# Engine integration tests
Rust integration tests for `lemma-engine`. Each `*.rs` file here is a separate test binary (public API only). Unit tests belong in `engine/src/**` under `#[cfg(test)]` (or the shared `engine/src/tests/` harness for private helpers).
Run:
```bash
cargo nextest run -p lemma-engine --tests
```
## Layout elsewhere
| `engine/src/**` `mod tests` | Unit tests (private API) |
| `engine/src/tests/` | Unit tests wired from `lib.rs` (AST, bindings, engine, error, serializers) |
| `cli/tests/` | CLI / MCP / HTTP black-box tests |
| `engine/packages/npm/test.js` | WASM npm contract |
| `engine/packages/hex/test/` | Hex NIF ExUnit |
| `engine/fuzz/` | `cargo-fuzz` targets (workspace excluded) |
## Catalog by subsystem
### Evaluation, veto, missing data
| `veto.rs` | Veto propagation, unless interaction |
| `veto_inversion.rs` | Veto + inversion |
| `missing_data_propagation.rs` | Missing data through rules |
| `branch_aware_missing_data.rs` | `Response::missing_data` ordering |
| `explanation_e2e.rs` | Explanation / operation traces |
### Inversion
| `test_discount_inversion.rs` | Discount inversion scenario |
| `inversion_constraint_linear.rs` | Linear constraints |
| `inversion_display_serialize.rs` | Inversion result display/JSON |
### Temporal versioning & planning
| `temporal_slicing.rs` | Multi-slice execution plans |
| `temporal_range_references.rs` | Range refs across versions |
| `temporal_type_resolver_instant.rs` | Parent type resolution at slice instant |
| `temporal_timezone_ordering.rs` | TZ ordering across slices |
| `temporal_interface_deep_slice.rs` | Deep temporal interface |
| `temporal_boundary_explosion.rs` | Boundary cases |
| `temporal_cycle_panic.rs` | Invalid temporal cycles |
| `spec_reference_scenarios.rs` | Composability contracts (unpinned/pinned `uses`, coverage, self-ref, `fill`; see [spec_composability.md](../documentation/spec_composability.md)) |
| `temporal_self_uses.rs` | Cross-temporal same-name `uses` (implicit or explicit alias) |
| `type_import_temporal.rs` | Type-only deps + temporal versions |
| `spec_name_repository_plan_collision.rs` | Cross-repo spec name collision (regression) |
| `uses_lemma_compound_unit_planning.rs` | Compound units via `uses lemma si` |
| `validator_type_checking.rs` | Type validation at plan time |
| `semantic_validation.rs` | Semantic checks |
| `duration_trait_planning.rs` | Duration trait planning |
### Datetime, duration, calendar
| `datetime_sugar.rs` | `now`, `in past`, calendar sugar |
| `datetime_edge_cases.rs` | Date/time edge cases |
| `datetime_edge_hunting.rs` | Adversarial datetime scenarios |
| `timezone.rs` | Timezone parsing/eval |
| `date_range.rs` | Date ranges |
| `calendar_duration_split.rs` | Calendar vs duration split |
| `duration_trait_anonymous.rs` | Anonymous duration types |
| `duration_trait_arithmetic.rs` | Duration arithmetic |
| `duration_trait_temporal.rs` | Duration + temporal |
| `duration_trait_precision.rs` | Duration precision |
### Arithmetic, units, quantities, ratios
| `arithmetic_exactness.rs` | Division planning/runtime behavior; quantity wage integration |
| `arithmetic_type_combinations.rs` | Typed arithmetic matrix |
| `math_ops.rs` | Math operators |
| `modulo_power.rs` | `%` and power |
| `equal_operator.rs` | Equality |
| `type_aware_arithmetic.rs` | Type-aware ops |
| `quantity_unit_conversion.rs` | Unit conversion + validation bounds |
| `quantity_number_refactoring.rs` | Quantity/number interactions |
| `quantity_duration_arithmetic_types.rs` | Quantity + duration types |
| `ratio_quantity_units.rs` | Ratio vs quantity units |
| `ratio_runtime_input.rs` | Runtime ratio input grammar |
| `multidim_unit_system.rs` | Multi-dimensional units |
| `unit_percentage_operations.rs` | Percentage on units |
| `decimal_storage_pipeline.rs` | Decimal storage in results/JSON |
### Data, bindings, literals, schema
| `data_literals_coverage.rs` | Inline literal RHS on data |
| `data_binding_type_validation.rs` | Binding type checks |
| `data_type_declarations_coverage.rs` | Type declarations on data |
| `data_nested_bindings_coverage.rs` | Nested bindings |
| `data_references.rs` | Cross-data references |
| `data_with_values_contract.rs` | Data-with-values contract |
| `typed_values.rs` | Typed value handling |
| `schema_defaults_distinction.rs` | Schema vs defaults |
| `meta_fields.rs` | Meta fields on specs |
### Spec graph, imports, references
| `nested_spec_references.rs` | Nested `uses` / refs |
| `cross_spec_references.rs` | Cross-spec refs |
| `inline_type_imports.rs` | Inline type imports |
| `type_definitions.rs` | User type definitions |
| `required_data_names_nested_spec.rs` | Required data in nested specs |
| `repo_keyword.rs` | `repo` keyword |
### Parsing, syntax, errors
| `expression_syntax.rs` | Expression grammar |
| `error_messages.rs` | Error text/locations |
### Formatting
| `formatter.rs` | Formatter behavior |
| `format_weather_clothing_integration.rs` | Formatter integration example |
### End-to-end / examples
| `integration_comprehensive.rs` | Broad integration scenarios |
| `integration_examples.rs` | CLI example `.lemma` files |
| `documentation_examples.rs` | `documentation/examples/` |
| `coffee_order.rs` | Coffee order example |
### Registry, WASM, limits
| `load_batch_wasm_planning_parity.rs` | WASM vs native planning parity |
| `repro_finance_dual_slice_registry_uses.rs` | Registry + dual slice |
| `resource_limits_test.rs` | Resource limits |
### Fuzz smoke
| `duration_conversion.rs` | `uses lemma si` duration unit conversion |
| `fuzz_api_surface.rs` | Fuzz-related API smoke |
### Repro / regression guards
| `repro_named_types_panic.rs` | Named types panic |
| `repro_finance_dual_slice_registry_uses.rs` | Finance/registry slice |
---
## `engine/src` modules without inline unit tests
Coverage for these lives mainly in this directory (integration) or in `engine/src/tests/`:
| `computation/arithmetic.rs` | |
| `computation/comparison.rs` | |
| `computation/range.rs` | |
| `computation/decimal_math.rs` | |
| `computation/mod.rs` | |
| `evaluation/expression.rs` | |
| `evaluation/explanation.rs` | |
| `inversion/target.rs`, `inversion/derived.rs` | |
| `literals.rs` | Partial coverage via `src/tests/ast.rs` |
| `parsing/parser.rs` | Lexer/AST have unit tests |
| `serialization/mod.rs` | `json.rs` has unit tests |
| `limits.rs`, `deps.rs`, `stdlib.rs`, `wasm.rs` | |
Prefer adding **unit** tests beside the module when testing private helpers; add **integration** tests here when exercising full load → plan → run paths.
---
## Semantics audit
### Unit vs integration split
| Unit | `engine/src/**` `#[cfg(test)]`, `engine/src/tests/` | Private + public | Parser nodes, planner invariants, `LiteralValue`, serde |
| Integration (this dir) | `engine/tests/*.rs` | Public `lemma::*` only | `Engine::load` → `run` / `get_plan` / `invert` |
| CLI | `cli/tests/integrations/*` | `lemma` binary, MCP JSON | stdout, HTTP, tool payloads |
| WASM npm | `engine/packages/npm/test.js` | JS `Engine` | Shape of load/run/list/schema |
| Hex | `engine/packages/hex/test/` | `Lemma.*` NIF | Lifecycle, list groups, run JSON |
**~519** lib unit tests vs **~1470** integration `#[test]` functions across 76 binaries (run `cargo nextest run -p lemma-engine --lib` / `--tests` for current counts).
### Primary APIs used here
| `Engine::load_batch` | 4 | Dependency bundles (WASM-style) |
| `Engine::invert` | 3 | Inversion only in dedicated files |
| `parse` / `format_source` | 2 | `formatter.rs`, `format_weather_clothing_integration.rs` (no `Engine`) |
`inversion_display_serialize.rs` tests `Domain` JSON only (no engine).
### Unit-test coverage by production module
| `parsing/` (lexer, ast, mod) | Heavy (~110+) | `expression_syntax`, `error_messages`, `data_literals_coverage` |
| `planning/` (graph, normalize, semantics, execution_plan) | Heavy (~150+) | `temporal_*`, `validator_*`, `uses_lemma_*`, `type_definitions` |
| `inversion/` | Moderate (~45) | `inversion_*`, `test_discount_inversion`, `veto_inversion` |
| `computation/datetime`, `rational`, `units` | datetime/rational heavy | `datetime_*`, `arithmetic_*`, `quantity_*`, `ratio_*` |
| `computation/arithmetic`, `comparison`, `range` | **None** | `arithmetic_type_combinations`, `math_ops`, `range_*`, `equal_operator` |
| `evaluation/expression`, `explanation` | **None** | E2E via `run` in most files |
| `engine.rs` | ~32 | `integration_*`, `coffee_order`, registry repros |
| `formatting/` | ~25 | `formatter`, `format_weather_clothing_integration` |
| `registry` | ~8 | `repo_keyword`, `load_batch_wasm_planning_parity`, repros |
| `literals.rs` | via `src/tests/ast.rs` | `data_literals_coverage`, `ratio_runtime_input` |
### Overlap clusters (consolidation candidates)
When changing behavior, run the whole cluster — scenarios often duplicate.
| Datetime eval | `datetime_sugar`, `datetime_edge_cases`, `datetime_edge_hunting`, `timezone`, `date_range` |
| Duration trait | `duration_trait_anonymous`, `_arithmetic`, `_precision`, `_temporal`, `_planning` |
| Quantity / ratio | `quantity_unit_conversion`, `quantity_number_refactoring`, `quantity_duration_arithmetic_types`, `ratio_quantity_units`, `ratio_runtime_input`, `unit_percentage_operations`, `type_aware_arithmetic`, `multidim_unit_system` |
| Decimal eval precision | `quantity_unit_conversion` (`precision_*` stress: prime chains from 37, API unit toggles, mixed `*`/`/`); `arithmetic_exactness` (`runtime_data_ten_divide_three_*`) |
| Arithmetic | `arithmetic_type_combinations`, `arithmetic_exactness`, `math_ops`, `modulo_power`, `equal_operator` |
| Range | `range_generic`, `range_semantics_table`, `date_range` |
| Spec graph | `nested_spec_references`, `cross_spec_references`, `required_data_names_nested_spec`, `inline_type_imports` |
| Temporal | `temporal_slicing`, `type_import_temporal`, `temporal_range_references`, `temporal_type_resolver_instant`, `temporal_timezone_ordering`, `temporal_interface_deep_slice`, `temporal_boundary_explosion` |
| Registry / plan identity | `spec_name_repository_plan_collision`, `repro_finance_dual_slice_registry_uses`, `load_batch_wasm_planning_parity` |
| Example E2E | `coffee_order`, `documentation_examples`, `integration_examples`, `integration_comprehensive` |
| Data QA matrix | `data_literals_coverage`, `data_type_declarations_coverage`, `data_binding_type_validation`, `data_with_values_contract`, `data_nested_bindings_coverage`, `data_references` |
### Regression guards and intentional reds
| `spec_name_repository_plan_collision.rs` | **Passes** when `get_plan` returns distinct plans per repository for same basename; **fails** if `plan_sets` aliases by name only (see module doc) |
| `repro_finance_dual_slice_registry_uses.rs` | Same for WASM `load_batch` + duplicate `finance` basename |
| `data_literals_coverage.rs` | Pins literal RHS invariants; header says some cases may stay red — currently all green |
| `data_type_declarations_coverage.rs`, `data_binding_type_validation.rs`, `data_references.rs` | Constraint matrices; do not weaken assertions |
| `temporal_slicing.rs` | TDD guard for multi-slice planning |
| `repo_keyword.rs` | Signals until `repo` semantics complete |
| `resource_limits_test.rs` | `#[ignore]` on `performance_test_10k_rules`, `bench_deep_chains` (manual benches) |
No integration file uses `#[should_panic]`. `proptest` is not used in `engine/tests/` (dev-dependency exists on the crate for future use).
### Gaps and recommendations
| No `computation/*` unit tests except datetime/rational/units | Add unit tests beside `arithmetic.rs` / `comparison.rs` for edge cases; keep integration matrix |
| Thin `Engine::invert` integration surface | Extend when changing inversion UX; unit tests in `inversion/` are primary |
| `evaluation/expression.rs` untested in isolation | Unit-test eval of individual ops; integration already heavy via `run` |
| Duplicate example runners | `documentation_examples` vs `integration_examples` — different roots; keep both, share helpers if duplicated |
| Stale “must fail” comments | `spec_name_*` / `repro_finance_*` **pass** when fixed; comments describe failure mode if bug returns |
### CLI integration map
See [cli/tests/README.md](../../cli/tests/README.md): `run`, `mcp`, `server`, `examples` — 57 tests, black-box on the `lemma` binary.
### Shared integration helpers
[support/mod.rs](support/mod.rs) — `get_rule_value`, `eval_rule_bool`, `make_effective*` (import with `mod support;`). Used by `datetime_edge_cases`, `datetime_sugar`; extend to other datetime cluster files when editing them.
### Implementation report
[REPORT.md](REPORT.md) — what changed in the audit follow-up, quality assessment, and verification commands.