GdzieWjade/scripts/build-data.mjs

// Builds public/data/kk-snapshot.json from OpenStreetMap via Overpass.
// Raw responses are cached in scripts/.cache; delete it to refetch.
// Data © OpenStreetMap contributors, ODbL 1.0.
import { mkdir, readFile, writeFile } from 'node:fs/promises';
import { existsSync } from 'node:fs';
import { execFileSync } from 'node:child_process';
import { toPlace } from '../src/lib/osmPlace.ts';

// S,W,N,E: Wawel + Old Town + Kazimierz out east to Tauron Arena / Czyżyny
const BBOX = [50.044, 19.920, 50.074, 20.000];
const bb = BBOX.join(',');
const OVERPASS = 'https://overpass-api.de/api/interpreter';

const queries = {
  places: `[out:json][timeout:240];
(
  nwr["name"]["amenity"~"^(restaurant|cafe|bar|pub|fast_food|ice_cream|food_court|pharmacy|bank|clinic|doctors|dentist|hospital|post_office|cinema|theatre|library|arts_centre|townhall|community_centre|place_of_worship|university|college|school|police|courthouse|events_venue|conference_centre|exhibition_centre|marketplace)$"](${bb});
  nwr["name"]["leisure"~"^(stadium|sports_centre|swimming_pool|fitness_centre)$"](${bb});
  nwr["name"]["tourism"~"^(museum|gallery|hotel|hostel|guest_house|apartment|attraction|information)$"](${bb});
  nwr["name"]["shop"](${bb});
  nwr["name"]["historic"~"^(castle|monument|church|building|city_gate|fort)$"](${bb});
);
out meta center;`,
  support: `[out:json][timeout:240];
(
  nwr["amenity"="toilets"](${bb});
  nwr["highway"="elevator"](${bb});
  nwr["amenity"="bench"](${bb});
  node["highway"="bus_stop"]["name"](${bb});
  node["railway"="tram_stop"]["name"](${bb});
);
out meta center;`,
  net: `[out:json][timeout:300];
way["highway"~"^(footway|pedestrian|path|steps|living_street|residential|service|tertiary|unclassified|track|cycleway)$"](${bb});
out meta;
>;
out skel qt;`,
  // named green areas (parks, gardens, cemeteries...) as outlines: unnamed paths inside them get the area's name in route text
  parks: `[out:json][timeout:240];
(
  way["name"]["leisure"~"^(park|garden|playground)$"](${bb});
  relation["name"]["leisure"~"^(park|garden)$"](${bb});
  way["name"]["landuse"~"^(recreation_ground|cemetery)$"](${bb});
  way["name"]["natural"="wood"](${bb});
);
out geom;`,
  kerbs: `[out:json][timeout:240];
(
  node["kerb"](${bb});
  node["highway"="crossing"](${bb});
);
out tags;`,
};

// Public Overpass instances rate-limit hard (HTTP 429/504): fetch one query at a time, back off, and rotate mirrors.
const MIRRORS = [OVERPASS, 'https://overpass.kumi.systems/api/interpreter', 'https://overpass.private.coffee/api/interpreter'];
async function fetchRaw(name) {
  const file = `scripts/.cache/${name}-${BBOX.join('_')}.json`; // keyed by bbox so a changed area never reuses stale data
  if (existsSync(file)) return JSON.parse(await readFile(file, 'utf8'));
  for (let attempt = 1; attempt <= 8; attempt++) {
    const endpoint = MIRRORS[(attempt - 1) % MIRRORS.length];
    try {
      const res = await fetch(endpoint, {
        signal: AbortSignal.timeout(280_000), // a hung mirror must not stall the build forever
        method: 'POST',
        headers: {
          'User-Agent': 'krakow-bez-barier-hackathon/0.1',
          Accept: 'application/json',
          'Content-Type': 'application/x-www-form-urlencoded',
        },
        body: 'data=' + encodeURIComponent(queries[name]),
      });
      if (res.ok) {
        const text = await res.text();
        JSON.parse(text); // a mirror can answer 200 with an HTML error page
        await writeFile(file, text);
        return JSON.parse(text);
      }
      console.warn(`${name}: HTTP ${res.status} from ${endpoint}, retry ${attempt}`);
    } catch (e) {
      console.warn(`${name}: ${e.message} from ${endpoint}, retry ${attempt}`);
    }
    await new Promise((r) => setTimeout(r, 15000 * attempt));
  }
  throw new Error(`Overpass failed for ${name}`);
}

const round = (n) => Math.round(n * 1e6) / 1e6;
const day = (ts) => (ts ? ts.slice(0, 10) : null);

const raw = {};
for (const q of Object.keys(queries)) raw[q] = await fetchRaw(q);
const { places: placesRaw, support: supportRaw, net: netRaw, kerbs: kerbRaw, parks: parksRaw } = raw;

// ---- green areas: closed rings, relations stitched from their outer member ways ----
const r5 = (n) => Math.round(n * 1e5) / 1e5;
const pt = (g) => [r5(g.lat), r5(g.lon)];
const same = (a, b) => a[0] === b[0] && a[1] === b[1];
function stitch(segs) {
  const rings = [];
  const left = segs.map((s) => s.slice());
  while (left.length) {
    let ring = left.pop();
    for (let progress = true; progress && !same(ring[0], ring[ring.length - 1]);) {
      progress = false;
      for (let i = 0; i < left.length; i++) {
        const s = left[i], end = ring[ring.length - 1];
        if (same(s[0], end)) ring = ring.concat(s.slice(1));
        else if (same(s[s.length - 1], end)) ring = ring.concat(s.slice(0, -1).reverse());
        else continue;
        left.splice(i, 1);
        progress = true;
        break;
      }
    }
    if (ring.length >= 4 && same(ring[0], ring[ring.length - 1])) rings.push(ring);
  }
  return rings;
}
const parks = [];
for (const el of parksRaw.elements) {
  const n = el.tags?.name;
  if (!n) continue;
  if (el.type === 'way' && el.geometry) {
    const ring = el.geometry.map(pt);
    if (ring.length >= 4 && same(ring[0], ring[ring.length - 1])) parks.push({ n, p: ring });
  } else if (el.type === 'relation') {
    const outer = (el.members ?? []).filter((m) => m.type === 'way' && m.role !== 'inner' && m.geometry).map((m) => m.geometry.map(pt));
    for (const ring of stitch(outer)) parks.push({ n, p: ring });
  }
}

const places = placesRaw.elements.map((e) => toPlace(e)).filter(Boolean);
const toilets = [], elevators = [], benches = [], stops = [];
for (const el of supportRaw.elements) {
  const t = el.tags ?? {};
  if (t.amenity === 'toilets') toilets.push(toPlace(el, 'toilets'));
  else if (t.highway === 'elevator') elevators.push(toPlace(el, 'elevator'));
  else if (t.amenity === 'bench') {
    const lat = el.lat ?? el.center?.lat, lon = el.lon ?? el.center?.lon;
    if (lat != null) benches.push([round(lat), round(lon), day(el.timestamp)]);
  } else stops.push(toPlace(el, 'transit'));
}
for (const p of toilets) {
  const src = supportRaw.elements.find((e) => `${e.type[0]}${e.id}` === p.id).tags;
  for (const k of ['wheelchair', 'fee', 'opening_hours', 'access', 'toilets:wheelchair', 'changing_table', 'name', 'check_date']) {
    if (src[k] != null) p.tags[k] = src[k];
  }
}


// ---- city data: ZTP Kraków GTFS (stops + wheelchair_boarding) ----
const GTFS = [
  { mode: 'tram', url: 'https://gtfs.ztp.krakow.pl/GTFS_KRK_T.zip' },
  { mode: 'bus', url: 'https://gtfs.ztp.krakow.pl/GTFS_KRK_A.zip' },
];

function parseCsv(text) {
  const rows = [];
  let row = [], cur = '', q = false;
  for (let i = 0; i < text.length; i++) {
    const c = text[i];
    if (q) {
      if (c === '"' && text[i + 1] === '"') { cur += '"'; i++; }
      else if (c === '"') q = false;
      else cur += c;
    } else if (c === '"') q = true;
    else if (c === ',') { row.push(cur); cur = ''; }
    else if (c === '\n') { row.push(cur.replace(/\r$/, '')); rows.push(row); row = []; cur = ''; }
    else cur += c;
  }
  if (cur || row.length) { row.push(cur); rows.push(row); }
  const [head, ...body] = rows;
  return body.filter((r) => r.length === head.length).map((r) => Object.fromEntries(head.map((h, i) => [h.replace(/^\uFEFF/, ''), r[i]])));
}

async function loadGtfs({ mode, url }) {
  const zip = `scripts/.cache/gtfs-${mode}.zip`;
  if (!existsSync(zip)) {
    const res = await fetch(url, { headers: { 'User-Agent': 'krakow-bez-barier-hackathon/0.1' } });
    if (!res.ok) throw new Error(`GTFS ${mode}: HTTP ${res.status}`);
    await writeFile(zip, Buffer.from(await res.arrayBuffer()));
  }
  const read = (f) => execFileSync('unzip', ['-p', zip, f], { maxBuffer: 256 * 1024 * 1024 }).toString('utf8');
  const feed = parseCsv(read('feed_info.txt'))[0];
  const stops = parseCsv(read('stops.txt'));
  return { mode, feedVersion: feed?.feed_version ?? null, stops };
}

const feeds = await Promise.all(GTFS.map(loadGtfs));
const osmStops = stops.filter((s) => s.tags);
const transit = [];
const feedDate = (v) => (v && /^\d{8}/.test(v) ? `${v.slice(0, 4)}-${v.slice(4, 6)}-${v.slice(6, 8)}` : null);
for (const f of feeds) {
  for (const s of f.stops) {
    const lat = Number(s.stop_lat), lon = Number(s.stop_lon);
    if (!(lat >= BBOX[0] && lat <= BBOX[2] && lon >= BBOX[1] && lon <= BBOX[3])) continue;
    let near = null, nd = 40;
    for (const o of osmStops) {
      const d = Math.hypot((o.lat - lat) * 111320, (o.lon - lon) * 71500);
      if (d < nd) { nd = d; near = o; }
    }
    const wb = s.wheelchair_boarding === '' || s.wheelchair_boarding == null ? null : Number(s.wheelchair_boarding);
    const tags = {};
    if (near?.tags.wheelchair) tags.wheelchair = near.tags.wheelchair;
    transit.push({
      id: `z${f.mode[0]}${s.stop_id}`, lat: round(lat), lon: round(lon), name: s.stop_name,
      cat: 'transit', kind: f.mode === 'tram' ? 'tram_stop' : 'bus_stop',
      ts: near?.ts ?? null, tags,
      ext: { gtfsWheelchair: wb, feedDate: feedDate(f.feedVersion), mode: f.mode, osmMatch: near?.id ?? null },
    });
  }
}

// ---- pedestrian network ----
const nodeCoord = new Map();
for (const el of netRaw.elements) if (el.type === 'node') nodeCoord.set(el.id, [el.lat, el.lon]);

const kerbById = new Map();
for (const el of kerbRaw.elements) {
  const k = el.tags?.kerb;
  if (k) kerbById.set(el.id, k);
}

const nodes = [];
const nodeIdx = new Map();
const idxOf = (id) => {
  let i = nodeIdx.get(id);
  if (i === undefined) {
    const c = nodeCoord.get(id);
    i = nodes.push([round(c[0]), round(c[1])]) - 1;
    nodeIdx.set(id, i);
  }
  return i;
};

const ways = [];
const kerbs = {};
const isNo = (v) => v === 'no' || v === 'private';
for (const el of netRaw.elements) {
  if (el.type !== 'way') continue;
  const t = el.tags;
  const hw = t.highway;
  if (t.indoor || t.level && t.highway === 'corridor') continue;
  if (isNo(t.access) && !['yes', 'designated', 'permissive'].includes(t.foot)) continue;
  if (t.foot === 'no') continue;
  if (hw === 'cycleway' && !['yes', 'designated', 'permissive'].includes(t.foot)) continue;
  if (hw === 'service' && (t.access === 'private' || t.service === 'parking_aisle' || t.service === 'drive-through')) continue;
  if (hw === 'track' && t.foot !== 'yes') continue;
  if (el.nodes.some((n) => !nodeCoord.has(n))) continue;
  const w = {
    id: el.id,
    hw,
    ts: day(el.timestamp),
    nodes: el.nodes.map(idxOf),
  };
  if (t.name) w.name = t.name;
  const surface = t.surface ?? t['footway:surface'] ?? t['path:surface'];
  if (surface) w.surface = surface;
  if (t.smoothness) w.smooth = t.smoothness;
  if (t.incline) w.incline = t.incline;
  if (t.width) w.width = t.width;
  if (t.step_count) w.stepCount = Number(t.step_count) || undefined;
  if (t.ramp) w.ramp = t.ramp;
  if (t['ramp:wheelchair']) w.rampWheelchair = t['ramp:wheelchair'];
  if (t['ramp:stroller']) w.rampStroller = t['ramp:stroller'];
  if (t.handrail) w.handrail = t.handrail;
  if (t.footway === 'crossing' || t.footway === 'sidewalk') w.footway = t.footway;
  if (t.tunnel) w.tunnel = true;
  if (t.bridge) w.bridge = true;
  if (t.wheelchair) w.wheelchair = t.wheelchair;
  if (t['check_date:surface'] || t.check_date) w.checked = t['check_date:surface'] ?? t.check_date;
  ways.push(w);
  if (w.footway === 'crossing' || hw === 'footway') {
    for (const n of [el.nodes[0], el.nodes[el.nodes.length - 1]]) {
      const k = kerbById.get(n);
      if (k) kerbs[idxOf(n)] = k;
    }
  }
}

const out = {
  meta: {
    fetchedAt: new Date().toISOString(),
    bbox: BBOX,
    source: 'OpenStreetMap contributors',
    license: 'ODbL 1.0',
    endpoint: OVERPASS,
    cityData: { name: 'ZTP Kraków GTFS (stops.txt: wheelchair_boarding)', url: 'https://gtfs.ztp.krakow.pl/', feedDates: feeds.map((f) => ({ mode: f.mode, version: f.feedVersion })), license: 'verify at otwartedane.um.krakow.pl before reuse' },
    counts: { places: places.length + transit.length, transit: transit.length, toilets: toilets.length, elevators: elevators.length, benches: benches.length, ways: ways.length, nodes: nodes.length },
  },
  places: [...places, ...transit], toilets, elevators, benches, nodes, ways, kerbs, parks,
};
await mkdir('public/data', { recursive: true });
await writeFile('public/data/kk-snapshot.json', JSON.stringify(out));
console.log(out.meta.counts);