Validate artifact candidates before final encoding

This commit is contained in:
2026-07-17 15:13:19 +00:00
parent 3772b308e9
commit 35bffdf336
5 changed files with 338 additions and 10 deletions

View File

@@ -0,0 +1,268 @@
package pipeline
import (
"context"
"encoding/json"
"errors"
"fmt"
"reflect"
"strings"
"testing"
"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
)
type observedNotesCodec struct {
candidateValues []codecNotes
finalValues []codecNotes
candidateError string
finalError string
}
func (*observedNotesCodec) Kind() contracts.ArtifactKind { return "test/notes" }
func (*observedNotesCodec) Schema() contracts.ArtifactSchema { return notesCodec().schema }
func (*observedNotesCodec) MediaType() string { return "application/json" }
func (c *observedNotesCodec) EncodeCandidate(value codecNotes) ([]byte, error) {
c.candidateValues = append(c.candidateValues, value)
if c.candidateError != "" && firstNote(value) == c.candidateError {
return nil, errors.New("candidate encoding failed")
}
return json.Marshal(value)
}
func (c *observedNotesCodec) Encode(value codecNotes) ([]byte, error) {
c.finalValues = append(c.finalValues, value)
if c.finalError != "" && firstNote(value) == c.finalError {
return nil, errors.New("final encoding failed")
}
if firstNote(value) == "invalid" {
return nil, errors.New("invalid note is not a final artifact")
}
return json.Marshal(value)
}
func (*observedNotesCodec) Decode(content []byte) (codecNotes, error) {
var value codecNotes
return value, json.Unmarshal(content, &value)
}
func firstNote(value codecNotes) string {
if len(value.Items) == 0 {
return ""
}
return value.Items[0]
}
func (c *observedNotesCodec) candidateCalls(value string) int {
return matchingNotes(c.candidateValues, value)
}
func (c *observedNotesCodec) finalCalls(value string) int {
return matchingNotes(c.finalValues, value)
}
func matchingNotes(values []codecNotes, value string) int {
count := 0
for _, candidate := range values {
if firstNote(candidate) == value {
count++
}
}
return count
}
type candidateCheckpointRecorder struct {
CheckpointRecorder
mergeSucceeded int
normalizeSucceeded int
mergeOutput CheckpointArtifact
normalizeOutput CheckpointArtifact
}
func (r *candidateCheckpointRecorder) MergeSucceeded(_ string, _ string, _ []CheckpointFingerprint, output CheckpointArtifact, _ []contracts.Warning) error {
r.mergeSucceeded++
r.mergeOutput = cloneCheckpointArtifact(output)
return nil
}
func (r *candidateCheckpointRecorder) NormalizeSucceeded(_ string, _ string, _ []CheckpointFingerprint, output CheckpointArtifact, _ []contracts.Warning) error {
r.normalizeSucceeded++
r.normalizeOutput = cloneCheckpointArtifact(output)
return nil
}
func installObservedNotesCodec(t *testing.T, prepared *PreparedPipeline, codec *observedNotesCodec) {
t.Helper()
registry := NewArtifactCodecRegistry()
if err := RegisterArtifactCodec(registry, codec); err != nil {
t.Fatalf("RegisterArtifactCodec() error = %v", err)
}
entry, _, err := registry.entry("test/notes")
if err != nil {
t.Fatalf("codec entry error = %v", err)
}
prepared.lanes[0].typed.codec = entry
}
func configureCandidateOperation(prepared *PreparedPipeline, target ModuleStage, value codecNotes) {
lane := &prepared.lanes[0]
switch target {
case StageMerge:
lane.typed.merge = func(context.Context, any, contracts.TypedMergeRequest[any]) (erasedTypedResult, error) {
return erasedTypedResult{Value: value}, nil
}
lane.typed.normalize = func(context.Context, any, contracts.TypedNormalizeRequest[any]) (erasedTypedResult, error) {
return erasedTypedResult{Value: codecNotes{Items: []string{"normalized-other"}}}, nil
}
case StageNormalize:
lane.typed.merge = func(context.Context, any, contracts.TypedMergeRequest[any]) (erasedTypedResult, error) {
return erasedTypedResult{Value: codecNotes{Items: []string{"merged-other"}}}, nil
}
lane.typed.normalize = func(context.Context, any, contracts.TypedNormalizeRequest[any]) (erasedTypedResult, error) {
return erasedTypedResult{Value: value}, nil
}
}
}
func setCandidateValidator(prepared *PreparedPipeline, target ModuleStage, approved bool) {
validator := preparedValidator{
resolved: ResolvedValidator{Binding: Binding("candidate-check"), Target: ValidatorTargetTyped, ArtifactKind: "test/notes"},
typedValidate: func(context.Context, any, typedValidationTarget) (contracts.ValidationResult, error) {
return contracts.ValidationResult{Approved: approved, ReasonCode: "candidate_rejected", Message: "candidate rejected by validator"}, nil
},
}
switch target {
case StageMerge:
prepared.lanes[0].mergeValidators.validators = []preparedValidator{validator}
case StageNormalize:
prepared.lanes[0].normalizeValidators.validators = []preparedValidator{validator}
}
}
func TestRunnerRejectsCandidatesBeforeFinalEncoding(t *testing.T) {
for _, target := range []ModuleStage{StageMerge, StageNormalize} {
t.Run(string(target), func(t *testing.T) {
prepared := preparedAttemptDebugPipeline(t)
codec := &observedNotesCodec{}
installObservedNotesCodec(t, prepared, codec)
configureCandidateOperation(prepared, target, codecNotes{Items: []string{"invalid"}})
setCandidateValidator(prepared, target, false)
recorder := &candidateCheckpointRecorder{CheckpointRecorder: NoopCheckpointRecorder()}
output, err := New().Run(context.Background(), RunInput{Prepared: prepared, RawInput: []byte("input"), Checkpoints: recorder})
if err != nil {
t.Fatalf("Run() error = %v, want validator rejection", err)
}
if len(output.Rejected) != 1 || output.Rejected[0].Stage != string(target) || output.Rejected[0].ReasonCode != "candidate_rejected" {
t.Fatalf("rejected outputs = %#v, want %s validator rejection", output.Rejected, target)
}
if codec.candidateCalls("invalid") != 1 || codec.finalCalls("invalid") != 0 {
t.Fatalf("invalid candidate calls = candidate %d, final %d; want 1, 0", codec.candidateCalls("invalid"), codec.finalCalls("invalid"))
}
if target == StageMerge && recorder.mergeSucceeded != 0 {
t.Fatalf("merge checkpoints = %d, want none", recorder.mergeSucceeded)
}
if target == StageNormalize && recorder.normalizeSucceeded != 0 {
t.Fatalf("normalize checkpoints = %d, want none", recorder.normalizeSucceeded)
}
})
}
}
func TestRunnerFinalEncodesAcceptedCandidatesOnce(t *testing.T) {
for _, target := range []ModuleStage{StageMerge, StageNormalize} {
t.Run(string(target), func(t *testing.T) {
prepared := preparedAttemptDebugPipeline(t)
codec := &observedNotesCodec{}
installObservedNotesCodec(t, prepared, codec)
value := "accepted-" + string(target)
configureCandidateOperation(prepared, target, codecNotes{Items: []string{value}})
setCandidateValidator(prepared, target, true)
recorder := &candidateCheckpointRecorder{CheckpointRecorder: NoopCheckpointRecorder()}
loader := &extractResultLoader{CheckpointLoader: NoopCheckpointLoader(), decision: CheckpointDecision{Reason: "not found"}}
output, err := New().Run(context.Background(), RunInput{Prepared: prepared, RawInput: []byte("input"), Checkpoints: recorder, Checkpoint: loader})
if err != nil {
t.Fatalf("Run() error = %v, want nil", err)
}
if codec.candidateCalls(value) != 1 || codec.finalCalls(value) != 1 {
t.Fatalf("accepted candidate calls = candidate %d, final %d; want 1, 1", codec.candidateCalls(value), codec.finalCalls(value))
}
if target == StageMerge && recorder.mergeSucceeded != 1 {
t.Fatalf("merge checkpoints = %d, want one", recorder.mergeSucceeded)
}
if target == StageNormalize && recorder.normalizeSucceeded != 1 {
t.Fatalf("normalize checkpoints = %d, want one", recorder.normalizeSucceeded)
}
checkpoint := recorder.mergeOutput
if target == StageNormalize {
checkpoint = recorder.normalizeOutput
}
var stored codecNotes
if err := json.Unmarshal(checkpoint.Artifact.Content, &stored); err != nil || !reflect.DeepEqual(stored, codecNotes{Items: []string{value}}) {
t.Fatalf("checkpoint content = %s, %v; want accepted value %q", checkpoint.Artifact.Content, err, value)
}
gotStages := make([]string, len(output.CheckpointEvents))
for i, event := range output.CheckpointEvents {
gotStages[i] = event.Stage
}
wantStages := []string{"source", string(StageChunk), string(StageExtract), string(StageMerge), string(StageNormalize)}
if !reflect.DeepEqual(gotStages, wantStages) {
t.Fatalf("checkpoint event stages = %#v, want %#v", gotStages, wantStages)
}
})
}
}
func TestRunnerRecordsCandidateAndFinalEncodingFailures(t *testing.T) {
tests := []struct {
name string
target ModuleStage
value string
candidate bool
wantError string
}{
{name: "merge candidate", target: StageMerge, value: "merge-candidate-failure", candidate: true, wantError: "serialize merge candidate"},
{name: "normalize candidate", target: StageNormalize, value: "normalize-candidate-failure", candidate: true, wantError: "serialize normalize candidate"},
{name: "merge final", target: StageMerge, value: "merge-final-failure", wantError: "serialize accepted merge output"},
{name: "normalize final", target: StageNormalize, value: "normalize-final-failure", wantError: "serialize accepted normalize output"},
}
for _, tc := range tests {
t.Run(tc.name, func(t *testing.T) {
prepared := preparedAttemptDebugPipeline(t)
codec := &observedNotesCodec{}
if tc.candidate {
codec.candidateError = tc.value
} else {
codec.finalError = tc.value
}
installObservedNotesCodec(t, prepared, codec)
configureCandidateOperation(prepared, tc.target, codecNotes{Items: []string{tc.value}})
setCandidateValidator(prepared, tc.target, true)
recorder := &candidateCheckpointRecorder{CheckpointRecorder: NoopCheckpointRecorder()}
debug := newCapturedDebugRecorder()
_, err := New().Run(context.Background(), RunInput{Prepared: prepared, RawInput: []byte("input"), Checkpoints: recorder, Debug: debug})
if err == nil || !strings.Contains(err.Error(), tc.wantError) {
t.Fatalf("Run() error = %v, want %q", err, tc.wantError)
}
attemptPath := fmt.Sprintf("%s/notes/attempt-01.json", tc.target)
envelope := debug.envelope(t, attemptPath)
if !strings.Contains(envelope.Error, tc.wantError) {
t.Fatalf("attempt error = %q, want %q", envelope.Error, tc.wantError)
}
if tc.candidate && codec.finalCalls(tc.value) != 0 {
t.Fatalf("final encode calls = %d, want none after candidate failure", codec.finalCalls(tc.value))
}
if tc.target == StageMerge && recorder.mergeSucceeded != 0 {
t.Fatalf("merge checkpoints = %d, want none", recorder.mergeSucceeded)
}
if tc.target == StageNormalize && recorder.normalizeSucceeded != 0 {
t.Fatalf("normalize checkpoints = %d, want none", recorder.normalizeSucceeded)
}
})
}
}

View File

@@ -111,6 +111,14 @@ func checkpointArtifact(codec artifactCodecEntry, laneID, moduleKey, sourceID st
return CheckpointArtifact{LaneID: laneID, ModuleKey: moduleKey, SourceID: sourceID, Artifact: serialized, SchemaDigest: contracts.DigestArtifactSchema(serialized.Schema)}, nil
}
func serializeCandidateArtifact(codec artifactCodecEntry, laneID, moduleKey, sourceID string, value any) (CheckpointArtifact, error) {
serialized, err := serializeArtifact(codec, value, true)
if err != nil {
return CheckpointArtifact{}, err
}
return CheckpointArtifact{LaneID: laneID, ModuleKey: moduleKey, SourceID: sourceID, Artifact: serialized, SchemaDigest: contracts.DigestArtifactSchema(serialized.Schema)}, nil
}
type laneRunError struct {
stage ModuleStage
err error
@@ -183,8 +191,8 @@ func (r *Runner) continueTypedLane(ctx context.Context, input RunInput, checkpoi
return false, nil, attemptErr
}
candidate := erasedMergeArtifact{LaneID: lane.ID, MergerKey: lane.Merge.Module, SourceID: doc.ID, Value: result.Value}
stored, encodeErr := checkpointArtifact(typed.codec, candidate.LaneID, candidate.MergerKey, candidate.SourceID, candidate.Value)
attemptWarnings := cloneWarnings(result.Warnings)
serializedCandidate, encodeErr := serializeCandidateArtifact(typed.codec, candidate.LaneID, candidate.MergerKey, candidate.SourceID, candidate.Value)
if encodeErr != nil {
attemptErr := fmt.Errorf("serialize merge candidate for lane %q: %w", lane.ID, encodeErr)
if debugErr := attemptEnvelope(map[string]any{"warnings": debugWarningEnvelopes(attemptWarnings)}, attemptErr); debugErr != nil {
@@ -192,15 +200,23 @@ func (r *Runner) continueTypedLane(ctx context.Context, input RunInput, checkpoi
}
return false, nil, attemptErr
}
warnings, rejected, validateErr := r.validateTypedArtifact(attemptCtx, typed.codec, typedValidationTarget{stage: StageMerge, laneID: lane.ID, moduleKey: lane.Merge.Module, source: doc, sourceID: doc.ID, sourceInput: sourceInput.Clone(), sessionID: sessionID, references: lane.MergeReferences.ReferenceSet, metadata: input.Metadata, value: result.Value}, prepared.mergeValidators, attempt, input.Debug)
warnings, rejected, validateErr := r.validateTypedArtifact(attemptCtx, typed.codec, typedValidationTarget{stage: StageMerge, laneID: lane.ID, moduleKey: lane.Merge.Module, source: doc, sourceID: doc.ID, sourceInput: sourceInput.Clone(), sessionID: sessionID, references: lane.MergeReferences.ReferenceSet, metadata: input.Metadata, value: result.Value, candidate: &serializedCandidate}, prepared.mergeValidators, attempt, input.Debug)
attemptWarnings = append(attemptWarnings, warnings...)
payload := map[string]any{"output": debugCheckpointArtifact(stored), "warnings": debugWarningEnvelopes(attemptWarnings), "rejection": debugRejectedOutputPtr(rejected)}
payload := map[string]any{"output": debugCheckpointArtifact(serializedCandidate), "warnings": debugWarningEnvelopes(attemptWarnings), "rejection": debugRejectedOutputPtr(rejected)}
if validateErr != nil || rejected != nil {
if debugErr := attemptEnvelope(payload, validateErr); debugErr != nil {
return false, nil, fmt.Errorf("write merge attempt debug artifact: %w", debugErr)
}
return false, rejected, validateErr
}
stored, encodeErr := checkpointArtifact(typed.codec, candidate.LaneID, candidate.MergerKey, candidate.SourceID, candidate.Value)
if encodeErr != nil {
attemptErr := fmt.Errorf("serialize accepted merge output for lane %q: %w", lane.ID, encodeErr)
if debugErr := attemptEnvelope(payload, attemptErr); debugErr != nil {
return false, nil, fmt.Errorf("write merge attempt debug artifact: %w", debugErr)
}
return false, nil, attemptErr
}
if debugErr := attemptEnvelope(payload, nil); debugErr != nil {
return false, nil, fmt.Errorf("write merge attempt debug artifact: %w", debugErr)
}
@@ -272,8 +288,8 @@ func (r *Runner) continueTypedLane(ctx context.Context, input RunInput, checkpoi
}
return false, nil, attemptErr
}
stored, encodeErr := checkpointArtifact(typed.codec, lane.ID, lane.Normalize.Module, doc.ID, result.Value)
attemptWarnings := cloneWarnings(result.Warnings)
serializedCandidate, encodeErr := serializeCandidateArtifact(typed.codec, lane.ID, lane.Normalize.Module, doc.ID, result.Value)
if encodeErr != nil {
attemptErr := fmt.Errorf("serialize normalize candidate for lane %q: %w", lane.ID, encodeErr)
if debugErr := attemptEnvelope(map[string]any{"warnings": debugWarningEnvelopes(attemptWarnings)}, attemptErr); debugErr != nil {
@@ -281,15 +297,23 @@ func (r *Runner) continueTypedLane(ctx context.Context, input RunInput, checkpoi
}
return false, nil, attemptErr
}
warnings, rejected, validateErr := r.validateTypedArtifact(attemptCtx, typed.codec, typedValidationTarget{stage: StageNormalize, laneID: lane.ID, moduleKey: lane.Normalize.Module, source: doc, sourceID: doc.ID, sourceInput: sourceInput.Clone(), sessionID: sessionID, references: lane.NormalizeReferences.ReferenceSet, metadata: input.Metadata, value: result.Value}, prepared.normalizeValidators, attempt, input.Debug)
warnings, rejected, validateErr := r.validateTypedArtifact(attemptCtx, typed.codec, typedValidationTarget{stage: StageNormalize, laneID: lane.ID, moduleKey: lane.Normalize.Module, source: doc, sourceID: doc.ID, sourceInput: sourceInput.Clone(), sessionID: sessionID, references: lane.NormalizeReferences.ReferenceSet, metadata: input.Metadata, value: result.Value, candidate: &serializedCandidate}, prepared.normalizeValidators, attempt, input.Debug)
attemptWarnings = append(attemptWarnings, warnings...)
payload := map[string]any{"output": debugCheckpointArtifact(stored), "warnings": debugWarningEnvelopes(attemptWarnings), "rejection": debugRejectedOutputPtr(rejected)}
payload := map[string]any{"output": debugCheckpointArtifact(serializedCandidate), "warnings": debugWarningEnvelopes(attemptWarnings), "rejection": debugRejectedOutputPtr(rejected)}
if validateErr != nil || rejected != nil {
if debugErr := attemptEnvelope(payload, validateErr); debugErr != nil {
return false, nil, fmt.Errorf("write normalize attempt debug artifact: %w", debugErr)
}
return false, rejected, validateErr
}
stored, encodeErr := checkpointArtifact(typed.codec, lane.ID, lane.Normalize.Module, doc.ID, result.Value)
if encodeErr != nil {
attemptErr := fmt.Errorf("serialize accepted normalize output for lane %q: %w", lane.ID, encodeErr)
if debugErr := attemptEnvelope(payload, attemptErr); debugErr != nil {
return false, nil, fmt.Errorf("write normalize attempt debug artifact: %w", debugErr)
}
return false, nil, attemptErr
}
if debugErr := attemptEnvelope(payload, nil); debugErr != nil {
return false, nil, fmt.Errorf("write normalize attempt debug artifact: %w", debugErr)
}
@@ -358,17 +382,17 @@ func (r *Runner) validateTypedArtifact(ctx context.Context, codec artifactCodecE
target.llmProfile = binding.LLMProfile
result, err = item.typedValidate(validatorCtx, item.typed, target)
case ValidatorTargetSerialized:
artifact, encodeErr := serializeArtifact(codec, target.value, true)
artifact, encodeErr := validationCandidateArtifact(codec, target)
if encodeErr != nil {
err = encodeErr
break
}
result, err = item.serialized.Validate(validatorCtx, contracts.SerializedValidationRequest{Stage: string(target.stage), LaneID: target.laneID, ModuleKey: target.moduleKey, Source: target.source, SourceID: target.sourceID, SourceInput: target.sourceInput.Clone(), SessionID: target.sessionID, References: CloneReferenceSet(target.references), LLMProfile: binding.LLMProfile, Metadata: cloneMetadata(target.metadata), Chunk: cloneSourceChunkPtr(target.chunk), Chunks: cloneSourceChunks(target.chunks), Schema: contracts.CloneArtifactSchema(artifact.Schema), MediaType: artifact.MediaType, Content: append([]byte(nil), artifact.Content...)})
result, err = item.serialized.Validate(validatorCtx, contracts.SerializedValidationRequest{Stage: string(target.stage), LaneID: target.laneID, ModuleKey: target.moduleKey, Source: target.source, SourceID: target.sourceID, SourceInput: target.sourceInput.Clone(), SessionID: target.sessionID, References: CloneReferenceSet(target.references), LLMProfile: binding.LLMProfile, Metadata: cloneMetadata(target.metadata), Chunk: cloneSourceChunkPtr(target.chunk), Chunks: cloneSourceChunks(target.chunks), Schema: contracts.CloneArtifactSchema(artifact.Artifact.Schema), MediaType: artifact.Artifact.MediaType, Content: append([]byte(nil), artifact.Artifact.Content...)})
default:
return nil, nil, fmt.Errorf("validator %q is incompatible with typed artifact validation", binding.Module)
}
artifact, _ := serializeArtifact(codec, target.value, true)
debugCall := debugValidationCall{ValidatorName: binding.Module, Request: map[string]any{"stage": string(target.stage), "lane_id": target.laneID, "module_key": target.moduleKey, "source_id": target.sourceID, "artifact": debugCheckpointArtifact(CheckpointArtifact{Artifact: artifact, SchemaDigest: contracts.DigestArtifactSchema(artifact.Schema)}), "metadata": redactSensitiveMap(target.metadata)}, Result: debugValidationResultEnvelope(result)}
artifact, _ := validationCandidateArtifact(codec, target)
debugCall := debugValidationCall{ValidatorName: binding.Module, Request: map[string]any{"stage": string(target.stage), "lane_id": target.laneID, "module_key": target.moduleKey, "source_id": target.sourceID, "artifact": debugCheckpointArtifact(artifact), "metadata": redactSensitiveMap(target.metadata)}, Result: debugValidationResultEnvelope(result)}
if err != nil {
debugCall.Error = err.Error()
}
@@ -403,3 +427,10 @@ func (r *Runner) validateTypedArtifact(ctx context.Context, codec artifactCodecE
}
return warnings, nil, nil
}
func validationCandidateArtifact(codec artifactCodecEntry, target typedValidationTarget) (CheckpointArtifact, error) {
if target.candidate != nil {
return cloneCheckpointArtifact(*target.candidate), nil
}
return serializeCandidateArtifact(codec, target.laneID, target.moduleKey, target.sourceID, target.value)
}

View File

@@ -41,6 +41,7 @@ type typedValidationTarget struct {
chunks []source.Chunk
ref source.SourceRef
value any
candidate *CheckpointArtifact
}
func exactTypedValue[T any](operation string, value any) (T, error) {