Merge branch 'ingestion-flow' into langflow-ingestion-modes
This commit is contained in:
commit
5dd9959dd0
5 changed files with 340 additions and 151 deletions
File diff suppressed because one or more lines are too long
|
|
@ -34,9 +34,7 @@ async def upload_user_file(
|
|||
logger.debug("JWT token status", jwt_present=jwt_token is not None)
|
||||
|
||||
logger.debug("Calling langflow_file_service.upload_user_file")
|
||||
result = await langflow_file_service.upload_user_file(
|
||||
file_tuple, jwt_token=jwt_token
|
||||
)
|
||||
result = await langflow_file_service.upload_user_file(file_tuple, jwt_token)
|
||||
logger.debug("Upload successful", result=result)
|
||||
return JSONResponse(result, status_code=201)
|
||||
except Exception as e:
|
||||
|
|
@ -99,15 +97,20 @@ async def run_ingestion(
|
|||
# (search parameters are for retrieval, not document processing)
|
||||
|
||||
logger.debug("Final tweaks with settings applied", tweaks=tweaks)
|
||||
|
||||
# Include user JWT if available
|
||||
jwt_token = getattr(request.state, "jwt_token", None)
|
||||
if jwt_token:
|
||||
# Set auth context for downstream services
|
||||
from auth_context import set_auth_context
|
||||
|
||||
user_id = getattr(request.state, "user_id", None)
|
||||
set_auth_context(user_id, jwt_token)
|
||||
|
||||
result = await langflow_file_service.run_ingestion_flow(
|
||||
file_paths=file_paths or [],
|
||||
jwt_token=jwt_token,
|
||||
session_id=session_id,
|
||||
tweaks=tweaks,
|
||||
jwt_token=jwt_token,
|
||||
)
|
||||
return JSONResponse(result)
|
||||
except Exception as e:
|
||||
|
|
|
|||
|
|
@ -322,6 +322,10 @@ class AppClients:
|
|||
existing_headers = kwargs.pop("headers", {})
|
||||
headers = {**default_headers, **existing_headers}
|
||||
|
||||
# Remove Content-Type if explicitly set to None (for file uploads)
|
||||
if headers.get("Content-Type") is None:
|
||||
headers.pop("Content-Type", None)
|
||||
|
||||
url = f"{LANGFLOW_URL}{endpoint}"
|
||||
|
||||
return await self.langflow_http_client.request(
|
||||
|
|
|
|||
|
|
@ -3,11 +3,12 @@ from typing import Any, Dict, List, Optional
|
|||
|
||||
from config.settings import LANGFLOW_INGEST_FLOW_ID, clients
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
|
||||
class LangflowFileService:
|
||||
def __init__(self):
|
||||
self.flow_id_ingest = LANGFLOW_INGEST_FLOW_ID
|
||||
self.logger = logging.getLogger(__name__)
|
||||
|
||||
async def upload_user_file(
|
||||
self, file_tuple, jwt_token: Optional[str] = None
|
||||
|
|
@ -15,15 +16,18 @@ class LangflowFileService:
|
|||
"""Upload a file using Langflow Files API v2: POST /api/v2/files.
|
||||
Returns JSON with keys: id, name, path, size, provider.
|
||||
"""
|
||||
self.logger.debug("[LF] Upload (v2) -> /api/v2/files")
|
||||
logger.debug("[LF] Upload (v2) -> /api/v2/files")
|
||||
resp = await clients.langflow_request(
|
||||
"POST", "/api/v2/files", files={"file": file_tuple}
|
||||
"POST",
|
||||
"/api/v2/files",
|
||||
files={"file": file_tuple},
|
||||
headers={"Content-Type": None},
|
||||
)
|
||||
self.logger.debug(
|
||||
logger.debug(
|
||||
"[LF] Upload response: %s %s", resp.status_code, resp.reason_phrase
|
||||
)
|
||||
if resp.status_code >= 400:
|
||||
self.logger.error(
|
||||
logger.error(
|
||||
"[LF] Upload failed: %s %s | body=%s",
|
||||
resp.status_code,
|
||||
resp.reason_phrase,
|
||||
|
|
@ -35,13 +39,13 @@ class LangflowFileService:
|
|||
async def delete_user_file(self, file_id: str) -> None:
|
||||
"""Delete a file by id using v2: DELETE /api/v2/files/{id}."""
|
||||
# NOTE: use v2 root, not /api/v1
|
||||
self.logger.debug("[LF] Delete (v2) -> /api/v2/files/%s", file_id)
|
||||
logger.debug("[LF] Delete (v2) -> /api/v2/files/%s", file_id)
|
||||
resp = await clients.langflow_request("DELETE", f"/api/v2/files/{file_id}")
|
||||
self.logger.debug(
|
||||
logger.debug(
|
||||
"[LF] Delete response: %s %s", resp.status_code, resp.reason_phrase
|
||||
)
|
||||
if resp.status_code >= 400:
|
||||
self.logger.error(
|
||||
logger.error(
|
||||
"[LF] Delete failed: %s %s | body=%s",
|
||||
resp.status_code,
|
||||
resp.reason_phrase,
|
||||
|
|
@ -52,15 +56,16 @@ class LangflowFileService:
|
|||
async def run_ingestion_flow(
|
||||
self,
|
||||
file_paths: List[str],
|
||||
jwt_token: str,
|
||||
session_id: Optional[str] = None,
|
||||
tweaks: Optional[Dict[str, Any]] = None,
|
||||
jwt_token: Optional[str] = None,
|
||||
) -> Dict[str, Any]:
|
||||
"""
|
||||
Trigger the ingestion flow with provided file paths.
|
||||
The flow must expose a File component path in input schema or accept files parameter.
|
||||
"""
|
||||
if not self.flow_id_ingest:
|
||||
logger.error("[LF] LANGFLOW_INGEST_FLOW_ID is not configured")
|
||||
raise ValueError("LANGFLOW_INGEST_FLOW_ID is not configured")
|
||||
|
||||
payload: Dict[str, Any] = {
|
||||
|
|
@ -68,19 +73,26 @@ class LangflowFileService:
|
|||
"input_type": "chat",
|
||||
"output_type": "text", # Changed from "json" to "text"
|
||||
}
|
||||
if not tweaks:
|
||||
tweaks = {}
|
||||
|
||||
# Pass files via tweaks to File component (File-PSU37 from the flow)
|
||||
if file_paths:
|
||||
if not tweaks:
|
||||
tweaks = {}
|
||||
tweaks["File-PSU37"] = {"path": file_paths}
|
||||
|
||||
# Pass JWT token via tweaks using the x-langflow-global-var- pattern
|
||||
if jwt_token:
|
||||
# Using the global variable pattern that Langflow expects for OpenSearch components
|
||||
tweaks["OpenSearchHybrid-Ve6bS"] = {"jwt_token": jwt_token}
|
||||
logger.error("[LF] Adding JWT token to tweaks for OpenSearch components")
|
||||
else:
|
||||
logger.error("[LF] No JWT token provided")
|
||||
if tweaks:
|
||||
payload["tweaks"] = tweaks
|
||||
if session_id:
|
||||
payload["session_id"] = session_id
|
||||
|
||||
self.logger.debug(
|
||||
logger.debug(
|
||||
"[LF] Run ingestion -> /run/%s | files=%s session_id=%s tweaks_keys=%s jwt_present=%s",
|
||||
self.flow_id_ingest,
|
||||
len(file_paths) if file_paths else 0,
|
||||
|
|
@ -90,16 +102,14 @@ class LangflowFileService:
|
|||
)
|
||||
|
||||
# Log the full payload for debugging
|
||||
self.logger.debug("[LF] Request payload: %s", payload)
|
||||
logger.debug("[LF] Request payload: %s", payload)
|
||||
|
||||
resp = await clients.langflow_request(
|
||||
"POST", f"/api/v1/run/{self.flow_id_ingest}", json=payload
|
||||
)
|
||||
self.logger.debug(
|
||||
"[LF] Run response: %s %s", resp.status_code, resp.reason_phrase
|
||||
)
|
||||
logger.debug("[LF] Run response: %s %s", resp.status_code, resp.reason_phrase)
|
||||
if resp.status_code >= 400:
|
||||
self.logger.error(
|
||||
logger.error(
|
||||
"[LF] Run failed: %s %s | body=%s",
|
||||
resp.status_code,
|
||||
resp.reason_phrase,
|
||||
|
|
|
|||
|
|
@ -5,7 +5,7 @@ import uuid
|
|||
from typing import Dict
|
||||
|
||||
from models.tasks import FileTask, TaskStatus, UploadTask
|
||||
from src.utils.gpu_detection import get_worker_count
|
||||
from utils.gpu_detection import get_worker_count
|
||||
from utils.logging_config import get_logger
|
||||
|
||||
logger = get_logger(__name__)
|
||||
|
|
|
|||
Loading…
Add table
Reference in a new issue