zlei6/ragflow
1
0
Fork 0
mirror of https://github.com/infiniflow/ragflow.git synced 2026-07-23 00:46:42 +08:00
Code Issues Packages Projects Releases Wiki Activity

Align Go parser backends and PDF pipeline with Python (#16676)

Browse Source 
Ports remaining Go parser wiring and PDF backends, adds tenant-aware VLM
dispatch, aligns post-processing with Python, and adds end-to-end
pipeline coverage with a generated six-page PDF.
This commit is contained in:
Zhichang Yu
2026-07-06 19:50:54 +08:00
committed by GitHub
parent 3044283442
commit 55af1d70f3
36 changed files with 4918 additions and 28 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
internal/ingestion/component/parser.go
View File

@@ -84,6 +84,7 @@ import (
"ragflow/internal/agent/runtime"
"ragflow/internal/ingestion/component/schema"
"ragflow/internal/utility"
)
const ComponentNameParser = "Parser"
@@ -291,8 +292,19 @@ func (c *ParserComponent) Invoke(ctx context.Context, inputs map[string]any) (ma
}
}
dispatched := dispatchParse(fileTypeExt, filename, binary, c.Param.Setups)
dispatched = hydrateEmptyDispatchPayload(dispatched, binary)
dispatched, handledVision, visionErr := maybeDispatchPDFVision(fileTypeExt, filename, binary, inputs, c.Param.Setups)
if visionErr != nil {
return nil, visionErr
}
if !handledVision {
dispatched = dispatchParse(fileTypeExt, filename, binary, c.Param.Setups)
dispatched = hydrateEmptyDispatchPayload(dispatched, binary)
}
// Known/supported families must fail loudly when dispatch or
// parsing breaks. Only unknown families keep the raw-text fallback.
if dispatched.Err != nil && fileTypeExt != utility.FileTypeOTHER {
return nil, dispatched.Err
}
// 3. Build the legacy `pages` slice. When the dispatch path
// produced a JSON payload, we re-shape it into the page