🐛 fix(chat_manager.py): fix is_first_message logic to correctly identify first message

🐛 fix(nodes.py): remove deepcopying of certain objects to improve performance
🐛 fix(loading.py): add error handling for empty documents in instantiate_textsplitter
The is_first_message logic in chat_manager.py was not correctly identifying the first message. The fix in this commit changes the comparison to <= 1 instead of == 0. In nodes.py, deepcopying of certain objects was removed to improve performance. This was done because deepcopying was not necessary for these objects. In loading.py, error handling was added to handle empty documents in instantiate_textsplitter. This was done to prevent errors when the source provided did not load correctly or was empty.

src/backend/langflow/api/chat_manager.py

@@ -110,7 +110,7 @@ class ChatManager:
         start_resp = ChatResponse(message=None, type="start", intermediate_steps="")
         await self.send_json(client_id, start_resp)
 
-        is_first_message = len(self.chat_history.get_history(client_id=client_id)) == 0
+        is_first_message = len(self.chat_history.get_history(client_id=client_id)) <= 1
         # Generate result and thought
         try:
             logger.debug("Generating result and thought")

src/backend/langflow/graph/nodes.py

@@ -36,7 +36,7 @@ class AgentNode(Node):
         #! Cannot deepcopy VectorStore, VectorStoreRouter, or SQL agents
         if self.node_type in ["VectorStoreAgent", "VectorStoreRouterAgent", "SQLAgent"]:
             return self._built_object
-        return deepcopy(self._built_object)
+        return self._built_object
 
 
 class ToolNode(Node):
@@ -81,7 +81,7 @@ class PromptNode(Node):
             self.params["input_variables"] = list(set(self.params["input_variables"]))
 
             self._build()
-        return deepcopy(self._built_object)
+        return self._built_object
 
 
 class ChainNode(Node):
@@ -105,13 +105,29 @@ class ChainNode(Node):
         #! Cannot deepcopy SQLDatabaseChain
         if self.node_type in ["SQLDatabaseChain"]:
             return self._built_object
-        return deepcopy(self._built_object)
+        return self._built_object
 
 
 class LLMNode(Node):
+    built_node_type = None
+    class_built_object = None
+
     def __init__(self, data: Dict):
         super().__init__(data, base_type="llms")
 
+    def build(self, force: bool = False) -> Any:
+        # LLM is different because some models might take up too much memory
+        # or time to load. So we only load them when we need them.ß
+        if self.node_type == self.built_node_type:
+            return self.class_built_object
+        if not self._built or force:
+            self._build()
+            self.built_node_type = self.node_type
+            self.class_built_object = self._built_object
+        # Avoid deepcopying the LLM
+        # that are loaded from a file
+        return self._built_object
+
 
 class ToolkitNode(Node):
     def __init__(self, data: Dict):
@@ -132,7 +148,7 @@ class WrapperNode(Node):
             if "headers" in self.params:
                 self.params["headers"] = eval(self.params["headers"])
             self._build()
-        return deepcopy(self._built_object)
+        return self._built_object
 
 
 class DocumentLoaderNode(Node):

src/backend/langflow/interface/loading.py

@@ -134,7 +134,13 @@ def instantiate_documentloader(class_object, params):
 
 
 def instantiate_textsplitter(class_object, params):
-    documents = params.pop("documents")
+    try:
+        documents = params.pop("documents")
+    except KeyError as e:
+        raise ValueError(
+            "The source you provided did not load correctly or was empty."
+            "Try changing the chunk_size of the Text Splitter."
+        ) from e
     text_splitter = class_object(**params)
     return text_splitter.split_documents(documents)