From 89bf26ba9dc41e49c0e9f7f7d7e6520db92860d8 Mon Sep 17 00:00:00 2001
From: codewarnab <dp950810@gmail.com>
Date: Mon, 19 Oct 2026 07:54:11 +0000
Subject: [PATCH] test: cover emoji and maximum query browser parity

---
 docs/IMPROVEMENT-PLAN.md   |  1 +
 scripts/test-regression.ts | 29 ++++++++++++++++++++++++++---
 2 files changed, 27 insertions(+), 3 deletions(-)

diff --git a/docs/IMPROVEMENT-PLAN.md b/docs/IMPROVEMENT-PLAN.md
index 201483c..de0a6f2 100644
--- a/docs/IMPROVEMENT-PLAN.md
+++ b/docs/IMPROVEMENT-PLAN.md
@@ -25,6 +25,7 @@ Low-risk housekeeping that clears the way for the bigger phases.
 There are about 300 tests, but several core modules have none of their own. This phase covers the guarantees that 1.0 commits to.
 
 ### 2.1 Tests that GPU and CPU return the same scores
+- **Browser coverage extended:** the existing `scripts/test-regression.ts` differential gate now includes joined emoji and component queries, plus 128-token ASCII and case-fold-expanded queries in substring and fuzzy modes. Matching records and last-token near misses exercise the actual WGSL path against the CPU reference. CI already runs this gate on every PR.
 - Add `test/scoring-parity.test.ts`. It runs a fixed set of queries through the mock WebGPU path and through `cpu-reference.ts` / `exact-scorer.ts`.
 - Check that both return the same `SearchResultItem[]`: same order, same integer scores, scores in descending order.
 - Cover ASCII, CJK, emoji/ZWJ, mixed case, and queries at the maximum length.
diff --git a/scripts/test-regression.ts b/scripts/test-regression.ts
index 41c7d85..649d6f4 100644
--- a/scripts/test-regression.ts
+++ b/scripts/test-regression.ts
@@ -1,7 +1,7 @@
 import puppeteer from 'puppeteer-core';
 import path from 'path';
 import fs from 'fs';
-import { normalizeText, scoreExactMatches } from '../packages/webgpu-search/src/index';
+import { normalizeText, scoreExactMatches, QUERY_TOKENS_MAX } from '../packages/webgpu-search/src/index';
 import {
     getChromeExecutablePath,
     getChromeLaunchArgs,
@@ -180,6 +180,12 @@ async function main() {
         // agree on NFC for this ASCII-heavy corpus (negligible risk here;
         // cross-runtime NFC is probed/reported, not gated).
         const FCP = String.fromCodePoint;
+        const joinedEmoji = FCP(0x1f469, 0x200d, 0x1f4bb);
+        // A distinct final token catches dispatch/scoring paths that truncate
+        // the query at 127 tokens. The folded query reaches the same cap with
+        // only 64 input characters, so the limit is measured after folding.
+        const maxQuery = 'a'.repeat(QUERY_TOKENS_MAX - 1) + 'z';
+        const foldedMaxQuery = FCP(0xdf).repeat(QUERY_TOKENS_MAX / 2);
         const parityStrings: string[] = [];
         for (let i = 0; i < 200; i++) {
             parityStrings.push(`src/services/Auth${i % 7 === 0 ? FCP(0xdf) : 's'}Controller_${i}.ts`);
@@ -189,7 +195,16 @@ async function main() {
             FCP(0x5317) + FCP(0x4eac) + 'city',
             FCP(0x1f600) + 'smile',
             'caf' + FCP(0xe9),
-            'plain'
+            'plain',
+            joinedEmoji,
+            'Dev ' + joinedEmoji + ' at work',
+            FCP(0x1f469) + ' ' + FCP(0x1f4bb), // Separate scalars are not the ZWJ sequence.
+            maxQuery,
+            'prefix ' + maxQuery + ' suffix',
+            'a'.repeat(QUERY_TOKENS_MAX - 1) + 'x', // Differs only at the last token.
+            foldedMaxQuery,
+            'ss'.repeat(QUERY_TOKENS_MAX / 2),
+            'ss'.repeat(QUERY_TOKENS_MAX / 2 - 1) + 'sx'
         );
         const parityCells = [
             { query: 'auth', mode: 'substring', limit: 50 },
@@ -200,6 +215,14 @@ async function main() {
             { query: 'SS', mode: 'substring', limit: 50 },
             { query: FCP(0x5317), mode: 'substring', limit: 10 },
             { query: FCP(0x1f600), mode: 'substring', limit: 10 },
+            { query: joinedEmoji, mode: 'substring', limit: 10 },
+            { query: joinedEmoji, mode: 'fuzzy', limit: 10 },
+            { query: FCP(0x1f4bb), mode: 'substring', limit: 10 },
+            { query: FCP(0x1f4bb), mode: 'fuzzy', limit: 10 },
+            { query: maxQuery, mode: 'substring', limit: 10 },
+            { query: maxQuery, mode: 'fuzzy', limit: 10 },
+            { query: foldedMaxQuery, mode: 'substring', limit: 10 },
+            { query: foldedMaxQuery, mode: 'fuzzy', limit: 10 },
             { query: 'zzz-no-match', mode: 'substring', limit: 50 },
             { query: 'zzz-no-match', mode: 'fuzzy', limit: 50 },
         ] as Array<{ query: string; mode: 'substring' | 'fuzzy'; limit: number }>;
@@ -258,7 +281,7 @@ async function main() {
                         }
                     }
                     (testResults as any).results.push({
-                        name: `parity ${cell.mode} ${JSON.stringify(cell.query.slice(0, 12))} (ordered index/score/text)`,
+                        name: `parity ${cell.mode} ${JSON.stringify(cell.query.slice(0, 12))} (${qT.length} tokens; ordered index/score/text)`,
                         passed: pass,
                         details: { totalMatches: got.totalMatches, candidateCount: got.candidateCount, expected: expected.totalMatches }
                     });