LoRA JSON Workflow Serialization

Overview

The LoRAInfoExtractor node now automatically saves its lora_json output value to the workflow JSON file. This enables automation scripts and external tools to access the structured LoRA metadata directly from the workflow without needing to re-execute the node.

Problem Solved

Before: Workflow JSON only contained node inputs and basic metadata:

{
    "1031": {
        "inputs": {
            "civitai_api_key": "9c80a8272431b8fae138c3a82838ed1a",
            "fallback_name": "",
            "use_civitai_api": true,
            "wan_model_type": "low",
            "lora": ["1030", 0]
        },
        "class_type": "LoRAInfoExtractor",
        "_meta": {
            "title": "[Low] LoRA Info Extractor"
        }
    }
}

After: Workflow JSON now includes the lora_json output:

{
    "1031": {
        "inputs": {
            "civitai_api_key": "9c80a8272431b8fae138c3a82838ed1a",
            "fallback_name": "",
            "use_civitai_api": true,
            "wan_model_type": "low",
            "lora": ["1030", 0]
        },
        "class_type": "LoRAInfoExtractor",
        "_meta": {
            "title": "[Low] LoRA Info Extractor"
        },
        "lora_json_output": "{\"loras\":[{\"index\":0,\"display_name\":\"Epic Realism\",\"hash\":\"abc123...\",\"file\":{\"exists\":true,\"path\":\"/path/to/lora.safetensors\"},\"strength\":1.0,\"civitai\":{\"civitai_name\":\"Epic Realism\",\"version_name\":\"v5.0\",\"creator\":\"Creator\"}}],\"summary\":{\"count\":1,\"civitai_matches\":1},\"combined_display\":\"Epic Realism\",\"wan_model_type\":\"low\"}"
    }
}

How It Works

JavaScript Widget Extension

A new ComfyUI extension (comfyui_swissarmyknife.lora_info_extractor) automatically:

Captures Output: Listens for LoRAInfoExtractor node execution results
Caches Data: Stores the lora_json output in the node instance
Serializes: Saves the cached data to workflow JSON when workflow is saved
Restores: Loads the cached data when workflow is opened

Technical Implementation

// Capture execution results
node.onExecuted = function (message) {
    if (message?.output?.lora_json?.[0]) {
        this._cached_lora_json = message.output.lora_json[0];
    }
};

// Save to workflow JSON
nodeType.prototype.onSerialize = function (o) {
    if (this._cached_lora_json) {
        o.lora_json_output = this._cached_lora_json;
    }
};

// Restore from workflow JSON
nodeType.prototype.onConfigure = function (o) {
    if (o.lora_json_output) {
        this._cached_lora_json = o.lora_json_output;
    }
};

Usage for Automation Scripts

Accessing LoRA JSON Data

import json

# Load workflow JSON
with open('workflow.json', 'r') as f:
    workflow = json.load(f)

# Find LoRAInfoExtractor nodes
for node_id, node_data in workflow.items():
    if node_data.get('class_type') == 'LoRAInfoExtractor':
        # Access the saved lora_json output
        if 'lora_json_output' in node_data:
            lora_json_str = node_data['lora_json_output']
            lora_data = json.loads(lora_json_str)

            print(f"Node {node_id} LoRA Data:")
            print(f"  Combined Display: {lora_data['combined_display']}")
            print(f"  LoRA Count: {lora_data['summary']['count']}")
            print(f"  Wan Model Type: {lora_data['wan_model_type']}")

            for lora in lora_data['loras']:
                print(f"  - {lora['display_name']} (strength: {lora['strength']})")

Example Automation Use Cases

Video Processing Pipelines: Extract LoRA metadata for video descriptions
Batch Processing: Identify which LoRAs were used in completed workflows
Quality Control: Verify LoRA configurations match requirements
Report Generation: Generate summaries of LoRA usage across workflows

Data Structure

The saved lora_json_output contains the complete JSON structure from LoRAInfoExtractor:

{
    "loras": [
        {
            "index": 0,
            "display_name": "Epic Realism",
            "hash": "abc123def456...",
            "hashes": {
                "sha256": "abc123def456...",
                "md5": "def456ghi789...",
                "blake3": "ghi789jkl012..."
            },
            "file": {
                "exists": true,
                "path": "/path/to/epic_realism.safetensors"
            },
            "strength": 1.0,
            "original": {
                "raw": {
                    "name": "epic_realism",
                    "strength": 1.0
                }
            },
            "civitai": {
                "civitai_name": "Epic Realism",
                "version_name": "v5.0",
                "civitai_url": "https://civitai.com/models/123456",
                "model_id": 123456,
                "version_id": 789012,
                "matched_hash_type": "sha256",
                "cache_hit": true
            }
        }
    ],
    "summary": {
        "count": 1,
        "missing_files": 0,
        "civitai_matches": 1,
        "civitai_cache_hits": 1,
        "local_only": 0,
        "tags": ["photorealistic", "portrait", "detailed"]
    },
    "combined_display": "Epic Realism",
    "wan_model_type": "low"
}

Compatibility

Browser Cache Refresh

After updating the extension, refresh your browser cache to ensure the new JavaScript code is loaded:

Chrome/Edge: Ctrl+F5 (Windows) or Cmd+Shift+R (Mac)
Firefox: Ctrl+Shift+R (Windows) or Cmd+Shift+R (Mac)

Existing Workflows

New Workflows: Automatically include lora_json_output after first execution
Existing Workflows: Need to be executed once to populate the lora_json_output field
Backward Compatibility: Workflows without lora_json_output continue to work normally

Troubleshooting

Missing lora_json_output in Workflow

Execute Node: Run the workflow to trigger LoRAInfoExtractor execution
Save Workflow: Save the workflow after execution to capture the output
Check Browser Console: Look for debug messages about caching and serialization

Debug Information

The extension logs detailed information to the browser console:

[DEBUG] LoRAInfoExtractor node created
[DEBUG] LoRAInfoExtractor onExecuted called with: {...}
[DEBUG] Cached lora_json from execution: {"loras":[...]...
[SERIALIZE] Saved lora_json to workflow: {"loras":[...]...
[CONFIGURE] Restored lora_json from workflow: {"loras":[...]...

Validation Steps

Open Browser DevTools: F12 in most browsers
Execute LoRAInfoExtractor: Run a workflow containing the node
Check Console: Look for cache and serialize debug messages
Save and Reload: Save workflow, reload page, check for restore messages

Future Enhancements

Potential Improvements

Compression: Compress large JSON outputs to reduce workflow file size
Selective Fields: Option to save only specific fields from the JSON output
Multiple Outputs: Support for saving all output values, not just lora_json
Export Tools: Built-in utilities to extract and format the JSON data

Integration Opportunities

API Endpoints: Expose LoRA data through ComfyUI API endpoints
Webhook Support: Send LoRA metadata to external services on completion
Database Integration: Store LoRA usage statistics in external databases