A story's label names each thing once (#205)

ddiff --git a/server/domain/clusters.js b/server/domain/clusters.js
iindex 30aba2a..7abafc5 100644
---- a/server/domain/clusters.js
++++ b/server/domain/clusters.js
@@@ -99,13 +99,40 @@ function compare(sig, cluster, n = 1, common = null) {
      return { sharedEntities, sharedTerms, score, joins, coreMin: min };
  }
  
++/**
++ * A label names each thing once (11-03, the text lint on /live: "department · doj · doj to monitor elections" and
++ * "ceconomy · ceconomy ag ceconomy"): in rank order, a name is left out when its words sit inside a name already
++ * kept, or a kept name's words sit inside it, or it repeats a word of its own (a headline caught as an entity).
++ */
++function distinctNames(list, max) {
++    const out = [];
++    const words = (s) => String(s || '').split(/\s+/).filter(Boolean);
++    const inside = (a, b) => ` ${b.join(' ')} `.includes(` ${a.join(' ')} `);
++    for (const name of list) {
++        const w = words(name);
++        if (!w.length || new Set(w).size < w.length) continue;
++        if (out.some((kept) => inside(w, words(kept)) || inside(words(kept), w))) continue;
++        out.push(name);
++        if (out.length >= max) break;
++    }
++    return out;
++}
++
  function labelOf(entities, terms) {
      const top = (obj) => Object.entries(obj).sort(([a, x], [b, y]) => (y - x) || (a < b ? -1 : 1)).map(([k]) => k);
--    const e = top(entities).slice(0, 3);
--    const words = e.length ? e : top(terms).slice(0, 4);
++    const e = distinctNames(top(entities), 3);
++    const words = e.length ? e : distinctNames(top(terms), 4);
      return words.length ? words.join(' · ') : 'unlabelled';
  }
  
++/** A stored label as it is shown: the same rule, for labels written before it (they change only on a new report). */
++function tidyLabel(label) {
++    const parts = String(label || '').split(' · ');
++    if (parts.length < 2) return label;
++    const kept = distinctNames(parts, parts.length);
++    return kept.length ? kept.join(' · ') : label;
++}
++
  function createClusters({ store, config, outbox }) {
      const { db } = store;
      const q = {
@@@ -242,7 +269,7 @@ function createClusters({ store, config, outbox }) {
      }
  
      const api = {
--        signature, compare, coreMin, commonWords, itemTime, labelOf, assign, resolve, recompute,
++        signature, compare, coreMin, commonWords, itemTime, labelOf, tidyLabel, assign, resolve, recompute,
  
          get: async (id) => await q.get.get(String(id || '')) || null,
          members: async (id) => await q.members.all(id),
@@@ -275,7 +302,7 @@ function createClusters({ store, config, outbox }) {
                  const items = byCluster.get(c.id);
                  if (!items || !items.length) continue;
                  items.sort((a, b) => String(b.published_at || '').localeCompare(String(a.published_at || '')));
--                out.push({ id: c.id, label: c.label, updated: c.window_end, items });
++                out.push({ id: c.id, label: tidyLabel(c.label), updated: c.window_end, items });
                  if (out.length >= limit) break;
              }
              return out;
@@@ -296,7 +323,7 @@ function createClusters({ store, config, outbox }) {
              const cutoff = store.now() + skewMs;
              for (const r of rows) {
                  if (r.published_at && Date.parse(r.published_at) > cutoff) continue;
--                if (!byCluster.has(r.cid)) byCluster.set(r.cid, { id: r.cid, label: r.label, updated: r.window_end, items: [], sources: new Set() });
++                if (!byCluster.has(r.cid)) byCluster.set(r.cid, { id: r.cid, label: tidyLabel(r.label), updated: r.window_end, items: [], sources: new Set() });
                  const c = byCluster.get(r.cid);
                  c.sources.add(r.source_key);
                  if (r.canonical_url) c.items.push({ id: r.id, headline: r.headline, url: r.canonical_url, outlet: r.outlet, published_at: r.published_at });
@@@ -453,4 +480,4 @@ function createClusters({ store, config, outbox }) {
      return api;
  }
  
--module.exports = { createClusters, signature, compare, coreMin, commonWords, itemTime, labelOf };
++module.exports = { createClusters, signature, compare, coreMin, commonWords, itemTime, labelOf, tidyLabel, distinctNames };
ddiff --git a/test/clusters.test.js b/test/clusters.test.js
iindex 2abcb3f..67145ff 100644
---- a/test/clusters.test.js
++++ b/test/clusters.test.js
@@@ -111,6 +111,16 @@ const { boot, check, done, launchReports } = require('./helpers/boot');
          assert.deepStrictEqual(audit, ['merge', 'reverse_merge', 'split', 'reverse_split']);
      });
  
++    await check('a label names each thing once (11-03: "department · doj · doj to monitor elections")', async () => {
++        const { labelOf, tidyLabel } = require('../server/domain/clusters');
++        assert.strictEqual(labelOf({ department: 9, doj: 8, 'doj to monitor elections': 7, justice: 3 }, {}), 'department · doj · justice', 'a name inside another is left out');
++        assert.strictEqual(labelOf({ 'ceconomy ag ceconomy': 5, ceconomy: 4, mediamarkt: 2 }, {}), 'ceconomy · mediamarkt', 'a name that repeats a word is left out');
++        assert.strictEqual(labelOf({}, { rail: 3, strike: 2, 'rail strike': 1 }), 'rail · strike', 'terms follow the same rule');
++        assert.strictEqual(labelOf({ 'new york': 3, york: 2, york2: 1 }, {}), 'new york · york2', 'whole words only: york2 is not york');
++        assert.strictEqual(tidyLabel('department · doj · doj to monitor elections'), 'department · doj', 'a stored label is shown tidy');
++        assert.strictEqual(tidyLabel('unlabelled'), 'unlabelled');
++    });
++
      await check('a cluster is matched on its core, not the union of its reports: it cannot snowball', async () => {
          const { compare, coreMin, signature } = require('../server/domain/clusters');
          assert.deepStrictEqual([1, 2, 3, 5, 6, 196].map(coreMin), [1, 1, 2, 2, 3, 66]);