Bump version: 0.3.11 → 0.3.12

fix(openrouter): handle rate limit errors with OpenWebUI fallback

When OpenRouter returns a 429 error due to rate limiting, the application now attempts to fall back to using an OpenWebUI model instead of returning a default prompt. This provides better resilience when external API limits are exceeded while maintaining functionality through local models.

The changes include:
- Adding RateLimitError import from openai
- Implementing fallback logic in create_prompt_on_openrouter function
- Using OpenWebUI as secondary source for prompts when rate limiting occurs
- Proper error handling and logging for both primary and fallback scenarios

This change improves the robustness of prompt generation by ensuring that users receive generated content even when external services are temporarily unavailable due to rate limits. The fallback mechanism prioritizes configured local models if available, with a final default prompt as backup.

The implementation follows the existing pattern of using random selection from configured OpenWebUI models and includes comprehensive error handling for both primary and fallback operations. This ensures that all failure modes are gracefully handled while maintaining backward compatibility.

.bumpversion.toml

@@ -1,5 +1,5 @@
 [tool.bumpversion]
-current_version = "0.3.10"
+current_version = "0.3.12"
 parse = "(?P<major>\\d+)\\.(?P<minor>\\d+)\\.(?P<patch>\\d+)"
 serialize = ["{major}.{minor}.{patch}"]
 replace = "{new_version}"

Dockerfile

@@ -4,7 +4,7 @@ FROM python:3.11-slim
 # Set the working directory in the container
 WORKDIR /app
 # Set version label
-ARG VERSION="0.3.10"
+ARG VERSION="0.3.12"
 LABEL version=$VERSION
 
 # Copy project files into the container

libs/generic.py

@@ -240,7 +240,6 @@ def create_prompt_with_random_model(base_prompt: str, topic: str = "random"):
             logging.error(f"Error with OpenRouter: {e}")
             return "A colorful abstract composition"  # Default fallback prompt
     
-    return "A colorful abstract composition"  # Default fallback prompt
 
 user_config = load_config()
 output_folder = user_config["comfyui"]["output_dir"]

libs/openrouter.py

@@ -1,8 +1,9 @@
 import random
 import logging
-from openai import OpenAI
+from openai import OpenAI, RateLimitError
 import nest_asyncio
 from libs.generic import load_recent_prompts, load_config
+from libs.openwebui import create_prompt_on_openwebui
 import re
 nest_asyncio.apply()
 
@@ -90,6 +91,20 @@ def create_prompt_on_openrouter(prompt: str, topic: str = "random", model: str =
             prompt = match.group(1)
         logging.debug(prompt)
         return prompt
+    except RateLimitError as e:
+        logging.warning(f"OpenRouter rate limit exceeded (429): {e}. Falling back to local OpenWebUI model.")
+        # Try to use OpenWebUI as fallback
+        openwebui_models = [m.strip() for m in user_config["openwebui"]["models"].split(",") if m.strip()] if "openwebui" in user_config and "models" in user_config["openwebui"] else []
+        if openwebui_models:
+            selected_model = random.choice(openwebui_models)
+            try:
+                return create_prompt_on_openwebui(user_content, topic, selected_model)
+            except Exception as e2:
+                logging.error(f"OpenWebUI fallback also failed: {e2}")
+                return "A colorful abstract composition"  # Final fallback
+        else:
+            logging.error("No OpenWebUI models configured for fallback.")
+            return "A colorful abstract composition"  # Final fallback
     except Exception as e:
         logging.error(f"Error generating prompt with OpenRouter: {e}")
         return ""

workflow_flux.json

@@ -38,7 +38,7 @@
     "inputs": {
       "filename_prefix": "ComfyUI",
       "images": [
-        "8",
+        "42",
         0
       ]
     },
@@ -201,5 +201,19 @@
     "_meta": {
       "title": "VAELoaderMultiGPU"
     }
+  },
+  "42": {
+    "inputs": {
+      "offload_model": true,
+      "offload_cache": true,
+      "anything": [
+        "8",
+        0
+      ]
+    },
+    "class_type": "VRAMCleanup",
+    "_meta": {
+      "title": "🎈VRAM-Cleanup"
+    }
   }
 }

workflow_qwen.json

@@ -98,7 +98,7 @@
   "102": {
     "inputs": {
       "images": [
-        "98",
+        "129",
         0
       ]
     },
@@ -143,5 +143,19 @@
     "_meta": {
       "title": "VAELoaderMultiGPU"
     }
+  },
+  "129": {
+    "inputs": {
+      "offload_model": true,
+      "offload_cache": true,
+      "anything": [
+        "98",
+        0
+      ]
+    },
+    "class_type": "VRAMCleanup",
+    "_meta": {
+      "title": "🎈VRAM-Cleanup"
+    }
   }
 }

workflow_sdxl.json

@@ -52,6 +52,12 @@
   "6": {
     "inputs": {
       "text": "A bustling cyberpunk street at night, filled with neon signs, rain-soaked pavement, and futuristic street vendors. High detail, vivid neon colors, and realistic reflections.",
+      "speak_and_recognation": {
+        "__value__": [
+          false,
+          true
+        ]
+      },
       "clip": [
         "4",
         1
@@ -65,6 +71,12 @@
   "7": {
     "inputs": {
       "text": "text, watermark, deformed Avoid flat colors, poor lighting, and artificial elements. No unrealistic elements, low resolution, or flat colors. Avoid generic objects, poor lighting, and inconsistent styles, blurry, low-quality, distorted faces, overexposed lighting, extra limbs, bad anatomy, low contrast",
+      "speak_and_recognation": {
+        "__value__": [
+          false,
+          true
+        ]
+      },
       "clip": [
         "4",
         1
@@ -95,7 +107,7 @@
     "inputs": {
       "filename_prefix": "ComfyUI",
       "images": [
-        "8",
+        "10",
         0
       ]
     },
@@ -103,5 +115,19 @@
     "_meta": {
       "title": "Save Image"
     }
+  },
+  "10": {
+    "inputs": {
+      "offload_model": true,
+      "offload_cache": true,
+      "anything": [
+        "8",
+        0
+      ]
+    },
+    "class_type": "VRAMCleanup",
+    "_meta": {
+      "title": "🎈VRAM-Cleanup"
+    }
   }
 }