Bound D&D normalization diagnostics

This commit is contained in:
2026-08-09 02:22:23 +00:00
parent a705ba74a1
commit 2a75f40871
8 changed files with 218 additions and 30 deletions

View File

@@ -16,8 +16,12 @@ import (
)
const (
Key = "dnd/scene-descriptions"
normalizerPolicy = "dnd.scene_descriptions.normalizer.v1"
Key = "dnd/scene-descriptions"
normalizerPolicy = "dnd.scene_descriptions.normalizer.v1"
ReasonCodeProseNormalized = "scene_description_prose_normalized"
ReasonCodeOrderNormalized = "scene_description_order_normalized"
ReasonCodeDuplicateCollapsed = "scene_description_duplicate_collapsed"
ReasonCodeNormalizationWarningsOmitted = "scene_description_normalization_warnings_omitted"
)
var requiredCapabilities = []string{"merged"}
@@ -60,70 +64,106 @@ func (n *Normalizer) Normalize(ctx context.Context, req contracts.TypedNormalize
if err := ctx.Err(); err != nil {
return contracts.TypedNormalizeResult[dnd.SceneDescriptionList]{}, normalizerErrorf("context error before normalize: %w", err)
}
value, err := normalizeList(req.MergeOutput.Value, req.Source)
value, warnings, err := normalizeList(req.MergeOutput.Value, req.Source)
if err != nil {
return contracts.TypedNormalizeResult[dnd.SceneDescriptionList]{}, normalizerErrorf("normalize scenes: %w", err)
}
return contracts.TypedNormalizeResult[dnd.SceneDescriptionList]{Value: value}, nil
return contracts.TypedNormalizeResult[dnd.SceneDescriptionList]{
Value: value,
Warnings: diagnostics.LimitWarnings(warnings, "scenes", ReasonCodeNormalizationWarningsOmitted),
}, nil
}
func normalizeList(input dnd.SceneDescriptionList, doc *source.SourceDocument) (dnd.SceneDescriptionList, error) {
type normalizedScene struct {
scene dnd.SceneDescription
inputIndex int
}
func normalizeList(input dnd.SceneDescriptionList, doc *source.SourceDocument) (dnd.SceneDescriptionList, []contracts.Warning, error) {
if doc == nil {
return dnd.SceneDescriptionList{}, fmt.Errorf("source document must not be nil")
return dnd.SceneDescriptionList{}, nil, fmt.Errorf("source document must not be nil")
}
if input.Scenes == nil {
return dnd.SceneDescriptionList{}, fmt.Errorf("scenes must be present")
return dnd.SceneDescriptionList{}, nil, fmt.Errorf("scenes must be present")
}
if len(input.Scenes) == 0 {
return dnd.SceneDescriptionList{}, fmt.Errorf("scenes must not be empty")
return dnd.SceneDescriptionList{}, nil, fmt.Errorf("scenes must not be empty")
}
documentIndex := source.NewDocumentIndex(doc)
output := dnd.SceneDescriptionList{Scenes: make([]dnd.SceneDescription, len(input.Scenes))}
records := make([]normalizedScene, len(input.Scenes))
warnings := make([]contracts.Warning, 0)
for sceneIndex, scene := range input.Scenes {
originalTitle, originalSummary := scene.Title, scene.Summary
scene.Title = strings.TrimSpace(scene.Title)
scene.Summary = strings.TrimSpace(scene.Summary)
if err := shape.Validate(dnd.SceneDescriptionList{Scenes: []dnd.SceneDescription{scene}}); err != nil {
return dnd.SceneDescriptionList{}, fmt.Errorf("scenes[%d]: %w", sceneIndex, err)
return dnd.SceneDescriptionList{}, nil, fmt.Errorf("scenes[%d]: %w", sceneIndex, err)
}
if err := documentIndex.ValidateRef(scene.SourceRef); err != nil {
return dnd.SceneDescriptionList{}, fmt.Errorf("scenes[%d].source_ref: %s", sceneIndex, diagnostics.Truncate(err.Error()))
return dnd.SceneDescriptionList{}, nil, fmt.Errorf("scenes[%d].source_ref: %s", sceneIndex, diagnostics.Truncate(err.Error()))
}
output.Scenes[sceneIndex] = scene
if originalTitle != scene.Title || originalSummary != scene.Summary {
warnings = append(warnings, contracts.Warning{
Scope: sceneScope(sceneIndex),
ReasonCode: ReasonCodeProseNormalized,
Message: fmt.Sprintf("input index %d: title and/or summary whitespace normalized", sceneIndex),
})
}
records[sceneIndex] = normalizedScene{scene: scene, inputIndex: sceneIndex}
}
sort.SliceStable(output.Scenes, func(left, right int) bool {
leftStart, _ := documentIndex.Position(output.Scenes[left].SourceRef.StartUnitID)
rightStart, _ := documentIndex.Position(output.Scenes[right].SourceRef.StartUnitID)
sort.SliceStable(records, func(left, right int) bool {
leftStart, _ := documentIndex.Position(records[left].scene.SourceRef.StartUnitID)
rightStart, _ := documentIndex.Position(records[right].scene.SourceRef.StartUnitID)
if leftStart != rightStart {
return leftStart < rightStart
}
return output.Scenes[left].ID < output.Scenes[right].ID
return records[left].scene.ID < records[right].scene.ID
})
for position, record := range records {
if position == record.inputIndex {
continue
}
warnings = append(warnings, contracts.Warning{
Scope: sceneScope(record.inputIndex),
ReasonCode: ReasonCodeOrderNormalized,
Message: fmt.Sprintf("input index %d moved to normalized position %d by canonical scene order",
record.inputIndex, position),
})
}
unique := make([]dnd.SceneDescription, 0, len(output.Scenes))
byID := make(map[string]dnd.SceneDescription, len(output.Scenes))
byRange := make(map[source.SourceRef]dnd.SceneDescription, len(output.Scenes))
seen := make(map[dnd.SceneDescription]struct{}, len(output.Scenes))
for _, scene := range output.Scenes {
unique := make([]dnd.SceneDescription, 0, len(records))
byID := make(map[string]dnd.SceneDescription, len(records))
byRange := make(map[source.SourceRef]dnd.SceneDescription, len(records))
seen := make(map[dnd.SceneDescription]int, len(records))
for _, record := range records {
scene := record.scene
if previous, ok := byID[scene.ID]; ok && !identical(previous, scene) {
return dnd.SceneDescriptionList{}, fmt.Errorf("scene ID %s has conflicting records", diagnostics.Quote(scene.ID))
return dnd.SceneDescriptionList{}, nil, fmt.Errorf("scene ID %s has conflicting records", diagnostics.Quote(scene.ID))
}
if previous, ok := byRange[scene.SourceRef]; ok && !sameModelContent(previous, scene) {
return dnd.SceneDescriptionList{}, fmt.Errorf("source range %s has conflicting records", sourceRefLabel(scene.SourceRef))
return dnd.SceneDescriptionList{}, nil, fmt.Errorf("source range %s has conflicting records", sourceRefLabel(scene.SourceRef))
}
if _, ok := seen[scene]; ok {
if retainedIndex, ok := seen[scene]; ok {
warnings = append(warnings, contracts.Warning{
Scope: sceneScope(record.inputIndex),
ReasonCode: ReasonCodeDuplicateCollapsed,
Message: fmt.Sprintf("input index %d: exact duplicate scene collapsed; retained input index %d",
record.inputIndex, retainedIndex),
})
continue
}
byID[scene.ID] = scene
byRange[scene.SourceRef] = scene
seen[scene] = struct{}{}
seen[scene] = record.inputIndex
unique = append(unique, scene)
}
output.Scenes = unique
return output, nil
return dnd.SceneDescriptionList{Scenes: unique}, warnings, nil
}
func sceneScope(index int) string { return fmt.Sprintf("scenes[%d]", index) }
func identical(left, right dnd.SceneDescription) bool {
return left == right
}