fifteen/worker/src/jobs/compute-routing.ts

import type { Job } from "bullmq";
import { getSql } from "../db.js";
import type { JobProgress } from "@transportationer/shared";
import { fetchMatrix } from "../valhalla.js";

export type ComputeRoutingData = {
  type: "compute-routing";
  citySlug: string;
  mode: "walking" | "cycling" | "driving";
  category: string;
};

/** Number of nearest POI candidates per grid point. */
const K = 6;
/** Grid points per Valhalla matrix call. */
const BATCH_SIZE = 20;
/** Concurrent Valhalla calls within this job. */
const BATCH_CONCURRENCY = 4;
/** Rows per INSERT. */
const INSERT_CHUNK = 2000;

async function asyncPool<T>(
  concurrency: number,
  items: T[],
  fn: (item: T) => Promise<void>,
): Promise<void> {
  const queue = [...items];
  async function worker(): Promise<void> {
    while (queue.length > 0) await fn(queue.shift()!);
  }
  await Promise.all(Array.from({ length: Math.min(concurrency, items.length) }, worker));
}

/**
 * Fetch Valhalla routing times from every grid point to the K nearest POIs
 * in the given category, then persist nearest-per-subcategory into grid_poi_details.
 */
export async function handleComputeRouting(job: Job<ComputeRoutingData>): Promise<void> {
  const { citySlug, mode, category } = job.data;
  const sql = getSql();

  const gridPoints = await Promise.resolve(sql<{ id: string; lat: number; lng: number }[]>`
    SELECT id::text AS id, ST_Y(geom) AS lat, ST_X(geom) AS lng
    FROM grid_points
    WHERE city_slug = ${citySlug}
    ORDER BY id
  `);

  if (gridPoints.length === 0) return;

  const [{ count }] = await Promise.resolve(sql<{ count: number }[]>`
    SELECT COUNT(*)::int AS count
    FROM raw_pois
    WHERE city_slug = ${citySlug} AND category = ${category}
  `);
  if (count === 0) return;

  // Nearest POI per (gridPointId, subcategory).
  const result = new Map<string, Map<string, {
    poiId: string;
    poiName: string | null;
    distM: number;
    timeS: number | null;
  }>>();
  for (const gp of gridPoints) result.set(gp.id, new Map());

  const batches: Array<{ id: string; lat: number; lng: number }[]> = [];
  for (let i = 0; i < gridPoints.length; i += BATCH_SIZE) {
    batches.push(gridPoints.slice(i, i + BATCH_SIZE));
  }

  let batchesDone = 0;

  await asyncPool(BATCH_CONCURRENCY, batches, async (batch) => {
    const batchIds = batch.map((gp) => gp.id);

    const knnRows = await Promise.resolve(sql<{
      grid_point_id: string;
      poi_id: string;
      poi_lat: number;
      poi_lng: number;
      poi_name: string | null;
      dist_m: number;
      subcategory: string;
    }[]>`
      SELECT
        gp.id::text AS grid_point_id,
        p.osm_id::text AS poi_id,
        ST_Y(p.geom) AS poi_lat,
        ST_X(p.geom) AS poi_lng,
        p.poi_name,
        ST_Distance(gp.geom::geography, p.geom::geography) AS dist_m,
        p.subcategory
      FROM grid_points gp
      CROSS JOIN LATERAL (
        SELECT p.osm_id, p.geom, p.subcategory, p.name AS poi_name
        FROM raw_pois p
        WHERE p.city_slug = ${citySlug} AND p.category = ${category}
        ORDER BY gp.geom <-> p.geom
        LIMIT ${K}
      ) p
      WHERE gp.id = ANY(${batchIds}::bigint[])
      ORDER BY gp.id, dist_m
    `);

    batchesDone++;

    if (knnRows.length === 0) return;

    const targetIdx = new Map<string, number>();
    const targets: { lat: number; lng: number }[] = [];
    for (const row of knnRows) {
      if (!targetIdx.has(row.poi_id)) {
        targetIdx.set(row.poi_id, targets.length);
        targets.push({ lat: row.poi_lat, lng: row.poi_lng });
      }
    }

    const sources = batch.map((gp) => ({ lat: gp.lat, lng: gp.lng }));
    let matrix: (number | null)[][];
    try {
      matrix = await fetchMatrix(sources, targets, mode);
    } catch (err) {
      console.error(
        `[compute-routing] Valhalla failed (${mode}/${category}, batch ${batchesDone}):`,
        (err as Error).message,
      );
      return;
    }

    type KnnRow = (typeof knnRows)[number];
    const gpKnn = new Map<string, KnnRow[]>();
    for (const row of knnRows) {
      const list = gpKnn.get(row.grid_point_id) ?? [];
      list.push(row);
      gpKnn.set(row.grid_point_id, list);
    }

    for (let bi = 0; bi < batch.length; bi++) {
      const gp = batch[bi];
      const knn = gpKnn.get(gp.id);
      if (!knn || knn.length === 0) continue;

      const subcatMap = result.get(gp.id)!;
      for (const row of knn) {
        if (!subcatMap.has(row.subcategory)) {
          const idx = targetIdx.get(row.poi_id);
          subcatMap.set(row.subcategory, {
            poiId: row.poi_id,
            poiName: row.poi_name,
            distM: row.dist_m,
            timeS: idx !== undefined ? (matrix[bi]?.[idx] ?? null) : null,
          });
        }
      }
    }

    await job.updateProgress({
      stage: `Routing ${mode}/${category}`,
      pct: Math.round((batchesDone / batches.length) * 100),
      message: `Batch ${batchesDone}/${batches.length}`,
    } satisfies JobProgress);
  });

  // Bulk-insert nearest POI per subcategory into grid_poi_details.
  const gpIdArr: string[] = [];
  const subcatArr: string[] = [];
  const poiIdArr: (string | null)[] = [];
  const poiNameArr: (string | null)[] = [];
  const distArr: (number | null)[] = [];
  const timeArr: (number | null)[] = [];

  for (const [gpId, subcatMap] of result) {
    for (const [subcategory, detail] of subcatMap) {
      gpIdArr.push(gpId);
      subcatArr.push(subcategory);
      poiIdArr.push(detail.poiId);
      poiNameArr.push(detail.poiName);
      distArr.push(detail.distM);
      timeArr.push(detail.timeS);
    }
  }

  for (let i = 0; i < gpIdArr.length; i += INSERT_CHUNK) {
    const end = Math.min(i + INSERT_CHUNK, gpIdArr.length);
    await Promise.resolve(sql`
      INSERT INTO grid_poi_details (
        grid_point_id, category, subcategory, travel_mode,
        nearest_poi_id, nearest_poi_name, distance_m, travel_time_s
      )
      SELECT
        gp_id::bigint,
        ${category},
        subcat,
        ${mode},
        CASE WHEN poi_id IS NULL THEN NULL ELSE poi_id::bigint END,
        poi_name,
        dist,
        time_s
      FROM unnest(
        ${gpIdArr.slice(i, end)}::text[],
        ${subcatArr.slice(i, end)}::text[],
        ${poiIdArr.slice(i, end)}::text[],
        ${poiNameArr.slice(i, end)}::text[],
        ${distArr.slice(i, end)}::float8[],
        ${timeArr.slice(i, end)}::float8[]
      ) AS t(gp_id, subcat, poi_id, poi_name, dist, time_s)
      ON CONFLICT (grid_point_id, category, subcategory, travel_mode)
      DO UPDATE SET
        nearest_poi_id   = EXCLUDED.nearest_poi_id,
        nearest_poi_name = EXCLUDED.nearest_poi_name,
        distance_m       = EXCLUDED.distance_m,
        travel_time_s    = EXCLUDED.travel_time_s,
        computed_at      = now()
    `);
  }
}