Lane 11 PR 2: the sheet language — expressions, evaluator, dice, definitions, validator, conformance fixtures (#737) #961

Merged

claude-bot merged 3 commits from build/lane11-2-evaluator into main 2026-09-30 07:59:55 +00:00

Conversation 0 Commits 3 Files changed 39 +9093

claude-bot commented 2026-09-30 07:25:49 +00:00

Collaborator

Copy link

Lane 11 (#737) PR 2. This is the engine every sheet is computed with (IS-405 server, IS-431, IS-425). It adds no database, no routes and no UI; nothing outside app/sheets/ imports it yet.

webapp/backend/app/sheets/:

• expr.py: the expression language from the contract's §3.4. It is total and side-effect free.
  • Syntax: paths @a.b.c, [*] lists, repeat keys, and row./entry. locals.
  • Functions: if, the math functions, lookup, rank (plus a two-argument form), has, and sum/count/any/all/min/max.
  • Numbers: floored %, round half away from zero, and normalised output.
  • Nulls: a missing path is null, arithmetic treats null as 0, and a comparison with null is false.
  • Errors: typed, with their offsets. There are limits of 2,000 characters and 64 levels of nesting.
  • Dice are rejected here; they belong to generators only.
• evaluator.py: evaluates a definition over data/entries/resources/overrides/ext.
  • Derivations: the terms a sheet prints under a value.
  • Declarative modifiers: add, multiply, upgrade, downgrade, cap and override, stacked bonus (best per type) or flat.
  • Also: when, repeat groups, entries with provenance and a snapshot, derived rows, resources and tracks.
  • Order and caching: evaluation follows the dependency order, with a cycle guard, and compiled definitions are cached by digest.
• dice.py: NdM, keep/drop, an injectable RNG, and the individual dice kept. It's for creation and levelling only (D85).
• definition_model.py and definition.schema.json: the Pydantic model and the published JSON Schema.
• validator.py: the §3.2 checks.
  • Every @path/ref resolves, with did-you-mean.
  • No cycles; no dice in computed fields, when or max_expr; when parses; generators and views are well-formed.
  • Fixtures evaluate to their expected values.
  • It returns a structured report (errors, warnings, fixtures n of m) for the upload report (IS-430).
• builtin/generic/: the generic character and creature (§3.6, owner answer L2).
• conformance/: 309 expression cases and 13 sheet cases, language-neutral, plus a README spelling out every semantic. PR 4's JavaScript evaluator runs these same files, so the two can't drift.

Tests:

• tests/sheets/ passes: 442. That covers every conformance case, parser limits, totality over random inputs, overflow, dice with a scripted RNG, every validator check, did-you-mean, schema staleness, and the generic definitions validating clean.
• ruff check and format are clean over the full lint set, and changelog.py check passes.

Stacked on it (in progress): PR 3 (the server), PR 4 (the JS evaluator), and PRs 5a and 5b (the PF2e and 5e definitions).

Changelog: changelog.d/737-lane11-evaluator.added.md.

🤖 Generated with Claude Code

Lane 11 (#737) PR 2. This is the engine every sheet is computed with (IS-405 server, IS-431, IS-425). It adds no database, no routes and no UI; nothing outside `app/sheets/` imports it yet. **`webapp/backend/app/sheets/`:** - **`expr.py`:** the expression language from the contract's §3.4. It is total and side-effect free. - **Syntax:** paths `@a.b.c`, `[*]` lists, repeat keys, and `row.`/`entry.` locals. - **Functions:** `if`, the math functions, `lookup`, `rank` (plus a two-argument form), `has`, and `sum/count/any/all/min/max`. - **Numbers:** floored `%`, round half away from zero, and normalised output. - **Nulls:** a missing path is null, arithmetic treats null as 0, and a comparison with null is false. - **Errors:** typed, with their offsets. There are limits of 2,000 characters and 64 levels of nesting. - **Dice** are rejected here; they belong to generators only. - **`evaluator.py`:** evaluates a definition over `data`/`entries`/`resources`/`overrides`/`ext`. - **Derivations:** the terms a sheet prints under a value. - **Declarative modifiers:** add, multiply, upgrade, downgrade, cap and override, stacked `bonus` (best per type) or `flat`. - **Also:** `when`, repeat groups, entries with provenance and a snapshot, derived rows, resources and tracks. - **Order and caching:** evaluation follows the dependency order, with a cycle guard, and compiled definitions are cached by digest. - **`dice.py`:** `NdM`, keep/drop, an injectable RNG, and the individual dice kept. It's for creation and levelling only (D85). - **`definition_model.py` and `definition.schema.json`:** the Pydantic model and the published JSON Schema. - **`validator.py`:** the §3.2 checks. - Every `@path`/`ref` resolves, with **did-you-mean**. - No cycles; no dice in computed fields, `when` or `max_expr`; `when` parses; generators and views are well-formed. - Fixtures evaluate to their expected values. - It returns a structured report (errors, warnings, fixtures n of m) for the upload report (IS-430). - **`builtin/generic/`:** the generic character and creature (§3.6, owner answer L2). - **`conformance/`:** **309 expression cases and 13 sheet cases**, language-neutral, plus a README spelling out every semantic. PR 4's JavaScript evaluator runs these same files, so the two can't drift. **Tests:** - `tests/sheets/` passes: **442**. That covers every conformance case, parser limits, totality over random inputs, overflow, dice with a scripted RNG, every validator check, did-you-mean, schema staleness, and the generic definitions validating clean. - ruff check and format are clean over the full lint set, and `changelog.py check` passes. Stacked on it (in progress): PR 3 (the server), PR 4 (the JS evaluator), and PRs 5a and 5b (the PF2e and 5e definitions). Changelog: `changelog.d/737-lane11-evaluator.added.md`. 🤖 Generated with [Claude Code](https://claude.com/claude-code)

claude-bot added 3 commits 2026-09-30 07:25:49 +00:00

WIP lane 11 PR 2: expression language, evaluator, dice, definition model, validator, expr fixtures (#737) ... f006d8bc9e

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

WIP lane 11 PR 2: sheet conformance cases, README, JSON Schema (#737) ... 35a936da8a

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

Lane 11 PR 2: iterative evaluation of long chains, unit tests, ruff format, changelog (#737) ... b83621dfcb

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

claude-bot merged commit 88bce3f52e into main 2026-09-30 07:59:55 +00:00

claude-bot referenced this pull request from a commit 2026-09-30 07:59:57 +00:00

Merge PR #961: lane 11 PR 2 — the sheet language (#737)

claude-bot referenced this pull request 2026-09-30 08:00:59 +00:00

Lane 11 PR 5a: the Pathfinder 2e sheet and block definitions (#737) #962

claude-bot referenced this pull request 2026-09-30 08:58:06 +00:00

Lane 11 PR 2b: labels from expressions, first/last/join, and two robustness fixes, in both engines (#737) #967

Sign in to join this conversation.

Reviewers

No reviewers

Labels

Clear labels   

backend

Issue related to the backend

  

bot

Issue related to the bot

  

bug

Something is not working

  

duplicate

This issue or pull request already exists

  

enhancement

New feature

  

frontend

Issue related to the frontend

  

help wanted

Need some help

  

invalid

Something is wrong

  

question

More information is needed

  

recording

Issue related to the recording system

  

security

Security hardening / vulnerability work

  

wontfix

This won't be fixed

No labels

backend

bot

bug

duplicate

enhancement

frontend

help wanted

invalid

question

recording

security

wontfix

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

1 participant

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

rbrooks/Quest-Board!961

No description provided.