release: OpenNexus 0.5.0
This commit is contained in:
@@ -33,7 +33,7 @@ def get_settings() -> Settings:
|
||||
data_dir = Path(os.getenv("APP_DATA_DIR", str(BACKEND_DIR / "data")))
|
||||
return Settings(
|
||||
name=os.getenv("APP_NAME", "OpenNexus AI Core"),
|
||||
version=os.getenv("APP_VERSION", "0.1.0"),
|
||||
version=os.getenv("APP_VERSION", "0.5.0"),
|
||||
environment=os.getenv("APP_ENVIRONMENT", "development"),
|
||||
host=os.getenv("APP_HOST", "127.0.0.1"),
|
||||
port=int(os.getenv("APP_PORT", "8000")),
|
||||
|
||||
@@ -1250,6 +1250,12 @@ class TranscriptNoteRequest(Contract):
|
||||
include_speakers: bool = True
|
||||
|
||||
|
||||
class TranscriptArtifactsRequest(TranscriptNoteRequest):
|
||||
provider_id: str = Field(min_length=1, max_length=128)
|
||||
model: str = Field(min_length=1, max_length=256)
|
||||
knowledge_title: str | None = Field(default=None, min_length=1, max_length=200)
|
||||
|
||||
|
||||
class IndexStatus(Contract):
|
||||
running_jobs: int = 0
|
||||
active_searches: int = 0
|
||||
|
||||
@@ -5,11 +5,14 @@ import os
|
||||
import shutil
|
||||
import subprocess
|
||||
|
||||
from app.config import BACKEND_DIR
|
||||
from app.config import BACKEND_DIR, get_settings
|
||||
from app.errors import ApiError
|
||||
from app.local_models.process import ThreadedProcess
|
||||
|
||||
ROOT = BACKEND_DIR / '.venv-models-cuda'
|
||||
# 模型运行环境会在安装与升级时写入大量文件，必须位于应用数据目录，
|
||||
# 不能写入受完整性清单保护的 Core 发布目录。
|
||||
DEFAULT_ROOT = get_settings().data_dir / 'model-runtime'
|
||||
ROOT = DEFAULT_ROOT
|
||||
state = {'status': 'unchecked', 'stage': '', 'cuda_available': None}
|
||||
task = None
|
||||
|
||||
|
||||
@@ -11,7 +11,7 @@ from uuid import uuid4
|
||||
from fastapi import APIRouter, Header, Query, Request
|
||||
from fastapi.responses import FileResponse, StreamingResponse
|
||||
|
||||
from app.contracts import TranscriptEditRequest, TranscriptNoteRequest, TranscriptionJob
|
||||
from app.contracts import TranscriptArtifactsRequest, TranscriptEditRequest, TranscriptNoteRequest, TranscriptionJob
|
||||
from app.database.db import connect, transaction
|
||||
from app.errors import ApiError
|
||||
from app.services import transcription_service as jobs
|
||||
@@ -160,6 +160,12 @@ async def create_note(job_id: str, request: TranscriptNoteRequest):
|
||||
return await create_transcript_note(job_id, request)
|
||||
|
||||
|
||||
@router.post("/transcriptions/{job_id}/artifacts", status_code=201)
|
||||
async def create_artifacts(job_id: str, request: TranscriptArtifactsRequest):
|
||||
from app.services.media_notes import create_transcript_artifacts
|
||||
return await create_transcript_artifacts(job_id, request)
|
||||
|
||||
|
||||
@router.get("/attachments/{attachment_id}/cleanup-impact")
|
||||
async def cleanup_impact(attachment_id: str):
|
||||
attachment_path(attachment_id)
|
||||
|
||||
@@ -41,7 +41,7 @@ async def execute(call, request):
|
||||
run = await container.agent.create_run(AgentRunCreateRequest(
|
||||
input=task, provider_id=request.provider_id, model=request.model,
|
||||
skill_id=skill_id,
|
||||
allowed_tools=ALLOWED_TOOLS, max_steps=10, token_budget=16000,
|
||||
allowed_tools=ALLOWED_TOOLS, max_steps=10, token_budget=None,
|
||||
allow_network=False, metadata={'source': 'chat', 'conversation_id': request.conversation_id},
|
||||
))
|
||||
elif call.name == 'agent.status':
|
||||
|
||||
@@ -4,8 +4,11 @@ import hashlib
|
||||
from contextlib import closing
|
||||
|
||||
from app.config import get_settings
|
||||
from app.contracts import Message, MessageRole, ModelRequest, TranscriptNoteRequest
|
||||
from app.database.db import connect, transaction
|
||||
from app.errors import ApiError
|
||||
from app.providers.base import ProviderError
|
||||
from app.providers.registry import ProviderNotFoundError
|
||||
from app.services import note_service
|
||||
from app.services.transcription_service import require_job
|
||||
|
||||
@@ -66,6 +69,136 @@ async def create_transcript_note(job_id, options):
|
||||
return note
|
||||
|
||||
|
||||
def _transcript_text(job) -> str:
|
||||
if job.segments:
|
||||
rows = []
|
||||
for segment in job.segments:
|
||||
speaker = job.speaker_names.get(segment.speaker, segment.speaker) if segment.speaker else ""
|
||||
stamp = f"{int(segment.start_time // 60):02}:{int(segment.start_time % 60):02}"
|
||||
rows.append(f"[{stamp}] {speaker}：{segment.text}" if speaker else f"[{stamp}] {segment.text}")
|
||||
return "\n".join(rows)
|
||||
return job.text or ""
|
||||
|
||||
|
||||
def _chunks(text: str, limit: int = 12000) -> list[str]:
|
||||
"""按段落切分长转录，避免在中间截断句子。"""
|
||||
paragraphs = [part.strip() for part in text.splitlines() if part.strip()]
|
||||
if not paragraphs:
|
||||
return []
|
||||
chunks: list[str] = []
|
||||
current: list[str] = []
|
||||
size = 0
|
||||
for paragraph in paragraphs:
|
||||
if current and size + len(paragraph) + 1 > limit:
|
||||
chunks.append("\n".join(current))
|
||||
current, size = [], 0
|
||||
if len(paragraph) > limit:
|
||||
if current:
|
||||
chunks.append("\n".join(current))
|
||||
current, size = [], 0
|
||||
chunks.extend(paragraph[index:index + limit] for index in range(0, len(paragraph), limit))
|
||||
continue
|
||||
current.append(paragraph)
|
||||
size += len(paragraph) + 1
|
||||
if current:
|
||||
chunks.append("\n".join(current))
|
||||
return chunks
|
||||
|
||||
|
||||
async def _complete(provider_id: str, model: str, system: str, content: str) -> str:
|
||||
from app.container import container
|
||||
try:
|
||||
provider = container.providers.get(provider_id).adapter
|
||||
except ProviderNotFoundError as exc:
|
||||
raise ApiError(404, "PROVIDER_NOT_FOUND", "所选模型提供商不存在或未启用。",
|
||||
{"provider_id": provider_id}) from exc
|
||||
try:
|
||||
turn = await provider.complete(ModelRequest(
|
||||
provider_id=provider_id,
|
||||
model=model,
|
||||
system=system,
|
||||
messages=[Message(role=MessageRole.user, content=content)],
|
||||
temperature=0.2,
|
||||
))
|
||||
except ProviderError as exc:
|
||||
raise ApiError(502, exc.code, exc.message, {"provider_id": provider_id}) from exc
|
||||
if not turn.text or not turn.text.strip():
|
||||
raise ApiError(502, "KNOWLEDGE_NOTE_EMPTY", "模型没有返回知识点笔记。")
|
||||
return turn.text.strip().removeprefix("```markdown").removeprefix("```").removesuffix("```").strip()
|
||||
|
||||
|
||||
async def _knowledge_markdown(job, provider_id: str, model: str, title: str) -> str:
|
||||
transcript = _transcript_text(job)
|
||||
if not transcript.strip():
|
||||
raise ApiError(409, "TRANSCRIPT_EMPTY", "转录内容为空，无法提取知识点。")
|
||||
system = (
|
||||
"你是一名严谨的课程笔记整理助手。只能依据提供的转录内容整理，不补写未出现的事实。"
|
||||
"输出中文 Markdown 正文，使用清晰的二级、三级标题；包含课程主题、核心概念、关键论证或步骤、"
|
||||
"重要例子、待复习问题。合并口语重复，保留专业术语和必要条件。不要使用代码围栏，也不要写处理说明。"
|
||||
)
|
||||
parts = _chunks(transcript)
|
||||
summaries: list[str] = []
|
||||
for index, part in enumerate(parts, 1):
|
||||
summaries.append(await _complete(
|
||||
provider_id, model, system,
|
||||
f"这是课程转录的第 {index}/{len(parts)} 部分。请提取可供最终整合的知识点：\n\n{part}",
|
||||
))
|
||||
if len(summaries) == 1:
|
||||
body = summaries[0]
|
||||
else:
|
||||
body = await _complete(
|
||||
provider_id, model, system,
|
||||
"请将以下分段知识点合并成一篇完整课程笔记，消除重复并保持逻辑顺序：\n\n"
|
||||
+ "\n\n".join(f"### 分段 {index}\n{summary}" for index, summary in enumerate(summaries, 1)),
|
||||
)
|
||||
return "\n".join([
|
||||
f"<!-- knowledge-note:{job.job_id}:{job.revision}:{provider_id}:{model} -->",
|
||||
f"# {title}", "", f"[查看完整转录稿](/#/media?job={job.job_id})", "", body,
|
||||
])
|
||||
|
||||
|
||||
async def create_transcript_artifacts(job_id, options):
|
||||
"""为完成的转录生成可回听的全文和模型整理的知识点笔记。"""
|
||||
transcript_options = TranscriptNoteRequest(
|
||||
title=options.title,
|
||||
folder=options.folder,
|
||||
update_existing=options.update_existing,
|
||||
include_timestamps=options.include_timestamps,
|
||||
include_speakers=options.include_speakers,
|
||||
)
|
||||
transcript_note = await create_transcript_note(job_id, transcript_options)
|
||||
job = require_job(job_id)
|
||||
knowledge_title = options.knowledge_title or f"{options.title} · 知识点"
|
||||
identity = (str(get_settings().db_path), job_id, "knowledge")
|
||||
lock = _locks.setdefault(identity, asyncio.Lock())
|
||||
async with lock:
|
||||
signature = "knowledge:" + hashlib.sha256(options.model_copy(update={
|
||||
"update_existing": False,
|
||||
"knowledge_title": knowledge_title,
|
||||
}).model_dump_json(exclude={"update_existing"}).encode()).hexdigest()
|
||||
with closing(connect()) as conn:
|
||||
row = conn.execute(
|
||||
"SELECT note_id FROM media_notes WHERE job_id=? AND revision=? AND options_hash=?",
|
||||
(job_id, job.revision, signature),
|
||||
).fetchone()
|
||||
if row:
|
||||
knowledge_note = await note_service.get_note(row[0])
|
||||
if knowledge_note is not None:
|
||||
return {"transcript": transcript_note, "knowledge_note": knowledge_note}
|
||||
markdown = await _knowledge_markdown(job, options.provider_id, options.model, knowledge_title)
|
||||
note_title = f"{knowledge_title} · {job_id[-8:]}-r{job.revision}-{signature[-6:]}"
|
||||
knowledge_note = await note_service.create_note(
|
||||
title=note_title,
|
||||
markdown=markdown,
|
||||
folder=options.folder,
|
||||
tags=["课程笔记", "知识点"],
|
||||
)
|
||||
with closing(connect()) as conn, transaction(conn):
|
||||
conn.execute("INSERT OR IGNORE INTO media_notes VALUES (?,?,?,?)",
|
||||
(job_id, job.revision, signature, knowledge_note.note_id))
|
||||
return {"transcript": transcript_note, "knowledge_note": knowledge_note}
|
||||
|
||||
|
||||
async def _create_note(title, markdown, options, marker):
|
||||
try:
|
||||
note = await note_service.create_note(title=title, markdown=markdown, folder=options.folder, tags=["转写"])
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
[project]
|
||||
name = "notes-agent-backend"
|
||||
version = "0.1.0"
|
||||
version = "0.5.0"
|
||||
description = "Notes Agent 的 FastAPI 基础壳子"
|
||||
readme = "README.md"
|
||||
requires-python = ">=3.11"
|
||||
|
||||
@@ -96,6 +96,18 @@ def test_terminology_export_and_privacy_cleanup():
|
||||
first = client.post(f'/api/media/transcriptions/{job_id}/notes', json={'title':'课程'}).json()
|
||||
again = client.post(f'/api/media/transcriptions/{job_id}/notes', json={'title':'课程'}).json()
|
||||
assert first['note_id'] == again['note_id']
|
||||
artifacts = client.post(f'/api/media/transcriptions/{job_id}/artifacts', json={
|
||||
'title': '课程', 'knowledge_title': '课程知识点',
|
||||
'provider_id': 'mock', 'model': 'mock-1',
|
||||
})
|
||||
assert artifacts.status_code == 201
|
||||
assert artifacts.json()['transcript']['note_id'] == first['note_id']
|
||||
assert artifacts.json()['knowledge_note']['note_id'] != first['note_id']
|
||||
repeated = client.post(f'/api/media/transcriptions/{job_id}/artifacts', json={
|
||||
'title': '课程', 'knowledge_title': '课程知识点',
|
||||
'provider_id': 'mock', 'model': 'mock-1',
|
||||
})
|
||||
assert repeated.json()['knowledge_note']['note_id'] == artifacts.json()['knowledge_note']['note_id']
|
||||
response = client.delete('/api/media/attachments/lecture.txt')
|
||||
assert first['note_id'] in response.json()['retained_note_ids']
|
||||
cleaned = client.get(f'/api/media/transcriptions/{job_id}').json()
|
||||
|
||||
@@ -13,8 +13,6 @@ def isolate(monkeypatch, tmp_path):
|
||||
monkeypatch.setattr(components, 'ROOT', tmp_path / 'cuda')
|
||||
monkeypatch.setattr(components, 'state', {'status': 'unchecked', 'stage': '', 'cuda_available': None})
|
||||
monkeypatch.setattr(components, 'task', None)
|
||||
|
||||
|
||||
def test_status_checks_without_installing_and_detects_existing_cuda(monkeypatch):
|
||||
python = components.ROOT / 'Scripts/python.exe'
|
||||
python.parent.mkdir(parents=True)
|
||||
|
||||
Generated
+1
-1
@@ -1108,7 +1108,7 @@ wheels = [
|
||||
|
||||
[[package]]
|
||||
name = "notes-agent-backend"
|
||||
version = "0.1.0"
|
||||
version = "0.5.0"
|
||||
source = { virtual = "." }
|
||||
dependencies = [
|
||||
{ name = "cryptography" },
|
||||
|
||||
Reference in New Issue
Block a user