Consolidate duplicate item registry records
This commit is contained in:
File diff suppressed because it is too large
Load Diff
@@ -212,13 +212,16 @@ func preprocessRecords(input dnd.ItemRegistry, order shared.SourceRefOrder) ([]n
|
||||
warnings = append(warnings, contracts.Warning{Scope: itemScope(index), ReasonCode: ReasonCodeItemIDRecomputed, Message: fmt.Sprintf("input index %d: item ID recomputed from %s", index, diagnostics.Quote(item.Name))})
|
||||
}
|
||||
}
|
||||
groups := exactDuplicateGroups(records)
|
||||
groups := comparisonNameGroups(records)
|
||||
output := make([]normalizedRecord, 0, len(groups))
|
||||
for _, members := range groups {
|
||||
retained := cloneRecord(records[members[0]])
|
||||
for _, member := range members[1:] {
|
||||
retained.item.SourceRefs = append(retained.item.SourceRefs, records[member].item.SourceRefs...)
|
||||
retained.inputIndexes = append(retained.inputIndexes, records[member].inputIndexes...)
|
||||
}
|
||||
retained.item.SourceRefs = order.Canonicalize(retained.item.SourceRefs)
|
||||
retained.item.ID = identity.DeriveID(retained.item.Name)
|
||||
retained.inputIndexes = sortedUniqueIndexes(retained.inputIndexes)
|
||||
output = append(output, retained)
|
||||
if len(members) > 1 {
|
||||
@@ -236,14 +239,14 @@ func normalizeRecord(input dnd.Item, order shared.SourceRefOrder) (dnd.Item, boo
|
||||
return output, input.Name != output.Name, !reflect.DeepEqual(input.SourceRefs, output.SourceRefs)
|
||||
}
|
||||
|
||||
func exactDuplicateGroups(records []normalizedRecord) [][]int {
|
||||
func comparisonNameGroups(records []normalizedRecord) [][]int {
|
||||
groups := make([][]int, 0, len(records))
|
||||
for index, record := range records {
|
||||
key := identity.ComparisonKey(record.item.Name)
|
||||
found := false
|
||||
for groupIndex, members := range groups {
|
||||
first := records[members[0]]
|
||||
if identity.ComparisonKey(first.item.Name) == key && reflect.DeepEqual(first.item.SourceRefs, record.item.SourceRefs) {
|
||||
if identity.ComparisonKey(first.item.Name) == key {
|
||||
groups[groupIndex] = append(groups[groupIndex], index)
|
||||
found = true
|
||||
break
|
||||
|
||||
@@ -17,6 +17,7 @@ import (
|
||||
"gitea.maximumdirect.net/eric/notarius/internal/modules/dnd/items/identity"
|
||||
"gitea.maximumdirect.net/eric/notarius/internal/modules/dnd/shared/diagnostics"
|
||||
"gitea.maximumdirect.net/eric/notarius/internal/modules/dnd/shared/entityreconcile"
|
||||
identityvalidator "gitea.maximumdirect.net/eric/notarius/internal/modules/dnd/validate/itemregistry/identity"
|
||||
"gitea.maximumdirect.net/eric/promptkit"
|
||||
)
|
||||
|
||||
@@ -38,19 +39,48 @@ func TestModuleContractAndMetadata(t *testing.T) {
|
||||
}
|
||||
}
|
||||
|
||||
func TestNormalizePreprocessesExactDuplicatesWithoutMutation(t *testing.T) {
|
||||
input := dnd.ItemRegistry{Items: []dnd.Item{
|
||||
{Name: " Rope ", SourceRefs: []source.SourceRef{{SourceID: "session", StartUnitID: 1, EndUnitID: 1}}},
|
||||
{Name: "rope", SourceRefs: []source.SourceRef{{SourceID: "session", StartUnitID: 1, EndUnitID: 1}}},
|
||||
{Name: "Rope", SourceRefs: []source.SourceRef{{SourceID: "session", StartUnitID: 2, EndUnitID: 2}}},
|
||||
func TestNormalizeConsolidatesEqualNamesAcrossEvidenceWithoutMutation(t *testing.T) {
|
||||
doc := &source.SourceDocument{ID: "session", Units: []source.SourceUnit{
|
||||
{ID: 1, Text: "The rope is secured."},
|
||||
{ID: 2, Text: "The party takes the rope."},
|
||||
{ID: 3, Text: "The rope is packed away."},
|
||||
{ID: 4, Text: "A lantern lights the path."},
|
||||
}}
|
||||
before := dnd.ItemRegistry{Items: append([]dnd.Item(nil), input.Items...)}
|
||||
result, err := newNormalizer(t, &recordingNormalizerClient{}).Normalize(context.Background(), normalizeRequest(input))
|
||||
if err != nil || len(result.Value.Items) != 2 || !reflect.DeepEqual(input, before) {
|
||||
t.Fatalf("Normalize() = %#v, %v; want non-mutating exact deduplication", result, err)
|
||||
input := dnd.ItemRegistry{Items: []dnd.Item{
|
||||
{Name: " Rope ", SourceRefs: []source.SourceRef{{SourceID: "session", StartUnitID: 3, EndUnitID: 3}, {SourceID: "session", StartUnitID: 1, EndUnitID: 1}}},
|
||||
{Name: "rope", SourceRefs: []source.SourceRef{{SourceID: "session", StartUnitID: 2, EndUnitID: 2}, {SourceID: "session", StartUnitID: 3, EndUnitID: 3}}},
|
||||
{Name: "Lantern", SourceRefs: []source.SourceRef{{SourceID: "session", StartUnitID: 4, EndUnitID: 4}}},
|
||||
}}
|
||||
before := dnd.ItemRegistry{Items: []dnd.Item{
|
||||
{Name: " Rope ", SourceRefs: []source.SourceRef{{SourceID: "session", StartUnitID: 3, EndUnitID: 3}, {SourceID: "session", StartUnitID: 1, EndUnitID: 1}}},
|
||||
{Name: "rope", SourceRefs: []source.SourceRef{{SourceID: "session", StartUnitID: 2, EndUnitID: 2}, {SourceID: "session", StartUnitID: 3, EndUnitID: 3}}},
|
||||
{Name: "Lantern", SourceRefs: []source.SourceRef{{SourceID: "session", StartUnitID: 4, EndUnitID: 4}}},
|
||||
}}
|
||||
client := &recordingNormalizerClient{}
|
||||
result, err := newNormalizer(t, client).Normalize(context.Background(), normalizeRequestWithSource(input, doc))
|
||||
if err != nil || len(result.Value.Items) != 2 || !reflect.DeepEqual(input, before) || len(client.requests) != 1 {
|
||||
t.Fatalf("Normalize() = %#v, %v; want deterministic non-mutating consolidation", result, err)
|
||||
}
|
||||
if result.Value.Items[0].ID != result.Value.Items[1].ID || !hasWarning(result.Warnings, ReasonCodeDuplicateItemCollapsed) {
|
||||
t.Fatalf("items = %#v, warnings = %#v; want type identity and duplicate warning", result.Value.Items, result.Warnings)
|
||||
rope := result.Value.Items[0]
|
||||
wantRefs := []source.SourceRef{{SourceID: "session", StartUnitID: 1, EndUnitID: 1}, {SourceID: "session", StartUnitID: 2, EndUnitID: 2}, {SourceID: "session", StartUnitID: 3, EndUnitID: 3}}
|
||||
if rope.Name != "Rope" || rope.ID != identity.DeriveID("Rope") || !reflect.DeepEqual(rope.SourceRefs, wantRefs) || result.Value.Items[1].Name != "Lantern" || !hasWarning(result.Warnings, ReasonCodeDuplicateItemCollapsed) {
|
||||
t.Fatalf("items = %#v, warnings = %#v; want earliest display name, canonical evidence union, and stable placement", result.Value.Items, result.Warnings)
|
||||
}
|
||||
validation, validationErr := identityvalidator.New(identityvalidator.Options{}).Validate(context.Background(), contracts.TypedValidationRequest[dnd.ItemRegistry]{Value: result.Value})
|
||||
if validationErr != nil || !validation.Approved {
|
||||
t.Fatalf("default item-registry identity validation = %#v, %v; want normalized cross-chunk candidates accepted", validation, validationErr)
|
||||
}
|
||||
var candidates struct {
|
||||
Candidates []struct {
|
||||
Name string `json:"name"`
|
||||
} `json:"candidates"`
|
||||
}
|
||||
if err := json.Unmarshal(client.requests[0].Inputs["candidates"].Content, &candidates); err != nil || len(candidates.Candidates) != 2 || candidates.Candidates[0].Name != "Rope" || candidates.Candidates[1].Name != "Lantern" {
|
||||
t.Fatalf("semantic candidates = %#v, %v; want one candidate per comparison name", candidates, err)
|
||||
}
|
||||
repeated, repeatErr := newNormalizer(t, &recordingNormalizerClient{}).Normalize(context.Background(), normalizeRequestWithSource(result.Value, doc))
|
||||
if repeatErr != nil || !reflect.DeepEqual(repeated.Value, result.Value) {
|
||||
t.Fatalf("repeated normalization = %#v, %v; want stable output", repeated, repeatErr)
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
Reference in New Issue
Block a user