feat(tools-pve/asr-pipeline): SQLite model config, task delete/rerefine, multi-recording review, lexicon CRUD

servers/tools-pve/asr-pipeline/asrservice/worker.py

@@ -4,6 +4,7 @@ import fcntl
 import hashlib
 import json
 import os
+import re
 import shutil
 import sqlite3
 import time
@@ -11,9 +12,10 @@ import uuid
 from pathlib import Path
 
 from asrpipe import graph, lexicon, providers, refine, review, revise, vault
+from asrpipe.config import CFG
 from . import db
 from .config import (
-    DATA, DEFAULT_PROVIDER, DRAFTS, FAILED, INBOX, LOGS, MAX_ATTEMPTS,
+    DATA, DRAFTS, FAILED, INBOX, LOGS, MAX_ATTEMPTS,
     POLL_SECONDS, PROCESSED, PROCESSING, PUBLIC_URL, REVISION_QUEUE,
     STABLE_SECONDS, STATE, SUPPORTED, TRANSCRIPTS, ensure_dirs,
 )
@@ -50,7 +52,7 @@ def transcribe(
 ) -> tuple[Path, providers.TranscribeResult]:
     res = providers.transcribe_audio(
         audio,
-        provider=provider or DEFAULT_PROVIDER,
+        provider=provider or CFG.asr_provider,
         model=model,
         title=title,
         user_context=user_context,
@@ -71,7 +73,7 @@ def transcribe(
 def process_audio(source: Path, job: dict) -> None:
     job_id = job["id"]
     attempts = int(job.get("attempts") or 0) + 1
-    prov = providers.normalize_provider_name(job.get("asr_provider") or DEFAULT_PROVIDER)
+    prov = providers.normalize_provider_name(job.get("asr_provider") or CFG.asr_provider)
     model = str(job.get("asr_model") or "")
     digest = sha256(source)
     prov_label = providers.PROVIDERS.get(prov, {}).get("label", prov)
@@ -120,6 +122,47 @@ def process_audio(source: Path, job: dict) -> None:
          f"{PUBLIC_URL}/jobs/{job_id}/review/")
 
 
+def rerefine_job(job: dict) -> None:
+    """Re-run LLM refinement on an existing raw transcript without re-running ASR."""
+    job_id = job["id"]
+    transcript = Path(str(job.get("transcript") or ""))
+    if not transcript.is_file():
+        raise FileNotFoundError(f"transcript file missing for job {job_id}: {transcript}")
+    raw = transcript.read_text(encoding="utf-8", errors="ignore")
+    intake = {
+        "job_id": job_id,
+        "title": job.get("title", ""),
+        "context": job.get("context", ""),
+        "source": job.get("source", ""),
+        "asr_provider": job.get("asr_provider") or CFG.asr_provider,
+        "asr_model": job.get("asr_model") or "",
+    }
+    intake_line = f"<!-- intake: {json.dumps(intake, ensure_ascii=False)} -->"
+    if re.match(r"^<!--\s*intake:\s*\{.*?\}\s*-->", raw):
+        raw = re.sub(r"^<!--\s*intake:\s*\{.*?\}\s*-->", intake_line, raw, count=1)
+    else:
+        raw = intake_line + "\n" + raw
+    transcript.write_text(raw, encoding="utf-8")
+
+    db.update_job(job_id, "refining", "正在基于现有转写重新生成精修草稿")
+    review.delete_for_job(
+        transcript_path=str(transcript),
+        draft_path=str(job.get("draft") or ""),
+        job_id=job_id,
+    )
+    result = refine.refine(
+        raw, date=time.strftime("%Y-%m-%d"),
+        transcript_path=str(transcript), queue_review=True,
+    )
+    draft = DRAFTS / f"{job_id}.md"
+    vault.write_draft_file(draft, result.draft)
+    questions = review.enqueue_draft(str(transcript), draft)
+    db.update_job(
+        job_id, "review", f"整理稿已重新生成，{questions} 个问题待确认",
+        transcript=str(transcript), draft=str(draft),
+    )
+
+
 def claim(path: Path) -> tuple[Path, dict]:
     job = db.find_by_stored(path.name)
     if not job:
@@ -127,7 +170,7 @@ def claim(path: Path) -> tuple[Path, dict]:
         job = db.create_job(
             job_id, path.name, path.name, title=path.stem,
             size=path.stat().st_size, source="ssh",
-            asr_provider=DEFAULT_PROVIDER,
+            asr_provider=CFG.asr_provider,
         )
     target = PROCESSING / path.name
     if target.exists():
@@ -164,8 +207,15 @@ def process_revisions() -> None:
     for marker in sorted(REVISION_QUEUE.glob("*.json")):
         try:
             payload = json.loads(marker.read_text(encoding="utf-8"))
-            draft = Path(payload["draft"])
+            action = str(payload.get("action") or "revise")
             job_id = str(payload.get("job_id") or "")
+            if action == "rerefine":
+                job = db.get_job(job_id)
+                if job:
+                    rerefine_job(job)
+                marker.unlink(missing_ok=True)
+                continue
+            draft = Path(payload["draft"])
             db.update_job(job_id, "publishing", "正在应用人工确认")
             revise.revise(draft)
             current = db.get_job(job_id) or {}
@@ -181,6 +231,14 @@ def process_revisions() -> None:
             marker.unlink(missing_ok=True)
         except Exception as exc:
             log(f"revision marker={marker.name} failed type={type(exc).__name__} err={exc}")
+            try:
+                payload = json.loads(marker.read_text(encoding="utf-8"))
+                jid = str(payload.get("job_id") or "")
+                if jid:
+                    db.update_job(jid, "failed", f"精修/修订失败：{exc}", error_code=type(exc).__name__)
+            except Exception:
+                pass
+            marker.unlink(missing_ok=True)
 
 
 def recover() -> None: