fix(asr): 过滤 GLM-ASR 偶发的纯井号占位转写（#787）

[H-Chris233]

User description

修复 issue #787：智谱 GLM-ASR 作为供应商时，静音/弱音片段偶发把整段识别为单个 #\（或 ##\…），当前 Whisper 兼容客户端仅 trim 后当作有效转写，导致占位符被插入用户输入。

改动

• 在 \�sr/whisper.rs\ 新增 \is_placeholder_heading\：整段仅由 #\ 组成（#/##/###\…）即视为占位结果。
• \WhisperBatchASR::transcribe_chunk\ 普通分支与 \�xtract_confident_text\ 回退分支均应用该判定，命中归一化为空转写。
• 空转写自然进入既有空转写护栏（「没有识别到语音」+ emptyTranscript 历史），占位符不再落到输入框；多分片场景仅丢弃异常分片，\C#\、# 你好\ 等真实内容不受影响。

测试

新增 4 个单元测试（纯 #/##/### 归一化为空、混合分片丢弃占位、全占位整体为空、正常内容不受影响），\cargo test -p openless asr::whisper\ 33 个测试全部通过。

---

PR Type

Bug fix, Tests

---

Description

• Filter GLM-ASR pure # placeholder transcriptions.

• Normalize placeholder chunks to empty transcripts.

• Add unit tests covering placeholder scenarios.

• Guard real transcripts with other characters.

---

Diagram Walkthrough

flowchart LR
A["ASR response text"] --> B{"is_placeholder_heading?"}
B -- "yes" --> C["Empty transcript"]
B -- "no" --> D["Normal transcript"]
C --> E["Existing empty guard"]

Loading

File Walkthrough

	Relevant files
Bug fix	whisper.rs Filter pure hash placeholders in ASR transcriptions openless-all/app/src-tauri/src/asr/whisper.rs Added is_placeholder_heading helper for pure # text. Applied empty normalization in transcribe_chunk and fallback. Added tests for hash-only, mixed, and all-placeholder chunks. Preserved real content like C# and # 你好. +103/-2

---

[github-actions]

PR Reviewer Guide 🔍

Here are some key observations to aid the review process:

🎫 Ticket compliance analysis ✅ 787 - PR Code Verified Compliant requirements: 非 verbose_json 分支对纯 '#' 文本归一化为空转写。 extract_confident_text 无 segments 回退分支同样对纯 '#' 归一化。 空转写走既有护栏，占位符不会插入输入框。 新增 4 个单元测试覆盖纯 '#/##/###'、混合分片丢弃、全占位为空、正常内容不受影响。 Requires further human verification: 需要在实际 GLM-ASR 供应商上做端到端/UI 验证，确认多次语音输入不再出现 '#'。 若 GLM-ASR 在 verbose_json 模式下会返回包含纯 '#' 文本的 segments，需确认现有 segments 置信度过滤是否仍会漏过；当前 filter 未在 segments 循环内应用。

⏱️ Estimated effort to review: 2 🔵🔵⚪⚪⚪

🧪 PR contains tests

🔒 No security concerns identified

⚡ Recommended focus areas for review Coverage Gap extract_confident_text only applies is_placeholder_heading in the no-segments fallback. If a verbose_json response contains a segment whose text is #/##, that segment can still be appended to kept and surface a placeholder unless the existing confidence filters happen to drop it. This is a coverage gap rather than a regression; it matters only if GLM-ASR can return such segments. Consider applying is_placeholder_heading per segment, or to the final joined text, to close this path. letmut kept = String::new();for seg in segments {let text = seg.get("text").and_then(|t| t.as_str()).unwrap_or("");if text.trim().is_empty(){continue;}let no_speech = seg .get("no_speech_prob").and_then(|v| v.as_f64()).unwrap_or(0.0);let avg_logprob = seg .get("avg_logprob").and_then(|v| v.as_f64()).unwrap_or(0.0);let compression = seg .get("compression_ratio").and_then(|v| v.as_f64()).unwrap_or(1.0);let is_hallucination = (no_speech > 0.6 && avg_logprob < -0.5) || compression > 2.4 || avg_logprob < -1.0;if is_hallucination { log::warn!("[whisper] 丢弃疑似幻听段落: no_speech={:.2} avg_logprob={:.2} compression={:.2} text={:?}", no_speech, avg_logprob, compression, text.trim());continue;} kept.push_str(text);}let kept = kept.trim().to_string();