Unify previous source resolution

This commit is contained in:
2026-08-10 21:20:59 +00:00
parent d9fa1d9328
commit b39b68add7
20 changed files with 406 additions and 76 deletions

View File

@@ -4,7 +4,6 @@ import (
"context"
"errors"
"fmt"
"path"
"path/filepath"
"sort"
"strings"
@@ -40,9 +39,13 @@ type Record struct {
Size int64
Generation string
S3Bucket string
VerifiedContent []byte
}
func BuildPlan(
// Resolve evaluates previous-session requirements against the one selected
// remote current state. Optional absence is represented in SkippedMissing;
// required absence returns an error.
func Resolve(
ctx context.Context,
cfg *config.Config,
paths artifacts.SessionPaths,
@@ -137,45 +140,44 @@ func BuildPlan(
manifestRecord.SHA256 = committedManifest.SHA256
manifestRecord.Size = committedManifest.Size
manifestRecord.Generation = committedManifest.Generation
manifestRecord.VerifiedContent = append([]byte(nil), current.ManifestData...)
}
result.Records = append(result.Records, manifestRecord)
for _, requirement := range orderedRequirements {
candidates := artifactRelativePathCandidates(requirement.Name, previousManifest, cfg)
if len(candidates) == 0 {
if requirement.Required {
return nil, fmt.Errorf(
"required previous-session artifact %q is unavailable in previous-session manifest/published-state",
requirement.Name,
)
}
result.SkippedMissing = append(result.SkippedMissing, requirement.Name)
continue
}
selectedRel := ""
selectedKey := ""
var selectedArtifact *artifacts.RemoteArtifact
for _, candidate := range candidates {
if current.Commit != nil {
artifact, ok := committedPublishedArtifact(current.Commit, previousSessionPrefix, candidate)
if !ok {
continue
if current.Commit != nil {
sourceID := previousArtifactSourceID(requirement.Name)
artifact, ok := committedPublishedArtifactBySource(current.Commit, sourceID)
if ok {
selectedRel, ok = sourceRelativePath(sourceID, requirement.Name, previousManifest, cfg)
if ok {
selectedKey = artifact.DestinationKey
selectedArtifact = &artifact
}
selectedRel = candidate
selectedKey = artifact.DestinationKey
selectedArtifact = &artifact
break
}
remoteKey := artifacts.S3PublishedOutputKey(previousSessionPrefix, candidate)
exists, err := store.Exists(ctx, remoteKey)
if err != nil {
return nil, fmt.Errorf("check previous-session artifact object %q: %w", remoteKey, err)
} else {
// Legacy current state predates immutable source-to-destination
// mappings. Keep its compatibility candidates isolated here.
matchedCandidates := make([]string, 0, 1)
for _, candidate := range legacyArtifactRelativePathCandidates(requirement.Name, previousManifest, cfg) {
remoteKey := artifacts.S3PublishedOutputKey(previousSessionPrefix, candidate)
exists, err := store.Exists(ctx, remoteKey)
if err != nil {
return nil, fmt.Errorf("check previous-session artifact object %q: %w", remoteKey, err)
}
if exists {
matchedCandidates = append(matchedCandidates, candidate)
if len(matchedCandidates) == 1 {
selectedRel = candidate
selectedKey = remoteKey
}
}
}
if exists {
selectedRel = candidate
selectedKey = remoteKey
break
if len(matchedCandidates) > 1 {
return nil, fmt.Errorf("legacy previous-session artifact %q matches multiple published candidates: %s", requirement.Name, strings.Join(matchedCandidates, ", "))
}
}
if selectedRel == "" {
@@ -224,13 +226,27 @@ func BuildPlan(
return result, nil
}
func committedPublishedArtifact(commit *artifacts.RemoteCommitManifest, sessionPrefix, relativePath string) (artifacts.RemoteArtifact, bool) {
// BuildPlan is retained for callers that need the historical name.
func BuildPlan(
ctx context.Context,
cfg *config.Config,
paths artifacts.SessionPaths,
requirements []artifacts.PreviousArtifactRequirement,
store storage.ObjectStore,
) (*Plan, error) {
return Resolve(ctx, cfg, paths, requirements, store)
}
func committedPublishedArtifactBySource(commit *artifacts.RemoteCommitManifest, source string) (artifacts.RemoteArtifact, bool) {
if commit == nil {
return artifacts.RemoteArtifact{}, false
}
want := artifacts.S3RunRelativeDestinationKey(artifacts.S3RunPrefix(sessionPrefix, commit.RunID), relativePath)
want := strings.TrimSpace(source)
if want == "" {
return artifacts.RemoteArtifact{}, false
}
for _, artifact := range commit.Artifacts {
if artifact.Type == artifacts.RemoteArtifactTypePublishedOutput && artifact.DestinationKey == want {
if artifact.Type == artifacts.RemoteArtifactTypePublishedOutput && artifact.Source == want {
return artifact, true
}
}
@@ -260,7 +276,30 @@ func optionalPreviousArtifactNames(requirements []artifacts.PreviousArtifactRequ
return names
}
func artifactRelativePathCandidates(
func previousArtifactSourceID(artifactName string) string {
sourceID := artifactpolicy.ConfiguredSourceID(artifactName)
if sourceDescriptor, err := artifactpolicy.PreviousSessionSourceDescriptorForConfiguredKey(artifactName); err == nil {
sourceID = sourceDescriptor.ConfiguredSourceID
}
return sourceID
}
func sourceRelativePath(sourceID, artifactName string, previousManifest *manifest.Manifest, cfg *config.Config) (string, bool) {
if rel, ok := manifestArtifactRelativePathBySourceID(previousManifest, sourceID); ok {
return rel, true
}
if cfg != nil && cfg.Pipeline != nil && cfg.Pipeline.Scriptorium != nil {
if artifactCfg, ok := cfg.Pipeline.Scriptorium.Artifacts[artifactName]; ok {
normalized, err := pathsafe.NormalizeRelativeDestination(artifactCfg.OutputPath)
if err == nil {
return normalized, true
}
}
}
return "", false
}
func legacyArtifactRelativePathCandidates(
artifactName string,
previousManifest *manifest.Manifest,
cfg *config.Config,
@@ -274,16 +313,12 @@ func artifactRelativePathCandidates(
candidates = append(candidates, normalized)
}
sourceDescriptor, err := artifactpolicy.PreviousSessionSourceDescriptorForConfiguredKey(artifactName)
sourceID := artifactpolicy.ConfiguredSourceID(artifactName)
if err == nil {
sourceID = sourceDescriptor.ConfiguredSourceID
}
sourceID := previousArtifactSourceID(artifactName)
if rel, ok := manifestArtifactRelativePathBySourceID(previousManifest, sourceID); ok {
appendCandidate(rel)
base := path.Base(rel)
base := filepath.Base(rel)
for _, published := range manifestPublishedPaths(previousManifest) {
if path.Base(published) == base {
if filepath.Base(published) == base {
appendCandidate(published)
}
}

View File

@@ -2,6 +2,8 @@ package previouscache
import (
"context"
"crypto/sha256"
"encoding/hex"
"encoding/json"
"path/filepath"
"strings"
@@ -58,6 +60,76 @@ func TestBuildPlanFallsBackToConfiguredOutputPath(t *testing.T) {
}
}
func TestResolveUsesCommittedSourceMappingForCustomDestination(t *testing.T) {
cfg, paths := previousCacheTestConfig(t)
store := &storage.FakeBackend{}
previous := previousManifestWithOutput(t, cfg, "artifacts/session_recap.md", nil)
customKey := seedCommittedPreviousState(t, store, cfg, previous, []artifacts.RemoteArtifact{
committedPublishedArtifactForTest("narratio.artifact.session_recap", "history/recap-v2.txt", []byte("# recap\n"), "recap-generation"),
committedPublishedArtifactForTest("narratio.artifact.player_handout", "other/recap-v2.txt", []byte("handout"), "handout-generation"),
})
plan, err := Resolve(context.Background(), cfg, paths, []artifacts.PreviousArtifactRequirement{{Name: "session_recap", Required: true}}, store)
if err != nil {
t.Fatalf("Resolve() error = %v", err)
}
if len(plan.Records) != 2 {
t.Fatalf("records = %#v, want manifest and artifact", plan.Records)
}
if manifestRecord := findRecord(plan.Records, InputKindManifest); manifestRecord == nil || len(manifestRecord.VerifiedContent) == 0 {
t.Fatalf("manifest record = %#v, want retained verified content", manifestRecord)
}
artifact := findRecord(plan.Records, InputKindArtifact)
if artifact == nil {
t.Fatal("artifact record is missing")
}
if artifact.RemoteKey != customKey {
t.Fatalf("remote key = %q, want exact configured source destination %q", artifact.RemoteKey, customKey)
}
if artifact.LocalRelativePath != "previous/artifacts/session_recap.md" {
t.Fatalf("local relative path = %q, want source-local cache path", artifact.LocalRelativePath)
}
}
func TestResolveDoesNotUseCommittedBasenameFallback(t *testing.T) {
cfg, paths := previousCacheTestConfig(t)
store := &storage.FakeBackend{}
previous := previousManifestWithOutput(t, cfg, "artifacts/session_recap.md", nil)
seedCommittedPreviousState(t, store, cfg, previous, []artifacts.RemoteArtifact{
committedPublishedArtifactForTest("narratio.artifact.player_handout", "history/session_recap.md", []byte("wrong source"), "handout-generation"),
})
_, err := Resolve(context.Background(), cfg, paths, []artifacts.PreviousArtifactRequirement{{Name: "session_recap", Required: true}}, store)
if err == nil || !strings.Contains(err.Error(), `required previous-session artifact "session_recap" object missing`) {
t.Fatalf("Resolve() error = %v, want source-mapping absence", err)
}
}
func TestResolveRejectsAmbiguousLegacyCandidates(t *testing.T) {
cfg, paths := previousCacheTestConfig(t)
store := &storage.FakeBackend{}
seedPreviousCurrent(t, store, cfg, previousManifestWithOutput(t, cfg, "artifacts/session_recap.md", []string{"archive/session_recap.md"}))
prefix := artifacts.S3SessionPrefix("dnd", cfg.Session.Campaign, cfg.Session.PreviousSessionID)
store.SeedObject(storage.FakeObject{Key: prefix + "artifacts/session_recap.md", Data: []byte("source candidate")})
store.SeedObject(storage.FakeObject{Key: prefix + "archive/session_recap.md", Data: []byte("published candidate")})
_, err := Resolve(context.Background(), cfg, paths, []artifacts.PreviousArtifactRequirement{{Name: "session_recap", Required: true}}, store)
if err == nil || !strings.Contains(err.Error(), "matches multiple published candidates") {
t.Fatalf("Resolve() error = %v, want legacy ambiguity error", err)
}
}
func TestResolveHonorsCancellation(t *testing.T) {
cfg, paths := previousCacheTestConfig(t)
ctx, cancel := context.WithCancel(context.Background())
cancel()
_, err := Resolve(ctx, cfg, paths, []artifacts.PreviousArtifactRequirement{{Name: "session_recap", Required: true}}, &storage.FakeBackend{})
if err == nil || !strings.Contains(err.Error(), "context canceled") {
t.Fatalf("Resolve() error = %v, want cancellation", err)
}
}
func TestBuildPlanSkipsMissingOptionalPreviousArtifact(t *testing.T) {
cfg, paths := previousCacheTestConfig(t)
store := &storage.FakeBackend{}
@@ -238,3 +310,91 @@ func recordRelPaths(records []Record) []string {
}
return out
}
func findRecord(records []Record, kind string) *Record {
for index := range records {
if records[index].Kind == kind {
return &records[index]
}
}
return nil
}
func seedCommittedPreviousState(t *testing.T, store *storage.FakeBackend, cfg *config.Config, m *manifest.Manifest, published []artifacts.RemoteArtifact) string {
t.Helper()
prefix := artifacts.S3SessionPrefix(cfg.Pipeline.Storage.S3.RootPrefix, cfg.Session.Campaign, cfg.Session.PreviousSessionID)
runID := "previous-run"
manifestKey := artifacts.S3RunSessionManifestKey(prefix, runID)
manifestData, err := marshalManifestForPreviousCacheTest(m)
if err != nil {
t.Fatalf("marshal previous manifest: %v", err)
}
manifestGeneration := "manifest-generation"
commit := artifacts.RemoteCommitManifest{
FormatVersion: artifacts.RemoteCommitFormatVersion,
Campaign: cfg.Session.Campaign,
SessionID: cfg.Session.PreviousSessionID,
RunID: runID,
Artifacts: append([]artifacts.RemoteArtifact{{
Type: artifacts.RemoteArtifactTypeSessionManifest,
Source: "session.manifest",
DestinationKey: manifestKey,
SHA256: sha256ForPreviousCacheTest(manifestData),
Size: int64(len(manifestData)),
Generation: manifestGeneration,
}}, published...),
}
commitData, err := artifacts.EncodeRemoteCommitManifest(commit)
if err != nil {
t.Fatalf("encode committed previous state: %v", err)
}
commitKey := artifacts.S3RunCommitKey(prefix, runID)
commitGeneration := "commit-generation"
pointerData, err := artifacts.EncodeCurrentCommitPointer(artifacts.CurrentCommitPointer{
FormatVersion: artifacts.RemoteCommitFormatVersion,
Campaign: cfg.Session.Campaign,
SessionID: cfg.Session.PreviousSessionID,
RunID: runID,
CommitKey: commitKey,
CommitSHA256: sha256ForPreviousCacheTest(commitData),
CommitSize: int64(len(commitData)),
CommitGeneration: commitGeneration,
})
if err != nil {
t.Fatalf("encode committed previous pointer: %v", err)
}
store.SeedObject(storage.FakeObject{Key: artifacts.S3CurrentCommitPointerKey(prefix), Data: pointerData, ETag: "pointer-generation"})
store.SeedObject(storage.FakeObject{Key: commitKey, Data: commitData, ETag: commitGeneration})
store.SeedObject(storage.FakeObject{Key: manifestKey, Data: manifestData, ETag: manifestGeneration})
for _, artifact := range published {
store.SeedObject(storage.FakeObject{Key: artifact.DestinationKey, Data: publishedDataForTest(artifact.Source), ETag: artifact.Generation})
}
return published[0].DestinationKey
}
func committedPublishedArtifactForTest(source, destination string, data []byte, generation string) artifacts.RemoteArtifact {
return artifacts.RemoteArtifact{
Type: artifacts.RemoteArtifactTypePublishedOutput,
Source: source,
DestinationKey: "dnd/campaigns/sample-campaign/sessions/2026-04-26/runs/previous-run/published/" + destination,
SHA256: sha256ForPreviousCacheTest(data),
Size: int64(len(data)),
Generation: generation,
}
}
func publishedDataForTest(source string) []byte {
if source == "narratio.artifact.session_recap" {
return []byte("# recap\n")
}
if source == "narratio.artifact.player_handout" {
return []byte("handout")
}
return []byte("published")
}
func sha256ForPreviousCacheTest(data []byte) string {
sum := sha256.Sum256(data)
return hex.EncodeToString(sum[:])
}