bugfix: modify root node's global metadata (#392)

lazyllm/tools/rag/chroma_store.py

@@ -170,7 +170,7 @@ def _make_chroma_metadata(self, node: DocNode) -> Dict[str, Any]:
             "metadata": obj2str(node._metadata),
         }
 
-        if not node.parent:
+        if node.is_root_node:
             metadata["global_metadata"] = obj2str(node.global_metadata)
 
         return metadata

lazyllm/tools/rag/doc_node.py

@@ -83,6 +83,10 @@ def root_node(self) -> Optional["DocNode"]:
             root = root.parent
         return root or self
 
+    @property
+    def is_root_node(self) -> bool:
+        return (not self.parent)
+
     @property
     def global_metadata(self) -> Dict[str, Any]:
         return self.root_node._global_metadata

lazyllm/tools/rag/milvus_store.py

@@ -280,7 +280,7 @@ def _deserialize_node_partial(self, result: Dict) -> DocNode:
             if k.startswith(self._embedding_key_prefix):
                 doc.embedding[k[len(self._embedding_key_prefix):]] = v
             elif k.startswith(self._global_metadata_key_prefix):
-                if doc.parent:
+                if doc.is_root_node:
                     doc._global_metadata[k[len(self._global_metadata_key_prefix):]] = v
 
         return doc

requirements.full.txt

@@ -31,7 +31,7 @@ psutil
 pypdf
 pytest
 numpy==1.26.4
-pymilvus
+pymilvus>=2.4.7, <2.5.0
 async-timeout
 httpx<0.28.0
 redis>=5.0.4

requirements.txt

@@ -31,6 +31,6 @@ psutil
 pypdf
 pytest
 numpy==1.26.4
-pymilvus
+pymilvus>=2.4.7, <2.5.0
 async-timeout
 httpx<0.28.0

scripts/check_requirements.py

@@ -28,7 +28,7 @@ def parse_requirement(line):
     return None, None
 
 def compare_versions(version_spec, req_version):
-    if version_spec.startswith('^') and req_version == '*':
+    if version_spec.startswith('^'):
         return True
     return version_spec == req_version
 

tests/basic_tests/test_store.py

@@ -192,6 +192,9 @@ def setUp(self):
         }
         self.global_metadata_desc = {
             'comment': GlobalMetadataDesc(data_type=GlobalMetadataDesc.DTYPE_VARCHAR, max_size=65535, default_value=' '),
+            'signature': GlobalMetadataDesc(data_type=GlobalMetadataDesc.DTYPE_VARCHAR, max_size=256, default_value=' '),
+            'tags': GlobalMetadataDesc(data_type=GlobalMetadataDesc.DTYPE_ARRAY,
+                                       element_type=GlobalMetadataDesc.DTYPE_INT32, max_size=128, default_value=[]),
         }
 
         self.node_groups = [LAZY_ROOT_NAME, "group1", "group2"]
@@ -278,6 +281,7 @@ def test_query_with_filter_non_exist(self):
 
     def test_reload(self):
         self.store.update_nodes([self.node1, self.node2, self.node3])
+        # reload from storage
         del self.store
         self.store = MilvusStore(group_embed_keys=self.group_embed_keys, embed=self.mock_embed,
                                  embed_dims=self.embed_dims, global_metadata_desc=self.global_metadata_desc,
@@ -291,6 +295,10 @@ def test_reload(self):
             for orig_node in orig_nodes:
                 if node._uid == orig_node._uid:
                     self.assertEqual(node.text, orig_node.text)
+                    # builtin fields are not in orig node, so we can not use
+                    # node.global_metadata == orig_node.global_metadata
+                    for k, v in orig_node.global_metadata.items():
+                        self.assertEqual(node.global_metadata[k], v)
                     break
 
     # XXX `array_contains_any` is not supported in local(aka lite) mode. skip this ut

tests/requirements.txt

@@ -3,4 +3,4 @@ docx2txt
 olefile
 pytest-rerunfailures
 pytest-order
-pymilvus
+pymilvus>=2.4.7, <2.5.0