narratio/internal/previouscache/previouscache.go

package previouscache

import (
	"context"
	"errors"
	"fmt"
	"path/filepath"
	"sort"
	"strings"

	"gitea.maximumdirect.net/eric/narratio/internal/adapters/storage"
	"gitea.maximumdirect.net/eric/narratio/internal/artifactpolicy"
	"gitea.maximumdirect.net/eric/narratio/internal/artifacts"
	"gitea.maximumdirect.net/eric/narratio/internal/config"
	"gitea.maximumdirect.net/eric/narratio/internal/pathsafe"
)

const (
	InputKindManifest = "previous_manifest"
	InputKindArtifact = "previous_artifact"
	InputSource       = "previous_session_publish.current"
)

type Plan struct {
	Records        []Record
	SkippedMissing []string
	PreviousRunID  string
}

type Record struct {
	Kind              string
	RequirementName   string
	Required          bool
	LocalRelativePath string
	LocalPath         string
	RemoteKey         string
	SHA256            string
	Size              int64
	Generation        string
	S3Bucket          string
	VerifiedContent   []byte
}

// 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,
	requirements []artifacts.PreviousArtifactRequirement,
	store storage.ObjectStore,
) (*Plan, error) {
	if len(requirements) == 0 {
		return &Plan{}, nil
	}
	if cfg == nil || cfg.Session == nil || cfg.Pipeline == nil {
		return nil, fmt.Errorf("resolved config with session/pipeline is required")
	}

	orderedRequirements := append([]artifacts.PreviousArtifactRequirement(nil), requirements...)
	sort.Slice(orderedRequirements, func(i, j int) bool {
		return orderedRequirements[i].Name < orderedRequirements[j].Name
	})

	requiredNames := requiredPreviousArtifactNames(orderedRequirements)
	optionalNames := optionalPreviousArtifactNames(orderedRequirements)
	previousSessionID := strings.TrimSpace(cfg.Session.PreviousSessionID)
	if previousSessionID == "" {
		if len(requiredNames) > 0 {
			return nil, fmt.Errorf(
				"previous_session_id is required for required previous-session artifacts: %s",
				strings.Join(requiredNames, ", "),
			)
		}
		return &Plan{SkippedMissing: optionalNames}, nil
	}

	if store == nil {
		return nil, fmt.Errorf("previous-session artifact hydration requires object store backend")
	}
	if cfg.Pipeline.Storage.S3 == nil {
		return nil, fmt.Errorf("pipeline.storage.s3 configuration is required for previous-session artifact hydration")
	}

	campaign := strings.TrimSpace(cfg.Session.Campaign)
	if campaign == "" {
		return nil, fmt.Errorf("session campaign is required for previous-session artifact hydration")
	}
	bucket := strings.TrimSpace(cfg.Pipeline.Storage.S3.Bucket)
	if bucket == "" {
		return nil, fmt.Errorf("pipeline.storage.s3.bucket is required for previous-session artifact hydration")
	}

	previousSessionPrefix := artifacts.S3SessionPrefix(
		cfg.Pipeline.Storage.S3.RootPrefix,
		campaign,
		previousSessionID,
	)

	result := &Plan{}
	current, err := artifacts.LoadCurrentState(ctx, store, previousSessionPrefix, artifacts.CurrentStateValidation{
		ExpectedSessionID: previousSessionID,
		ExpectedCampaign:  campaign,
		ValidateRunID:     true,
	})
	if err != nil {
		var pointerMissing *artifacts.CurrentCommitPointerMissingError
		if errors.As(err, &pointerMissing) {
			if len(requiredNames) > 0 {
				return nil, fmt.Errorf("required previous-session artifacts unavailable: remote %w", err)
			}
			result.SkippedMissing = optionalNames
			return result, nil
		}
		return nil, fmt.Errorf("load previous-session current state: %w", err)
	}
	result.PreviousRunID = strings.TrimSpace(current.RunID)
	currentManifestKey := current.CurrentManifestKey
	if current.Commit == nil {
		return nil, fmt.Errorf("previous-session current state does not include a committed remote declaration")
	}

	manifestRel, err := relativeToSession(paths, paths.PreviousManifestPath)
	if err != nil {
		return nil, err
	}
	manifestRecord := Record{
		Kind:              InputKindManifest,
		LocalRelativePath: manifestRel,
		LocalPath:         paths.PreviousManifestPath,
		RemoteKey:         currentManifestKey,
		S3Bucket:          bucket,
	}
	committedManifest, ok := current.Commit.Artifact(artifacts.RemoteArtifactTypeSessionManifest)
	if !ok || committedManifest.DestinationKey != currentManifestKey {
		return nil, fmt.Errorf("previous-session remote commit does not declare its session manifest")
	}
	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 {
		sourceID := previousArtifactSourceID(requirement.Name)
		selectedArtifact, ok := committedPublishedArtifactBySource(current.Commit, sourceID)
		if !ok {
			if requirement.Required {
				return nil, fmt.Errorf(
					"required previous-session artifact %q is not declared by the selected remote commit",
					requirement.Name,
				)
			}
			result.SkippedMissing = append(result.SkippedMissing, requirement.Name)
			continue
		}
		selectedRel, err := publishedOutputRelativePath(previousSessionPrefix, current.RunID, selectedArtifact.DestinationKey)
		if err != nil {
			return nil, fmt.Errorf("resolve declared previous-session artifact %q: %w", requirement.Name, err)
		}

		localPath, err := artifacts.SessionPreviousArtifactPath(paths, selectedRel)
		if err != nil {
			return nil, fmt.Errorf("resolve previous-session artifact path: %w", err)
		}
		localRel, err := relativeToSession(paths, localPath)
		if err != nil {
			return nil, err
		}
		record := Record{
			Kind:              InputKindArtifact,
			RequirementName:   requirement.Name,
			Required:          requirement.Required,
			LocalRelativePath: localRel,
			LocalPath:         localPath,
			RemoteKey:         selectedArtifact.DestinationKey,
			S3Bucket:          bucket,
			SHA256:            selectedArtifact.SHA256,
			Size:              selectedArtifact.Size,
			Generation:        selectedArtifact.Generation,
		}
		result.Records = append(result.Records, record)
	}

	sort.Strings(result.SkippedMissing)
	sort.Slice(result.Records, func(i, j int) bool {
		if result.Records[i].LocalRelativePath != result.Records[j].LocalRelativePath {
			return result.Records[i].LocalRelativePath < result.Records[j].LocalRelativePath
		}
		return result.Records[i].RemoteKey < result.Records[j].RemoteKey
	})
	return result, nil
}

func committedPublishedArtifactBySource(commit *artifacts.RemoteCommitManifest, source string) (artifacts.RemoteArtifact, bool) {
	if commit == nil {
		return artifacts.RemoteArtifact{}, false
	}
	want := strings.TrimSpace(source)
	if want == "" {
		return artifacts.RemoteArtifact{}, false
	}
	for _, artifact := range commit.Artifacts {
		if artifact.Type == artifacts.RemoteArtifactTypePublishedOutput && artifact.Source == want {
			return artifact, true
		}
	}
	return artifacts.RemoteArtifact{}, false
}

func publishedOutputRelativePath(sessionPrefix, runID, destinationKey string) (string, error) {
	prefix := artifacts.S3RunRelativeDestinationKey(artifacts.S3RunPrefix(sessionPrefix, runID), "published") + "/"
	rel, ok := strings.CutPrefix(strings.TrimSpace(destinationKey), prefix)
	if !ok || strings.TrimSpace(rel) == "" {
		return "", fmt.Errorf("destination %q is outside selected run published-output prefix %q", destinationKey, prefix)
	}
	normalized, err := pathsafe.NormalizeRelativeDestination(rel)
	if err != nil {
		return "", fmt.Errorf("published output destination %q: %w", destinationKey, err)
	}
	return normalized, nil
}

func requiredPreviousArtifactNames(requirements []artifacts.PreviousArtifactRequirement) []string {
	names := make([]string, 0, len(requirements))
	for _, requirement := range requirements {
		if requirement.Required {
			names = append(names, strings.TrimSpace(requirement.Name))
		}
	}
	sort.Strings(names)
	return names
}

func optionalPreviousArtifactNames(requirements []artifacts.PreviousArtifactRequirement) []string {
	names := make([]string, 0, len(requirements))
	for _, requirement := range requirements {
		if requirement.Required {
			continue
		}
		names = append(names, strings.TrimSpace(requirement.Name))
	}
	sort.Strings(names)
	return names
}

func previousArtifactSourceID(artifactName string) string {
	sourceID := artifactpolicy.ConfiguredSourceID(artifactName)
	if sourceDescriptor, err := artifactpolicy.PreviousSessionSourceDescriptorForConfiguredKey(artifactName); err == nil {
		sourceID = sourceDescriptor.ConfiguredSourceID
	}
	return sourceID
}

func relativeToSession(paths artifacts.SessionPaths, localPath string) (string, error) {
	root := filepath.Clean(paths.Root)
	if strings.TrimSpace(root) == "" {
		return "", fmt.Errorf("session root is required")
	}
	normalized, err := pathsafe.SlashRelativeFromRoot(root, localPath)
	if err != nil {
		return "", fmt.Errorf("resolve previous-cache relative path: %w", err)
	}
	return normalized, nil
}