- implemented workflow converter and browser

- fixed workflow ui link custom node

__init__.py

@@ -1,15 +1,3 @@
-"""
-ComfyUI custom node: WorkflowUI plugin.
-
-Registers API routes so WorkflowUI (and other clients) can:
-- Delete files from ComfyUI output/input/temp (POST /delete, POST /workflowui/media/delete)
-- List and browse media (GET /workflowui/media/list, GET /workflowui/media/tree)
-- Detect plugin (GET /workflowui/media/capabilities)
-- Installed modules/versions for run metadata (GET /workflowui/version_info)
-
-Adds WorkflowUILink node: configurable inputs/outputs for WorkflowUI schema.
-"""
-
 from .version_info import __version__
 from .routes import register_routes
 from .workflow_ui_link import NODE_CLASS_MAPPINGS as LINK_MAPPINGS
@@ -17,7 +5,6 @@ from .workflow_ui_link import NODE_DISPLAY_NAME_MAPPINGS as LINK_DISPLAY_NAMES
 
 register_routes()
 
-# Startup banner
 _banner_text = f"  WorkflowUI Plugin v{__version__}  "
 _width = max(len(_banner_text) + 2, 24)
 _border = "+" + "-" * (_width - 2) + "+"

routes.py

@@ -1,36 +1,64 @@
-"""
-WorkflowUI plugin: ComfyUI API routes for media access and deletion.
-
-- POST /delete and POST /workflowui/media/delete — delete a file by filename/subfolder/type
-  (WorkflowUI calls POST /delete with JSON body).
-- GET /workflowui/media/view — serve a file from output/input/temp; ?preview=webp|jpeg|SIZE for compressed image previews.
-- GET /workflowui/media/list — list files in output/input/temp (optional subfolder).
-- GET /workflowui/media/tree — tree of subfolders and files for browsing.
-- GET /workflowui/media/capabilities — report supported features (delete, list, view).
-- GET /workflowui/version_info — installed custom node modules and versions (for run metadata).
-"""
-
 from __future__ import annotations
 
+import hashlib
+import importlib.util
+import json
 import io
 import logging
 import mimetypes
 import os
+import sys
 from pathlib import Path
 
 from aiohttp import web
 
 logger = logging.getLogger(__name__)
 
+
+def _load_workflow_converter():
+    try:
+        from .workflow_converter import WorkflowConverter
+        return WorkflowConverter
+    except SyntaxError as e:
+        if "null bytes" not in str(e).lower():
+            raise
+        logger.warning(
+            "WorkflowUIPlugin: workflow_converter.py contains null bytes in source; "
+            "loading with null bytes stripped (check deployment/sync)."
+        )
+        pkg_dir = os.path.dirname(os.path.abspath(__file__))
+        path = os.path.join(pkg_dir, "workflow_converter.py")
+        with open(path, "rb") as f:
+            source = f.read().replace(b"\x00", b"").decode("utf-8", errors="replace")
+        fullname = (__package__ or "WorkflowUIPlugin") + ".workflow_converter"
+
+        class _Loader:
+            def __init__(self, src, pathname):
+                self.src = src
+                self.pathname = pathname
+            def create_module(self, spec):
+                return None
+            def exec_module(self, module):
+                code = compile(self.src, self.pathname, "exec")
+                exec(code, module.__dict__)
+
+        spec = importlib.util.spec_from_loader(fullname, _Loader(source, path), origin=path)
+        module = importlib.util.module_from_spec(spec)
+        sys.modules[fullname] = module
+        spec.loader.exec_module(module)
+        return module.WorkflowConverter
+
 MEDIA_TYPES = ("output", "input", "temp")
 
 PREVIEW_MAX_DIM_DEFAULT = 512
 PREVIEW_QUALITY_DEFAULT = 85
 IMAGE_EXTENSIONS = {".png", ".jpg", ".jpeg", ".webp", ".gif", ".bmp", ".tiff", ".tif"}
+UPLOAD_IMAGE_EXTENSIONS = {".png", ".jpg", ".jpeg", ".webp", ".gif"}
+UPLOAD_VIDEO_EXTENSIONS = {".mp4", ".webm", ".mkv", ".mov"}
+UPLOAD_AUDIO_EXTENSIONS = {".mp3", ".wav", ".ogg", ".flac", ".m4a"}
 
 
 def _get_media_root(folder_type: str):
-    """Resolve ComfyUI media root for output/input/temp. Returns None if invalid."""
     try:
         import folder_paths
         out = folder_paths.get_directory_by_type(folder_type)
@@ -40,20 +68,14 @@ def _get_media_root(folder_type: str):
 
 
 def _resolve_media_path(folder_type: str, subfolder: str, filename: str) -> tuple[str | None, str | None]:
-    """
-    Resolve full_path for a media file. Ensures path is under base (no traversal).
-    Returns (full_path, None) on success; (None, error_message) on validation failure.
-    """
     if folder_type not in MEDIA_TYPES:
         return None, "Invalid type; use output, input, or temp"
     base = _get_media_root(folder_type)
     if not base:
         return None, "Could not resolve media directory"
-    # Normalize subfolder: no leading/trailing slashes, no '..'
     sub = (subfolder or "").strip().replace("\\", "/").strip("/")
     if ".." in sub or sub.startswith("/"):
         return None, "Invalid subfolder"
-    # Filename must be a single segment (no path)
     name = (filename or "").strip()
     if not name or "/" in name or "\\" in name or ".." in name:
         return None, "Invalid filename"
@@ -65,7 +87,6 @@ def _resolve_media_path(folder_type: str, subfolder: str, filename: str) -> tupl
 
 
 def _delete_file(folder_type: str, subfolder: str, filename: str) -> tuple[bool, str, int]:
-    """Delete one file. Returns (success, message, http_status)."""
     full, err = _resolve_media_path(folder_type, subfolder, filename)
     if err:
         return False, err, 400
@@ -92,7 +113,6 @@ def _make_image_preview(
     max_dim: int,
     quality: int,
 ) -> tuple[bytes | None, str | None]:
-    """Return (body, content_type) for a resized/compressed preview, or (None, None) on failure."""
     try:
         from PIL import Image
     except ImportError:
@@ -124,7 +144,6 @@ def _make_image_preview(
 
 
 async def _handle_view(request: web.Request) -> web.Response:
-    """GET ?filename=...&subfolder=...&type=output&preview=webp|jpeg|SIZE — serve file; optional compressed image preview."""
     filename = request.query.get("filename", "").strip()
     subfolder = request.query.get("subfolder", "")
     folder_type = (request.query.get("type") or "output").strip().lower()
@@ -142,7 +161,6 @@ async def _handle_view(request: web.Request) -> web.Response:
     if not content_type:
         content_type = "application/octet-stream"
 
-    # Optional preview: only for images; format webp (default) or jpeg, optional max dimension
     if preview and _is_image_file(filename, content_type):
         max_dim = PREVIEW_MAX_DIM_DEFAULT
         quality = PREVIEW_QUALITY_DEFAULT
@@ -168,7 +186,6 @@ async def _handle_view(request: web.Request) -> web.Response:
 
 
 async def _handle_delete(request: web.Request) -> web.Response:
-    """POST body: JSON { filename, subfolder?, type? }. type defaults to 'output'."""
     try:
         body = await request.json()
     except Exception:
@@ -186,8 +203,74 @@ async def _handle_delete(request: web.Request) -> web.Response:
     return web.json_response({"ok": False, "error": msg}, status=status)
 
 
+def _content_hash_name(content: bytes, ext: str) -> str:
+    h = hashlib.sha256(content).hexdigest()[:16]
+    return f"{h}{ext}"
+
+
+async def _handle_upload(request: web.Request) -> web.Response:
+    try:
+        reader = await request.multipart()
+    except Exception as e:
+        logger.debug("WorkflowUIPlugin: upload multipart error: %s", e)
+        return web.json_response({"error": "Invalid multipart body"}, status=400)
+    media_type = (request.query.get("type") or "image").strip().lower()
+    if media_type not in ("image", "video", "audio"):
+        return web.json_response({"error": "type must be image, video, or audio"}, status=400)
+    if media_type == "image":
+        allowed_ext = UPLOAD_IMAGE_EXTENSIONS
+    elif media_type == "video":
+        allowed_ext = UPLOAD_VIDEO_EXTENSIONS
+    else:
+        allowed_ext = UPLOAD_AUDIO_EXTENSIONS
+    part = None
+    while True:
+        try:
+            p = await reader.next()
+        except Exception:
+            p = None
+        if p is None:
+            break
+        name = getattr(p, "name", None) or ""
+        if name in ("file", "image"):
+            part = p
+            break
+    if part is None:
+        return web.json_response({"error": "Missing form field 'file' or 'image'"}, status=400)
+    filename = getattr(part, "filename", None) or "upload"
+    ext = Path(filename).suffix.lower()
+    if ext not in allowed_ext:
+        return web.json_response(
+            {"error": f"Unsupported {media_type} format. Allowed: {', '.join(sorted(allowed_ext))}"},
+            status=400,
+        )
+    try:
+        content = await part.read()
+    except Exception as e:
+        logger.warning("WorkflowUIPlugin: upload read failed: %s", e)
+        return web.json_response({"error": "Failed to read uploaded file"}, status=400)
+    if not content:
+        return web.json_response({"error": "Empty file"}, status=400)
+    base = _get_media_root("input")
+    if not base:
+        return web.json_response({"error": "Could not resolve input directory"}, status=500)
+    safe_name = _content_hash_name(content, ext)
+    full = os.path.join(base, safe_name)
+    try:
+        with open(full, "wb") as f:
+            f.write(content)
+    except OSError as e:
+        logger.warning("WorkflowUIPlugin: upload write failed for %s: %s", full, e)
+        return web.json_response({"error": f"Failed to save file: {e}"}, status=500)
+    return web.json_response({
+        "filename": safe_name,
+        "name": safe_name,
+        "subfolder": "",
+        "type": "input",
+    })
+
+
 def _list_dir(folder_type: str, subfolder: str) -> tuple[list | None, str | None]:
-    """List files in type/subfolder. Returns (list of {name, size, mtime}, error)."""
     if folder_type not in MEDIA_TYPES:
         return None, "Invalid type"
     base = _get_media_root(folder_type)
@@ -223,7 +306,6 @@ def _list_dir(folder_type: str, subfolder: str) -> tuple[list | None, str | None
 
 
 async def _handle_list(request: web.Request) -> web.Response:
-    """GET ?type=output&subfolder= (subfolder optional)."""
     folder_type = (request.query.get("type") or "output").strip().lower()
     subfolder = request.query.get("subfolder", "")
     data, err = _list_dir(folder_type, subfolder)
@@ -233,7 +315,6 @@ async def _handle_list(request: web.Request) -> web.Response:
 
 
 def _tree_dir(folder_type: str, subfolder: str) -> tuple[dict | None, str | None]:
-    """Return tree { folders: [name], files: [{filename, size, mtime}] } for one level."""
     if folder_type not in MEDIA_TYPES:
         return None, "Invalid type"
     base = _get_media_root(folder_type)
@@ -268,7 +349,6 @@ def _tree_dir(folder_type: str, subfolder: str) -> tuple[dict | None, str | None
 
 
 async def _handle_tree(request: web.Request) -> web.Response:
-    """GET ?type=output&subfolder= (one level: folders + files)."""
     folder_type = (request.query.get("type") or "output").strip().lower()
     subfolder = request.query.get("subfolder", "")
     data, err = _tree_dir(folder_type, subfolder)
@@ -283,7 +363,6 @@ async def _handle_tree(request: web.Request) -> web.Response:
 
 
 async def _handle_capabilities(request: web.Request) -> web.Response:
-    """GET /workflowui/media/capabilities — report what this plugin supports."""
     return web.json_response({
         "workflowui_plugin": True,
         "delete": True,
@@ -291,29 +370,152 @@ async def _handle_capabilities(request: web.Request) -> web.Response:
         "tree": True,
         "view": True,
         "view_preview": True,
+        "upload": True,
     })
 
 
 async def _handle_version_info(request: web.Request) -> web.Response:
-    """GET /workflowui/version_info — installed custom node modules and versions for run metadata."""
     from .version_info import get_version_info_payload
     return web.json_response(get_version_info_payload())
 
 
+async def _handle_workflows_list(request: web.Request) -> web.Response:
+    from .workflow_export import list_workflows
+    try:
+        items = list_workflows()
+        return web.json_response(items)
+    except Exception as e:
+        logger.warning("WorkflowUIPlugin: workflows list failed: %s", e)
+        return web.json_response([])
+
+
+async def _handle_workflow_by_id(request: web.Request) -> web.Response:
+    from .workflow_export import load_workflow
+    workflow_id = request.match_info.get("id", "").strip()
+    if not workflow_id:
+        return web.json_response({"error": "Missing workflow id"}, status=400)
+    try:
+        name, graph = load_workflow(workflow_id)
+        return web.json_response({"name": name, "graph": graph})
+    except FileNotFoundError as e:
+        return web.json_response({"error": str(e)}, status=404)
+    except ValueError as e:
+        return web.json_response({"error": str(e)}, status=400)
+    except Exception as e:
+        logger.exception("WorkflowUIPlugin: load workflow %s failed", workflow_id)
+        return web.json_response({"error": str(e)}, status=500)
+
+
+WORKFLOWCONVERTER_MAX_CONTENT_LENGTH = 1 * 1024 * 1024
+
+
+def _strip_null_bytes(obj):
+    if isinstance(obj, dict):
+        out = {}
+        for k, v in obj.items():
+            key = k
+            if isinstance(k, str) and ("\x00" in k or "\u0000" in k):
+                key = k.replace("\x00", "").replace("\u0000", "")
+            out[key] = _strip_null_bytes(v)
+        return out
+    if isinstance(obj, list):
+        return [_strip_null_bytes(v) for v in obj]
+    if isinstance(obj, str):
+        if "\x00" in obj or "\u0000" in obj:
+            return obj.replace("\x00", "").replace("\u0000", "")
+        return obj
+    return obj
+
+
+async def _handle_workflowconverter_convert_post(request: web.Request) -> web.Response:
+    logger.info("WorkflowUIPlugin: workflowconverter/convert POST received")
+    if request.content_length is not None and request.content_length > WORKFLOWCONVERTER_MAX_CONTENT_LENGTH:
+        return web.json_response({
+            "success": False,
+            "error": f"Request too large. Maximum size is {WORKFLOWCONVERTER_MAX_CONTENT_LENGTH // (1024 * 1024)} MB",
+        }, status=413)
+    try:
+        raw = await request.read()
+        raw = raw.replace(b"\x00", b"")
+        body = json.loads(raw.decode("utf-8", errors="replace"))
+    except Exception as e:
+        logger.debug("WorkflowUIPlugin: workflowconverter/convert invalid JSON: %s", e)
+        return web.json_response({"success": False, "error": f"Invalid JSON: {e!s}"}, status=400)
+    if not isinstance(body, dict):
+        return web.json_response({"success": False, "error": "Body must be a JSON object"}, status=400)
+    body = _strip_null_bytes(body)
+    try:
+        WorkflowConverter = _load_workflow_converter()
+        if WorkflowConverter.is_api_format(body):
+            return web.json_response(body)
+        if "nodes" not in body or "links" not in body:
+            return web.json_response({
+                "success": False,
+                "error": "Invalid workflow format - missing nodes or links",
+            }, status=400)
+        if not isinstance(body.get("nodes"), list):
+            return web.json_response({
+                "success": False,
+                "error": "Invalid workflow format - 'nodes' must be a list",
+            }, status=400)
+        if not isinstance(body.get("links"), list):
+            return web.json_response({
+                "success": False,
+                "error": "Invalid workflow format - 'links' must be a list",
+            }, status=400)
+        api_prompt = WorkflowConverter.convert_to_api(body)
+        num_nodes = len(body.get("nodes", []))
+        num_links = len(body.get("links", []))
+        num_converted = len(api_prompt)
+        logger.info(
+            "WorkflowUIPlugin: workflowconverter converted %s nodes, %s links -> %s API nodes",
+            num_nodes, num_links, num_converted,
+        )
+        return web.json_response(api_prompt)
+    except Exception as e:
+        import traceback
+        tb = traceback.format_exc()
+        logger.error("WorkflowUIPlugin: workflowconverter/convert failed: %s", e)
+        logger.error("Traceback: %s", tb)
+        err_msg = str(e).strip() or "Internal server error during conversion"
+        if len(err_msg) > 500:
+            err_msg = err_msg[:497] + "..."
+        return web.json_response({
+            "success": False,
+            "error": err_msg,
+            "traceback": tb,
+        }, status=500)
+
+
+async def _handle_workflowconverter_convert_get(request: web.Request) -> web.Response:
+    return web.json_response({
+        "name": "WorkflowUI Workflow Converter",
+        "version": "2.1.0",
+        "description": "Converts non-API workflow format to API format for execution (subgraphs, bypassed nodes, GetNode/SetNode). Part of WorkflowUIPlugin.",
+        "usage": "POST a workflow JSON to this endpoint to convert it to API format",
+        "endpoint": "/workflowui/workflowconverter/convert",
+        "source": "Based on Seth A. Robinson's comfyui-workflow-to-api-converter-endpoint",
+        "repository": "https://github.com/SethRobinson/comfyui-workflow-to-api-converter-endpoint",
+    })
+
+
 def register_routes():
-    """Register routes with ComfyUI PromptServer. Call from __init__.py."""
     try:
         from server import PromptServer
         routes = PromptServer.instance.routes
     except Exception as e:
         logger.warning("WorkflowUIPlugin: could not get PromptServer routes: %s", e)
         return
-    # WorkflowUI backend calls POST /delete with JSON { filename, subfolder, type }
     routes.post("/delete")(_handle_delete)
     routes.post("/workflowui/media/delete")(_handle_delete)
+    routes.post("/workflowui/media/upload")(_handle_upload)
     routes.get("/workflowui/media/view")(_handle_view)
     routes.get("/workflowui/media/list")(_handle_list)
     routes.get("/workflowui/media/tree")(_handle_tree)
     routes.get("/workflowui/media/capabilities")(_handle_capabilities)
     routes.get("/workflowui/version_info")(_handle_version_info)
-    logger.info("WorkflowUIPlugin: registered media routes (delete, view, list, tree, capabilities, version_info)")
+    routes.get("/workflowui/workflows")(_handle_workflows_list)
+    routes.get("/workflowui/workflows/{id}")(_handle_workflow_by_id)
+    routes.post("/workflowui/workflowconverter/convert")(_handle_workflowconverter_convert_post)
+    routes.get("/workflowui/workflowconverter/convert")(_handle_workflowconverter_convert_get)
+    logger.info("WorkflowUIPlugin: registered media + workflow export + workflowconverter routes")

version_info.py

@@ -4,7 +4,7 @@ import json
 import re
 from pathlib import Path
 
-__version__ = "1.0.6"
+__version__ = "1.0.10"
 
 _PLUGIN_DIR = Path(__file__).resolve().parent
 _CUSTOM_NODES_DIR = _PLUGIN_DIR.parent
@@ -14,7 +14,6 @@ _PYPROJECT_VERSION_RE = re.compile(r'^\s*version\s*=\s*["\']([^"\']+)["\']', re.
 
 
 def _version_from_init_py(init_path: Path) -> str | None:
-    """Try to read __version__ from a package __init__.py without importing."""
     if not init_path.is_file():
         return None
     try:
@@ -26,7 +25,6 @@ def _version_from_init_py(init_path: Path) -> str | None:
 
 
 def _version_from_metadata(package_name: str) -> str | None:
-    """Try to get version from importlib.metadata (pip-installed packages)."""
     try:
         from importlib.metadata import version
         for name in (package_name, package_name.replace("_", "-"), package_name.replace("-", "_")):
@@ -40,7 +38,6 @@ def _version_from_metadata(package_name: str) -> str | None:
 
 
 def _version_from_pyproject_toml(package_dir: Path) -> str | None:
-    """Try to read version from pyproject.toml ([project] or [tool.poetry] version)."""
     path = package_dir / "pyproject.toml"
     if not path.is_file():
         return None
@@ -54,7 +51,6 @@ def _version_from_pyproject_toml(package_dir: Path) -> str | None:
 
 
 def _version_from_package_json(package_dir: Path) -> str | None:
-    """Try to read version from package.json."""
     path = package_dir / "package.json"
     if not path.is_file():
         return None
@@ -68,7 +64,6 @@ def _version_from_package_json(package_dir: Path) -> str | None:
 
 
 def _version_from_version_file(package_dir: Path) -> str | None:
-    """Try __version__ in version.py (or similar) in the package dir."""
     for name in ("version.py", "version.json"):
         path = package_dir / name
         if not path.is_file():
@@ -89,20 +84,10 @@ def _version_from_version_file(package_dir: Path) -> str | None:
 
 
 def _is_git_repo(package_dir: Path) -> bool:
-    """True if package_dir is the root of a git repo (has .git)."""
     return (package_dir / ".git").exists()
 
 
 def get_installed_modules() -> list[dict]:
-    """
-    List installed custom node modules (directories under custom_nodes with __init__.py)
-    and their version if discoverable.
-
-    Returns a list of {"name": str, "version": str | null} sorted by name.
-    Version is resolved from: __init__.py, importlib.metadata, pyproject.toml,
-    package.json, version.py; if still unknown and the dir is a git repo, "nightly".
-    Does not import custom node code.
-    """
     result: list[dict] = []
     if not _CUSTOM_NODES_DIR.is_dir():
         return result
@@ -129,13 +114,6 @@ def get_installed_modules() -> list[dict]:
 
 
 def get_version_info_payload() -> dict:
-    """
-    Build the JSON payload for GET /workflowui/version_info.
-
-    Includes:
-    - workflowui_plugin_version: this plugin's version
-    - installed_custom_nodes: list of { name, version } for each custom node package
-    """
     return {
         "workflowui_plugin_version": __version__,
         "installed_custom_nodes": get_installed_modules(),

workflow_converter.py

@@ -0,0 +1,800 @@
+import json
+import logging
+from typing import Dict, Any, List, Tuple, Optional, Union
+logger = logging.getLogger(__name__)
+try:
+    import nodes
+except ImportError as e:
+    raise ImportError('Cannot import ComfyUI nodes module. This converter must be run within the ComfyUI environment. Make sure ComfyUI is properly initialized before using the converter.') from e
+_node_info_cache = {}
+
+def _strip_nulls(s: str) -> str:
+    if not isinstance(s, str):
+        return s
+    return s.replace('\x00', '').replace('\x00', '')
+
+def get_node_info_for_type(node_type: str) -> Dict[str, Any]:
+    global _node_info_cache
+    node_type = _strip_nulls(node_type)
+    if node_type not in _node_info_cache:
+        if node_type in nodes.NODE_CLASS_MAPPINGS:
+            try:
+                obj_class = nodes.NODE_CLASS_MAPPINGS[node_type]
+                info = {}
+                info['input'] = obj_class.INPUT_TYPES()
+                info['input_order'] = {key: list(value.keys()) for key, value in obj_class.INPUT_TYPES().items()}
+                _node_info_cache[node_type] = info
+            except ValueError as e:
+                if 'null bytes' in str(e).lower():
+                    logger.error("WorkflowConverter: node type '%s' triggered 'source code string cannot contain null bytes'. Treating as unknown (using fallback input order). Check this custom node's source or ComfyUI.", node_type, exc_info=True)
+                    _node_info_cache[node_type] = None
+                    return None
+                raise
+            except Exception as e:
+                logger.debug('Could not get node info for %s: %s', node_type, e)
+                _node_info_cache[node_type] = None
+        else:
+            _node_info_cache[node_type] = None
+    return _node_info_cache.get(node_type)
+
+class WorkflowConverter:
+
+    @staticmethod
+    def is_subgraph_uuid(node_type: str) -> bool:
+        if not node_type or not isinstance(node_type, str):
+            return False
+        if len(node_type) == 36 and node_type.count('-') == 4:
+            parts = node_type.split('-')
+            if len(parts) == 5 and all((len(p) in [8, 4, 4, 4, 12] for i, p in enumerate(parts) if i == 0 or i == 4 or len(p) == 4)):
+                return True
+        return False
+
+    @staticmethod
+    def expand_subgraph(subgraph_node_id: int, subgraph_def: Dict[str, Any], workflow_links: List) -> Tuple[List[Dict], List]:
+        expanded_nodes = []
+        expanded_links = []
+        internal_nodes = subgraph_def.get('nodes', [])
+        internal_links = subgraph_def.get('links', [])
+        max_link_id = 0
+        for link in workflow_links:
+            if isinstance(link, (list, tuple)) and len(link) > 0:
+                link_id = link[0]
+                if isinstance(link_id, int) and link_id > max_link_id:
+                    max_link_id = link_id
+        link_id_remap = {}
+        next_link_id = max_link_id + 1
+        for link in internal_links:
+            if isinstance(link, dict):
+                old_id = link.get('id')
+                if old_id is not None:
+                    link_id_remap[old_id] = next_link_id
+                    next_link_id += 1
+        internal_link_map = {}
+        for link in internal_links:
+            if isinstance(link, dict):
+                link_id = link.get('id')
+                internal_link_map[link_id] = link
+        subgraph_inputs = subgraph_def.get('inputs', [])
+        subgraph_outputs = subgraph_def.get('outputs', [])
+        input_slot_to_internal = {}
+        internal_to_output_slot = {}
+        for idx, input_def in enumerate(subgraph_inputs):
+            input_link_ids = input_def.get('linkIds', [])
+            targets = []
+            for link_id in input_link_ids:
+                if link_id in internal_link_map:
+                    link = internal_link_map[link_id]
+                    target_id = link.get('target_id')
+                    target_slot = link.get('target_slot')
+                    targets.append((target_id, target_slot))
+            if targets:
+                input_slot_to_internal[idx] = targets
+        for idx, output_def in enumerate(subgraph_outputs):
+            output_link_ids = output_def.get('linkIds', [])
+            for link_id in output_link_ids:
+                if link_id in internal_link_map:
+                    link = internal_link_map[link_id]
+                    origin_id = link.get('origin_id')
+                    origin_slot = link.get('origin_slot')
+                    internal_to_output_slot[origin_id, origin_slot] = idx
+        for node in internal_nodes:
+            internal_id = node.get('id')
+            expanded_node = node.copy()
+            expanded_node['id'] = f'{subgraph_node_id}:{internal_id}'
+            if 'inputs' in expanded_node:
+                updated_inputs = []
+                for input_info in expanded_node['inputs']:
+                    input_link = input_info.get('link')
+                    if input_link in internal_link_map:
+                        link_data = internal_link_map[input_link]
+                        if link_data.get('origin_id') == -10:
+                            input_copy = input_info.copy()
+                            input_copy['link'] = None
+                            updated_inputs.append(input_copy)
+                        else:
+                            input_copy = input_info.copy()
+                            if input_link in link_id_remap:
+                                input_copy['link'] = link_id_remap[input_link]
+                            updated_inputs.append(input_copy)
+                    else:
+                        updated_inputs.append(input_info)
+                expanded_node['inputs'] = updated_inputs
+            expanded_nodes.append(expanded_node)
+        for link in internal_links:
+            if isinstance(link, dict):
+                origin_id = link.get('origin_id')
+                target_id = link.get('target_id')
+                if origin_id in [-10, -20] or target_id in [-10, -20]:
+                    continue
+                old_link_id = link.get('id')
+                new_link_id = link_id_remap.get(old_link_id, old_link_id)
+                expanded_link = [new_link_id, f'{subgraph_node_id}:{origin_id}', link.get('origin_slot'), f'{subgraph_node_id}:{target_id}', link.get('target_slot'), link.get('type')]
+                expanded_links.append(expanded_link)
+        return (expanded_nodes, expanded_links, input_slot_to_internal, internal_to_output_slot)
+
+    @staticmethod
+    def is_api_format(workflow: Dict[str, Any]) -> bool:
+        if 'nodes' in workflow and 'links' in workflow:
+            return False
+        for key, value in workflow.items():
+            if key in ['prompt', 'extra_data', 'client_id']:
+                continue
+            if isinstance(value, dict) and 'class_type' in value:
+                return True
+        return False
+
+    @staticmethod
+    def _strip_null_bytes(obj: Any) -> Any:
+        if isinstance(obj, dict):
+            out = {}
+            for k, v in obj.items():
+                key = k.replace('\x00', '').replace('\x00', '') if isinstance(k, str) else k
+                out[key] = WorkflowConverter._strip_null_bytes(v)
+            return out
+        if isinstance(obj, list):
+            return [WorkflowConverter._strip_null_bytes(v) for v in obj]
+        if isinstance(obj, str):
+            return obj.replace('\x00', '').replace('\x00', '')
+        return obj
+
+    @staticmethod
+    def convert_to_api(workflow: Dict[str, Any]) -> Dict[str, Any]:
+        workflow = WorkflowConverter._strip_null_bytes(workflow)
+        if WorkflowConverter.is_api_format(workflow):
+            return workflow
+        workflow_nodes = workflow.get('nodes', [])
+        links = workflow.get('links', [])
+        node_types_in_workflow = sorted(set((n.get('type') or n.get('class_type') for n in workflow_nodes if isinstance(n, dict) and (n.get('type') or n.get('class_type')))))
+        logger.info('WorkflowConverter: starting conversion, %s nodes, %s links. Node types: %s', len(workflow_nodes), len(links), node_types_in_workflow[:40])
+        subgraph_defs = {}
+        definitions = workflow.get('definitions', {})
+        for subgraph in definitions.get('subgraphs', []):
+            subgraph_id = subgraph.get('id')
+            if subgraph_id:
+                subgraph_defs[subgraph_id] = subgraph
+        subgraph_input_mappings = {}
+        subgraph_output_mappings = {}
+        subgraph_slot_to_input_idx = {}
+        max_iterations = 10
+        iteration = 0
+        while iteration < max_iterations:
+            iteration += 1
+            expanded_nodes = []
+            found_subgraph = False
+            for node in workflow_nodes:
+                node_type = _strip_nulls(node.get('type') or '')
+                node_id = node.get('id')
+                if WorkflowConverter.is_subgraph_uuid(node_type) and node_type in subgraph_defs:
+                    found_subgraph = True
+                    logger.debug(f'Expanding subgraph node {node_id} (type: {node_type}) - iteration {iteration}')
+                    subgraph_def = subgraph_defs[node_type]
+                    sg_nodes, sg_links, input_map, output_map = WorkflowConverter.expand_subgraph(node_id, subgraph_def, links)
+                    expanded_nodes.extend(sg_nodes)
+                    links.extend(sg_links)
+                    subgraph_input_mappings[str(node_id)] = input_map
+                    subgraph_output_mappings[str(node_id)] = output_map
+                    outer_inputs = node.get('inputs', [])
+                    subgraph_inputs = subgraph_def.get('inputs', [])
+                    subgraph_input_name_to_idx = {}
+                    for idx, sg_input in enumerate(subgraph_inputs):
+                        input_name = sg_input.get('name')
+                        if input_name:
+                            subgraph_input_name_to_idx[input_name] = idx
+                    slot_mapping = {}
+                    for outer_slot, outer_input in enumerate(outer_inputs):
+                        outer_name = outer_input.get('name')
+                        if outer_name and outer_name in subgraph_input_name_to_idx:
+                            slot_mapping[outer_slot] = subgraph_input_name_to_idx[outer_name]
+                            logger.debug(f"Subgraph {node_id}: outer slot {outer_slot} ('{outer_name}') -> subgraph input idx {subgraph_input_name_to_idx[outer_name]}")
+                    subgraph_slot_to_input_idx[str(node_id)] = slot_mapping
+                else:
+                    expanded_nodes.append(node)
+            workflow_nodes = expanded_nodes
+            if not found_subgraph:
+                logger.debug(f'Subgraph expansion complete after {iteration} iteration(s)')
+                break
+        if iteration >= max_iterations:
+            logger.warning(f'Reached maximum subgraph expansion iterations ({max_iterations}). There may be circular subgraph references.')
+
+        def resolve_subgraph_output(node_id_str, slot, depth=0):
+            if depth > 100:
+                logger.warning(f'Max recursion depth reached resolving subgraph output for node {node_id_str}')
+                return (node_id_str, slot)
+            if node_id_str in subgraph_output_mappings:
+                output_map = subgraph_output_mappings[node_id_str]
+                for (internal_node, internal_slot), out_slot in output_map.items():
+                    if out_slot == slot:
+                        new_node_id = f'{node_id_str}:{internal_node}'
+                        return resolve_subgraph_output(new_node_id, internal_slot, depth + 1)
+            return (node_id_str, slot)
+
+        def resolve_subgraph_input(node_id_str, slot, depth=0):
+            if depth > 100:
+                logger.warning(f'Max recursion depth reached resolving subgraph input for node {node_id_str}')
+                return (node_id_str, slot)
+            if node_id_str in subgraph_input_mappings:
+                input_map = subgraph_input_mappings[node_id_str]
+                subgraph_input_idx = slot
+                if node_id_str in subgraph_slot_to_input_idx:
+                    slot_mapping = subgraph_slot_to_input_idx[node_id_str]
+                    if slot in slot_mapping:
+                        subgraph_input_idx = slot_mapping[slot]
+                if subgraph_input_idx in input_map:
+                    targets = input_map[subgraph_input_idx]
+                    if targets:
+                        internal_node, internal_slot = targets[0]
+                        new_node_id = f'{node_id_str}:{internal_node}'
+                        return resolve_subgraph_input(new_node_id, internal_slot, depth + 1)
+            return (node_id_str, slot)
+
+        def resolve_subgraph_input_all(node_id_str, slot, depth=0):
+            if depth > 100:
+                logger.warning(f'Max recursion depth reached resolving subgraph input for node {node_id_str}')
+                return [(node_id_str, slot)]
+            if node_id_str in subgraph_input_mappings:
+                input_map = subgraph_input_mappings[node_id_str]
+                subgraph_input_idx = slot
+                if node_id_str in subgraph_slot_to_input_idx:
+                    slot_mapping = subgraph_slot_to_input_idx[node_id_str]
+                    if slot in slot_mapping:
+                        subgraph_input_idx = slot_mapping[slot]
+                if subgraph_input_idx in input_map:
+                    targets = input_map[subgraph_input_idx]
+                    results = []
+                    for internal_node, internal_slot in targets:
+                        new_node_id = f'{node_id_str}:{internal_node}'
+                        results.extend(resolve_subgraph_input_all(new_node_id, internal_slot, depth + 1))
+                    return results if results else [(node_id_str, slot)]
+            return [(node_id_str, slot)]
+        node_input_updates = {}
+        updated_links = []
+        for link in links:
+            if len(link) >= 6:
+                link_id = link[0]
+                source_id = link[1]
+                source_slot = link[2]
+                target_id = link[3]
+                target_slot = link[4]
+                link_type = link[5] if len(link) > 5 else None
+                source_id_str = str(source_id)
+                source_id, source_slot = resolve_subgraph_output(source_id_str, source_slot)
+                target_id_str = str(target_id)
+                all_targets = resolve_subgraph_input_all(target_id_str, target_slot)
+                for resolved_target_id, resolved_target_slot in all_targets:
+                    if resolved_target_id != target_id_str:
+                        if resolved_target_id not in node_input_updates:
+                            node_input_updates[resolved_target_id] = {}
+                        node_input_updates[resolved_target_id][resolved_target_slot] = link_id
+                resolved_target_id, resolved_target_slot = all_targets[0]
+                target_id = resolved_target_id
+                target_slot = resolved_target_slot
+                updated_links.append([link_id, source_id, source_slot, target_id, target_slot, link_type])
+        links = updated_links
+        for node in workflow_nodes:
+            node_id_str = str(node.get('id'))
+            if node_id_str in node_input_updates and 'inputs' in node:
+                slot_to_link = node_input_updates[node_id_str]
+                inputs = node.get('inputs', [])
+                for i, input_info in enumerate(inputs):
+                    if i in slot_to_link:
+                        input_info['link'] = slot_to_link[i]
+        link_map = {}
+        nodes_with_connected_outputs = set()
+        for link in links:
+            if len(link) >= 6:
+                link_id = link[0]
+                source_id = link[1]
+                source_slot = link[2]
+                target_id = link[3]
+                target_slot = link[4]
+                link_type = link[5] if len(link) > 5 else None
+                link_map[link_id] = {'source_id': source_id, 'source_slot': source_slot, 'target_id': target_id, 'target_slot': target_slot, 'type': link_type}
+                nodes_with_connected_outputs.add(source_id)
+        primitive_values = {}
+        nodes_to_exclude = set()
+        bypassed_nodes = set()
+        set_node_sources = {}
+        get_node_vars = {}
+        node_by_id = {}
+        for node in workflow_nodes:
+            node_by_id[str(node.get('id'))] = node
+        for node in workflow_nodes:
+            node_id = node.get('id')
+            node_type = _strip_nulls(node.get('type') or '')
+            node_mode = node.get('mode', 0)
+            if node_mode == 4:
+                bypassed_nodes.add(str(node_id))
+                logger.debug(f'Tracking bypassed node {node_id} ({node_type})')
+            if node_type == 'PrimitiveNode':
+                node_id_str = str(node_id)
+                widget_values = node.get('widgets_values')
+                if widget_values and len(widget_values) > 0:
+                    primitive_values[node_id_str] = widget_values[0]
+            elif node_type == 'SetNode':
+                widget_values = node.get('widgets_values')
+                if widget_values and len(widget_values) > 0:
+                    var_name = widget_values[0]
+                    node_inputs = node.get('inputs', [])
+                    if node_inputs:
+                        for input_info in node_inputs:
+                            input_link = input_info.get('link')
+                            if input_link is not None and input_link in link_map:
+                                link_data = link_map[input_link]
+                                set_node_sources[var_name] = (link_data['source_id'], link_data['source_slot'])
+                                logger.debug(f"SetNode '{var_name}' receives from node {link_data['source_id']} slot {link_data['source_slot']}")
+                                break
+            elif node_type == 'GetNode':
+                widget_values = node.get('widgets_values')
+                if widget_values and len(widget_values) > 0:
+                    var_name = widget_values[0]
+                    get_node_vars[str(node_id)] = var_name
+                    logger.debug(f"GetNode {node_id} provides variable '{var_name}'")
+            outputs = node.get('outputs', [])
+            has_connected_output = False
+            for output in outputs:
+                output_links = output.get('links', [])
+                if output_links and len(output_links) > 0:
+                    has_connected_output = True
+                    break
+            if node_type == 'LoadImageOutput':
+                nodes_to_exclude.add(str(node_id))
+                logger.debug(f'Marking node {node_id} ({node_type}) for exclusion - UI-only node type')
+            elif not outputs or not has_connected_output:
+                node_class = nodes.NODE_CLASS_MAPPINGS.get(node_type) if hasattr(nodes, 'NODE_CLASS_MAPPINGS') else None
+                is_output_node = node_class and hasattr(node_class, 'OUTPUT_NODE') and node_class.OUTPUT_NODE
+                has_connected_input = False
+                if not is_output_node and (not node_class):
+                    node_inputs = node.get('inputs', [])
+                    for input_info in node_inputs:
+                        if input_info.get('link') is not None:
+                            has_connected_input = True
+                            break
+                if not is_output_node and (not has_connected_input):
+                    nodes_to_exclude.add(str(node_id))
+                    logger.debug(f'Marking node {node_id} ({node_type}) for exclusion - no connected outputs')
+                else:
+                    logger.debug(f'Keeping node {node_id} ({node_type}) - OUTPUT_NODE=True or has connected inputs despite no connected outputs')
+
+        def trace_through_get_set_nodes(source_node_id, source_slot, visited=None):
+            if visited is None:
+                visited = set()
+            source_node_id_str = str(source_node_id)
+            if source_node_id_str in visited:
+                return (source_node_id, source_slot)
+            visited.add(source_node_id_str)
+            if source_node_id_str in get_node_vars:
+                var_name = get_node_vars[source_node_id_str]
+                if var_name in set_node_sources:
+                    actual_source_id, actual_source_slot = set_node_sources[var_name]
+                    logger.debug(f"Tracing GetNode {source_node_id} (var '{var_name}') -> actual source {actual_source_id} slot {actual_source_slot}")
+                    return trace_through_get_set_nodes(actual_source_id, actual_source_slot, visited)
+                else:
+                    logger.warning(f"GetNode {source_node_id} references variable '{var_name}' but no SetNode found for it")
+            return (source_node_id, source_slot)
+
+        def trace_through_bypassed(source_node_id, source_slot, visited=None):
+            if visited is None:
+                visited = set()
+            if source_node_id in visited:
+                return (source_node_id, source_slot)
+            visited.add(source_node_id)
+            if source_node_id not in bypassed_nodes:
+                return (source_node_id, source_slot)
+            for node in workflow_nodes:
+                if str(node.get('id')) == str(source_node_id):
+                    node_type = node.get('type', 'unknown')
+                    node_inputs = node.get('inputs', [])
+                    node_outputs = node.get('outputs', [])
+                    output_type = None
+                    if node_outputs and source_slot < len(node_outputs):
+                        output_type = node_outputs[source_slot].get('type')
+                    if node_inputs:
+                        linked_input = None
+                        fallback_linked_input = None
+                        for idx, input_info in enumerate(node_inputs):
+                            input_link = input_info.get('link')
+                            input_type = input_info.get('type')
+                            if input_link is not None and input_link in link_map:
+                                if fallback_linked_input is None:
+                                    fallback_linked_input = input_link
+                                if output_type and input_type == output_type:
+                                    linked_input = input_link
+                                    break
+                        if linked_input is None:
+                            linked_input = fallback_linked_input
+                        if linked_input is not None:
+                            link_data = link_map[linked_input]
+                            upstream_source_id = link_data['source_id']
+                            upstream_source_slot = link_data['source_slot']
+                            upstream_source_id, upstream_source_slot = trace_through_get_set_nodes(upstream_source_id, upstream_source_slot)
+                            return trace_through_bypassed(upstream_source_id, upstream_source_slot, visited)
+                    break
+            return (source_node_id, source_slot)
+        api_prompt = {}
+        for node in workflow_nodes:
+            node_id = str(node.get('id'))
+            node_type = _strip_nulls(node.get('type') or '')
+            if not node_type:
+                continue
+            node_mode = node.get('mode', 0)
+            if node_mode == 2:
+                logger.debug(f'Skipping muted node {node_id} ({node_type})')
+                continue
+            elif node_mode == 4:
+                logger.debug(f'Skipping bypassed/disabled node {node_id} ({node_type})')
+                continue
+            if node_type in ['Note', 'PrimitiveNode', 'GetNode', 'SetNode']:
+                logger.debug(f'Skipping {node_type} node {node_id}')
+                continue
+            if node_id in nodes_to_exclude:
+                logger.debug(f'Skipping {node_type} node {node_id} - no connected outputs')
+                continue
+            api_node = {'inputs': {}, 'class_type': node_type}
+            if 'title' in node:
+                api_node['_meta'] = {'title': node['title']}
+            elif hasattr(nodes, 'NODE_DISPLAY_NAME_MAPPINGS') and node_type in nodes.NODE_DISPLAY_NAME_MAPPINGS:
+                api_node['_meta'] = {'title': nodes.NODE_DISPLAY_NAME_MAPPINGS[node_type]}
+            else:
+                api_node['_meta'] = {'title': node_type}
+            link_inputs = {}
+            primitive_inputs = {}
+            node_inputs = node.get('inputs', [])
+            if node_inputs:
+                for input_info in node_inputs:
+                    input_name = input_info.get('name')
+                    input_link = input_info.get('link')
+                    if input_link is not None and input_link in link_map:
+                        link_data = link_map[input_link]
+                        source_node_id = link_data['source_id']
+                        source_slot = link_data['source_slot']
+                        actual_source_id, actual_source_slot = trace_through_get_set_nodes(source_node_id, source_slot)
+                        if str(actual_source_id) in bypassed_nodes:
+                            actual_source_id, actual_source_slot = trace_through_bypassed(actual_source_id, actual_source_slot)
+                            if str(actual_source_id) in bypassed_nodes:
+                                logger.debug(f'Could not trace through bypassed node {source_node_id} for input {input_name}, falling back to widget value')
+                                continue
+                        actual_source_id, actual_source_slot = trace_through_get_set_nodes(actual_source_id, actual_source_slot)
+                        actual_source_id, actual_source_slot = resolve_subgraph_output(str(actual_source_id), actual_source_slot)
+                        source_node_id_str = str(actual_source_id)
+                        if source_node_id_str in primitive_values:
+                            primitive_inputs[input_name] = primitive_values[source_node_id_str]
+                        elif actual_source_id in nodes_to_exclude:
+                            logger.debug(f'Skipping input {input_name} from excluded node {source_node_id_str}')
+                        elif actual_source_id in bypassed_nodes:
+                            logger.warning(f'Could not resolve bypassed node {source_node_id} for input {input_name}, skipping connection')
+                        else:
+                            if actual_source_id != source_node_id:
+                                logger.info(f'Bypassed disabled node {source_node_id}, connecting {input_name} to {actual_source_id} (slot {actual_source_slot}) instead')
+                            link_inputs[input_name] = [str(actual_source_id), actual_source_slot]
+            logger.info('WorkflowConverter: processing node id=%s type=%s', node_id, node_type)
+            ordered_inputs = WorkflowConverter._get_ordered_inputs(node_type, node)
+            widget_values = node.get('widgets_values')
+            widget_inputs = {}
+            if widget_values is not None:
+                if isinstance(widget_values, dict):
+                    for key, value in widget_values.items():
+                        if key in ['videopreview', 'preview']:
+                            continue
+                        if key not in link_inputs:
+                            widget_inputs[key] = value
+                elif isinstance(widget_values, list):
+                    has_dict_widgets = any((isinstance(v, dict) for v in widget_values))
+                    if has_dict_widgets:
+                        WorkflowConverter._process_dict_widget_values(widget_values, widget_inputs, link_inputs)
+                    else:
+                        widget_mappings = WorkflowConverter._get_widget_mappings(node_type, node)
+                        filtered_values = WorkflowConverter._filter_control_values(widget_values)
+                        if widget_mappings:
+                            for i, value in enumerate(filtered_values):
+                                if i < len(widget_mappings):
+                                    widget_name = widget_mappings[i]
+                                    if widget_name and widget_name not in link_inputs:
+                                        widget_inputs[widget_name] = value
+                        elif filtered_values:
+                            logger.warning(f"Could not map widget values for unknown node type '{node_type}' (node {node_id})")
+            default_inputs = WorkflowConverter._get_default_inputs(node_type, widget_inputs, primitive_inputs, link_inputs)
+            if ordered_inputs:
+                for input_name in ordered_inputs:
+                    if input_name in widget_inputs:
+                        api_node['inputs'][input_name] = widget_inputs[input_name]
+                    elif input_name in primitive_inputs:
+                        api_node['inputs'][input_name] = primitive_inputs[input_name]
+                    elif input_name in default_inputs:
+                        api_node['inputs'][input_name] = default_inputs[input_name]
+                for input_name in ordered_inputs:
+                    if input_name in link_inputs and input_name not in api_node['inputs']:
+                        api_node['inputs'][input_name] = link_inputs[input_name]
+                for key, value in widget_inputs.items():
+                    if key not in api_node['inputs']:
+                        api_node['inputs'][key] = value
+                for key, value in primitive_inputs.items():
+                    if key not in api_node['inputs']:
+                        api_node['inputs'][key] = value
+                for key, value in default_inputs.items():
+                    if key not in api_node['inputs']:
+                        api_node['inputs'][key] = value
+                for key, value in link_inputs.items():
+                    if key not in api_node['inputs']:
+                        api_node['inputs'][key] = value
+            else:
+                for key, value in widget_inputs.items():
+                    api_node['inputs'][key] = value
+                for key, value in primitive_inputs.items():
+                    if key not in api_node['inputs']:
+                        api_node['inputs'][key] = value
+                for key, value in default_inputs.items():
+                    if key not in api_node['inputs']:
+                        api_node['inputs'][key] = value
+                for key, value in link_inputs.items():
+                    if key not in api_node['inputs']:
+                        api_node['inputs'][key] = value
+            api_prompt[node_id] = api_node
+        return api_prompt
+
+    @staticmethod
+    def _process_dict_widget_values(widget_values: List[Any], widget_inputs: Dict[str, Any], link_inputs: Dict[str, Any]) -> None:
+        lora_counter = 0
+        for value in widget_values:
+            if isinstance(value, dict):
+                if not value:
+                    continue
+                elif 'type' in value:
+                    widget_name = value.get('type')
+                    if widget_name and widget_name not in link_inputs:
+                        widget_inputs[widget_name] = value
+                elif 'lora' in value:
+                    lora_counter += 1
+                    widget_name = f'lora_{lora_counter}'
+                    if widget_name not in link_inputs:
+                        clean_value = {k: v for k, v in value.items() if k != 'strengthTwo' or v is not None}
+                        widget_inputs[widget_name] = clean_value
+                else:
+                    logger.debug(f'Unknown dict widget value structure: {value}')
+            elif isinstance(value, str):
+                if value == '':
+                    widget_inputs['➕ Add Lora'] = value
+
+    @staticmethod
+    def _filter_control_values(widget_values: List[Any]) -> List[Any]:
+        filtered = []
+        skip_next = False
+        for i, value in enumerate(widget_values):
+            if skip_next:
+                skip_next = False
+                continue
+            if value in ['fixed', 'increment', 'decrement', 'randomize']:
+                continue
+            if i + 1 < len(widget_values):
+                next_val = widget_values[i + 1]
+                if next_val in ['fixed', 'increment', 'decrement', 'randomize']:
+                    filtered.append(value)
+                    continue
+            filtered.append(value)
+        return filtered
+
+    @staticmethod
+    def _get_ordered_inputs(node_type: str, node: Dict[str, Any]) -> List[str]:
+        node_type = _strip_nulls(node_type)
+        properties = node.get('properties', {})
+        if 'Node name for S&R' in properties:
+            node_type = _strip_nulls(properties['Node name for S&R'])
+        node_info = get_node_info_for_type(node_type)
+        if node_info and 'input_order' in node_info:
+            input_order = node_info['input_order']
+            input_names = []
+            for section in ['required', 'optional']:
+                if section in input_order:
+                    input_names.extend(input_order[section])
+            if input_names:
+                return input_names
+        if hasattr(nodes, 'NODE_CLASS_MAPPINGS') and node_type in nodes.NODE_CLASS_MAPPINGS:
+            try:
+                node_class = nodes.NODE_CLASS_MAPPINGS[node_type]
+                input_types = node_class.INPUT_TYPES()
+                input_names = []
+                for section in ['required', 'optional']:
+                    if section in input_types:
+                        for input_name in input_types[section].keys():
+                            input_names.append(input_name)
+                if input_names:
+                    return input_names
+            except ValueError as e:
+                if 'null bytes' in str(e).lower():
+                    logger.warning("WorkflowConverter: node type '%s' raised null bytes in INPUT_TYPES(); using fallback order", node_type)
+                return []
+            except Exception as e:
+                logger.debug('Could not get input order from node class for %s: %s', node_type, e)
+        return []
+
+    @staticmethod
+    def _get_dynamic_combo_sub_inputs(input_name: str, input_spec: Any, widget_values: List[Any], current_idx: int) -> List[str]:
+        if not isinstance(input_spec, (list, tuple)) or len(input_spec) < 2:
+            return []
+        input_type = input_spec[0]
+        if not isinstance(input_type, str) or not input_type.startswith('COMFY_') or 'COMBO' not in input_type:
+            return []
+        spec_options = input_spec[1] if len(input_spec) > 1 else {}
+        if not isinstance(spec_options, dict):
+            return []
+        options = spec_options.get('options', [])
+        if not options or current_idx >= len(widget_values):
+            return []
+        selected_value = widget_values[current_idx]
+        for option in options:
+            if isinstance(option, dict) and option.get('key') == selected_value:
+                sub_inputs = option.get('inputs', {})
+                sub_input_names = []
+                for section in ['required', 'optional']:
+                    if section in sub_inputs:
+                        for sub_name in sub_inputs[section].keys():
+                            sub_input_names.append(f'{input_name}.{sub_name}')
+                return sub_input_names
+        return []
+
+    @staticmethod
+    def _get_widget_mappings(node_type: str, node: Dict[str, Any]) -> List[Optional[str]]:
+        node_type = _strip_nulls(node_type)
+        properties = node.get('properties', {})
+        if 'Node name for S&R' in properties:
+            node_type = _strip_nulls(properties['Node name for S&R'])
+        widget_values = node.get('widgets_values', [])
+        if isinstance(widget_values, dict):
+            widget_values = []
+        node_info = get_node_info_for_type(node_type)
+        if node_info and 'input' in node_info:
+            try:
+                input_def = node_info['input']
+                widget_names = []
+                widget_idx = 0
+                for section in ['required', 'optional']:
+                    if section in input_def:
+                        for input_name, input_spec in input_def[section].items():
+                            if isinstance(input_spec, (list, tuple)) and len(input_spec) >= 1:
+                                input_type = input_spec[0]
+                                is_widget = False
+                                is_dynamic_combo = False
+                                if isinstance(input_type, list):
+                                    is_widget = True
+                                elif input_type in ['INT', 'FLOAT', 'STRING', 'BOOLEAN', 'COMBO']:
+                                    is_widget = True
+                                elif isinstance(input_type, str) and input_type.startswith('COMFY_') and ('COMBO' in input_type):
+                                    is_widget = True
+                                    is_dynamic_combo = True
+                                elif isinstance(input_type, str) and (not input_type.isupper()):
+                                    is_widget = True
+                                if is_widget:
+                                    widget_names.append(input_name)
+                                    if is_dynamic_combo and widget_values:
+                                        sub_inputs = WorkflowConverter._get_dynamic_combo_sub_inputs(input_name, input_spec, widget_values, widget_idx)
+                                        widget_names.extend(sub_inputs)
+                                        widget_idx += 1 + len(sub_inputs)
+                                    else:
+                                        widget_idx += 1
+                if widget_names:
+                    return widget_names
+            except Exception as e:
+                logger.debug(f'Could not get widget mappings from node info for {node_type}: {e}')
+        if hasattr(nodes, 'NODE_CLASS_MAPPINGS') and node_type in nodes.NODE_CLASS_MAPPINGS:
+            try:
+                node_class = nodes.NODE_CLASS_MAPPINGS[node_type]
+                input_types = node_class.INPUT_TYPES()
+                widget_names = []
+                widget_idx = 0
+                for section in ['required', 'optional']:
+                    if section in input_types:
+                        for input_name, input_spec in input_types[section].items():
+                            if isinstance(input_spec, tuple) and len(input_spec) >= 1:
+                                input_type = input_spec[0]
+                                is_widget = False
+                                is_dynamic_combo = False
+                                if isinstance(input_type, list):
+                                    is_widget = True
+                                elif input_type in ['INT', 'FLOAT', 'STRING', 'BOOLEAN', 'COMBO']:
+                                    is_widget = True
+                                elif isinstance(input_type, str) and input_type.startswith('COMFY_') and ('COMBO' in input_type):
+                                    is_widget = True
+                                    is_dynamic_combo = True
+                                elif isinstance(input_type, str) and (not input_type.isupper()):
+                                    is_widget = True
+                                if is_widget:
+                                    widget_names.append(input_name)
+                                    if is_dynamic_combo and widget_values:
+                                        sub_inputs = WorkflowConverter._get_dynamic_combo_sub_inputs(input_name, input_spec, widget_values, widget_idx)
+                                        widget_names.extend(sub_inputs)
+                                        widget_idx += 1 + len(sub_inputs)
+                                    else:
+                                        widget_idx += 1
+                if widget_names:
+                    return widget_names
+            except ValueError as e:
+                if 'null bytes' in str(e).lower():
+                    logger.warning("WorkflowConverter: node type '%s' raised null bytes in INPUT_TYPES(); using fallback widget order", node_type)
+                return []
+            except Exception as e:
+                logger.debug('Could not get widget mappings from node class for %s: %s', node_type, e)
+        widget_values = node.get('widgets_values', [])
+        if not isinstance(widget_values, list) or len(widget_values) == 0:
+            return []
+        properties = node.get('properties', {})
+        ue_properties = properties.get('ue_properties', {})
+        widget_ue_connectable = ue_properties.get('widget_ue_connectable', {})
+        if widget_ue_connectable and isinstance(widget_ue_connectable, dict):
+            widget_names = list(widget_ue_connectable.keys())
+            if widget_names and len(widget_names) >= len(widget_values):
+                return widget_names[:len(widget_values)]
+        all_inputs = []
+        connected_inputs = set()
+        widget_flagged_inputs = []
+        for input_info in node.get('inputs', []):
+            input_name = input_info.get('name')
+            if input_name:
+                all_inputs.append(input_name)
+                if input_info.get('link') is not None:
+                    connected_inputs.add(input_name)
+                if input_info.get('widget'):
+                    widget_flagged_inputs.append(input_name)
+        if widget_flagged_inputs:
+            if len(widget_values) > len(widget_flagged_inputs):
+                potential_widgets = [inp for inp in all_inputs if inp not in connected_inputs and inp not in widget_flagged_inputs]
+                return widget_flagged_inputs + potential_widgets[:len(widget_values) - len(widget_flagged_inputs)]
+            return widget_flagged_inputs
+        unconnected = [inp for inp in all_inputs if inp not in connected_inputs]
+        if unconnected and len(unconnected) >= len(widget_values):
+            return unconnected[:len(widget_values)]
+        return []
+
+    @staticmethod
+    def _get_default_inputs(node_type: str, widget_inputs: Dict[str, Any], primitive_inputs: Dict[str, Any], link_inputs: Dict[str, Any]) -> Dict[str, Any]:
+        default_inputs = {}
+        node_type = _strip_nulls(node_type)
+        if not hasattr(nodes, 'NODE_CLASS_MAPPINGS') or node_type not in nodes.NODE_CLASS_MAPPINGS:
+            return default_inputs
+        try:
+            node_class = nodes.NODE_CLASS_MAPPINGS[node_type]
+            input_types = node_class.INPUT_TYPES()
+            for section in ['required', 'optional']:
+                if section not in input_types:
+                    continue
+                for input_name, input_spec in input_types[section].items():
+                    if input_name in widget_inputs or input_name in primitive_inputs or input_name in link_inputs:
+                        continue
+                    if isinstance(input_spec, (list, tuple)) and len(input_spec) >= 1:
+                        input_type = input_spec[0]
+                        spec_options = input_spec[1] if len(input_spec) >= 2 else {}
+                        if isinstance(spec_options, dict) and 'default' in spec_options:
+                            default_inputs[input_name] = spec_options['default']
+                            logger.debug(f"Adding default value for {node_type}.{input_name}: {spec_options['default']}")
+                        elif isinstance(input_type, list) and len(input_type) > 0:
+                            default_inputs[input_name] = input_type[0]
+                            logger.debug(f'Adding implicit combo default for {node_type}.{input_name}: {input_type[0]}')
+                        elif input_type == 'COMBO' and isinstance(spec_options, dict) and ('options' in spec_options):
+                            options = spec_options['options']
+                            if isinstance(options, list) and len(options) > 0:
+                                default_inputs[input_name] = options[0]
+                                logger.debug(f'Adding implicit COMBO default for {node_type}.{input_name}: {options[0]}')
+        except ValueError as e:
+            if 'null bytes' in str(e).lower():
+                logger.warning("WorkflowConverter: node type '%s' raised null bytes in _get_default_inputs; skipping defaults", node_type)
+        except Exception as e:
+            logger.debug('Could not get default inputs for %s: %s', node_type, e)
+        return default_inputs

workflow_export.py

@@ -0,0 +1,248 @@
+from __future__ import annotations
+
+import json
+import logging
+import os
+from pathlib import Path
+from typing import Any
+
+logger = logging.getLogger(__name__)
+
+_PLUGIN_DIR = Path(__file__).resolve().parent
+_COMFYUI_ROOT = _PLUGIN_DIR.parent.parent
+
+def _workflow_ui_link_widget_order() -> list[str]:
+    order = ["form_label"]
+    for i in range(8):
+        order.append(f"type_{i}")
+        order.append(f"name_{i}")
+    for i in range(8):
+        order.append(f"input_text_{i}")
+    for i in range(8):
+        order.append(f"input_number_{i}")
+    for i in range(8):
+        order.append(f"input_image_{i}")
+    for i in range(8):
+        order.append(f"input_video_{i}")
+    for i in range(8):
+        order.append(f"input_audio_{i}")
+    for i in range(8):
+        order.append(f"input_boolean_{i}")
+    return order
+
+_WIDGET_ORDER: dict[str, list[str]] = {
+    "WorkflowUILink": _workflow_ui_link_widget_order(),
+    "WorkflowUI Link": _workflow_ui_link_widget_order(),
+    "CLIPTextEncode": ["text"],
+    "Text Multiline": ["text"],
+    "PrimitiveStringMultiline": ["value"],
+    "EmptyLatentImage": ["width", "height", "batch_size"],
+    "EmptySD3LatentImage": ["width", "height", "batch_size"],
+    "SDXLEmptyLatentSizePicker+": ["resolution", "batch_size", "width_override", "height_override"],
+    "KSampler": ["seed", "steps", "cfg", "sampler_name", "scheduler", "denoise"],
+    "BasicScheduler": ["scheduler", "steps", "denoise"],
+    "UNETLoader": ["unet_name", "weight_dtype"],
+    "DiffusionModelLoader": ["unet_name", "weight_dtype"],
+    "LoadDiffusionModel": ["unet_name", "weight_dtype"],
+    "StableCascadeCheckpointLoader": ["key_opt_b", "key_opt_c", "cache_mode"],
+    "StableCascade_CheckpointLoader": ["key_opt_b", "key_opt_c", "cache_mode"],
+    "SD3CheckpointLoader": ["ckpt_name", "shift"],
+    "SD3LoadCheckpoint": ["ckpt_name", "shift"],
+    "SaveImage": ["filename_prefix"],
+    "Save Image": ["filename_prefix"],
+    "Save Image (api)": ["filename_prefix"],
+    "SaveImageNode": ["filename_prefix"],
+}
+
+
+def get_workflows_directory() -> Path | None:
+    env_path = os.environ.get("WORKFLOWUI_WORKFLOWS_DIR", "").strip()
+    if env_path:
+        p = Path(env_path).resolve()
+        if p.is_dir():
+            return p
+        return None
+    default = _COMFYUI_ROOT / "user" / "default" / "workflows"
+    if default.is_dir():
+        return default
+    return None
+
+
+def list_workflows() -> list[dict[str, str]]:
+    wf_dir = get_workflows_directory()
+    if not wf_dir:
+        return []
+    result: list[dict[str, str]] = []
+    try:
+        for path in sorted(wf_dir.glob("*.json")):
+            if path.is_file():
+                stem = path.stem
+                label = stem.replace("_", " ").strip() or stem
+                result.append({"id": stem, "label": label})
+    except OSError as e:
+        logger.warning("WorkflowUIPlugin: list_workflows failed: %s", e)
+    return result
+
+
+def _is_api_format(data: dict[str, Any]) -> bool:
+    if not data or not isinstance(data, dict):
+        return False
+    for v in data.values():
+        if isinstance(v, dict) and ("class_type" in v or "inputs" in v):
+            return True
+    return False
+
+
+def _extract_graph(workflow: dict[str, Any]) -> dict[str, Any]:
+    if not isinstance(workflow, dict):
+        return workflow
+    if workflow.get("nodes") is not None:
+        return workflow
+    if _is_api_format(workflow):
+        return workflow
+    for key in ("graph", "workflow", "prompt"):
+        inner = workflow.get(key)
+        if isinstance(inner, dict):
+            if inner.get("nodes") is not None or _is_api_format(inner):
+                return inner
+            deeper = inner.get("graph") or inner.get("workflow")
+            if isinstance(deeper, dict):
+                return deeper
+    return workflow
+
+
+def _build_link_map(links: list[Any]) -> dict[int | str, tuple[str, int]]:
+    result: dict[int | str, tuple[str, int]] = {}
+    if not isinstance(links, list):
+        return result
+    for link in links:
+        if not isinstance(link, (list, tuple)) or len(link) < 6:
+            continue
+        link_id, origin_id, origin_slot = link[0], link[1], link[2]
+        try:
+            origin_slot_int = int(origin_slot)
+        except (TypeError, ValueError):
+            continue
+        result[link_id] = (str(origin_id), origin_slot_int)
+    return result
+
+
+def _inputs_from_ui_node(
+    node: dict[str, Any],
+    class_type: str,
+    link_map: dict[int | str, tuple[str, int]],
+) -> dict[str, Any]:
+    inputs: dict[str, Any] = {}
+    node_inputs = node.get("inputs")
+    widgets_values = (
+        node.get("widgets_values")
+        or node.get("widgetsValues")
+        or node.get("widget_values")
+    )
+    widget_order = _WIDGET_ORDER.get(class_type)
+
+    if isinstance(widgets_values, dict):
+        inputs = dict(widgets_values)
+    elif isinstance(widgets_values, list) and widget_order:
+        for idx, field in enumerate(widget_order):
+            if idx < len(widgets_values):
+                inputs[field] = widgets_values[idx]
+
+    if isinstance(node_inputs, list):
+        widget_only_names: list[str] = []
+        for inp in node_inputs:
+            if not isinstance(inp, dict):
+                continue
+            name = inp.get("name")
+            if not name or not isinstance(name, str):
+                continue
+            link = inp.get("link")
+            if link is not None:
+                conn = link_map.get(link)
+                if conn is not None:
+                    inputs[name] = [conn[0], conn[1]]
+            else:
+                widget_only_names.append(name)
+        for idx, name in enumerate(widget_only_names):
+            if name not in inputs and isinstance(widgets_values, list) and idx < len(widgets_values):
+                inputs[name] = widgets_values[idx]
+    elif isinstance(node_inputs, dict) and not inputs:
+        inputs = dict(node_inputs)
+    return inputs
+
+
+def _ensure_saveimage_defaults(api_graph: dict[str, Any]) -> None:
+    save_types = ("SaveImage", "Save Image", "Save Image (api)", "SaveImageNode")
+    for node in api_graph.values():
+        if not isinstance(node, dict) or node.get("class_type") not in save_types:
+            continue
+        inputs = node.get("inputs")
+        if isinstance(inputs, dict) and not inputs.get("filename_prefix"):
+            inputs["filename_prefix"] = "ComfyUI"
+
+_NODE_MODE_DISABLED = 4
+
+
+def ui_to_api_format(workflow: dict[str, Any]) -> dict[str, Any]:
+    workflow = _extract_graph(workflow)
+    nodes = workflow.get("nodes")
+    if not isinstance(nodes, list):
+        if _is_api_format(workflow):
+            return workflow
+        return {}
+    links = workflow.get("links")
+    link_map = _build_link_map(links) if isinstance(links, list) else {}
+    out: dict[str, Any] = {}
+    for node in nodes:
+        if not isinstance(node, dict):
+            continue
+        if node.get("mode") == _NODE_MODE_DISABLED:
+            continue
+        nid = node.get("id")
+        if nid is None:
+            continue
+        node_id = str(nid)
+        class_type = node.get("type") or node.get("class_type")
+        if not class_type:
+            continue
+        inputs = _inputs_from_ui_node(node, class_type, link_map)
+        meta = node.get("_meta")
+        if not meta and isinstance(node.get("properties"), dict):
+            s_r_name = (node.get("properties") or {}).get("Node name for S&R")
+            if isinstance(s_r_name, str) and s_r_name.strip():
+                meta = {"title": s_r_name.strip()}
+        normalized: dict[str, Any] = {"class_type": class_type, "inputs": inputs}
+        if meta:
+            normalized["_meta"] = meta
+        out[node_id] = normalized
+    _ensure_saveimage_defaults(out)
+    return out
+
+
+def load_workflow(workflow_id: str) -> tuple[str, dict[str, Any]]:
+    wf_dir = get_workflows_directory()
+    if not wf_dir:
+        raise FileNotFoundError("Workflow directory not configured")
+    safe_id = (workflow_id or "").strip()
+    if not safe_id or "/" in safe_id or "\\" in safe_id or ".." in safe_id:
+        raise ValueError("Invalid workflow id")
+    path = (wf_dir / safe_id).with_suffix(".json")
+    if not path.is_file():
+        raise FileNotFoundError(f"Workflow not found: {safe_id}")
+    try:
+        raw = path.read_text(encoding="utf-8", errors="replace")
+        data = json.loads(raw)
+    except json.JSONDecodeError as e:
+        raise ValueError(f"Invalid workflow JSON: {e}") from e
+    if not isinstance(data, dict):
+        raise ValueError("Workflow must be a JSON object")
+    name = data.get("name")
+    if not isinstance(name, str) and isinstance(data.get("workflow"), dict):
+        name = data["workflow"].get("name")
+    if not isinstance(name, str) or not name.strip():
+        name = safe_id.replace("_", " ").strip() or safe_id
+
+    graph = _extract_graph(data)
+    if not graph or (isinstance(graph, dict) and not graph.get("nodes") and not _is_api_format(graph)):
+        raise ValueError("Workflow has no executable nodes")
+    return name, graph

workflow_ui_link.py

@@ -1,16 +1,3 @@
-"""
-WorkflowUILink: A ComfyUI custom node that defines which form fields WorkflowUI shows.
-
-IMPORTANT - form_label: This is displayed as the header/section title above the inputs in the
-WorkflowUI app. This is the main text users see when filling out the form. Set it to describe
-your inputs (e.g. "Generation parameters", "Image settings").
-
-Configuration:
-- form_label: Header shown above inputs in the WorkflowUI app (displayed as section title)
-- type_0..7: Dropdown to select field type per slot (text, number, seed, image, video, audio, boolean, select)
-- name_0..7: Field name and label shown in the app (use "seed" for master seed binding; reserved convention)
-"""
-
 from __future__ import annotations
 
 import logging
@@ -29,12 +16,10 @@ except ImportError:
 
 MAX_INPUT_SLOTS = 8
 
-# Type options for the per-slot dropdown; empty = unused slot
 TYPE_OPTIONS = ["", "text", "number", "seed", "image", "video", "audio", "boolean", "select"]
 
 
 def _input_field_for_type(slot: int, typ: str) -> str:
-    """Map definition type to actual input field name for binding."""
     if typ == "image":
         return f"input_image_{slot}"
     if typ == "video":
@@ -49,7 +34,6 @@ def _input_field_for_type(slot: int, typ: str) -> str:
 
 
 def _get_media_files(content_types: list[str]) -> list[str]:
-    """Get sorted list of filenames from input directory filtered by content type."""
     try:
         import folder_paths
         input_dir = folder_paths.get_input_directory()
@@ -63,7 +47,6 @@ def _get_media_files(content_types: list[str]) -> list[str]:
 
 
 def _build_input_types() -> dict:
-    """Build INPUT_TYPES with title widgets and file pickers for image/video/audio."""
     try:
         import folder_paths
         input_dir = folder_paths.get_input_directory()
@@ -87,7 +70,6 @@ def _build_input_types() -> dict:
             "placeholder": f"Name and label in app (e.g. 'Positive prompt'; use 'seed' for master seed)",
         })
 
-    # Ensure '' is always first so unused slots can send empty and pass validation
     image_opts = [""] + (image_files or [])
     video_opts = [""] + (video_files or [])
     audio_opts = [""] + (audio_files or [])
@@ -132,7 +114,6 @@ def _build_return_names() -> tuple[str, ...]:
 
 
 def _load_image_from_path(image: str) -> Any:
-    """Load image from filename (ComfyUI input dir) and return IMAGE tensor [1,H,W,3]."""
     if not _HAS_IMAGE_DEPS or not image or not isinstance(image, str):
         return None
     try:
@@ -162,7 +143,6 @@ def _empty_image_placeholder() -> Any:
 
 
 class WorkflowUILink:
-    """Pass-through node: defines app form inputs with per-slot labels and file pickers for media."""
 
     @classmethod
     def INPUT_TYPES(cls) -> dict:
@@ -201,7 +181,6 @@ class WorkflowUILink:
         return tuple(results)
 
 
-# Registration
 NODE_CLASS_MAPPINGS = {
     "WorkflowUILink": WorkflowUILink,
 }