fix(bailian): 修复 end_time fallback 将 interim 误判为 final 的 bug

[H-Chris233]

User description

背景

PR #535 合并后，@github-actions 的 review comment 指出两个 concern：

1. ID collision — sentence_id == 0 覆盖 → 理论正确，实际 DashScope 始终提供递增 ID，低风险
2. Finality fallback — end_time > 0 误判 interim → 真实 bug

这个 fix 解决的是 concern 2

问题

原代码 line 397：

let is_sentence_final = sentence_end || end_time > 0;

DashScope 的 interim result-generated 事件同时携带：

• sentence_end: false
• end_time: 1500 （正数，随音频推进增长）

因此 is_sentence_final = false || true = true，interim 结果被误判为 final
存入 final_segments，reintroduce 了 #530 要修的累积文本重复。

修复

sentence_end 字段存在时完全信任它；end_time > 0 仅在该字段不存在时 fallback：

let sentence_end_val = sentence.get("sentence_end");
let is_sentence_final = match sentence_end_val {
    Some(_) => sentence_end,
    None => end_time > 0,
};

测试修正

make_result_event 之前给 interim 设 end_time: 0，不匹配真实 API 行为，
导致测试没有覆盖这个路径。修复后 interim 也带正数 end_time，
并新增 interim_with_positive_end_time_not_mistaken_for_final 测试。

测试结果

21 passed; 0 failed

---

PR Type

Bug fix, Tests

---

Description

• Skip heartbeat ASR events

• Trust sentence_end for finality

• Cache partials by sentence_id

• Merge overlaps and expand tests

---

Diagram Walkthrough

flowchart LR
  A["DashScope ASR events"] --> B["Filter heartbeat events"]
  B --> C["Detect finality via sentence_end"]
  C --> D["Track partials per sentence_id"]
  D --> E["Store finals in BTreeMap"]
  E --> F["Merge overlapping segments"]
  F --> G["Emit deduplicated transcript"]

Loading

File Walkthrough

Relevant files

Bug fix

bailian.rs Harden DashScope ASR deduplication flow                                    openless-all/app/src-tauri/src/asr/bailian.rs Skips heartbeat events before transcript processing. Uses sentence_end as the primary finality signal, with end_time as compatibility fallback. Tracks interim text in partial_segments and clears it after final results arrive. Adds merge_segments() to prevent overlapping duplicate text in the final transcript. +324/-9

---

[github-actions]

PR Reviewer Guide 🔍

Here are some key observations to aid the review process:

🎫 Ticket compliance analysis ✅ 535 - Fully compliant Compliant requirements: Skip heartbeat-only ASR events. Use sentence_end as the primary finality signal, with end_time as a fallback only when sentence_end is missing. Track interim results per sentence_id and clear them when the final result arrives. Keep final results even when sentence_id == 0. Merge overlapping final segments to avoid duplicated transcript text. Add tests covering heartbeat filtering, interim updates, duplicate finals, zero sentence_id, and ordered multi-segment assembly.

⏱️ Estimated effort to review: 3 🔵🔵🔵⚪⚪

🧪 PR contains tests

🔒 No security concerns identified

⚡ No major issues detected