OpenRAG 0.5.0.dev34__py3-none-any.whl → 0.5.0.dev36__py3-none-any.whl

api/langflow_files.py

@@ -1,43 +1,43 @@
 import json
 import os
 import tempfile
-from typing import List, Optional
 
 from fastapi import Depends, File, Form, UploadFile
-from pydantic import BaseModel
 from fastapi.responses import JSONResponse
+from pydantic import BaseModel
 
 from dependencies import (
+    get_current_user,
     get_langflow_file_service,
+    get_optional_user,
     get_session_manager,
     get_task_service,
-    get_current_user,
-    get_optional_user,
 )
 from session_manager import User
+from utils.file_utils import langflow_safe_filename_and_mimetype
 from utils.logging_config import get_logger
 
 logger = get_logger(__name__)
 
 
 class RunIngestionBody(BaseModel):
-    file_paths: List[str] = []
-    file_ids: Optional[List[str]] = None
-    file_metadata: Optional[List[dict]] = None
-    session_id: Optional[str] = None
-    tweaks: Optional[dict] = None
-    settings: Optional[dict] = None
+    file_paths: list[str] = []
+    file_ids: list[str] | None = None
+    file_metadata: list[dict] | None = None
+    session_id: str | None = None
+    tweaks: dict | None = None
+    settings: dict | None = None
 
 
 class DeleteFilesBody(BaseModel):
-    file_ids: List[str]
+    file_ids: list[str]
 
 
 async def upload_user_file(
     file: UploadFile = File(...),
     langflow_file_service=Depends(get_langflow_file_service),
     session_manager=Depends(get_session_manager),
-    user: Optional[User] = Depends(get_optional_user),
+    user: User | None = Depends(get_optional_user),
 ):
     """Upload a file to Langflow's Files API"""
     try:
@@ -45,11 +45,11 @@ async def upload_user_file(
         logger.debug("Processing file", filename=file.filename, size=file.size)
 
         content = await file.read()
-        file_tuple = (
-            file.filename,
-            content,
-            file.content_type or "application/octet-stream",
+        # Langflow's docling chokes on text/plain — rename .txt -> .md.
+        upload_filename, upload_mimetype = langflow_safe_filename_and_mimetype(
+            file.filename, file.content_type
         )
+        file_tuple = (upload_filename, content, upload_mimetype)
 
         jwt_token = user.jwt_token if user else session_manager.get_effective_jwt_token(None, None)
 
@@ -60,6 +60,7 @@ async def upload_user_file(
     except Exception as e:
         logger.error("upload_user_file endpoint failed", error_type=type(e).__name__, error=str(e))
         import traceback
+
         logger.error("Full traceback", traceback=traceback.format_exc())
         return JSONResponse({"error": str(e)}, status_code=500)
 
@@ -111,6 +112,7 @@ async def run_ingestion(
 
     if jwt_token:
         from auth_context import set_auth_context
+
         set_auth_context(user.user_id, jwt_token)
 
     try:
@@ -132,9 +134,9 @@ async def run_ingestion(
 
 async def upload_and_ingest_user_file(
     file: UploadFile = File(...),
-    session_id: Optional[str] = Form(None),
-    settings_json: Optional[str] = Form(None, alias="settings"),
-    tweaks_json: Optional[str] = Form(None, alias="tweaks"),
+    session_id: str | None = Form(None),
+    settings_json: str | None = Form(None, alias="settings"),
+    tweaks_json: str | None = Form(None, alias="tweaks"),
     delete_after_ingest: str = Form("true"),
     langflow_file_service=Depends(get_langflow_file_service),
     session_manager=Depends(get_session_manager),
@@ -193,6 +195,7 @@ async def upload_and_ingest_user_file(
             )
         except Exception:
             from utils.file_utils import safe_unlink
+
             safe_unlink(temp_path)
             raise
 
@@ -221,4 +224,4 @@ async def delete_user_files(
             "errors": errors,
         },
         status_code=status,
-    )
+    )

connectors/langflow_connector_service.py

@@ -3,11 +3,15 @@ from typing import Any, Dict, List, Optional
 # Create custom processor for connector files using Langflow
 from models.processors import LangflowConnectorFileProcessor
 from services.langflow_file_service import LangflowFileService
+from utils.file_utils import (
+    clean_connector_filename,
+    get_file_extension,
+    langflow_safe_filename_and_mimetype,
+)
 from utils.logging_config import get_logger
 
 from .base import BaseConnector, ConnectorDocument
 from .connection_manager import ConnectionManager
-from utils.file_utils import get_file_extension, clean_connector_filename
 
 logger = get_logger(__name__)
 
@@ -73,11 +77,15 @@ class LangflowConnectorService:
             
             # Clean filename and ensure we don't add a double extension
             processed_filename = clean_connector_filename(document.filename, document.mimetype)
+            # Langflow's docling chokes on text/plain — rename .txt -> .md.
+            processed_filename, processed_mimetype = langflow_safe_filename_and_mimetype(
+                processed_filename, document.mimetype
+            )
 
             file_tuple = (
                 processed_filename,
                 content,
-                document.mimetype or "application/octet-stream",
+                processed_mimetype,
             )
 
             # Step 0: Delete existing chunks for this file before re-ingesting

models/processors.py

@@ -5,6 +5,7 @@ from utils.file_utils import (
     get_file_extension,
     clean_connector_filename,
     get_filename_aliases,
+    langflow_safe_filename_and_mimetype,
 )
 
 logger = get_logger(__name__)
@@ -868,14 +869,10 @@ class LangflowFileProcessor(TaskProcessor):
             if not content_type:
                 content_type = 'application/octet-stream'
 
-            # Rename .txt to .md for Langflow compatibility
-            # Langflow has issues processing text/plain files
-            langflow_filename = original_filename
-            if original_filename.lower().endswith('.txt'):
-                langflow_filename = original_filename[:-4] + '.md'
-                content_type = 'text/markdown'
-                logger.debug(f"Renamed {original_filename} to {langflow_filename} for Langflow")
-
+            # Langflow's docling chokes on text/plain — rename .txt -> .md.
+            langflow_filename, content_type = langflow_safe_filename_and_mimetype(
+                original_filename, content_type
+            )
             file_tuple = (langflow_filename, content, content_type)
 
             # Get JWT token using same logic as DocumentFileProcessor

{openrag-0.5.0.dev34.dist-info → openrag-0.5.0.dev36.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: openrag
-Version: 0.5.0.dev34
+Version: 0.5.0.dev36
 Summary: OpenRAG is a comprehensive Retrieval-Augmented Generation platform that enables intelligent document search and AI-powered conversations.
 Classifier: Development Status :: 4 - Beta
 Classifier: Environment :: Console

{openrag-0.5.0.dev34.dist-info → openrag-0.5.0.dev36.dist-info}/RECORD

@@ -14,7 +14,7 @@ api/documents.py,sha256=5H3yiFrU4JKSxuICvyk0xdXqOWijflzBZW829OLBtaw,6167
 api/flows.py,sha256=jMpfs9dV3ciz0fMAmMp80FOMxXphOiERyBweu3-t33c,1828
 api/keys.py,sha256=xml08ZVmciwKg-7_v5raxQ5mOkVpWB1vORNIXh120dU,3382
 api/knowledge_filter.py,sha256=krJTX83FdL2umArbmTosI8E4x_5SooiDABwqiODYpJ0,15515
-api/langflow_files.py,sha256=VS5WdeDM-AecuLPhBTk4EmRUwQyWStJ_WP3-QSVfFFE,7959
+api/langflow_files.py,sha256=vXPucrspWRryoEndqy1xnqjUi9SRouUutr3z7q7JFoE,8099
 api/models.py,sha256=QvvSaTAM1KKx-RGjQrjhZ6b6zD9PTbyHQbZ4oiSQOk4,5582
 api/nudges.py,sha256=hVWnZN4aegl7tTnEv2NjqDoFs78r8eX9zkhe0NrorIY,1921
 api/oidc.py,sha256=WJRnditsha1CqZQCaKgzKVGAJuQqRgho1danLhUEhDE,3455
@@ -43,7 +43,7 @@ config/settings.py,sha256=zhvJ3TAmdZwoWk55BPX4OuAAJiDTHlJg4_MoSWPhZMI,39298
 connectors/__init__.py,sha256=6ijOhF7jG-zJ3bIxQ4HIHZHqpT127cNvCDaaqdN_Pm8,390
 connectors/base.py,sha256=s9gBEIxl8_nZVtCeMKz7LrlOV-H0rCxvN_cFH8B9DMg,4532
 connectors/connection_manager.py,sha256=LrfiuoBB-hqNGjXhH-5x4uL6msU-OVOp7--4V302a14,28187
-connectors/langflow_connector_service.py,sha256=HjVgTmlpZCWsoR2msEuCouw3snX0HBHNv6Szm4A0Iik,17251
+connectors/langflow_connector_service.py,sha256=7kK66gs4RXpTDO58FG9SBxEpHadwQr_FZNMQOzp74ic,17512
 connectors/service.py,sha256=_qyRDskMiNfPa2OyOan9gAmOgB5JjcLTidzzg-rhLWs,18229
 connectors/aws_s3/__init__.py,sha256=ZRXSKmYvYK7Kr7Rzcd0_YIMpqfAm5qjm06zF-YfPPhA,369
 connectors/aws_s3/api.py,sha256=CDpsdOZaOM7hT93Je4O_JD7GxJjPN0-_qgQPPSPiy40,6861
@@ -70,10 +70,10 @@ connectors/sharepoint/utils.py,sha256=OBNEbXb8uAfR-iR-ML0aOojGfGj03vJnKm9DbYf2hW
 mcp_http/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 mcp_http/server.py,sha256=JutP94fQRCizeCN69kKkdYDFE71L3DqLTCcb397T8Eo,7738
 models/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-models/processors.py,sha256=Vc6a1HIHpIe9t3Jx7HJ_lCRrSmWtzw63C5bzJpxTz8s,35793
+models/processors.py,sha256=IzyYgov0VWmyDKxgEdzo9DumbbqDJWfa8BgdTCDCQqA,35614
 models/tasks.py,sha256=9iINXQZxA9-hcYzG8xC_baBfLa2w6gcRPJdQHsx_Kms,1643
 models/url.py,sha256=92rsrLLyIqUErYipYz5QwUNTdHwJ2jBRV7PIVpncTyo,5764
-openrag-0.5.0.dev34.dist-info/licenses/LICENSE,sha256=jWcnPXkUxEinX60Yn-_zjsY9i84vz2QSb0fi53djOkg,11334
+openrag-0.5.0.dev36.dist-info/licenses/LICENSE,sha256=jWcnPXkUxEinX60Yn-_zjsY9i84vz2QSb0fi53djOkg,11334
 services/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 services/api_key_service.py,sha256=NMz2ohJo0jcuHjDN3tHuzhCTVo6lsOqAkUhozzmWSv0,11720
 services/auth_service.py,sha256=nAKTvMyX-YUtiNfKF06Dq4E4iwHPBhfKR6nliqGpOSs,22090
@@ -144,7 +144,7 @@ utils/embedding_fields.py,sha256=l4cjVKNbe6hM75oLwyyZ7t94JJXRIzvgfzPmKcvrCbE,740
 utils/embeddings.py,sha256=luyWDU6Y4gw116XaF3_wHfdyKEv8zi3qRrMUtOgP1X4,1357
 utils/encryption.py,sha256=aTsiuhiG9rM6UmvnttfNyMFIBLpPSvEFb054fMT1Ohc,7758
 utils/env_utils.py,sha256=slkoqbTStWn0xJQJ9Ws0qtv6bUH_bC-rBJbWsgDw5pk,890
-utils/file_utils.py,sha256=n0Tp8OiBQ7QygRljl6-b8dNgqtWg1P_SB0GwOnra-y8,4345
+utils/file_utils.py,sha256=NRDfj4e-VloPay3iF7MKAKIZHLufxEqtiyDSOvnA-Hw,5241
 utils/gpu_detection.py,sha256=yU8PtO9Fp0rfHhY6JW0212cS_SVMA12BtWUXHXpQyp0,1697
 utils/hash_utils.py,sha256=LIEss0uTBNJcgOvMLoOoj0pIVkC3A7Wzs1IqfNx_23s,2200
 utils/langflow_headers.py,sha256=X8vZNYRv9jvwBu7Grd8S396dalPTZP9Ecb6owHMU2KE,3248
@@ -159,8 +159,8 @@ utils/telemetry/__init__.py,sha256=xoStY8hoYWnVe0B_jNthSdzRb7AxCT0rhM8ba9Q0Obg,2
 utils/telemetry/category.py,sha256=wei3V5WyUaiadYjEE5s0ys9PY0TjnebIvTykQU0P_2k,1114
 utils/telemetry/client.py,sha256=RhjW1a27w7W6gHt9X7AxdRPdZGmCEQDPskkxj5nsuiQ,12500
 utils/telemetry/message_id.py,sha256=IAu0wBVNOGPcSlQ0bAzvqIYr1Ac_n8wSTSxXzq_tiV0,9634
-openrag-0.5.0.dev34.dist-info/METADATA,sha256=f0L0xZ-KNdGeg-BriYou_3rtWkkzl-xa7P-trb86p1k,7819
-openrag-0.5.0.dev34.dist-info/WHEEL,sha256=aeYiig01lYGDzBgS8HxWXOg3uV61G9ijOsup-k9o1sk,91
-openrag-0.5.0.dev34.dist-info/entry_points.txt,sha256=ArKNHB63I3K8vynNkogvZOLmmQFpIsHDL7MvreCmju4,45
-openrag-0.5.0.dev34.dist-info/top_level.txt,sha256=5vfLBIkzuwBJrizhCSZiXV7NTzTRIMlNLsyzYDEeCF8,125
-openrag-0.5.0.dev34.dist-info/RECORD,,
+openrag-0.5.0.dev36.dist-info/METADATA,sha256=Vb6XbbV_JOqF6PkPkiZlDtECtd5YbumtFvMlL-mGI1A,7819
+openrag-0.5.0.dev36.dist-info/WHEEL,sha256=aeYiig01lYGDzBgS8HxWXOg3uV61G9ijOsup-k9o1sk,91
+openrag-0.5.0.dev36.dist-info/entry_points.txt,sha256=ArKNHB63I3K8vynNkogvZOLmmQFpIsHDL7MvreCmju4,45
+openrag-0.5.0.dev36.dist-info/top_level.txt,sha256=5vfLBIkzuwBJrizhCSZiXV7NTzTRIMlNLsyzYDEeCF8,125
+openrag-0.5.0.dev36.dist-info/RECORD,,

utils/file_utils.py

@@ -101,6 +101,25 @@ def clean_connector_filename(filename: str, mimetype: str) -> str:
     return clean_name
 
 
+def langflow_safe_filename_and_mimetype(filename: str, mimetype: str | None) -> tuple[str, str]:
+    """Apply the .txt -> .md workaround for Langflow ingestion.
+
+    Langflow's docling component fails on text/plain (see commit f6b9fe0).
+    The workaround is to rename .txt files to .md with text/markdown before
+    handing the file to Langflow. Centralised here so every Langflow-upload
+    site applies it identically — local user uploads, connector-driven
+    ingestion, and the direct upload route.
+
+    Returns the (possibly rewritten) (filename, mimetype) pair. Pass-through
+    for all other extensions; falls back to application/octet-stream when
+    mimetype is missing.
+    """
+    safe_mimetype = mimetype or "application/octet-stream"
+    if filename and filename.lower().endswith(".txt"):
+        return filename[:-4] + ".md", "text/markdown"
+    return filename, safe_mimetype
+
+
 def get_filename_aliases(filename: str) -> list[str]:
     """Return equivalent filename variants used by ingestion/indexing.