Merge branch 'master' of http://192.168.0.200/knowledge-base-bj/deerflow-server into dhk-1
This commit is contained in:
commit
3ea8e73389
61 files changed
+734
-429
No files matched your search
@@ -27,7 +27,7 @@ from app.gateway.routers.thread_runs import RunCreateRequest
|
||||
from app.gateway.routers.threads import ThreadCreateRequest, create_thread
|
||||
from app.gateway.routers.uploads import _make_file_sandbox_writable
|
||||
from app.gateway.services import start_run
|
||||
from deerflow.browser_context.tasks import claim_next_browser_task, complete_browser_task, prune_browser_tasks
|
||||
from deerflow.browser_context.tasks import claim_next_browser_task, complete_browser_task, prune_browser_tasks, tasks_dir
|
||||
from deerflow.config.app_config import AppConfig
|
||||
from deerflow.config.browser_context_config import BrowserContextConfig
|
||||
from deerflow.runtime import serialize_channel_values
|
||||
@@ -307,10 +307,19 @@ async def claim_browser_context_task(
|
||||
config: AppConfig = Depends(get_config),
|
||||
) -> BrowserContextTaskNextResponse:
|
||||
settings = _require_enabled(config)
|
||||
effective_user_id = get_effective_user_id()
|
||||
effective_worker_id = worker_id or "chrome-extension"
|
||||
logger.info(
|
||||
"Browser task next poll user_id=%s worker_id=%s tasks_dir=%s client=%s",
|
||||
effective_user_id,
|
||||
effective_worker_id,
|
||||
tasks_dir(),
|
||||
getattr(request.client, "host", None),
|
||||
)
|
||||
prune_browser_tasks()
|
||||
task = claim_next_browser_task(
|
||||
user_id=get_effective_user_id(),
|
||||
worker_id=worker_id or "chrome-extension",
|
||||
user_id=effective_user_id,
|
||||
worker_id=effective_worker_id,
|
||||
lease_seconds=settings.task_lease_seconds,
|
||||
)
|
||||
if task is None:
|
||||
|
||||
@@ -5,7 +5,9 @@
|
||||
|
||||
入库目标接口的字段名不固定:标题字段可能叫「名称」也可能叫「标题」,因此先按
|
||||
「名称」发送一次,若判定失败则用 **同一个 nodeName** 把标题字段换成「标题」再
|
||||
发一次。nodeName 是入库记录的唯一标识,取「md 标题 + 时间戳」。
|
||||
发一次。nodeName 是入库记录的唯一标识,**只由 md 标题决定**(md5(标题)),不含
|
||||
时间戳——同一篇 md(同标题)每次入库得到的 nodeName 都一致,避免重复入库时产生
|
||||
多条记录。
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
@@ -56,7 +58,7 @@ class KnowledgeIngestAttempt(BaseModel):
|
||||
class KnowledgeIngestPublishResponse(BaseModel):
|
||||
success: bool
|
||||
template_name: str
|
||||
node_name: str = Field(..., description="入库唯一标识(md 标题 + 时间戳),两次尝试共用")
|
||||
node_name: str = Field(..., description="入库唯一标识(md5(标题),同标题每次入库一致),两次尝试共用")
|
||||
batch_number: str
|
||||
attempts: list[KnowledgeIngestAttempt]
|
||||
|
||||
@@ -99,10 +101,14 @@ def _build_payload(template_name: str, batch_number: str, node_name: str, title_
|
||||
}
|
||||
|
||||
|
||||
def _build_node_name(title: str, timestamp: str) -> str:
|
||||
"""Build a stable node name as md5(title) + timestamp + .md."""
|
||||
def _build_node_name(title: str) -> str:
|
||||
"""Build a stable node name = md5(title) + .md.
|
||||
|
||||
nodeName 只由标题决定(不含时间戳),所以同一篇 md(同标题)每次入库都得到
|
||||
完全相同的 nodeName —— 入库接口据此把重复上传当作同一条记录,而不是每次新增。
|
||||
"""
|
||||
title_md5 = hashlib.md5(title.encode("utf-8")).hexdigest()
|
||||
return f"{title_md5}-{timestamp}.md"
|
||||
return f"{title_md5}.md"
|
||||
|
||||
|
||||
def _resolve_ingest_url() -> str:
|
||||
@@ -132,10 +138,10 @@ async def publish_to_knowledge_base(request: KnowledgeIngestPublishRequest) -> K
|
||||
raise HTTPException(status_code=400, detail=f"智能体 '{request.agent_id}' 未配置入库模板名(template_name),无法发布")
|
||||
|
||||
now = datetime.now()
|
||||
# nodeName 是入库记录的唯一标识:md 标题 + 时间戳。两次尝试必须共用同一个
|
||||
# nodeName,避免第一次实际已入库时第二次产生重复记录。
|
||||
timestamp = now.strftime("%Y%m%d%H%M%S")
|
||||
node_name = _build_node_name(title, timestamp)
|
||||
# nodeName 是入库记录的唯一标识:只取 md5(标题),不含时间戳。同一篇 md(同标题)
|
||||
# 每次入库得到一致的 nodeName —— 既保证两次尝试(名称/标题字段)共用同一个,也
|
||||
# 保证重复上传同一篇 md 不会产生多条记录。
|
||||
node_name = _build_node_name(title)
|
||||
batch_number = f"md{now.strftime('%Y%m%d')}"
|
||||
ingest_url = _resolve_ingest_url()
|
||||
|
||||
|
||||
Reference in new issue
Block a user