more words2

scripts/find-missing-combo-words.mjs

@@ -0,0 +1,67 @@
+// Finds words that can be built from 1-4 distinct tiles in the puzzle pool
+// but are missing from our dictionary, using a large reference word list
+// (an-array-of-english-words, ~275k words) as the source of truth.
+//
+// Usage:
+//   node scripts/find-missing-combo-words.mjs          # report only
+//   node scripts/find-missing-combo-words.mjs --write  # save scripts/combo-words.txt
+import fs from 'node:fs';
+import { createRequire } from 'node:module';
+import { buildWordSet } from './lexicon-sources.mjs';
+import { TARGET_WORDS_POOL, CURATED_PUZZLES } from '../src/data/puzzles.js';
+
+const require = createRequire(import.meta.url);
+const REFERENCE = require('an-array-of-english-words');
+const OUT_FILE = 'scripts/combo-words.txt';
+
+// Every tile that can appear on a board (random + daily puzzles)
+const chunks = new Set();
+TARGET_WORDS_POOL.forEach(t => t.tiles.forEach(c => chunks.add(c.toLowerCase())));
+CURATED_PUZZLES.forEach(p => p.quartets.flat().forEach(c => chunks.add(c.toLowerCase())));
+const maxChunkLen = Math.max(...[...chunks].map(c => c.length));
+
+// Base dictionary WITHOUT the previously generated combo list, so re-runs are stable
+const dictionary = buildWordSet({ includeComboWords: false });
+
+// Can `word` be split into 1-4 distinct pool tiles? Returns the split or null.
+function segment(word, start = 0, used = []) {
+  if (start === word.length) return used.length ? [...used] : null;
+  if (used.length === 4) return null;
+  for (let len = 1; len <= maxChunkLen && start + len <= word.length; len++) {
+    const piece = word.slice(start, start + len);
+    if (!chunks.has(piece) || used.includes(piece)) continue;
+    used.push(piece);
+    const result = segment(word, start + len, used);
+    used.pop();
+    if (result) return result;
+  }
+  return null;
+}
+
+const found = [];
+for (const raw of REFERENCE) {
+  const w = raw.toLowerCase();
+  if (w.length < 3 || w.length > 16 || !/^[a-z]+$/.test(w)) continue; // 2-letter words use the fixed allowlist
+  if (dictionary.has(w)) continue;
+  const split = segment(w);
+  if (split) found.push({ word: w, split });
+}
+found.sort((a, b) => a.word.localeCompare(b.word));
+
+console.log(`Pool tiles: ${chunks.size}`);
+console.log(`Missing words buildable from pool tiles: ${found.length}`);
+const byTiles = { 1: 0, 2: 0, 3: 0, 4: 0 };
+found.forEach(f => byTiles[f.split.length]++);
+console.log('By tile count:', byTiles);
+console.log('Sample:', found.slice(0, 40).map(f => `${f.word}(${f.split.join('+')})`).join(' '));
+
+if (process.argv.includes('--write')) {
+  const header = [
+    '# Words buildable from pool tiles that are missing from wordlist-english.',
+    '# Generated by scripts/find-missing-combo-words.mjs (source: an-array-of-english-words).',
+    '# These are accepted in play but never shown in the 30-word solution lists.',
+    '# Delete a line to stop accepting that word.'
+  ];
+  fs.writeFileSync(OUT_FILE, [...header, ...found.map(f => f.word)].join('\n') + '\n', 'utf8');
+  console.log(`Saved ${found.length} words to ${OUT_FILE}`);
+}