mirror of
https://github.com/got-feedBack/feedBack.git
synced 2026-09-12 05:08:31 +00:00
4.1 KiB
4.1 KiB
name, description
| name | description |
|---|---|
| plugin-validate | Validate a Slopsmith plugin's plugin.json against the schema and run structural checks. USE WHEN the user asks to validate a plugin, check the manifest, plugin.json errors, lint plugin, verify plugin structure, plugin license check, or audit plugin contract. Runs JSON Schema validation, capability metadata checks, file-existence checks for declared script/routes/settings.html/tour paths, and license-allowlist check. |
plugin-validate
Runs the same checks .github/workflows/validate-plugins.yml + tests/test_plugin_schema.py run in CI, but locally and instantly — useful for catching errors before push.
When to invoke
The user says one of:
- "validate plugins/"
- "check the manifest for X"
- "lint this plugin"
- "verify my plugin.json"
- "plugin license check"
Works on either a specific plugin (e.g. plugins/highway_3d/) or all in-tree plugins if none is specified.
What to run
# All in-tree plugins (default)
python <<'PY'
import json, glob, sys, jsonschema, pathlib
schema = json.load(open('schema/plugin.schema.json'))
jsonschema.Draft202012Validator.check_schema(schema)
ok = True
for path in sorted(glob.glob('plugins/*/plugin.json')):
plugin_dir = pathlib.Path(path).parent
plugin_id = plugin_dir.name
m = json.load(open(path))
plugin_ok = True # per-iteration flag so we don't print OK after a later FAIL
# 1. Schema, including capability-pipelines.v1 metadata
try:
jsonschema.validate(m, schema)
except jsonschema.ValidationError as e:
print(f"FAIL {path}: {e.message} (at {list(e.absolute_path)})")
ok = False
continue
# 2. id == directory name
if m['id'] != plugin_id:
print(f"FAIL {path}: id={m['id']!r} but directory is {plugin_id!r}")
ok = False
plugin_ok = False
# 3. Declared files exist
for field in ('script', 'routes', 'tour'):
if field in m and not (plugin_dir / m[field]).exists():
print(f"FAIL {path}: {field}={m[field]!r} but file missing")
ok = False
plugin_ok = False
if 'settings' in m and 'html' in m['settings']:
h = m['settings']['html']
if not (plugin_dir / h).exists():
print(f"FAIL {path}: settings.html={h!r} but file missing")
ok = False
plugin_ok = False
for field in ('settings', 'diagnostics'):
if field in m and 'server_files' in m[field]:
for relpath in m[field]['server_files']:
# server_files paths live under context["config_dir"], NOT the
# plugin dir, so we don't existence-check them here. We only
# confirm the path *looks* safe (already enforced by schema).
if '..' in relpath or relpath.startswith('/') or '\\' in relpath:
print(f"FAIL {path}: {field}.server_files contains unsafe path {relpath!r}")
ok = False
plugin_ok = False
if plugin_ok:
print(f"OK {path}")
sys.exit(0 if ok else 1)
PY
Targeted invocation
If the user names a specific plugin, swap the glob for plugins/<id>/plugin.json and report on just that one.
License-allowlist check
If the user specifically asks for a license check (or the plugin declares license in plugin.json), additionally run:
pytest tests/test_plugin_schema.py::test_schema_license_enum_subset_of_contributing_allowlist -v --noconftest
(or skip --noconftest if structlog is installed locally).
Output
Use the format from the script: OK <path> per validated manifest, FAIL <path>: <reason> per failure. Add a one-line summary:
Result: 3/3 plugins valid (OK app_tour_library, app_tour_settings, highway_3d)
or
Result: 2/3 plugins valid; 1 FAIL (see above)