Add D&D scene description validation

This commit is contained in:
2026-07-23 20:17:18 +00:00
parent cd29265d5d
commit ad89782c9b
10 changed files with 1024 additions and 0 deletions

View File

@@ -0,0 +1,119 @@
// Package invariants validates normalized D&D scene description artifacts.
package invariants
import (
"context"
"fmt"
"sort"
"gitea.maximumdirect.net/eric/notarius/internal/core/source"
"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
"gitea.maximumdirect.net/eric/notarius/internal/modules/dnd"
"gitea.maximumdirect.net/eric/notarius/internal/modules/dnd/shared/diagnostics"
"gitea.maximumdirect.net/eric/notarius/internal/modules/dnd/validate/scenedescriptions/shape"
)
const (
Key = "normalize/dnd/scene-descriptions/invariants"
ReasonCode = "invalid_scene_description_normalization"
policy = "dnd.scene_descriptions.validator.invariants.v1"
)
type Options struct{}
type Validator struct{}
var _ contracts.TypedValidator[dnd.SceneDescriptionList] = (*Validator)(nil)
var _ pipeline.CheckpointFingerprintProvider = (*Validator)(nil)
func New(Options) *Validator { return &Validator{} }
func (v *Validator) Name() string { return Key }
func (v *Validator) ExecutionClass() contracts.ExecutionClass {
return contracts.ExecutionClassDeterministic
}
func (v *Validator) CheckpointFingerprints() []pipeline.CheckpointFingerprint {
return []pipeline.CheckpointFingerprint{{Name: "policy", Value: policy}}
}
func (v *Validator) Validate(_ context.Context, req contracts.TypedValidationRequest[dnd.SceneDescriptionList]) (contracts.ValidationResult, error) {
if shape.Validate(req.Value) != nil || !sourceRefsValid(req.Source, req.Value) {
return contracts.ValidationResult{Approved: true}, nil
}
issues := issuesFor(req.Source, req.Value)
if len(issues) == 0 {
return contracts.ValidationResult{Approved: true}, nil
}
return contracts.ValidationResult{
Approved: false, ReasonCode: ReasonCode,
Message: diagnostics.Aggregate("invalid scene description normalization", issues),
}, nil
}
func sourceRefsValid(doc *source.SourceDocument, value dnd.SceneDescriptionList) bool {
for _, scene := range value.Scenes {
if source.ValidateRef(doc, scene.SourceRef) != nil {
return false
}
}
return true
}
func issuesFor(doc *source.SourceDocument, value dnd.SceneDescriptionList) []string {
issues := make([]string, 0)
if !sort.SliceIsSorted(value.Scenes, func(left, right int) bool {
leftStart, _ := source.UnitIndex(doc, value.Scenes[left].SourceRef.StartUnitID)
rightStart, _ := source.UnitIndex(doc, value.Scenes[right].SourceRef.StartUnitID)
if leftStart != rightStart {
return leftStart < rightStart
}
return value.Scenes[left].ID < value.Scenes[right].ID
}) {
issues = append(issues, "scenes are not in canonical source order")
}
byID := make(map[string]int, len(value.Scenes))
byRange := make(map[source.SourceRef]int, len(value.Scenes))
for index, scene := range value.Scenes {
if index > 0 && value.Scenes[index-1] == scene {
issues = append(issues, fmt.Sprintf("scenes[%d] duplicates scenes[%d]", index, index-1))
}
if previous, ok := byID[scene.ID]; ok && value.Scenes[previous] != scene {
issues = append(issues, fmt.Sprintf("scenes[%d] conflicts with scenes[%d] for scene ID", index, previous))
} else if !ok {
byID[scene.ID] = index
}
if previous, ok := byRange[scene.SourceRef]; ok && !sameModelContent(value.Scenes[previous], scene) {
issues = append(issues, fmt.Sprintf("scenes[%d] conflicts with scenes[%d] for source range", index, previous))
} else if !ok {
byRange[scene.SourceRef] = index
}
}
return issues
}
func sameModelContent(left, right dnd.SceneDescription) bool {
return left.Kind == right.Kind && left.Title == right.Title && left.Summary == right.Summary
}
func Spec() pipeline.ValidatorSpec {
return pipeline.ValidatorSpec{Key: Key, ExecutionClass: contracts.ExecutionClassDeterministic}
}
func Register(registry *pipeline.ValidatorRegistry) error {
return pipeline.RegisterTypedValidatorBuilder(registry, dnd.SceneDescriptionListKind, Spec(), validateOptions, func(request pipeline.BuildRequest) (contracts.TypedValidator[dnd.SceneDescriptionList], error) {
options, err := DecodeOptions(request.Options)
if err != nil {
return nil, err
}
return New(options), nil
})
}
func DecodeOptions(options map[string]any) (Options, error) {
if err := pipeline.RejectUnknownOptions(options); err != nil {
return Options{}, err
}
return Options{}, nil
}
func validateOptions(options map[string]any) error { _, err := DecodeOptions(options); return err }

View File

@@ -0,0 +1,69 @@
package invariants
import (
"context"
"reflect"
"testing"
"gitea.maximumdirect.net/eric/notarius/internal/core/source"
"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
"gitea.maximumdirect.net/eric/notarius/internal/modules/dnd"
)
func TestValidatorEnforcesOrderDuplicatesAndConflicts(t *testing.T) {
doc := document()
first := scene("first", 1, dnd.SceneKindNarrative, "Arrival", "The party arrives.")
second := scene("second", 2, dnd.SceneKindCombat, "Ambush", "Bandits attack.")
validator := New(Options{})
for _, test := range []struct {
name string
value dnd.SceneDescriptionList
approved bool
}{
{name: "normalized", value: dnd.SceneDescriptionList{Scenes: []dnd.SceneDescription{first, second}}, approved: true},
{name: "out of order", value: dnd.SceneDescriptionList{Scenes: []dnd.SceneDescription{second, first}}},
{name: "exact duplicate", value: dnd.SceneDescriptionList{Scenes: []dnd.SceneDescription{first, first}}},
{name: "same ID conflict", value: dnd.SceneDescriptionList{Scenes: []dnd.SceneDescription{first, scene("first", 2, dnd.SceneKindCombat, "Ambush", "Bandits attack.")}}},
{name: "same range conflict", value: dnd.SceneDescriptionList{Scenes: []dnd.SceneDescription{first, scene("other", 1, dnd.SceneKindMeta, "Rules", "The table checks rules.")}}},
{name: "same range same content different ID", value: dnd.SceneDescriptionList{Scenes: []dnd.SceneDescription{scene("a", 1, dnd.SceneKindNarrative, "Arrival", "The party arrives."), scene("b", 1, dnd.SceneKindNarrative, "Arrival", "The party arrives.")}}, approved: true},
} {
t.Run(test.name, func(t *testing.T) {
result, err := validator.Validate(context.Background(), contracts.TypedValidationRequest[dnd.SceneDescriptionList]{Source: doc, Value: test.value})
if err != nil || result.Approved != test.approved {
t.Fatalf("Validate() = %#v, %v; want approved=%t", result, err, test.approved)
}
})
}
}
func TestValidatorDefersOwnedFailuresAndRegistersPolicy(t *testing.T) {
doc := document()
result, err := New(Options{}).Validate(context.Background(), contracts.TypedValidationRequest[dnd.SceneDescriptionList]{Source: doc, Value: dnd.SceneDescriptionList{}})
if err != nil || !result.Approved {
t.Fatalf("Validate() = %#v, %v; want shape deferral", result, err)
}
invalidRef := dnd.SceneDescriptionList{Scenes: []dnd.SceneDescription{scene("one", 9, dnd.SceneKindNarrative, "Arrival", "The party arrives.")}}
result, err = New(Options{}).Validate(context.Background(), contracts.TypedValidationRequest[dnd.SceneDescriptionList]{Source: doc, Value: invalidRef})
if err != nil || !result.Approved {
t.Fatalf("Validate() = %#v, %v; want source-reference deferral", result, err)
}
registry := pipeline.NewValidatorRegistry()
if err := Register(registry); err != nil {
t.Fatal(err)
}
if _, ok := registry.Spec(Key); !ok {
t.Fatalf("registry missing %q", Key)
}
if got, want := New(Options{}).CheckpointFingerprints(), []pipeline.CheckpointFingerprint{{Name: "policy", Value: policy}}; !reflect.DeepEqual(got, want) {
t.Fatalf("fingerprints = %#v, want %#v", got, want)
}
}
func document() *source.SourceDocument {
return &source.SourceDocument{ID: "session", Units: []source.SourceUnit{{ID: 1, Text: "The party arrives."}, {ID: 2, Text: "Bandits attack."}}}
}
func scene(id string, unit int, kind dnd.SceneKind, title, summary string) dnd.SceneDescription {
return dnd.SceneDescription{ID: id, SourceRef: source.SourceRef{SourceID: "session", StartUnitID: unit, EndUnitID: unit}, Kind: kind, Title: title, Summary: summary}
}

View File

@@ -0,0 +1,112 @@
// Package shape validates required D&D scene description fields.
package shape
import (
"context"
"fmt"
"strings"
"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
"gitea.maximumdirect.net/eric/notarius/internal/modules/dnd"
"gitea.maximumdirect.net/eric/notarius/internal/modules/dnd/shared/diagnostics"
)
const (
Key = "extract/dnd/scene-descriptions/shape"
ReasonCode = "invalid_scene_description_shape"
policy = "dnd.scene_descriptions.validator.shape.v1"
)
type Options struct{}
type Validator struct{}
var _ contracts.TypedValidator[dnd.SceneDescriptionList] = (*Validator)(nil)
var _ pipeline.CheckpointFingerprintProvider = (*Validator)(nil)
func New(Options) *Validator { return &Validator{} }
func (v *Validator) Name() string { return Key }
func (v *Validator) ExecutionClass() contracts.ExecutionClass {
return contracts.ExecutionClassDeterministic
}
func (v *Validator) CheckpointFingerprints() []pipeline.CheckpointFingerprint {
return []pipeline.CheckpointFingerprint{{Name: "policy", Value: policy}}
}
func (v *Validator) Validate(_ context.Context, req contracts.TypedValidationRequest[dnd.SceneDescriptionList]) (contracts.ValidationResult, error) {
if err := ValidateForStage(req.Value, req.Stage); err != nil {
return contracts.ValidationResult{Approved: false, ReasonCode: ReasonCode, Message: err.Error()}, nil
}
return contracts.ValidationResult{Approved: true}, nil
}
func Validate(value dnd.SceneDescriptionList) error { return validate(value, false) }
func ValidateForStage(value dnd.SceneDescriptionList, stage string) error {
return validate(value, stage == string(pipeline.StageExtract))
}
func validate(value dnd.SceneDescriptionList, exactlyOne bool) error {
issues := make([]string, 0)
if value.Scenes == nil {
issues = append(issues, "scenes must be present")
} else if len(value.Scenes) == 0 {
issues = append(issues, "scenes must not be empty")
} else if exactlyOne && len(value.Scenes) != 1 {
issues = append(issues, "extraction must contain exactly one scene")
}
for index, scene := range value.Scenes {
prefix := fmt.Sprintf("scenes[%d]", index)
if strings.TrimSpace(scene.ID) == "" || scene.ID != strings.TrimSpace(scene.ID) {
issues = append(issues, prefix+".id must be non-empty and trimmed")
}
if !ValidKind(scene.Kind) {
issues = append(issues, prefix+".kind is unsupported: "+diagnostics.Quote(string(scene.Kind)))
}
if strings.TrimSpace(scene.Title) == "" || scene.Title != strings.TrimSpace(scene.Title) {
issues = append(issues, prefix+".title must be non-empty and trimmed")
}
if strings.TrimSpace(scene.Summary) == "" || scene.Summary != strings.TrimSpace(scene.Summary) {
issues = append(issues, prefix+".summary must be non-empty and trimmed")
}
if strings.TrimSpace(scene.SourceRef.SourceID) == "" || scene.SourceRef.SourceID != strings.TrimSpace(scene.SourceRef.SourceID) || scene.SourceRef.StartUnitID <= 0 || scene.SourceRef.EndUnitID <= 0 {
issues = append(issues, prefix+".source_ref must have a trimmed source ID and positive unit IDs")
}
}
if len(issues) == 0 {
return nil
}
return fmt.Errorf("%s", diagnostics.Aggregate("invalid scene description shape", issues))
}
func ValidKind(value dnd.SceneKind) bool {
switch value {
case dnd.SceneKindCombat, dnd.SceneKindNarrative, dnd.SceneKindRecap, dnd.SceneKindMeta:
return true
default:
return false
}
}
func Spec() pipeline.ValidatorSpec {
return pipeline.ValidatorSpec{Key: Key, ExecutionClass: contracts.ExecutionClassDeterministic}
}
func Register(registry *pipeline.ValidatorRegistry) error {
return pipeline.RegisterTypedValidatorBuilder(registry, dnd.SceneDescriptionListKind, Spec(), validateOptions, func(request pipeline.BuildRequest) (contracts.TypedValidator[dnd.SceneDescriptionList], error) {
options, err := DecodeOptions(request.Options)
if err != nil {
return nil, err
}
return New(options), nil
})
}
func DecodeOptions(options map[string]any) (Options, error) {
if err := pipeline.RejectUnknownOptions(options); err != nil {
return Options{}, err
}
return Options{}, nil
}
func validateOptions(options map[string]any) error { _, err := DecodeOptions(options); return err }

View File

@@ -0,0 +1,65 @@
package shape
import (
"context"
"reflect"
"testing"
"gitea.maximumdirect.net/eric/notarius/internal/core/source"
"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
"gitea.maximumdirect.net/eric/notarius/internal/modules/dnd"
)
func TestValidatorRequiresSceneShapeAndExactlyOneExtractionRecord(t *testing.T) {
valid := list(scene("one", dnd.SceneKindNarrative, "Arrival", "The party arrives.", 1, 1))
validator := New(Options{})
for _, test := range []struct {
name string
value dnd.SceneDescriptionList
stage string
approved bool
}{
{name: "valid normalization", value: valid, approved: true},
{name: "valid extraction", value: valid, stage: string(pipeline.StageExtract), approved: true},
{name: "multiple extraction records", value: list(valid.Scenes[0], scene("two", dnd.SceneKindMeta, "Rules", "The table checks rules.", 2, 2)), stage: string(pipeline.StageExtract)},
{name: "nil list", value: dnd.SceneDescriptionList{}},
{name: "empty list", value: dnd.SceneDescriptionList{Scenes: []dnd.SceneDescription{}}},
{name: "blank ID", value: list(scene(" ", dnd.SceneKindNarrative, "Arrival", "The party arrives.", 1, 1))},
{name: "unsupported kind", value: list(scene("one", "other", "Arrival", "The party arrives.", 1, 1))},
{name: "untrimmed prose", value: list(scene("one", dnd.SceneKindNarrative, " Arrival ", "The party arrives.", 1, 1))},
{name: "empty source reference", value: dnd.SceneDescriptionList{Scenes: []dnd.SceneDescription{{ID: "one", Kind: dnd.SceneKindNarrative, Title: "Arrival", Summary: "The party arrives."}}}},
} {
t.Run(test.name, func(t *testing.T) {
result, err := validator.Validate(context.Background(), contracts.TypedValidationRequest[dnd.SceneDescriptionList]{Stage: test.stage, Value: test.value})
if err != nil || result.Approved != test.approved {
t.Fatalf("Validate() = %#v, %v; want approved=%t", result, err, test.approved)
}
})
}
}
func TestValidatorContractAndRegistration(t *testing.T) {
registry := pipeline.NewValidatorRegistry()
if err := Register(registry); err != nil {
t.Fatalf("Register() error = %v", err)
}
if _, ok := registry.Spec(Key); !ok {
t.Fatalf("registry has no spec for %q", Key)
}
if _, err := DecodeOptions(map[string]any{"unexpected": true}); err == nil {
t.Fatal("DecodeOptions() accepted unknown options")
}
validator := New(Options{})
if got, want := validator.CheckpointFingerprints(), []pipeline.CheckpointFingerprint{{Name: "policy", Value: policy}}; !reflect.DeepEqual(got, want) {
t.Fatalf("fingerprints = %#v, want %#v", got, want)
}
}
func list(scenes ...dnd.SceneDescription) dnd.SceneDescriptionList {
return dnd.SceneDescriptionList{Scenes: scenes}
}
func scene(id string, kind dnd.SceneKind, title, summary string, start, end int) dnd.SceneDescription {
return dnd.SceneDescription{ID: id, Kind: kind, Title: title, Summary: summary, SourceRef: source.SourceRef{SourceID: "session", StartUnitID: start, EndUnitID: end}}
}

View File

@@ -0,0 +1,90 @@
// Package sourcerefs validates D&D scene-description source identity.
package sourcerefs
import (
"context"
"fmt"
"gitea.maximumdirect.net/eric/notarius/internal/core/source"
"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
"gitea.maximumdirect.net/eric/notarius/internal/modules/dnd"
"gitea.maximumdirect.net/eric/notarius/internal/modules/dnd/shared/diagnostics"
"gitea.maximumdirect.net/eric/notarius/internal/modules/dnd/validate/scenedescriptions/shape"
)
const (
Key = "extract/dnd/scene-descriptions/source_refs"
ReasonCode = "invalid_scene_description_source_refs"
policy = "dnd.scene_descriptions.validator.source_refs.v1"
)
type Options struct{}
type Validator struct{}
var _ contracts.TypedValidator[dnd.SceneDescriptionList] = (*Validator)(nil)
var _ pipeline.CheckpointFingerprintProvider = (*Validator)(nil)
func New(Options) *Validator { return &Validator{} }
func (v *Validator) Name() string { return Key }
func (v *Validator) ExecutionClass() contracts.ExecutionClass {
return contracts.ExecutionClassDeterministic
}
func (v *Validator) CheckpointFingerprints() []pipeline.CheckpointFingerprint {
return []pipeline.CheckpointFingerprint{{Name: "policy", Value: policy}}
}
func (v *Validator) Validate(_ context.Context, req contracts.TypedValidationRequest[dnd.SceneDescriptionList]) (contracts.ValidationResult, error) {
if shape.ValidateForStage(req.Value, req.Stage) != nil {
return contracts.ValidationResult{Approved: true}, nil
}
issues := make([]string, 0)
if req.Stage == string(pipeline.StageExtract) && req.Chunk == nil {
issues = append(issues, "current extraction chunk must not be nil")
}
for index, scene := range req.Value.Scenes {
prefix := fmt.Sprintf("scenes[%d]", index)
if err := source.ValidateRef(req.Source, scene.SourceRef); err != nil {
issues = append(issues, prefix+".source_ref: "+diagnostics.Truncate(err.Error()))
continue
}
if req.Stage == string(pipeline.StageExtract) && req.Chunk != nil {
if scene.ID != req.Chunk.ID {
issues = append(issues, prefix+".id must equal the current extraction chunk ID")
}
if scene.SourceRef != req.Chunk.Ref {
issues = append(issues, prefix+".source_ref must equal the current extraction chunk range")
}
}
}
if len(issues) == 0 {
return contracts.ValidationResult{Approved: true}, nil
}
return contracts.ValidationResult{
Approved: false, ReasonCode: ReasonCode,
Message: diagnostics.Aggregate("invalid scene description source references", issues),
}, nil
}
func Spec() pipeline.ValidatorSpec {
return pipeline.ValidatorSpec{Key: Key, ExecutionClass: contracts.ExecutionClassDeterministic}
}
func Register(registry *pipeline.ValidatorRegistry) error {
return pipeline.RegisterTypedValidatorBuilder(registry, dnd.SceneDescriptionListKind, Spec(), validateOptions, func(request pipeline.BuildRequest) (contracts.TypedValidator[dnd.SceneDescriptionList], error) {
options, err := DecodeOptions(request.Options)
if err != nil {
return nil, err
}
return New(options), nil
})
}
func DecodeOptions(options map[string]any) (Options, error) {
if err := pipeline.RejectUnknownOptions(options); err != nil {
return Options{}, err
}
return Options{}, nil
}
func validateOptions(options map[string]any) error { _, err := DecodeOptions(options); return err }

View File

@@ -0,0 +1,59 @@
package sourcerefs
import (
"context"
"reflect"
"testing"
"gitea.maximumdirect.net/eric/notarius/internal/core/source"
"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
"gitea.maximumdirect.net/eric/notarius/internal/modules/dnd"
)
func TestValidatorRequiresExactExtractionChunkIdentity(t *testing.T) {
doc := document()
chunk := &source.Chunk{ID: "chunk-1", SourceID: doc.ID, Ref: source.SourceRef{SourceID: doc.ID, StartUnitID: 1, EndUnitID: 2}}
valid := dnd.SceneDescriptionList{Scenes: []dnd.SceneDescription{{ID: chunk.ID, SourceRef: chunk.Ref, Kind: dnd.SceneKindNarrative, Title: "Arrival", Summary: "The party arrives."}}}
validator := New(Options{})
for _, test := range []struct {
name string
request contracts.TypedValidationRequest[dnd.SceneDescriptionList]
approved bool
}{
{name: "exact extraction identity", request: contracts.TypedValidationRequest[dnd.SceneDescriptionList]{Stage: string(pipeline.StageExtract), Source: doc, Chunk: chunk, Value: valid}, approved: true},
{name: "wrong ID", request: contracts.TypedValidationRequest[dnd.SceneDescriptionList]{Stage: string(pipeline.StageExtract), Source: doc, Chunk: chunk, Value: dnd.SceneDescriptionList{Scenes: []dnd.SceneDescription{{ID: "other", SourceRef: chunk.Ref, Kind: dnd.SceneKindNarrative, Title: "Arrival", Summary: "The party arrives."}}}}},
{name: "contained but unequal range", request: contracts.TypedValidationRequest[dnd.SceneDescriptionList]{Stage: string(pipeline.StageExtract), Source: doc, Chunk: chunk, Value: dnd.SceneDescriptionList{Scenes: []dnd.SceneDescription{{ID: chunk.ID, SourceRef: source.SourceRef{SourceID: doc.ID, StartUnitID: 1, EndUnitID: 1}, Kind: dnd.SceneKindNarrative, Title: "Arrival", Summary: "The party arrives."}}}}},
{name: "missing chunk", request: contracts.TypedValidationRequest[dnd.SceneDescriptionList]{Stage: string(pipeline.StageExtract), Source: doc, Value: valid}},
{name: "normalization needs no chunk", request: contracts.TypedValidationRequest[dnd.SceneDescriptionList]{Stage: string(pipeline.StageNormalize), Source: doc, Value: valid}, approved: true},
{name: "invalid current source", request: contracts.TypedValidationRequest[dnd.SceneDescriptionList]{Source: doc, Value: dnd.SceneDescriptionList{Scenes: []dnd.SceneDescription{{ID: "chunk", SourceRef: source.SourceRef{SourceID: doc.ID, StartUnitID: 7, EndUnitID: 7}, Kind: dnd.SceneKindNarrative, Title: "Arrival", Summary: "The party arrives."}}}}},
} {
t.Run(test.name, func(t *testing.T) {
result, err := validator.Validate(context.Background(), test.request)
if err != nil || result.Approved != test.approved {
t.Fatalf("Validate() = %#v, %v; want approved=%t", result, err, test.approved)
}
})
}
}
func TestValidatorDefersMalformedShapeAndRegistersPolicy(t *testing.T) {
result, err := New(Options{}).Validate(context.Background(), contracts.TypedValidationRequest[dnd.SceneDescriptionList]{Value: dnd.SceneDescriptionList{}})
if err != nil || !result.Approved {
t.Fatalf("Validate() = %#v, %v; want malformed shape deferred", result, err)
}
registry := pipeline.NewValidatorRegistry()
if err := Register(registry); err != nil {
t.Fatal(err)
}
if _, ok := registry.Spec(Key); !ok {
t.Fatalf("registry missing %q", Key)
}
if got, want := New(Options{}).CheckpointFingerprints(), []pipeline.CheckpointFingerprint{{Name: "policy", Value: policy}}; !reflect.DeepEqual(got, want) {
t.Fatalf("fingerprints = %#v, want %#v", got, want)
}
}
func document() *source.SourceDocument {
return &source.SourceDocument{ID: "session", Units: []source.SourceUnit{{ID: 1, Text: "The party arrives."}, {ID: 2, Text: "They investigate."}}}
}

View File

@@ -0,0 +1,139 @@
// Package sourcerelatedness warns when scene prose has no lexical grounding.
package sourcerelatedness
import (
"context"
"fmt"
"unicode/utf8"
"gitea.maximumdirect.net/eric/notarius/internal/core/source"
"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
"gitea.maximumdirect.net/eric/notarius/internal/modules/dnd"
"gitea.maximumdirect.net/eric/notarius/internal/modules/dnd/shared"
"gitea.maximumdirect.net/eric/notarius/internal/modules/dnd/shared/diagnostics"
"gitea.maximumdirect.net/eric/notarius/internal/modules/dnd/validate/scenedescriptions/shape"
)
const (
Key = "extract/dnd/scene-descriptions/source_relatedness"
WarningReasonCode = "scene_description_not_near_source"
OmittedReasonCode = "scene_description_relatedness_warnings_omitted"
policy = "dnd.scene_descriptions.validator.source_relatedness.v1"
)
var stopwords = map[string]struct{}{
"a": {}, "an": {}, "and": {}, "are": {}, "as": {}, "at": {}, "be": {}, "but": {}, "by": {}, "for": {}, "from": {},
"had": {}, "has": {}, "have": {}, "he": {}, "her": {}, "him": {}, "his": {}, "in": {}, "into": {}, "is": {}, "it": {}, "its": {},
"of": {}, "on": {}, "or": {}, "she": {}, "that": {}, "the": {}, "their": {}, "them": {}, "they": {}, "this": {}, "to": {},
"was": {}, "were": {}, "with": {},
}
type Options struct{}
type Validator struct{}
var _ contracts.TypedValidator[dnd.SceneDescriptionList] = (*Validator)(nil)
var _ pipeline.CheckpointFingerprintProvider = (*Validator)(nil)
func New(Options) *Validator { return &Validator{} }
func (v *Validator) Name() string { return Key }
func (v *Validator) ExecutionClass() contracts.ExecutionClass {
return contracts.ExecutionClassDeterministic
}
func (v *Validator) CheckpointFingerprints() []pipeline.CheckpointFingerprint {
return []pipeline.CheckpointFingerprint{{Name: "policy", Value: policy}}
}
func (v *Validator) Validate(_ context.Context, req contracts.TypedValidationRequest[dnd.SceneDescriptionList]) (contracts.ValidationResult, error) {
if shape.Validate(req.Value) != nil || !sourceRefsValid(req.Source, req.Value) {
return contracts.ValidationResult{Approved: true}, nil
}
warnings := make([]contracts.Warning, 0)
for index, scene := range req.Value.Scenes {
citedText, err := shared.CitedText(req.Source, []source.SourceRef{scene.SourceRef})
if err != nil {
return contracts.ValidationResult{Approved: true}, nil
}
citedTokens := tokenSet(citedText)
if !hasGroundedToken(citedTokens, scene.Title) {
warnings = append(warnings, warning(index, "title"))
}
if !hasGroundedToken(citedTokens, scene.Summary) {
warnings = append(warnings, warning(index, "summary"))
}
}
return contracts.ValidationResult{
Approved: true,
Warnings: diagnostics.LimitWarnings(warnings, "scenes", OmittedReasonCode),
}, nil
}
func sourceRefsValid(doc *source.SourceDocument, value dnd.SceneDescriptionList) bool {
for _, scene := range value.Scenes {
if source.ValidateRef(doc, scene.SourceRef) != nil {
return false
}
}
return true
}
func tokenSet(value string) map[string]struct{} {
tokens := make(map[string]struct{})
for _, token := range shared.NormalizedTokens(value) {
if significant(token) {
tokens[token] = struct{}{}
}
}
return tokens
}
func hasGroundedToken(cited map[string]struct{}, value string) bool {
for _, token := range shared.NormalizedTokens(value) {
if !significant(token) {
continue
}
if _, ok := cited[token]; ok {
return true
}
}
return false
}
func significant(token string) bool {
if utf8.RuneCountInString(token) < 3 {
return false
}
_, ignored := stopwords[token]
return !ignored
}
func warning(index int, field string) contracts.Warning {
return contracts.Warning{
Scope: fmt.Sprintf("scenes[%d].%s", index, field),
ReasonCode: WarningReasonCode,
Message: "scene description " + field + " has no significant token in cited source text",
}
}
func Spec() pipeline.ValidatorSpec {
return pipeline.ValidatorSpec{Key: Key, ExecutionClass: contracts.ExecutionClassDeterministic}
}
func Register(registry *pipeline.ValidatorRegistry) error {
return pipeline.RegisterTypedValidatorBuilder(registry, dnd.SceneDescriptionListKind, Spec(), validateOptions, func(request pipeline.BuildRequest) (contracts.TypedValidator[dnd.SceneDescriptionList], error) {
options, err := DecodeOptions(request.Options)
if err != nil {
return nil, err
}
return New(options), nil
})
}
func DecodeOptions(options map[string]any) (Options, error) {
if err := pipeline.RejectUnknownOptions(options); err != nil {
return Options{}, err
}
return Options{}, nil
}
func validateOptions(options map[string]any) error { _, err := DecodeOptions(options); return err }

View File

@@ -0,0 +1,78 @@
package sourcerelatedness
import (
"context"
"reflect"
"strings"
"testing"
"gitea.maximumdirect.net/eric/notarius/internal/core/source"
"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
"gitea.maximumdirect.net/eric/notarius/internal/modules/dnd"
)
func TestValidatorWarnsIndependentlyForUngroundedTitleAndSummary(t *testing.T) {
doc := document()
value := dnd.SceneDescriptionList{Scenes: []dnd.SceneDescription{
scene("one", "Watchtower arrival", "The group enters the ruined watchtower."),
scene("two", "Moonlit council", "Whispers linger."),
scene("three", "The and to", "An or the"),
}}
result, err := New(Options{}).Validate(context.Background(), contracts.TypedValidationRequest[dnd.SceneDescriptionList]{Source: doc, Value: value})
if err != nil || !result.Approved {
t.Fatalf("Validate() = %#v, %v", result, err)
}
wantScopes := []string{"scenes[1].title", "scenes[1].summary", "scenes[2].title", "scenes[2].summary"}
if len(result.Warnings) != len(wantScopes) {
t.Fatalf("warnings = %#v, want %d", result.Warnings, len(wantScopes))
}
for index, scope := range wantScopes {
item := result.Warnings[index]
if item.Scope != scope || item.ReasonCode != WarningReasonCode {
t.Fatalf("warning[%d] = %#v, want scope %q", index, item, scope)
}
if len(item.Message) > 4096 || strings.Contains(item.Message, doc.Units[0].Text) {
t.Fatalf("warning[%d] is not safely bounded: %#v", index, item)
}
}
}
func TestValidatorUsesTranscriptOnlyAndDefersInvalidInputs(t *testing.T) {
doc := document()
value := dnd.SceneDescriptionList{Scenes: []dnd.SceneDescription{scene("one", "Greencloak", "Greencloak arrives.")}}
result, err := New(Options{}).Validate(context.Background(), contracts.TypedValidationRequest[dnd.SceneDescriptionList]{
Source: doc, Value: value,
References: contracts.ReferenceSet{Slots: map[string]contracts.ResolvedReferenceSlot{
"glossary": {Items: []contracts.ReferenceItem{{Content: []byte("Greencloak: title")}}},
}},
})
if err != nil || len(result.Warnings) != 2 {
t.Fatalf("Validate() = %#v, %v; want transcript-only warnings", result, err)
}
malformed, err := New(Options{}).Validate(context.Background(), contracts.TypedValidationRequest[dnd.SceneDescriptionList]{Source: doc, Value: dnd.SceneDescriptionList{}})
if err != nil || !malformed.Approved || len(malformed.Warnings) != 0 {
t.Fatalf("malformed Validate() = %#v, %v; want deferral", malformed, err)
}
}
func TestValidatorRegistrationAndPolicyFingerprint(t *testing.T) {
registry := pipeline.NewValidatorRegistry()
if err := Register(registry); err != nil {
t.Fatal(err)
}
if _, ok := registry.Spec(Key); !ok {
t.Fatalf("registry missing %q", Key)
}
if got, want := New(Options{}).CheckpointFingerprints(), []pipeline.CheckpointFingerprint{{Name: "policy", Value: policy}}; !reflect.DeepEqual(got, want) {
t.Fatalf("fingerprints = %#v, want %#v", got, want)
}
}
func document() *source.SourceDocument {
return &source.SourceDocument{ID: "session", Units: []source.SourceUnit{{ID: 1, Text: "The group enters the ruined watchtower."}}}
}
func scene(id, title, summary string) dnd.SceneDescription {
return dnd.SceneDescription{ID: id, SourceRef: source.SourceRef{SourceID: "session", StartUnitID: 1, EndUnitID: 1}, Kind: dnd.SceneKindNarrative, Title: title, Summary: summary}
}