Fix JSON node types cache: commit missing file and repair union deserialization

[Copilot]

The JSON cache was never actually loading — the missing node_types_cache.json file caused fallback to full parse on every startup, and a silent KeyError bug in the union deserialization meant the cache would have failed even if the file existed.

Root cause: DirectConnection | PositionalConnections union always failed

TypeAdapter tries DirectConnection first on all connection dicts. Connection.__init__ calls super().__init__(**data) which triggers pydantic's internal field extractor — a compiled lambda that raises KeyError: 'role' (not ValidationError) when a PositionalConnections dict is processed. The original code silently swallowed this via except KeyError, making the cache a permanent no-op.

# Before: union TypeAdapter always KeyErrors on PositionalConnections dicts
connections: list[DirectConnection | PositionalConnections]  # ← never works

# After: validate as raw dicts, reconstruct per-item using 'role' as discriminator
connections: list[dict]  # in TypedDict
# then:
DirectConnection.model_validate(c) if isinstance(c, dict) and "role" in c
else PositionalConnections.model_validate(c)

Changes

• node_types_cache.json — generated and committed (13 MB compact JSON); added REUSE license sidecar
• node_type_parser.py — fix union deserialization; move ValidationError/TypeAdapter imports to module scope (fixes fragile import-inside-try); extract _ensure_cache_adapter(), _reconstruct_cache(), _clear_stale_cached_properties() helpers; cache TypeAdapter as ClassVar[TypeAdapter[Any] | None] built once per class lifetime; remove redundant json.JSONDecodeError (covered by ValidationError)
• preprocess-node-types.py — emit compact JSON (no indentation) to stay under the 13 MB VCS limit
• hk.pkl — raise check-added-large-files limit from 10 000 → 15 000 KB to accommodate the cache artifact

---

💬 Send tasks to Copilot coding agent from Slack and Teams to turn conversations into code. Copilot posts an update in your thread when it's finished.

[Copilot]

Just as a heads up, I was blocked by some firewall rules while working on your feedback. Expand below for details.

Warning

Firewall rules blocked me from connecting to one or more addresses (expand for details)

I tried to connect to the following addresses, but was blocked by firewall rules:

• openaipublic.blob.core.windows.net
  • Triggering command: /home/REDACTED/work/codeweaver/codeweaver/.venv/bin/python python -m pytest tests/unit/ -v --timeout=60 --no-cov -q (dns block)

If you need me to access, download, or install something from one of these locations, you can either:

• Configure Actions setup steps to set up my environment, which run before the firewall is enabled
• Add the appropriate URLs or hosts to the custom allowlist in this repository's Copilot coding agent settings (admins only)

[Copilot]

Pull request overview

Fixes the semantic node-types JSON cache so it can be loaded reliably at startup (instead of silently falling back to full JSON parsing), and commits the generated cache artifact needed for runtime loading.

Changes:

• Repair cache deserialization by validating cached connections as raw dicts and reconstructing DirectConnection vs PositionalConnections using "role" as a discriminator.
• Add and license the generated node_types_cache.json artifact; update preprocessing to emit compact JSON.
• Bump the check-added-large-files hook limit and pin hk in mise.toml to accommodate the cache artifact.

Reviewed changes

Copilot reviewed 5 out of 6 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
src/codeweaver/semantic/node_type_parser.py	Fixes cache load/deserialization; adds cached TypeAdapter and reconstruction helpers.
src/codeweaver/semantic/data/node_types_cache.json	Adds the missing runtime cache artifact (large JSON).
src/codeweaver/semantic/data/node_types_cache.json.license	Adds REUSE/Spdx licensing sidecar for the cache artifact.
scripts/build/preprocess-node-types.py	Writes compact JSON to reduce cache size.
mise.toml	Pins hk tool version.
hk.pkl	Raises large-file hook threshold to allow committing the cache file.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.