From 52f5880d213ee369c1fce3699055980b0518e366 Mon Sep 17 00:00:00 2001
From: Magicbook1108 <newyorkupperbay@gmail.com>
Date: Thu, 9 Apr 2026 20:20:11 +0800
Subject: [PATCH] Fix: support vlm fall back in pipeline (#14007)

### What problem does this PR solve?

Fix: support vlm fall back in pipeline for img/table parsing

### Type of change

- [x] Bug Fix (non-breaking change which fixes an issue)
---
 rag/flow/parser/parser.py | 41 +++++++++++++++++----------------------
 rag/flow/parser/utils.py  |  2 +-
 2 files changed, 19 insertions(+), 24 deletions(-)

diff --git a/rag/flow/parser/parser.py b/rag/flow/parser/parser.py
index 5179356f7..182f07595 100644
--- a/rag/flow/parser/parser.py
+++ b/rag/flow/parser/parser.py
@@ -625,15 +625,12 @@ class Parser(ProcessBase):
             if abstract_idx is not None:
                 bboxes[abstract_idx]["abstract"] = True
 
-        print(conf.get("vlm"))
-
-        if conf.get("vlm"):
-            enhance_media_sections_with_vision(
-                bboxes,
-                self._canvas._tenant_id,
-                conf["vlm"],
-                callback=self.callback,
-            )
+        enhance_media_sections_with_vision(
+            bboxes,
+            self._canvas._tenant_id,
+            conf.get("vlm"),
+            callback=self.callback,
+        )
 
         # Emit the requested final PDF output format.
         if conf.get("output_format") == "json":
@@ -804,13 +801,12 @@ class Parser(ProcessBase):
                             "doc_type_kwd": "table",
                         }
                     )
-            if conf.get("vlm"):
-                enhance_media_sections_with_vision(
-                    sections,
-                    self._canvas._tenant_id,
-                    conf["vlm"],
-                    callback=self.callback,
-                )
+            enhance_media_sections_with_vision(
+                sections,
+                self._canvas._tenant_id,
+                conf.get("vlm"),
+                callback=self.callback,
+            )
 
             self.set_output("json", sections)
 
@@ -930,13 +926,12 @@ class Parser(ProcessBase):
                 json_result["doc_type_kwd"] = "image" if json_result.get("image") is not None else "text"
                 json_results.append(json_result)
 
-            if conf.get("vlm"):
-                enhance_media_sections_with_vision(
-                    json_results,
-                    self._canvas._tenant_id,
-                    conf["vlm"],
-                    callback=self.callback,
-                )
+            enhance_media_sections_with_vision(
+                json_results,
+                self._canvas._tenant_id,
+                conf.get("vlm"),
+                callback=self.callback,
+            )
             self.set_output("json", json_results)
         else:
             self.set_output("text", "\n".join([section_text for section_text, _ in sections]))
diff --git a/rag/flow/parser/utils.py b/rag/flow/parser/utils.py
index b897ab170..96bfaff29 100644
--- a/rag/flow/parser/utils.py
+++ b/rag/flow/parser/utils.py
@@ -125,7 +125,7 @@ def enhance_media_sections_with_vision(
     vlm_conf=None,
     callback=None,
 ):
-    if not sections or not tenant_id or not vlm_conf:
+    if not sections or not tenant_id:
         return sections
 
     try: