Skip to content

Transpiler.md

Latest commit

 

History

History
99 lines (85 loc) · 3 KB

Transpiler.md

File metadata and controls

99 lines (85 loc) · 3 KB

Transpiler

The transpiler can translate ComfyUI's workflows to ComfyScript.

Installing as custom nodes

When ComfyScript is installed as custom nodes, SaveImage and similar nodes will be hooked to automatically save the script as the image's metadata. The script will also be printed to the terminal.

To control these features, see settings.example.toml.

From Python code

from comfy_script.transpile import WorkflowToScriptTranspiler

# PNG / Web UI JSON / API JSON
script = WorkflowToScriptTranspiler.from_file(
  r'tests/transpile/default.json',
  comfyui_api='http://127.0.0.1:8188/'
).to_script()
print(script)
# model, clip, vae = CheckpointLoaderSimple('v1-5-pruned-emaonly.ckpt')
# ...
# SaveImage(image, 'ComfyUI')

CLI

The command line interface.

Usage:

Usage: python -m comfy_script.transpile [OPTIONS] WORKFLOW

  Transpile workflow to ComfyScript.

Options:
  --api TEXT  [default: http://127.0.0.1:8188/]
  --runtime   Wrap the script with runtime imports and workflow context.
  --args [pos|pos2orkwd|kwd]  Format node inputs as positional or keyword
                              arguments.  [default: Pos2OrKwd]
  --help      Show this message and exit.

Example:

python -m comfy_script.transpile "D:\workflow.json"

Or without installing ComfyScript, directly with uv:

uvx --from "comfy-script[default]" python -m comfy_script.transpile "D:\workflow.json"

Output:

model, clip, vae = CheckpointLoaderSimple('v1-5-pruned-emaonly.ckpt')
conditioning = CLIPTextEncode('beautiful scenery nature glass bottle landscape, , purple galaxy bottle,', clip)
conditioning2 = CLIPTextEncode('text, watermark', clip)
latent = EmptyLatentImage(512, 512, 1)
latent = KSampler(model, 156680208700286, 20, 8, 'euler', 'normal', conditioning, conditioning2, latent, 1)
image = VAEDecode(latent, vae)
SaveImage(image, 'ComfyUI')

Wrap the script with runtime imports and workflow context:

python -m comfy_script.transpile "tests\transpile\default.json" --runtime

Output:

from comfy_script.runtime import *
load()
from comfy_script.runtime.nodes import *

with Workflow():
    model, clip, vae = CheckpointLoaderSimple('v1-5-pruned-emaonly.ckpt')
    conditioning = CLIPTextEncode('beautiful scenery nature glass bottle landscape, , purple galaxy bottle,', clip)
    conditioning2 = CLIPTextEncode('text, watermark', clip)
    latent = EmptyLatentImage(512, 512, 1)
    latent = KSampler(model, 156680208700286, 20, 8, 'euler', 'normal', conditioning, conditioning2, latent, 1)
    image = VAEDecode(latent, vae)
    SaveImage(image, 'ComfyUI')

Save the code to script.py:

python -m comfy_script.transpile "tests\transpile\default.json" --runtime > script.py

MetadataViewer

Jupyter Notebook / web, see MetadataViewer.