Files
narratio/internal/manifest/analyze_state.go

375 lines
13 KiB
Go

package manifest
import (
"encoding/hex"
"fmt"
"sort"
"strings"
"time"
"gitea.maximumdirect.net/eric/narratio/internal/artifactpolicy"
"gitea.maximumdirect.net/eric/narratio/internal/pathsafe"
)
const (
// AnalyzeStateContractVersion identifies the supported per-artifact state
// representation owned by the analyze stage.
AnalyzeStateContractVersion = 1
// AnalyzeFingerprintContractVersion identifies the fingerprint representation
// stored by the supported analyze state contract.
AnalyzeFingerprintContractVersion = 1
)
const (
maxAnalyzeArtifactErrorLength = 512
maxAnalyzeArtifactTextLength = 4096
maxAnalyzeArtifactListEntries = 128
)
// AnalyzeArtifactStatus describes whether one configured analysis artifact is
// currently available or why it is not.
type AnalyzeArtifactStatus string
const (
AnalyzeArtifactCurrent AnalyzeArtifactStatus = "current"
AnalyzeArtifactStale AnalyzeArtifactStatus = "stale"
AnalyzeArtifactMissing AnalyzeArtifactStatus = "missing"
AnalyzeArtifactFailed AnalyzeArtifactStatus = "failed"
AnalyzeArtifactUnselected AnalyzeArtifactStatus = "unselected"
)
// AnalyzeArtifactProvenance records useful non-secret Scriptorium invocation
// identity without making adapter diagnostics part of the generic artifact schema.
type AnalyzeArtifactProvenance struct {
PromptID string `json:"prompt_id,omitempty"`
ProfileID string `json:"profile_id,omitempty"`
CommandMode string `json:"command_mode,omitempty"`
}
// AnalyzeArtifactRecord is analyze-owned state for one configured artifact.
// Output is present only while the record is current.
type AnalyzeArtifactRecord struct {
Key string `json:"key"`
Status AnalyzeArtifactStatus `json:"status"`
FingerprintVersion int `json:"fingerprint_version,omitempty"`
Fingerprint string `json:"fingerprint,omitempty"`
Dependencies []string `json:"dependencies,omitempty"`
Output *ArtifactRecord `json:"output,omitempty"`
OutputSize int64 `json:"output_size,omitempty"`
ProducerRunID string `json:"producer_run_id"`
UpdatedAt time.Time `json:"updated_at"`
Error string `json:"error,omitempty"`
Scriptorium *AnalyzeArtifactProvenance `json:"scriptorium,omitempty"`
Logs []string `json:"logs,omitempty"`
GeneratedConfigs []string `json:"generated_configs,omitempty"`
}
// HasVersionedAnalyzeState reports whether an analyze stage record carries the
// supported per-artifact authority. A legacy aggregate-only record returns false.
func (s *StageRecord) HasVersionedAnalyzeState() bool {
return s != nil && s.Name == "analyze" && s.AnalyzeStateVersion == AnalyzeStateContractVersion
}
// ValidateAnalyzeArtifactCollection validates one complete session or
// invocation collection independently of its containing manifest.
func ValidateAnalyzeArtifactCollection(version int, records map[string]AnalyzeArtifactRecord) error {
if version != AnalyzeStateContractVersion {
return fmt.Errorf("unsupported analyze state version %d", version)
}
for key, record := range records {
if err := validateAnalyzeArtifactRecord(key, record); err != nil {
return fmt.Errorf("analyze artifact %q: %w", key, err)
}
}
return nil
}
// CloneAnalyzeArtifactCollection returns a deep copy in canonical dependency
// order so projections cannot be mutated after application.
func CloneAnalyzeArtifactCollection(records map[string]AnalyzeArtifactRecord) map[string]AnalyzeArtifactRecord {
if records == nil {
return nil
}
cloned := make(map[string]AnalyzeArtifactRecord, len(records))
for key, record := range records {
record.Dependencies = cloneStrings(record.Dependencies)
record.Logs = cloneStrings(record.Logs)
record.GeneratedConfigs = cloneStrings(record.GeneratedConfigs)
record.Output = cloneArtifactRecord(record.Output)
record.Scriptorium = cloneAnalyzeProvenance(record.Scriptorium)
cloned[key] = record
}
normalizeAnalyzeArtifactCollection(cloned)
return cloned
}
func validateAnalyzeStageState(stageName string, version int, records map[string]AnalyzeArtifactRecord) error {
if stageName != "analyze" {
if version != 0 || records != nil {
return fmt.Errorf("stage %q cannot contain analyze-owned state", stageName)
}
return nil
}
if version == 0 {
if records != nil {
return fmt.Errorf("legacy analyze stage without a state version cannot contain analyze_artifacts")
}
return nil
}
return ValidateAnalyzeArtifactCollection(version, records)
}
func validateAnalyzeArtifactRecord(mapKey string, record AnalyzeArtifactRecord) error {
if !isNormalizedAnalyzeArtifactKey(mapKey) {
return fmt.Errorf("map key must match ^[a-z][a-z0-9_]*$ without normalization")
}
if record.Key != mapKey {
return fmt.Errorf("record key %q does not match map key", record.Key)
}
switch record.Status {
case AnalyzeArtifactCurrent, AnalyzeArtifactStale, AnalyzeArtifactMissing, AnalyzeArtifactFailed, AnalyzeArtifactUnselected:
default:
return fmt.Errorf("unsupported status %q", record.Status)
}
if err := validateAnalyzeDependencies(record.Dependencies); err != nil {
return err
}
if err := validateAnalyzeFingerprint(record.FingerprintVersion, record.Fingerprint, record.Status == AnalyzeArtifactCurrent); err != nil {
return err
}
if err := pathsafe.ValidateOpaqueSegment(record.ProducerRunID); err != nil {
return fmt.Errorf("producer_run_id is invalid: %w", err)
}
if record.UpdatedAt.IsZero() {
return fmt.Errorf("updated_at is required")
}
if len(record.Error) > maxAnalyzeArtifactErrorLength {
return fmt.Errorf("error exceeds %d bytes", maxAnalyzeArtifactErrorLength)
}
if strings.TrimSpace(record.Error) != record.Error {
return fmt.Errorf("error must be trimmed")
}
if record.Status == AnalyzeArtifactFailed {
if record.Error == "" {
return fmt.Errorf("failed status requires error")
}
} else if record.Error != "" {
return fmt.Errorf("status %q forbids error", record.Status)
}
if record.Status == AnalyzeArtifactCurrent {
if err := validateCurrentAnalyzeOutput(record); err != nil {
return err
}
} else if record.Output != nil || record.OutputSize != 0 {
return fmt.Errorf("status %q forbids output and output_size", record.Status)
}
if err := validateAnalyzeProvenance(record.Scriptorium); err != nil {
return err
}
if err := validateAnalyzeTextList("logs", record.Logs); err != nil {
return err
}
if err := validateAnalyzeTextList("generated_configs", record.GeneratedConfigs); err != nil {
return err
}
return nil
}
func validateCurrentAnalyzeOutput(record AnalyzeArtifactRecord) error {
if record.Output == nil {
return fmt.Errorf("current status requires output")
}
if record.OutputSize <= 0 {
return fmt.Errorf("current status requires positive output_size")
}
output := record.Output
if strings.TrimSpace(output.Kind) == "" {
return fmt.Errorf("current output kind is required")
}
if strings.TrimSpace(output.Kind) != output.Kind {
return fmt.Errorf("current output kind must be trimmed")
}
wantSource := artifactpolicy.ConfiguredSourceID(record.Key)
if output.SourceID != wantSource {
return fmt.Errorf("current output source_id %q must equal %q", output.SourceID, wantSource)
}
normalizedPath, err := pathsafe.NormalizeRelativeDestination(output.LocalPath)
if err != nil {
return fmt.Errorf("current output local_path is unsafe: %w", err)
}
if normalizedPath != output.LocalPath {
return fmt.Errorf("current output local_path %q is not canonical %q", output.LocalPath, normalizedPath)
}
if output.Contract == nil || strings.TrimSpace(output.Contract.MediaType) == "" || strings.TrimSpace(output.Contract.SchemaID) == "" || strings.TrimSpace(output.Contract.SchemaVersion) == "" {
return fmt.Errorf("current output contract media_type, schema_id, and schema_version are required")
}
for field, value := range map[string]string{
"media_type": output.Contract.MediaType, "schema_id": output.Contract.SchemaID,
"schema_version": output.Contract.SchemaVersion, "module_key": output.Contract.ModuleKey,
} {
if strings.TrimSpace(value) != value {
return fmt.Errorf("current output contract %s must be trimmed", field)
}
}
if err := validateSHA256("current output checksum", output.Checksum); err != nil {
return err
}
if output.ProducerRunID != "" && output.ProducerRunID != record.ProducerRunID {
return fmt.Errorf("current output producer_run_id %q does not match record", output.ProducerRunID)
}
for field, value := range map[string]string{
"output kind": output.Kind,
"output source_id": output.SourceID,
"output local_path": output.LocalPath,
"output contract media_type": output.Contract.MediaType,
"output contract schema_id": output.Contract.SchemaID,
"output contract schema_version": output.Contract.SchemaVersion,
"output contract module_key": output.Contract.ModuleKey,
} {
if err := validateAnalyzeText(field, value); err != nil {
return err
}
}
if output.ExternalProvenance != nil {
if strings.TrimSpace(output.ExternalProvenance.System) == "" {
return fmt.Errorf("output provenance system is required when provenance is present")
}
for field, value := range map[string]string{
"output provenance system": output.ExternalProvenance.System,
"output provenance run_id": output.ExternalProvenance.RunID,
"output provenance pipeline_id": output.ExternalProvenance.PipelineID,
"output provenance artifact_id": output.ExternalProvenance.ArtifactID,
} {
if err := validateAnalyzeText(field, value); err != nil {
return err
}
}
}
return nil
}
func validateAnalyzeDependencies(dependencies []string) error {
seen := make(map[string]struct{}, len(dependencies))
for index, dependency := range dependencies {
if !isNormalizedAnalyzeArtifactKey(dependency) {
return fmt.Errorf("dependencies[%d] must match ^[a-z][a-z0-9_]*$ without normalization", index)
}
if _, duplicate := seen[dependency]; duplicate {
return fmt.Errorf("duplicate dependency %q", dependency)
}
seen[dependency] = struct{}{}
}
return nil
}
func validateAnalyzeFingerprint(version int, fingerprint string, required bool) error {
if version == 0 && fingerprint == "" {
if required {
return fmt.Errorf("current status requires fingerprint version and fingerprint")
}
return nil
}
if version != AnalyzeFingerprintContractVersion {
return fmt.Errorf("unsupported fingerprint version %d", version)
}
return validateSHA256("fingerprint", fingerprint)
}
func validateSHA256(field, value string) error {
if len(value) != 64 || strings.ToLower(value) != value {
return fmt.Errorf("%s must be a canonical lowercase SHA-256 hex digest", field)
}
decoded, err := hex.DecodeString(value)
if err != nil || len(decoded) != 32 {
return fmt.Errorf("%s must be a canonical lowercase SHA-256 hex digest", field)
}
return nil
}
func validateAnalyzeProvenance(provenance *AnalyzeArtifactProvenance) error {
if provenance == nil {
return nil
}
if provenance.PromptID == "" && provenance.ProfileID == "" && provenance.CommandMode == "" {
return fmt.Errorf("scriptorium provenance must contain at least one identifier")
}
for field, value := range map[string]string{
"scriptorium prompt_id": provenance.PromptID,
"scriptorium profile_id": provenance.ProfileID,
"scriptorium command_mode": provenance.CommandMode,
} {
if err := validateAnalyzeText(field, value); err != nil {
return err
}
}
return nil
}
func validateAnalyzeTextList(field string, values []string) error {
if len(values) > maxAnalyzeArtifactListEntries {
return fmt.Errorf("%s exceeds %d entries", field, maxAnalyzeArtifactListEntries)
}
for index, value := range values {
if err := validateAnalyzeText(fmt.Sprintf("%s[%d]", field, index), value); err != nil {
return err
}
}
return nil
}
func validateAnalyzeText(field, value string) error {
if len(value) > maxAnalyzeArtifactTextLength {
return fmt.Errorf("%s exceeds %d bytes", field, maxAnalyzeArtifactTextLength)
}
if strings.ContainsRune(value, '\x00') {
return fmt.Errorf("%s contains a NUL byte", field)
}
return nil
}
func isNormalizedAnalyzeArtifactKey(key string) bool {
return strings.TrimSpace(key) == key && artifactpolicy.IsConfiguredKey(key)
}
func normalizeAnalyzeArtifactCollection(records map[string]AnalyzeArtifactRecord) {
for key, record := range records {
if len(record.Dependencies) > 1 {
record.Dependencies = append([]string(nil), record.Dependencies...)
sort.Strings(record.Dependencies)
}
record.Logs = cloneStrings(record.Logs)
record.GeneratedConfigs = cloneStrings(record.GeneratedConfigs)
record.Output = cloneArtifactRecord(record.Output)
record.Scriptorium = cloneAnalyzeProvenance(record.Scriptorium)
records[key] = record
}
}
func cloneStrings(values []string) []string {
return append([]string(nil), values...)
}
func cloneArtifactRecord(record *ArtifactRecord) *ArtifactRecord {
if record == nil {
return nil
}
clone := *record
if record.Contract != nil {
contract := *record.Contract
clone.Contract = &contract
}
if record.ExternalProvenance != nil {
provenance := *record.ExternalProvenance
clone.ExternalProvenance = &provenance
}
return &clone
}
func cloneAnalyzeProvenance(provenance *AnalyzeArtifactProvenance) *AnalyzeArtifactProvenance {
if provenance == nil {
return nil
}
clone := *provenance
return &clone
}