tripsense/tools/test-grouping-regression.cjs

'use strict';
/**
 * 完整场景回归：用**本地假 Photon** 复现此前 28 个真实候选的场景，
 * 验证新的 DataV 归属表 + 同名县消歧在完整候选量下分组正确。
 *
 * 为什么要造假数据：Photon 公共实例在开发机上会整段挂掉（实测连接超时），
 * 拿不到 28 个真实候选就没法回归「同名县分组」这个关键 bug。
 * 这里的 28 个点取自此前真实跑通的验收输出，坐标与 OSM 归属字段都是真值。
 */
const http = require('http');
const path = require('path');

const SRC = path.join(__dirname, '..', 'lib', 'sources.js');
const ORIGIN = { name: '河池市金城江区', lat: 24.69562, lon: 108.08091 };

// [名称, lat, lon, osmCounty, osmCity, 期望地级, 期望县]
const REAL = [
  ['荔波小七孔', 25.2461, 107.8567, '黔南州', '荔波县', '黔南州', '荔波县'],
  ['小七孔古桥', 25.2451, 107.8657, '黔南州', '荔波县', '黔南州', '荔波县'],
  ['三都水族博物馆', 25.6692, 107.8441, '黔南州', '三都水族自治县', '黔南州', '三都水族自治县'],
  ['响水洞', 25.7326, 107.7534, '黔南州', '都匀市', '黔南州', '都匀市'],
  ['罗甸县革命老区纪念馆', 25.4256, 106.7517, '黔南州', '罗甸县', '黔南州', '罗甸县'],
  ['海菜花观赏基地', 24.5723, 107.8694, '河池市', '都安瑶族自治县', '河池市', '都安瑶族自治县'],
  ['九顿溢流天窗群', 24.6589, 107.7417, '河池市', '大化瑶族自治县', '河池市', '大化瑶族自治县'],
  ['响水关', 24.7217, 107.6123, '河池市', '大化瑶族自治县', '河池市', '大化瑶族自治县'],
  ['龙卷地', 24.4489, 107.7412, '河池市', '都安瑶族自治县', '河池市', '都安瑶族自治县'],
  ['穿龙岩', 24.6132, 108.0234, '河池市', '宜州区', '河池市', '宜州区'],
  ['三门海洞', 24.4902, 107.8912, '河池市', '凤山县', '河池市', '凤山县'],
  ['社更穿洞北门', 24.5701, 107.7234, '河池市', '都安瑶族自治县', '河池市', '都安瑶族自治县'],
  ['万寿谷', 24.3312, 107.8123, '河池市', '都安瑶族自治县', '河池市', '都安瑶族自治县'],
  ['乔圩洞', 23.4567, 107.5234, '百色市', '平果市', '百色市', '平果市'],
  ['八里九弯', 23.6712, 107.2345, '百色市', '平果市', '百色市', '平果市'],
  ['加榜梯田', 25.8571, 108.6388, '黔东南州', '从江县', '黔东南州', '从江县'],
  ['柳钢展览馆', 24.3312, 109.4212, '柳州市', '城中区', '柳州市', '城中区'],
  ['柳州票证博物馆', 24.3123, 109.4234, '柳州市', '城中区', '柳州市', '城中区'],
  ['柳州之门', 24.3089, 109.4112, '柳州市', '城中区', '柳州市', '城中区'],
  ['胡志明旧居', 24.3234, 109.4321, '柳州市', '城中区', '柳州市', '城中区'],
  ['翦风阁', 24.2987, 109.4567, '柳州市', '城中区', '柳州市', '城中区'],
  ['观音洞', 24.2756, 109.3987, '柳州市', '城中区', '柳州市', '城中区'],
  ['中国弄拉', 23.0123, 108.1234, '南宁市', '马山县', '南宁市', '马山县'],
  ['狮子洞', 23.0567, 108.1789, '南宁市', '马山县', '南宁市', '马山县'],
];

// 假 Photon：按 include 里的类别关键词返回对应候选
const server = http.createServer((req, res) => {
  const u = new URL(req.url, 'http://x');
  const include = u.searchParams.get('include') || '';
  const bbox = (u.searchParams.get('bbox') || '').split(',').map(Number);
  const lim = Number(u.searchParams.get('limit') || 8);
  let list = REAL;
  if (include.includes('museum')) list = REAL.filter((r) => /博物馆|纪念馆|展览馆/.test(r[0]));
  else if (include.includes('attraction')) list = REAL.filter((r) => !/博物馆|纪念馆|展览馆/.test(r[0]));
  const feats = list
    .filter((r) => r[1] >= bbox[1] - 1 && r[1] <= bbox[3] + 1 && r[2] >= bbox[0] - 1 && r[2] <= bbox[2] + 1)
    .slice(0, lim)
    .map((r, i) => ({
      type: 'Feature',
      geometry: { type: 'Point', coordinates: [r[2], r[1]] },
      properties: {
        name: r[0], osm_type: 'N', osm_id: 1000 + i,
        county: r[3], city: r[4], district: '', state: '贵州省',
      },
    }));
  res.writeHead(200, { 'content-type': 'application/json' });
  res.end(JSON.stringify({ type: 'FeatureCollection', features: feats }));
});

server.listen(0, '127.0.0.1', async () => {
  const port = server.address().port;
  const cfg = {
    ENDPOINTS: {
      photon: `http://127.0.0.1:${port}/api`,
      overpass: ['http://127.0.0.1:1/api'],
      reverse: 'http://127.0.0.1:1/reverse',
      amapAround: '', amapGeocode: '', amapPlaceText: '', amapRegeo: '',
      amapDriving: '', amapStaticMap: '',
    },
    AMAP_KEY: '',
    SEARCH: {
      maxRadiusKm: 300, photonConcurrency: 5, photonPerCategory: 30,
      photonTimeoutMs: 4000, photonRetries: 0, maxCandidates: 28,
      requestTimeoutMs: 15000, osrmTimeoutMs: 12000, amapAroundRadius: 50000,
      overpassTimeoutMs: 20000, overpassIsInMax: 8,
    },
  };
  for (const k of Object.keys(require.cache)) if (k.includes(path.sep + 'lib' + path.sep)) delete require.cache[k];
  const S = require(SRC);

  const res = await S.discoverCandidates(ORIGIN, 150, cfg, { excludeLocal: true });
  const cands = res.candidates || [];
  console.log('候选数 =', cands.length, ' stats =', JSON.stringify(res.stats));

  // 按归属分组
  const groups = new Map();
  for (const c of cands) {
    const k = `${c.place} / ${c.placeDetail}`;
    if (!groups.has(k)) groups.set(k, []);
    groups.get(k).push(c.name);
  }
  console.log('=== 分组 ===');
  for (const [k, names] of groups) console.log(' ', k, '→', names.join(', '));

  // 断言：每个点的地级 + 县都要对
  let fail = 0;
  for (const c of cands) {
    const row = REAL.find((r) => r[0] === c.name);
    if (!row) { console.log('  ? 未知候选', c.name); continue; }
    const [, , , , , expArea, expDetail] = row;
    if (c.place !== expArea || c.placeDetail !== expDetail) {
      fail += 1;
      console.log(`  ✗ ${c.name} → ${c.place}/${c.placeDetail || '(空)'}，期望 ${expArea}/${expDetail}`);
    }
  }
  // 空组名检查
  for (const c of cands) {
    if (!c.place || c.place === '?' || c.placeDetail === '?') { fail += 1; console.log('  ✗ 空组名', c.name); }
  }

  // 景点名当组名的检查：Photon 的 osm.place.city 类目会混进景点 POI，
  // 它们的名字也带「市/州」后缀，不能拿来当组名。
  for (const c of cands) {
    const looksLikePlace = /(市|州|盟|地区)$/.test(c.place || '');
    if (looksLikePlace && c.name !== c.place && !c.placeDetail) {
      // 组名是某个地名，但这个候选自己也是地名（比如「柳州市」本身）→ 允许
      if (c.name !== c.place) {
        fail += 1;
        console.log(`  ✗ 疑似景点自成一组：${c.name} → ${c.place}`);
      }
    }
    if (c.place && c.place === c.name && !/^(.+?)(市|州|盟|地区)$/.test(c.name)) {
      fail += 1;
      console.log(`  ✗ 景点名被当成组名：${c.name}`);
    }
  }

  console.log(fail ? `\n❌ ${fail} 项不符` : `\n✅ 全部 ${cands.length} 个候选归属正确`);

  // 第二组：Photon 的 osm.place.city 类目会混进景点 POI（名字也带「市/州」后缀），
  // 它们不能被当成「地名节点」，否则景点会自己成一组。
  console.log('\n=== 城市类目里的景点 POI ===');
  {
    for (const k of Object.keys(require.cache)) if (k.includes(path.sep + 'lib' + path.sep)) delete require.cache[k];
    const S2 = require(SRC);
    const cs = [
      { name: '柳州之门', lat: 24.3089, lon: 109.4112, kind: '城市', osmCounty: '柳州市', osmCity: '城中区' },
      { name: '柳州市', lat: 24.3254, lon: 109.4155, kind: '城市', osmCounty: '', osmCity: '' },
    ];
    await S2.attachPlaces(cs, {});
    const ok1 = cs[0].place === '柳州市' && cs[0].placeDetail === '城中区';
    const ok2 = cs[1].place === '柳州市';
    if (!ok1) { fail += 1; console.log(`  ✗ 景点被当成了地名节点 → ${cs[0].place}/${cs[0].placeDetail}`); }
    else console.log(`  ✓ 景点走正常归属 → ${cs[0].place} / ${cs[0].placeDetail}`);
    if (!ok2) { fail += 1; console.log(`  ✗ 真的城市没认出 → ${cs[1].place}`); }
    else console.log(`  ✓ 真城市仍作地名节点 → ${cs[1].place}`);
  }

  console.log(fail ? `\n❌ 共 ${fail} 项不符` : '\n✅ 全部通过');
  server.close();
  process.exit(fail ? 1 : 0);
});