fix(#110 follow-up): align CLI to PR-2 server schema — read prep.code, widen regex

[s2agi]

Author

Agent: 通信工程马

Summary

测试马 PR-5 Case 2 caught a schema mismatch between PR-2 (server, landed) and PR-3 (my CLI in #225 / commit f28ffd9). anet node rename node-c node-d --force on a purely-created node — exactly the case #110 was meant to fix — still hard-rolled-back with rc=1 instead of doing the local-only rename.

Root cause

Server PR-2 (server/src/rename.ts:117-124) ships:

{
  "ok": false,
  "code": "node_local_only",
  "error": "node 'X' has no server session in this network",
  "suggested": "rename locally"
}

My PR-3 (this file, pre-fix) checked:

prep.error === "node_local_only"                  // wrong field — error is the message
|| /not found in this network/i.test(errStr)       // wrong wording
|| /node .* not found/i.test(errStr)               // generic

→ isLocalOnly = false → throw → PHASE 1 rollback → rc=1.

Fix

Switch to prep.code as the primary signal (matches PR-2 contract), keep prep.error checks as belt-and-braces, widen the regex to include the server's actual wording:

isLocalOnly =
    prep.code === "node_local_only"            // PR-2 contract (primary)
 || prep.error === "node_local_only"           // legacy / older builds
 || /node_local_only/i.test(errStr)            // belt-and-braces
 || /has no server session/i.test(errStr)      // server's actual phrase
 || /not found in this network/i.test(errStr)  // pre-PR-2 wording
 || /node .* not found/i.test(errStr)          // generic fallback

Smoke

Bundle marker verification (post npm run build w/ javascript-obfuscator):

String	Count	Source
node_local_only	2	new prep.code check + regex literal
has no server session	1	new regex literal (matches server's actual phrase)
not found in this network	1	legacy regex retained

Full integration verify deferred to 测试马 PR-5 Case 2 rerun (their harness has the live commhub-server + the create→rename sequence).

Refs

• [bug][P2] anet node rename 对 purely-created 节点失败（与 RFC §4.1 推荐路径矛盾） #110 (the RFC-010 §4.1 "created is recommended rename path" follow-up that this is supposed to honor)
• 🔴 P0 rename bug — rename 后消息发不到新 alias 节点 (root cause + Docker 5-case repro + 方案) #146 PR-3 (feat(#146 PR-3): CLI rename lane — local-only / codex+grok pid match / NODE_ID env #225 / commit f28ffd9 — root cause)
• 测试马 PR-5 Case 2 (the regression detector)

Test plan

• bunx tsc --noEmit clean
• npm run build (bun + obfuscator x3) 14s clean
• Bundle marker grep — all 4 schema strings present
• 测试马 PR-5 Case 2 rerun: anet node create node-c (no start) → anet node rename node-c node-d --force → expect rc=0, .anet/nodes/node-d exists, .anet/nodes/node-c gone

🤖 Generated with Claude Code

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: c4bd660bee

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[chatgpt-codex-connector]

Check remote alias conflicts before local-only rename

When the old node has no sessions row, accepting prep.code === "node_local_only" skips the only server-side check that new_alias is free. In the current server path, prepareRename returns node_local_only before testing newAlias against sessions, so a never-started local node can now be renamed to an alias that is already registered by another machine in the same network; when this renamed node is later started, report_status deletes other same-alias sessions in that network, effectively evicting the real node. Please either have the local-only path verify the target alias against the hub or require the server response to guarantee that the target alias is unused.

Useful? React with 👍 / 👎.