RC1 : Python API : Circular API import

[Arron-Clague]

Hi there :

Im using the 1.0.0.RC1 build : if I make a call to send a message I get :

2024-06-24 14:17:20 - Error running graph: Error building Component Empty Prompt:

cannot import name 'process_prompt_template' from partially initialized module 'langflow.base.prompts.api_utils' (most likely due to a circular import) (/home/koni/.local/lib/python3.10/site-packages/langflow/base/prompts/api_utils.py)

Thanks,

Arron

[dosubot]

To resolve the circular import issue when calling send a message using the 1.0.0.RC1 build of the Python API, you can use the __getattr__ function to lazily import the necessary classes. This approach is already implemented in the provided code to avoid circular imports.

Here's the relevant part of the code that handles this:

from typing import Any

from .constants import (
    AgentExecutor,
    BaseChatMemory,
    BaseLanguageModel,
    BaseLLM,
    BaseLoader,
    BaseMemory,
    BaseOutputParser,
    BasePromptTemplate,
    BaseRetriever,
    Callable,
    Chain,
    ChatPromptTemplate,
    Data,
    Document,
    Embeddings,
    NestedDict,
    Object,
    Prompt,
    PromptTemplate,
    TextSplitter,
    Tool,
    VectorStore,
)
from .range_spec import RangeSpec


def _import_template_field():
    from langflow.template.field.base import TemplateField

    return TemplateField


def __getattr__(name: str) -> Any:
    # This is to avoid circular imports
    if name == "TemplateField":
        return _import_template_field()
    elif name == "RangeSpec":
        return RangeSpec
    # The other names should work as if they were imported from constants
    # Import the constants module langflow.field_typing.constants
    from . import constants

    return getattr(constants, name)


__all__ = [
    "NestedDict",
    "Data",
    "Tool",
    "PromptTemplate",
    "Chain",
    "BaseChatMemory",
    "BaseLLM",
    "BaseLanguageModel",
    "BaseLoader",
    "BaseMemory",
    "BaseOutputParser",
    "BaseRetriever",
    "VectorStore",
    "Embeddings",
    "TextSplitter",
    "Document",
    "AgentExecutor",
    "Object",
    "Callable",
    "BasePromptTemplate",
    "ChatPromptTemplate",
    "Prompt",
    "RangeSpec",
    "TemplateField",
]

This code ensures that the TemplateField and RangeSpec classes are only imported when they are actually accessed, thus preventing circular import issues ^[1][2].

To continue talking to Dosu, mention @dosu.

[nicoloboschi]

This has been fixed in the latest releases. 1.0.5 doesn't have the problem

[yuchieh8968]

Hello, I am currently encountering this issue while using the latest release. Are there any fixes for this? Many thanks.

[Nitewriter]

Also experiencing this issue with 1.0.5.

[nicoloboschi]

@yuchieh8968 @Nitewriter you need to remove and re-add all the broken components after the upgrade

[Nitewriter]

@nicoloboschi are you referring to components in the flow or python packages? I'm using the Memory Chatbot template as a simple example flow and attempting to run the exported JSON using the sample code provided in the UI.

from langflow.load import run_flow_from_json
TWEAKS = {
  "Memory-hzITR": {},
  "Prompt-viXZL": {},
  "ChatInput-tgsUe": {},
  "OpenAIModel-J2hiT": {},
  "ChatOutput-LH91h": {}
}

result = run_flow_from_json(flow="Memory Chatbot.json",
                            input_value="message",
                            fallback_to_env_vars=True, # False by default
                            tweaks=TWEAKS)

[nicoloboschi]

OK it wasn't clear you were using run_flow_from_json. I'll take a look and try to reproduce then.

[Nitewriter]

@nicoloboschi Oh shoot, ok I think I'm connecting the dots. The components have the imports that are causing the cycle and the template may have outdated components that I used as an example. I'll try recreating that flow from a blank template.

[Nitewriter]

new-memory-chat.json

@nicoloboschi Ok, I did try with a clean start on the flow. I'm still seeing the cycle issue. Let me know if there is anything I can provide to help track down the issue.