apunkt/SurfSense
1
0
Fork 0
mirror of https://github.com/MODSetter/SurfSense.git synced 2026-04-26 01:06:23 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

Route uploaded images to vision LLM with document-parser fallback

Browse source
This commit is contained in:
CREDO23 2026-04-09 14:33:33 +02:00
parent 78fa2d926a
commit 7e90a8ed3c
7 changed files with 199 additions and 5 deletions
Download patch file Download diff file Expand all files Collapse all files

11
surfsense_backend/app/tasks/document_processors/file_processors.py
View file

@ -333,6 +333,7 @@ async def process_file_in_background(
async def _extract_file_content(
file_path: str,
filename: str,
search_space_id: int,
session: AsyncSession,
user_id: str,
task_logger: TaskLoggingService,
@ -360,6 +361,7 @@ async def _extract_file_content(
FileCategory.PLAINTEXT: "Reading file",
FileCategory.DIRECT_CONVERT: "Converting file",
FileCategory.AUDIO: "Transcribing audio",
FileCategory.IMAGE: "Analyzing image",
FileCategory.UNSUPPORTED: "Unsupported file type",
FileCategory.DOCUMENT: "Extracting content",
}
@ -383,7 +385,13 @@ async def _extract_file_content(
estimated_pages = _estimate_pages_safe(page_limit_service, file_path)
await page_limit_service.check_page_limit(user_id, estimated_pages)
result = await EtlPipelineService().extract(
vision_llm = None
if category == FileCategory.IMAGE:
from app.services.llm_service import get_vision_llm
vision_llm = await get_vision_llm(session, search_space_id)
result = await EtlPipelineService(vision_llm=vision_llm).extract(
EtlRequest(
file_path=file_path,
filename=filename,
@ -439,6 +447,7 @@ async def process_file_in_background_with_document(
markdown_content, etl_service = await _extract_file_content(
file_path,
filename,
search_space_id,
session,
user_id,
task_logger,