From 406b36a4529dfa765b10d32a355fa8d74cbb7dee Mon Sep 17 00:00:00 2001
From: Shiyao Huang <102647710+Shiyao-Huang@users.noreply.github.com>
Date: Wed, 6 May 2026 14:28:25 +0800
Subject: [PATCH] fix(#14389): normalize list metadata values for in filters
 (#14410)

## Summary
- normalize string items for list-valued metadata filters in
`meta_filter`
- fix `in` / `not in` case asymmetry when document metadata is
lowercased but filter list values are not
- add regression tests that cover the original issue scenario using
uppercase list values

## Validation
- `PYTHONPATH=external/ragflow pytest
external/ragflow/test/unit_test/common/test_metadata_filter_operators.py
-q`

## Notes
- I commented on #14389 before opening this PR to claim the issue.
- The new tests use `value=["F2", "F11"]` so they fail on the old
implementation and pass with this fix.
- This also benefits other non-comparison operators that flow through
the same normalization path.

Co-authored-by: copizza <copizza@users.noreply.github.com>
Co-authored-by: Wang Qi <wangq8@outlook.com>
---
 common/metadata_utils.py                        | 17 +++++++++--------
 .../common/test_metadata_filter_operators.py    | 14 ++++++++++++++
 2 files changed, 23 insertions(+), 8 deletions(-)

diff --git a/common/metadata_utils.py b/common/metadata_utils.py
index f767b3bd5..79db193eb 100644
--- a/common/metadata_utils.py
+++ b/common/metadata_utils.py
@@ -42,6 +42,13 @@ def convert_conditions(metadata_condition):
 def meta_filter(metas: dict, filters: list[dict], logic: str = "and"):
     doc_ids = set([])
 
+    def normalize_string_values(value):
+        if isinstance(value, str):
+            return value.lower()
+        if isinstance(value, list):
+            return [item.lower() if isinstance(item, str) else item for item in value]
+        return value
+
     def filter_out(v2docs, operator, value):
         ids = []
         for input, docids in v2docs.items():
@@ -96,14 +103,8 @@ def meta_filter(metas: dict, filters: list[dict], logic: str = "and"):
                         value = value.lower()
             else:
                 # Non-comparison operators: maintain original logic
-                if isinstance(input, str):
-                    input = input.lower()
-                elif operator in ("in", "not in") and isinstance(input, list):
-                    input = [x.lower() if isinstance(x, str) else x for x in input]
-                if isinstance(value, str):
-                    value = value.lower()
-                elif operator in ("in", "not in") and isinstance(value, list):
-                    value = [x.lower() if isinstance(x, str) else x for x in value]
+                input = normalize_string_values(input)
+                value = normalize_string_values(value)
 
             matched = False
             try:
diff --git a/test/unit_test/common/test_metadata_filter_operators.py b/test/unit_test/common/test_metadata_filter_operators.py
index 90ee64e31..23f4c2b64 100644
--- a/test/unit_test/common/test_metadata_filter_operators.py
+++ b/test/unit_test/common/test_metadata_filter_operators.py
@@ -33,6 +33,20 @@ def test_not_in_operator():
     assert meta_filter(metas, filters) == ["doc3"]
 
 
+def test_in_operator_with_list_value_is_case_insensitive():
+    metas = {"product": {"F2": ["doc1"], "F11": ["doc2"], "G1": ["doc3"]}}
+    filters = [{"key": "product", "op": "in", "value": ["F2", "F11"]}]
+
+    assert set(meta_filter(metas, filters)) == {"doc1", "doc2"}
+
+
+def test_not_in_operator_with_list_value_is_case_insensitive():
+    metas = {"product": {"F2": ["doc1"], "F11": ["doc2"], "G1": ["doc3"]}}
+    filters = [{"key": "product", "op": "not in", "value": ["F2", "F11"]}]
+
+    assert meta_filter(metas, filters) == ["doc3"]
+
+
 def test_start_with():
     # returns chunk where the metadata starts with the value
     metas = {"name": {"prefix_value": ["doc1"], "other": ["doc2"]}}