From 0c41b879b34c0985f42482ea6f42434c40e080d8 Mon Sep 17 00:00:00 2001
From: Hassan Abdel-Rahman <hassan.abdelrahman@gmail.com>
Date: Tue, 30 Jun 2026 10:29:37 -0400
Subject: [PATCH] searchable codemod: add republish tool

republish.mjs republishes each published realm copy from its source
realm so the published snapshot picks up the source's searchable
annotations, then reads a known module back from the published realm to
assert the annotation landed. The env argument is checked against a
staging|prod allow-list before it is interpolated into the bash command
that sources the seed file. The run exits nonzero on a real failure or a
missing-searchable assertion, while a realm the server refuses to publish
("not publishable") is recorded as an expected skip and does not fail it.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0182qehwDYbwBmyYcAhMoij1
---
 .../scripts/codemod/searchable/republish.mjs  | 103 ++++++++++++++++++
 1 file changed, 103 insertions(+)
 create mode 100644 packages/realm-server/scripts/codemod/searchable/republish.mjs

diff --git a/packages/realm-server/scripts/codemod/searchable/republish.mjs b/packages/realm-server/scripts/codemod/searchable/republish.mjs
new file mode 100644
index 0000000000..b599d0e5fd
--- /dev/null
+++ b/packages/realm-server/scripts/codemod/searchable/republish.mjs
@@ -0,0 +1,103 @@
+// Republish each (source -> published) pair so the published snapshot picks up
+// the searchable annotations now in its source realm, then ASSERT the annotation
+// landed: `realm publish` waits for the published realm to be ready (fully
+// indexed), after which we read a known-changed module back from the published
+// realm and confirm it carries `searchable`.
+//
+//   node republish.mjs <pairs.json> <staging|prod> <out.jsonl> [--limit N]
+// You do not need to source the seed before running: each boxel-cli call sources
+// ~/.boxel-secrets/<env>.env itself, so the seed never enters this process.
+import { execFileSync } from 'node:child_process';
+import { readFileSync, writeFileSync, appendFileSync } from 'node:fs';
+import { fileURLToPath } from 'node:url';
+import { dirname, join } from 'node:path';
+
+let [pairsPath, env, out] = process.argv.slice(2);
+let limitIdx = process.argv.indexOf('--limit');
+let limit = limitIdx > -1 ? Number(process.argv[limitIdx + 1]) : Infinity;
+if (env !== 'staging' && env !== 'prod') {
+  // env is interpolated into the bash command that sources the seed file, so it
+  // must never carry arbitrary text.
+  throw new Error(
+    `env must be 'staging' or 'prod' (got ${JSON.stringify(env)})`,
+  );
+}
+const HERE = dirname(fileURLToPath(import.meta.url));
+// …/scripts/codemod/searchable -> repo root
+const REPO = join(HERE, '..', '..', '..', '..', '..');
+let BOXEL = join(REPO, 'packages', 'boxel-cli', 'bin', 'boxel.js');
+
+function boxel(args, timeout = 660000) {
+  let q = args.map((x) => `'${x.replace(/'/g, `'\\''`)}'`).join(' ');
+  let cmd = `set -a; . "$HOME/.boxel-secrets/${env}.env"; set +a; exec node ${JSON.stringify(BOXEL)} ${q}`;
+  return execFileSync('bash', ['-c', cmd], {
+    encoding: 'utf8',
+    timeout,
+    maxBuffer: 64 * 1024 * 1024,
+  });
+}
+
+let pairs = JSON.parse(readFileSync(pairsPath, 'utf8')).slice(0, limit);
+writeFileSync(out, '');
+let log = (o) => appendFileSync(out, JSON.stringify(o) + '\n');
+let ok = 0,
+  fail = 0,
+  skipped = 0,
+  assertFail = 0;
+process.stderr.write(`Republishing ${pairs.length} pair(s) on ${env}\n`);
+
+for (let [i, p] of pairs.entries()) {
+  try {
+    // Waits for the published realm to be ready (fully indexed) by default.
+    boxel([
+      'realm',
+      'publish',
+      p.source,
+      p.published,
+      '--realm-secret-seed',
+      '--force',
+      '--timeout',
+      '600000',
+    ]);
+    // Assert: the annotation is present in the published realm's source.
+    let r = JSON.parse(
+      boxel(['file', 'read', p.sampleModule, '--realm', p.published, '--json']),
+    );
+    let present = Boolean(r.ok && (r.content || '').includes('searchable'));
+    if (!present) assertFail++;
+    log({
+      published: p.published,
+      source: p.source,
+      sampleModule: p.sampleModule,
+      searchablePresent: present,
+    });
+    process.stderr.write(
+      `  ${present ? '✓' : '⚠'} ${p.published} (searchable in ${p.sampleModule}: ${present})\n`,
+    );
+    ok++;
+  } catch (e) {
+    let msg = String(e.message);
+    // A realm the server refuses to publish ("not publishable") is an expected
+    // skip, not a rollout failure — count it separately so it never trips the
+    // nonzero exit below.
+    let notPublishable = /not publishable/.test(msg);
+    if (notPublishable) skipped++;
+    else fail++;
+    log({
+      published: p.published,
+      source: p.source,
+      ...(notPublishable ? { skipped: true } : {}),
+      error: msg.split('\n').slice(0, 2).join(' | '),
+    });
+    process.stderr.write(
+      `  ${notPublishable ? '⊘ skip (not publishable):' : '✗'} ${p.published} — ${msg.split('\n')[0]}\n`,
+    );
+  }
+  process.stderr.write(`  …${i + 1}/${pairs.length}\n`);
+}
+process.stderr.write(
+  `\nDone. ${ok} republished, ${assertFail} missing-searchable, ${skipped} skipped (not publishable), ${fail} failed.\n`,
+);
+// Surface real failures to callers/CI; expected "not publishable" skips do not
+// fail the run.
+process.exit(fail > 0 || assertFail > 0 ? 1 : 0);