  unknown chunk "${chunkName}" — skipping`);
      continue;
    }
    const overpass = await fetchChunk(chunkName, bbox);
    // osmtogeojson is loosely typed; it accepts the raw Overpass JSON.
    const fc = osmtogeojson(overpass) as unknown as {
      features: Feature<Polygon | MultiPolygon | Point>[];
    };
    const parsed: ParsedFeature[] = [];
    for (const f of fc.features) {
      const p = parseFeature(f);
      if (p) parsed.push(p);
      else parseSkipped += 1;
    }
    console.log(
      `  ▸ ${chunkName}: ${parsed.length} named industrial parks ` +
        `(skipped ${fc.features.length - parsed.length} unnamed/malformed)`,
    );

    if (dryRun) {
      console.log(
        `  ▸ first 3 names: ${parsed
          .slice(0, 3)
          .map((p) => p.name)
          .join(', ')}`,
      );
      continue;
    }

    for (const p of parsed) {
      try {
        await upsertFeature(p, totalStats);
      } catch (err) {
        totalStats.skipped += 1;
        console.warn(
          `  ⚠️  upsert failed for ${p.osmType.toLowerCase()}/${p.osmId}: ` +
            `${err instanceof Error ? err.message : String(err)}`,
        );
      }
    }
  }

  console.log('\n✓ Sync complete.');
  console.log(
    `   inserted: ${totalStats.inserted} ` +
      `· updated: ${totalStats.updated} ` +
      `· locked-skipped: ${totalStats.locked} ` +
      `· errors: ${totalStats.skipped} ` +
      `· parse-skipped: ${parseSkipped}`,
  );
}

main()
  .catch((err) => {
    console.error('❌ Sync failed:', err);
    process.exit(1);
  })
  .finally(async () => {
    await prisma.$disconnect();
    await pool.end();
  });