dndwright 0.29.0

[0.29.0] — 2026-08-15

Changed — BREAKING

• evaluate_character now defaults to strict=True. Malformed input raises
  CharacterInputError instead of being coerced into a plausible-but-wrong sheet.
  Pass strict=False to restore the previous behaviour — the lenient path is unchanged
  and still available, it is simply no longer the default.

  Why this is worth a breaking change. A coerced sheet has every expected key and
  plausible numbers: missing ability scores default to 10, an omitted level to 1, a missing
  class zeroes HP and spellcasting. It is therefore indistinguishable from a correct sheet
  at every downstream surface. There was no error to swallow and nothing looked broken, so
  no consumer could report it — the signal was never generated rather than hidden. A new
  test, test_the_coerced_sheet_is_indistinguishable_from_a_good_one, pins exactly that.

  What to check before upgrading. Anything calling evaluate_character without the flag
  on data it did not itself validate. Callers behind a schema gate (e.g. a Pydantic request
  model that already requires ability scores, class and level) are unaffected — the data is
  complete by construction. Callers that build input internally, especially from
  LLM output, are the ones that may newly raise.

  ⚠️ If you depend on this package with a >= floor rather than an exact pin, a rebuild
  will pick this up automatically. That is the intended behaviour of a floor, but it means
  the change can arrive without anyone editing a version file.

dndwright 0.28.0

[0.28.0] — 2026-08-15

Added

• ClassMechanics and weapon_attack are now part of the public API (dndwright.__all__,
  72 → 74 names). Both were already reachable — ClassMechanics only via the private
  dndwright.rules.components, and weapon_attack via the published dndwright.combat
  sub-API — and both are now importable directly from dndwright. Additive: no existing
  import path was removed, so nothing breaks.
  • The API contract test pins all three of its surfaces accordingly: the exported name set,
    weapon_attack's signature (every public free function must be pinned), and
    ClassMechanics's field list (every public pydantic model must be pinned).
  • A minor version bump is owed at the next release under SemVer, since this widens the
    public surface. Deliberately not bumped here: bumping without releasing is what left
    the sibling package mapwright with a 0.28.0 that looks shipped and never was.

Changed

• Examples and the module docstring now teach published import paths only.
  examples/multiclass.py and the __init__ docstring imported ClassMechanics from
  dndwright.rules.components; examples/stat_diff.py imported compute_stat_diff from
  dndwright.rules.character_evaluator even though that name was already public. Since
  tests/test_examples.py executes every example, the package was effectively guaranteeing
  that private import paths keep working.

Added (2)

• dndwright.combat now re-exports both submodules' full declared surfaces — the 20 names
  in conditions.__all__ and the 7 in initiative.__all__. combat.__all__ goes 22 → 49, so
  from dndwright.combat import attempt_save, order_initiative works instead of needing a
  second import from the submodule. Additive: dndwright.combat.conditions and
  .initiative remain importable exactly as before.
  • Both submodule surfaces were ALREADY contract-pinned (EXPECTED_CONDITIONS_PUBLIC,
    EXPECTED_INITIATIVE_PUBLIC); this publishes them one level up and pins the wider
    EXPECTED_COMBAT_PUBLIC to match.
  • The full submodule surfaces were taken rather than only the nine names the examples
    happened to use, because four of those nine return types (SaveResult,
    ConditionChange, TurnAdvance, InitiativeRoll) would otherwise have stayed
    unpublished — leaving callers able to invoke a function but unable to name its result.
• examples/conditions.py and examples/initiative.py now import from dndwright.combat.
  No example imports a private path any more.

Known

• examples/conditions.py and examples/initiative.py still import names that are on no
  published surface — attempt_save, tick_conditions, ActiveCondition, ROUND_END,
  SAVE_ENDS, order_initiative, advance_turn, roll_initiative, InitiativeEntry.
  dndwright.combat.__all__ exports the conditions and initiative submodules but none
  of the functions within them. Widening the surface further is a separate decision and has
  not been taken.

Changed

• Adopted the wider lint families (I, SIM, UP, RUF) on top of the existing
  explicit rule set, in a deliberate pass rather than as a dependency side effect:
  import sorting, f-string conversions, __all__ sorting, and a deprecated import.
  PL is not adopted (262 findings, magic values that are the ruleset itself);
  RUF001-003, RUF005 and RUF012 are declined with reasons stated in
  pyproject.toml.

Added

• RUF046 is enabled here, and the reason is now checked. RUF046 rewrites
  int(round(x)) into round(x), which is only safe while no value is a numpy
  scalar — round(np.float64) returns a numpy float, not an int. dndwright has no
  numpy by rule, so the rewrite is safe; the sibling package mapwright is numpy-based
  and declines the same rule. Those two decisions are coupled and nothing said so, so
  tests/test_packaging_contract.py now fails if numpy is ever added while RUF046
  is still enforced, naming the reason.

Notes

• No behavioural change and no version bump. The public API is the same set of 72
  names (__all__ declaration order moved when it was sorted; the set did not), and
  seeded dice rolls and a computed character sheet hash identically against a detached
  worktree of the previous commit.

dndwright 0.27.0

[0.27.0] — 2026-07-29

Added

• clean_damage_types() and combatant_defenses() now accept an optional keyword-only
  allowed: frozenset[str] = DAMAGE_TYPES. By default they intersect with the SRD-13
  DAMAGE_TYPES exactly as before (fully backward-compatible); a caller that governs its own
  damage-type vocabulary can pass a wider set (e.g. SRD-13 ∪ custom) so caller-registered
  types survive the filter — at component assembly, the computed-stats read, and live combat
  damage application — instead of being silently dropped. dndwright stays domain-agnostic: it does
  not know where allowed comes from; members must be lower-cased (the caller owns canonicalisation).
  Enables host applications to support per-campaign/per-universe custom damage types without a
  split-brain between the character sheet and combat. Public API names (__all__) unchanged.

v0.26.0 — evaluate_character(components=...) compose

What's Changed

• feat(rules): evaluate_character(components=...) — compose item modifiers onto the sheet by @sligara7 in #9

Full Changelog: v0.25.2...v0.26.0

v0.25.2

Full Changelog: v0.25.1...v0.25.2

dndwright 0.23.2

Documentation

• Two new README/PyPI diagrams. assets/compose.svg shows lego-style components snapping
  onto the ruleset — typed set/add/union contributions that keep the target node's id, so
  one snap-in recomputes the whole downstream subtree. assets/theme-scaling.svg shows one
  computation graph re-skinned per setting via ThemeScalingLayer: the same plate node emerges
  as AC 18 (traditional / modern, re-flavored) or AC 19 (sci-fi / steampunk) through
  input_overrides, lookup_overrides, and display-only flavor_renames.
• New README sections — "Composable — snap mini-graphs onto the ruleset" and "Re-skin for
  any setting — theme scaling" (theme scaling was previously undocumented in the README), plus a
  apply_theme_scaling / ThemeScalingLayer row in the feature table. Docs-only; no code change.

dndwright 0.23.1

Fixed

• Critical damage preserves dice modifiers — DiceEngine.roll_damage(expr, is_critical=True)
  doubled die counts by rebuilding the expression as bare XdY, silently dropping per-group
  keep/drop/reroll/exploding flags. A crit on 1d6!+2 now stays exploding (2d6!+2)
  instead of becoming a plain 2d6+2. As part of the fix, str(DiceGroup) now serialises
  reroll-once as ro (was r), so the spec round-trips through parse_expression.
• Species damage immunities mis-categorised — _build_species_traits tagged damage
  immunities with category: "resistance" (condition immunities were already correct); they
  are now category: "immunity".

Changed

• compose() fails loudly on contradictory contributions — a target that receives both a
  union (set-valued) and an add/set (numeric) contribution now raises ValueError instead
  of silently dropping the numeric ones.

Performance

• Topological sort uses a binary heap (heapq) instead of re-sorting the ready-queue every
  iteration — same deterministic order, O(V·log V + E) instead of O(V²·log V). (Result is
  still cached per ruleset, so this only matters the first time a large graph is evaluated.)

Internal

• cli._read_json closes the file it opens (was leaked); cli content lookup calls
  categories() once instead of twice.
• assembler.assemble_inputs copies the ability-score dict once up front rather than on each
  increase.

dndwright 0.23.0

[0.23.0] — 2026-06-05

Added

• evaluate_character(..., scaling=...) — evaluate_character, compute_key_stats, and
  compute_stat_diff now take an optional scaling: ThemeScalingLayer keyword. When given,
  the sheet (or key-stat snapshot / before-after diff) is computed against the theme-scaled
  ruleset via apply_theme_scaling, so a campaign's setting (sci-fi, modern warfare, …) shapes
  the character's mechanical values. scaling=None (default) reproduces the stock 5e behaviour
  exactly — existing callers are unaffected. New tests/test_character_theme_scaling.py.

dndwright 0.22.0

[0.22.0] — 2026-06-05

Added

• apply_theme_scaling(ruleset, layer) (dndwright) — folds a ThemeScalingLayer's
  mechanical overrides onto a ruleset so the same computation graph yields
  setting-appropriate values (modern warfare, sci-fi, steampunk, …). lookup_overrides
  deep-merge into the ruleset's lookup tables (creating tables/keys, preserving the rest);
  input_overrides re-baseline a node's default_value (eval precedence stays explicit
  input > themed default > original default, so a theme sets the world's baseline without
  clobbering a value a character explicitly carries). flavor_renames are display-only and
  ignored here. Pure (the base ruleset is untouched) and composes with compose(). Raises
  KeyError on an input_overrides id that names no node, to catch authored-layer typos.
  This makes the previously inert ThemeScalingLayer / PREDEFINED_THEME_SCALING data a
  working feature. New tests/test_theme_scaling.py.

v0.21.0 — content structure contract

Formalizes dndwright's D&D 5e content structure as exported, strict Pydantic models — the single source of truth that downstream services conform to.

• New dndwright.content.models + exported Creature, CharClass, Species, Spell, MagicItem, Background, Feat, Weapon, Armor, Condition, Modifier, CONTENT_MODELS.
• All 1012 bundled assets validate against their models (extra="forbid").
• Non-breaking: asset shapes unchanged, only formalized.

See #2.