Update Vertex params with code type and add a my_conversation method to the MyPythonClass. Modify the instantiate_tool function to properly handle the CustomComponent node type. Add a dynamic field to the FrontendNode class.

src/backend/langflow/graph/vertex/base.py

@@ -110,7 +110,7 @@ class Vertex:
                 file_path = value.get("file_path")
 
                 params[key] = file_path
-            elif value.get("type") in ["str", "prompt"] and params.get(key) is None:
+            elif value.get("type") in ["code", "str", "prompt"] and params.get(key) is None:
                 params[key] = value.get("value")
         # Add _type to params
         self.params = params

src/backend/langflow/interface/importing/utils.py

@@ -169,25 +169,22 @@ def get_function_custom(code):
     class_name = "MyPythonClass"
 
     code = """
+from langchain.llms import OpenAI
 from langchain.chains import ConversationChain
+from langchain.memory import ConversationBufferMemory
+
 
 class MyPythonClass:
-    def __init__(self, title: str, author: str, year_published: int):
+    def my_conversation(self):
-        self.title = title
+        llm = OpenAI(temperature=0)
-        self.author = author
+        return ConversationChain(
-        self.year_published = year_published
+            llm=llm, verbose=True, memory=ConversationBufferMemory()
+        )
 
-    def get_details(self):
+    def build(self, name: str) -> ConversationChain:
-        return f"Title: {self.title}, Author: {self.author}, Year Published: {self.year_published}"
+        conversation = self.my_conversation()
-
+        
-    def update_year_published(self, new_year: int):
+        return conversation
-        self.year_published = new_year
-        print(f"The year of publication has been updated to {new_year}.")
-
-    def build(self, name, my_int, my_str, my_bool, no_type):
-        # do something...
-        print("x")
-        return ""
     """
 
     return validate.create_class(code, class_name)

src/backend/langflow/interface/initialize/loading.py

@@ -178,9 +178,8 @@ def instantiate_tool(node_type, class_object, params):
         params["func"] = get_function(params.get("code"))
         return class_object(**params)
     elif node_type == "CustomComponent":
-        return get_function_custom(params.get("code"))
+        params["func"] = get_function_custom(params.get("code"))
-        # params["func"] = get_function_custom(params.get("code"))
+        return class_object(**params)
-        # return class_object(**params)
     # For backward compatibility
     elif node_type == "PythonFunction":
         function_string = params["code"]

src/backend/langflow/template/frontend_node/base.py

@@ -16,6 +16,7 @@ class FrontendNode(BaseModel):
     name: str = ""
     display_name: str = ""
     custom_fields: List[str] = []
+    dynamic: bool = False
 
     def to_dict(self) -> dict:
         return {
@@ -25,6 +26,7 @@ class FrontendNode(BaseModel):
                 "base_classes": self.base_classes,
                 "display_name": self.display_name or self.name,
                 "custom_fields": self.custom_fields,
+                "dynamic": self.dynamic,
             },
         }
 
@@ -186,7 +188,8 @@ class FrontendNode(BaseModel):
     def should_be_password(key: str, show: bool) -> bool:
         """Determines whether the field should be a password field."""
         return (
-            any(text in key.lower() for text in {"password", "token", "api", "key"})
+            any(text in key.lower()
+                for text in {"password", "token", "api", "key"})
             and show
         )
 

src/backend/langflow/utils/constants.py

@@ -50,20 +50,22 @@ def python_function(text: str) -> str:
 """
 
 DEFAULT_CUSTOM_COMPONENT_CODE = """
+from langchain.llms import OpenAI
 from langchain.chains import ConversationChain
+from langchain.memory import ConversationBufferMemory
 
 
 class MyPythonClass:
-    def __init__(self, name: str, year: int):
+    def my_conversation(self):
-        self.name = name
+        llm = OpenAI(temperature=0)
-        self.year = year
+        return ConversationChain(
+            llm=llm, verbose=True, memory=ConversationBufferMemory()
+        )
 
-    def get_details(self):
+    def build(self, name: str) -> ConversationChain:
-        return f"Name: {self.name}, Year: {self.year}"
+        conversation = self.my_conversation()
-
+        
-    def build(self, name: str, year: int, true_or_false: bool, no_type) -> ConversationChain:
+        return conversation
-        # do something...
-        return ConversationChain()
 """
 
 DIRECT_TYPES = ["str", "bool", "code", "int", "float", "Any", "prompt"]

src/backend/langflow/utils/validate.py

@@ -186,6 +186,16 @@ def create_class(code, class_name):
                     raise ModuleNotFoundError(
                         f"Module {alias.name} not found. Please install it and try again."
                     ) from e
+        elif isinstance(node, ast.ImportFrom):
+            try:
+                imported_module = importlib.import_module(node.module)
+                for alias in node.names:
+                    exec_globals[alias.name] = getattr(
+                        imported_module, alias.name)
+            except ModuleNotFoundError as e:
+                raise ModuleNotFoundError(
+                    f"Module {node.module} not found. Please install it and try again."
+                ) from e
 
     class_code = next(
         node
@@ -209,6 +219,8 @@ def create_class(code, class_name):
         instance = exec_globals[class_name](*args, **kwargs)
         return instance
 
+    build.__globals__.update(exec_globals)
+
     return build