File size: 11,084 Bytes
0c85e62 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 | """Validate a ComfyUI workflow JSON for internal consistency.
Usage:
python scripts/validate_workflow.py <workflow.json> [--quiet]
Checks performed:
- The file parses as JSON.
- Every link referenced by node inputs/outputs is declared in the top-level
`links` table.
- Every entry in the `links` table is referenced by both endpoints (no
orphans).
- For each link, the source/destination nodes and slots exist and the slot
type matches the declared link type.
- `last_node_id >= max(node ids)` and `last_link_id >= max(link ids)`.
- No duplicate node IDs or link IDs.
Exit codes:
0 All checks passed.
1 Problems found (printed to stderr).
2 File not found or not valid JSON.
This catches the class of bugs that come from hand-editing workflow JSONs
(e.g. when an agent re-wires a graph to build a slim variant). Run it after
any non-trivial edit before re-loading the workflow into ComfyUI — Comfy's
loader will sometimes silently drop or mis-wire the bad parts, producing a
graph that looks fine but behaves wrong.
stdlib only, no third-party dependencies.
"""
from __future__ import annotations
import argparse
import json
import sys
from pathlib import Path
def _sanitize_slots(
node: dict, key: str, problems: list[str]
) -> list[dict]:
"""Return a list-of-dicts view of node[key], appending a problem for any
malformed entry. Non-list `node[key]` → reported and treated as empty.
Non-dict entries → reported and dropped.
"""
slots = node.get(key)
if slots is None:
return []
if not isinstance(slots, list):
problems.append(
f"node {node.get('id')!r}: {key!r} must be a list, got {type(slots).__name__}"
)
return []
cleaned: list[dict] = []
for j, slot in enumerate(slots):
if not isinstance(slot, dict):
problems.append(
f"node {node.get('id')!r}: {key}[{j}] must be an object, got {type(slot).__name__}"
)
continue
cleaned.append(slot)
return cleaned
def _slot_type_matches(slot_type, link_type) -> bool:
"""ComfyUI wildcard sockets (`*`) accept any concrete link type."""
return slot_type == link_type or slot_type == "*"
def validate(data) -> list[str]:
"""Return a list of problem strings; empty list means OK.
Defensive: every shape assumption is guarded so that malformed-but-valid
JSON produces problem lines, never an uncaught exception. Top-level not
being an object is the only short-circuit return (nothing else is safe
to do without a dict).
"""
problems: list[str] = []
# Top-level must be a JSON object.
if not isinstance(data, dict):
return [
f"top-level JSON must be an object, got {type(data).__name__}"
]
# `nodes` and `links` arrays — must be lists; non-list reported and skipped.
raw_nodes = data.get("nodes") or []
if not isinstance(raw_nodes, list):
problems.append(
f"'nodes' must be a list, got {type(raw_nodes).__name__}"
)
raw_nodes = []
raw_links = data.get("links") or []
if not isinstance(raw_links, list):
problems.append(
f"'links' must be a list, got {type(raw_links).__name__}"
)
raw_links = []
# Filter to well-shaped node entries, recording each bad one.
nodes: list[dict] = []
for i, n in enumerate(raw_nodes):
if not isinstance(n, dict):
problems.append(
f"nodes[{i}]: must be an object, got {type(n).__name__}"
)
continue
nodes.append(n)
# Filter to well-shaped link entries. Each must be a 6-element list.
links: list[list] = []
for i, link in enumerate(raw_links):
if not isinstance(link, list):
problems.append(
f"links[{i}]: must be a list of 6 elements, got {type(link).__name__}"
)
continue
if len(link) != 6:
problems.append(
f"links[{i}]: malformed — expected 6 elements "
f"[id, src, src_slot, dst, dst_slot, type], got {len(link)}"
)
continue
links.append(link)
# Pre-sanitize each node's inputs/outputs into list-of-dict views so
# downstream code can index safely. Caches per-node so problems aren't
# double-reported when a node is touched by multiple links.
sanitized_io: dict[int, dict[str, list[dict]]] = {}
for n in nodes:
sanitized_io[id(n)] = {
"inputs": _sanitize_slots(n, "inputs", problems),
"outputs": _sanitize_slots(n, "outputs", problems),
}
# Duplicate IDs (over good entries).
node_ids = [n.get("id") for n in nodes]
link_ids = [link[0] for link in links]
dup_nodes = sorted({i for i in node_ids if node_ids.count(i) > 1})
dup_links = sorted({i for i in link_ids if link_ids.count(i) > 1})
if dup_nodes:
problems.append(f"Duplicate node IDs: {dup_nodes}")
if dup_links:
problems.append(f"Duplicate link IDs: {dup_links}")
# Header counters should bound observed IDs (Comfy uses these to assign
# the next id when adding nodes; if they lag, the next add can collide).
int_node_ids = [i for i in node_ids if isinstance(i, int)]
int_link_ids = [i for i in link_ids if isinstance(i, int)]
if int_node_ids:
last_node_id = data.get("last_node_id", -1)
if not isinstance(last_node_id, int) or last_node_id < max(int_node_ids):
problems.append(
f"last_node_id={last_node_id!r} < max(node id)={max(int_node_ids)}"
)
if int_link_ids:
last_link_id = data.get("last_link_id", -1)
if not isinstance(last_link_id, int) or last_link_id < max(int_link_ids):
problems.append(
f"last_link_id={last_link_id!r} < max(link id)={max(int_link_ids)}"
)
# Cross-reference link declarations vs node IO references.
declared = set(int_link_ids)
referenced: set[int] = set()
for n in nodes:
io = sanitized_io[id(n)]
for inp in io["inputs"]:
link_val = inp.get("link")
if isinstance(link_val, int):
referenced.add(link_val)
for out in io["outputs"]:
out_links = out.get("links") or []
if not isinstance(out_links, list):
problems.append(
f"node {n.get('id')!r}: output.links must be a list, "
f"got {type(out_links).__name__}"
)
continue
for lk in out_links:
if isinstance(lk, int):
referenced.add(lk)
for missing in sorted(referenced - declared):
problems.append(
f"link {missing}: referenced by a node but not in the links table"
)
for orphan in sorted(declared - referenced):
problems.append(
f"link {orphan}: declared in the links table but unused by any node"
)
# Per-link endpoint validation.
# Build node_by_id over int ids only — non-int ids can't match a link's
# src/dst (which are ints in the link tuple), and treating them as keys
# would mask bad data.
node_by_id = {n.get("id"): n for n in nodes if isinstance(n.get("id"), int)}
for link in links:
lid, src, src_slot, dst, dst_slot, ltype = link
src_node = node_by_id.get(src)
dst_node = node_by_id.get(dst)
if src_node is None:
problems.append(f"link {lid}: src node {src!r} missing")
continue
if dst_node is None:
problems.append(f"link {lid}: dst node {dst!r} missing")
continue
src_outs = sanitized_io[id(src_node)]["outputs"]
dst_ins = sanitized_io[id(dst_node)]["inputs"]
# Slot index bounds + type checks.
if not isinstance(src_slot, int) or src_slot < 0 or src_slot >= len(src_outs):
problems.append(
f"link {lid}: src slot {src_slot!r} out of bounds on node {src} "
f"(has {len(src_outs)} outputs)"
)
elif not _slot_type_matches(src_outs[src_slot].get("type"), ltype):
problems.append(
f"link {lid}: src type mismatch — link declares {ltype!r}, "
f"slot is {src_outs[src_slot].get('type')!r}"
)
if not isinstance(dst_slot, int) or dst_slot < 0 or dst_slot >= len(dst_ins):
problems.append(
f"link {lid}: dst slot {dst_slot!r} out of bounds on node {dst} "
f"(has {len(dst_ins)} inputs)"
)
elif not _slot_type_matches(dst_ins[dst_slot].get("type"), ltype):
problems.append(
f"link {lid}: dst type mismatch — link declares {ltype!r}, "
f"slot is {dst_ins[dst_slot].get('type')!r}"
)
# Endpoint cross-check: node IO must reference this link id symmetrically.
if isinstance(src_slot, int) and 0 <= src_slot < len(src_outs):
out_links = src_outs[src_slot].get("links") or []
if isinstance(out_links, list) and lid not in out_links:
problems.append(
f"link {lid}: missing from node {src} output[{src_slot}].links={out_links}"
)
if isinstance(dst_slot, int) and 0 <= dst_slot < len(dst_ins):
in_link = dst_ins[dst_slot].get("link")
if in_link != lid:
problems.append(
f"link {lid}: node {dst} input[{dst_slot}].link={in_link} "
f"does not match"
)
return problems
def main(argv: list[str] | None = None) -> int:
parser = argparse.ArgumentParser(
description=__doc__.splitlines()[0],
)
parser.add_argument(
"path",
type=Path,
help="Path to the workflow JSON file.",
)
parser.add_argument(
"-q",
"--quiet",
action="store_true",
help="Suppress the OK summary; only print problems.",
)
args = parser.parse_args(argv)
if not args.path.exists():
print(f"error: file not found: {args.path}", file=sys.stderr)
return 2
try:
data = json.loads(args.path.read_text(encoding="utf-8"))
except json.JSONDecodeError as exc:
print(f"error: {args.path.name}: JSON parse error: {exc}", file=sys.stderr)
return 2
problems = validate(data)
if problems:
print(
f"FAIL: {args.path.name} ({len(problems)} problem(s))",
file=sys.stderr,
)
for p in problems:
print(f" - {p}", file=sys.stderr)
return 1
if not args.quiet:
print(
f"OK: {args.path.name} — "
f"{len(data.get('nodes') or [])} nodes, "
f"{len(data.get('links') or [])} links, "
f"{len(data.get('groups') or [])} groups"
)
return 0
if __name__ == "__main__":
sys.exit(main())
|