Superside ComfyUI Nodes
df2c9c0beb
Root cause of nodes not appearing in ComfyUI: this package's node classes originally reused the exact same class names as the internal fal-flux-nodes package it was migrated from (e.g. NanoBananaProEditNode, SeedreamV45EditNode). ComfyUI's NODE_CLASS_MAPPINGS is a flat global dict - when both packages are installed, whichever loads first wins and the other's registration for that key is silently dropped, which is why none of the "Superside "-prefixed nodes were showing up. Every class is now uniquely prefixed (e.g. SupersideNanoBananaProEditNode), so this package can never collide with fal-flux-nodes, ComfyUI-fal-API, or any other node pack regardless of install order. Updated __init__.py, web/js/show_text.js's TEXT_DISPLAY_NODES list, and README.md to match.
190 lines
6.4 KiB
Python
190 lines
6.4 KiB
Python
import logging
|
|
|
|
from .base_node import SupersideFalNode, APIClientMixin, API_KEY_INPUT_SPEC
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
|
|
class SupersideAnyLLMTextNode(SupersideFalNode, APIClientMixin):
|
|
"""
|
|
Any LLM Text Node: Run text-only chat/completion models via fal.ai OpenRouter.
|
|
|
|
This node is intended for pure text generation and reasoning tasks without image
|
|
inputs. It uses the OpenRouter-backed fal endpoint and exposes a curated list of
|
|
current text model options.
|
|
"""
|
|
|
|
MODEL_OPTIONS = [
|
|
# Google
|
|
"google/gemini-2.5-flash",
|
|
"google/gemini-2.5-flash-lite",
|
|
"google/gemini-2.5-pro",
|
|
# Anthropic
|
|
"anthropic/claude-sonnet-4.6",
|
|
"anthropic/claude-opus-4.6",
|
|
"anthropic/claude-sonnet-4.5",
|
|
"anthropic/claude-3.7-sonnet",
|
|
"anthropic/claude-3.5-sonnet",
|
|
# OpenAI
|
|
"openai/gpt-4o",
|
|
"openai/gpt-4.1",
|
|
"openai/gpt-5-chat",
|
|
"openai/gpt-oss-120b",
|
|
# Meta
|
|
"meta-llama/llama-4-maverick",
|
|
"meta-llama/llama-4-scout",
|
|
# Other
|
|
"x-ai/grok-4-fast",
|
|
"moonshotai/kimi-k2.5",
|
|
]
|
|
|
|
ENDPOINT_CANDIDATES = [
|
|
"openrouter/router",
|
|
"fal-ai/any-llm",
|
|
]
|
|
|
|
@classmethod
|
|
def INPUT_TYPES(cls):
|
|
return {
|
|
"required": {
|
|
"prompt": (
|
|
"STRING",
|
|
{
|
|
"multiline": True,
|
|
"placeholder": "Enter your prompt",
|
|
},
|
|
),
|
|
"api_key": API_KEY_INPUT_SPEC,
|
|
},
|
|
"optional": {
|
|
"system_prompt": (
|
|
"STRING",
|
|
{
|
|
"multiline": True,
|
|
"default": "",
|
|
"placeholder": "Optional system prompt",
|
|
"tooltip": "System prompt to provide context or instructions to the model",
|
|
},
|
|
),
|
|
"model": (
|
|
cls.MODEL_OPTIONS,
|
|
{
|
|
"default": "google/gemini-2.5-flash",
|
|
"tooltip": "Text model to use via fal OpenRouter.",
|
|
},
|
|
),
|
|
"reasoning": (
|
|
"BOOLEAN",
|
|
{
|
|
"default": False,
|
|
"tooltip": "Include reasoning when supported by the selected model.",
|
|
},
|
|
),
|
|
"temperature": (
|
|
"FLOAT",
|
|
{
|
|
"default": 1.0,
|
|
"min": 0.0,
|
|
"max": 2.0,
|
|
"step": 0.1,
|
|
"tooltip": "Lower values = more predictable, higher values = more creative.",
|
|
},
|
|
),
|
|
"max_tokens": (
|
|
"INT",
|
|
{
|
|
"default": 1024,
|
|
"min": 1,
|
|
"max": 32768,
|
|
"tooltip": "Maximum number of output tokens.",
|
|
},
|
|
),
|
|
},
|
|
"hidden": {
|
|
"unique_id": "UNIQUE_ID",
|
|
"extra_pnginfo": "EXTRA_PNGINFO",
|
|
},
|
|
}
|
|
|
|
RETURN_TYPES = ("STRING", "STRING")
|
|
RETURN_NAMES = ("output", "reasoning")
|
|
FUNCTION = "generate"
|
|
OUTPUT_NODE = True
|
|
DESCRIPTION = (
|
|
"Generate text with OpenRouter-backed LLMs on fal.ai. "
|
|
"Supports Gemini, GPT, Claude 4.6, Llama, Grok, and Kimi models."
|
|
)
|
|
|
|
def prepare_arguments(self, prompt, **kwargs):
|
|
arguments = {"prompt": prompt}
|
|
|
|
if kwargs.get("system_prompt"):
|
|
arguments["system_prompt"] = kwargs["system_prompt"]
|
|
|
|
if kwargs.get("model") is not None:
|
|
arguments["model"] = kwargs["model"]
|
|
|
|
if kwargs.get("reasoning") is not None:
|
|
arguments["reasoning"] = kwargs["reasoning"]
|
|
|
|
if kwargs.get("temperature") is not None:
|
|
arguments["temperature"] = kwargs["temperature"]
|
|
|
|
if kwargs.get("max_tokens") is not None:
|
|
arguments["max_tokens"] = kwargs["max_tokens"]
|
|
|
|
return arguments
|
|
|
|
def generate(self, prompt, api_key, unique_id=None, extra_pnginfo=None, **kwargs):
|
|
try:
|
|
client = self.get_client(api_key)
|
|
arguments = self.prepare_arguments(prompt, **kwargs)
|
|
model = kwargs.get("model", "google/gemini-2.5-flash")
|
|
logger.info(f"Any LLM Text: model={model}, prompt='{prompt[:50]}...'")
|
|
|
|
result = None
|
|
last_error = None
|
|
|
|
for endpoint in self.ENDPOINT_CANDIDATES:
|
|
try:
|
|
logger.info(f"Trying endpoint: {endpoint}")
|
|
result = self.call_api(client, endpoint, arguments)
|
|
break
|
|
except Exception as endpoint_error:
|
|
last_error = endpoint_error
|
|
logger.warning(f"Endpoint failed: {endpoint} - {str(endpoint_error)}")
|
|
|
|
if result is None:
|
|
raise RuntimeError(
|
|
f"All Any LLM Text endpoints failed. Last error: {str(last_error)}"
|
|
)
|
|
|
|
if "error" in result and result["error"]:
|
|
raise RuntimeError(f"API returned error: {result['error']}")
|
|
|
|
output = result.get("output", "")
|
|
reasoning = result.get("reasoning", "")
|
|
|
|
if not output:
|
|
raise RuntimeError("No output was generated by the API.")
|
|
|
|
if unique_id is not None and extra_pnginfo is not None:
|
|
if (
|
|
isinstance(extra_pnginfo, list)
|
|
and isinstance(extra_pnginfo[0], dict)
|
|
and "workflow" in extra_pnginfo[0]
|
|
):
|
|
workflow = extra_pnginfo[0]["workflow"]
|
|
node = next(
|
|
(x for x in workflow["nodes"] if str(x["id"]) == str(unique_id)),
|
|
None,
|
|
)
|
|
if node:
|
|
node["widgets_values"] = [output]
|
|
|
|
return {"ui": {"text": [output]}, "result": (output, reasoning)}
|
|
|
|
except Exception as e:
|
|
logger.error(f"Any LLM Text generation failed: {str(e)}")
|
|
raise RuntimeError(f"Any LLM Text generation failed: {str(e)}") from e
|