FL Studio 25.x: restore semantic mixer parsing for mixer diff

[aernw]

Describe the feature

Stemhub’s FL Studio mixer diff relies on pyflp semantic extraction (project.mixer).
With FL Studio 25.2.4 snapshots, pyflp.parse() succeeds but the mixer model is empty (NoModelsFound / 0 inserts), so downstream diff reports 0 changes even when mixer state changed (pan, insert FX, slot mix, routing/send, etc.).

We need full semantic mixer support for modern FL Studio snapshots so backend compare can produce real change types (insert_pan_changed, slot_plugin_changed, etc.) instead of fallback binary-only detection.

What version of PyFLP are you using?

stemhub-org/PyFLP_v2 (branch dev), currently vendored in Stemhub backend at commit 54d2b9628ee9e95909199e7230353577ebe8394c.

Screenshots, Additional info

Context and observed behavior

• Input FLP version in failing samples: 25.2.4.4960.
• pyflp.parse() returns a Project, but:
  • list(project.mixer) -> NoModelsFound / empty
  • no expected mixer event IDs (MixerID.Params, InsertID.Flags, InsertID.Name, SlotID.Index) in parsed events
• The FLP files themselves are different at binary level across versions (different SHA256 + size), confirming real project changes exist.

Goal

Implement robust mixer parsing for FL Studio 25.x artifacts in PyFLP_v2, with regression tests and compatibility guarantees for previously supported FLP versions.

Non-goals

• No DAW UI or plugin-side diff logic in this issue.
• No backend API schema change required here (handled in Stemhub).

---

Implementation plan (iterative)

Iteration 1 — Repro harness + failing fixtures

• Add minimal reproducible fixtures in tests/assets/ (or equivalent): at least two FLP files from FL 25.x with known mixer differences.
• Add a parser report script (or test helper) to inspect event IDs/chunks present in these fixtures.
• Add failing tests documenting expected mixer extraction behavior:
  • project.mixer must expose inserts
  • changed fixture pair must differ on at least one mixer semantic field (pan/slot/plugin/etc.).

Deliverable: deterministic failing tests proving current gap.

Iteration 2 — Event mapping reverse-engineering for FL 25.x mixer data

• Identify where FL 25.x stores mixer data when legacy IDs are absent/shifted.
• Document mapping candidates between raw events/chunks and mixer concepts:
  • insert identity/index/name/enabled/pan/volume
  • slot index/enabled/mix/plugin key
  • send/routing value where available
• Add inline parser documentation/comments with source evidence (event IDs/chunk shapes).

Deliverable: documented mapping table and decoded sample outputs.

Iteration 3 — Parser/model updates

• Update relevant event enums/constructs/model selectors to include FL 25.x mixer layout.
• Ensure parser remains tolerant:
  • unknown chunks continue as unknown
  • mixer extraction works when possible, without regressing parse for older versions.
• Keep existing public API stable (Project.mixer, Insert, Slot semantics).

Deliverable: parser returns non-empty mixer model for FL 25.x fixtures.

Iteration 4 — Regression and compatibility test suite

• Add/expand tests:
  • FL 25.x fixture semantic assertions
  • legacy fixtures still pass unchanged
  • malformed/partial chunks do not crash global parse path.
• Add focused tests for each mixer field used by Stemhub diff:
  • insert name/enabled/pan/volume
  • slot index/plugin/enabled/mix
  • (optional) routing/send where supported.

Deliverable: green test suite covering new and old behavior.

Iteration 5 — Release/integration handoff

• Publish changelog entry describing FL 25.x mixer support additions and limitations.
• Tag/release PyFLP_v2 version used by Stemhub.
• Provide integration notes for vendoring:
  • required dependency versions
  • any new parser caveats
  • expected behavior when mixer data still unavailable in edge files.

Deliverable: releasable parser update ready for Stemhub vendor bump.

---

Acceptance criteria

• For FL Studio 25.x fixture projects, project.mixer yields inserts/slots with stable semantic fields.
• A fixture pair with known mixer edits yields semantic differences at parser/model level (not only binary file hash changes).
• Existing supported fixture versions do not regress.
• Unknown/non-mixer chunks do not prevent parsing other valid mixer data.
• Documentation/changelog clearly state supported FL Studio versions and known limits.

Suggested task checklist

• Add FL 25.x fixtures + failing tests
• Map new mixer-related event/chunk layout
• Implement parser/model support
• Add regression tests for fields consumed by Stemhub diff
• Validate backward compatibility with existing fixtures
• Document + release for vendor update