narratio/internal/config/validate_notarius.go

package config

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

	"gitea.maximumdirect.net/eric/narratio/internal/artifactmodel"
	"gitea.maximumdirect.net/eric/narratio/internal/artifactpolicy"
	"gitea.maximumdirect.net/eric/narratio/internal/notariusref"
)

func validateNotarius(cfg *NotariusConfig, scriptorium *ScriptoriumConfig) error {
	if cfg == nil || !cfg.Enabled {
		return nil
	}

	binary := strings.TrimSpace(cfg.Binary)
	configPath := filepath.Clean(cfg.ConfigPath)
	pipelineID := strings.TrimSpace(cfg.PipelineID)
	timeout := strings.TrimSpace(cfg.Timeout)
	workingDirectory := filepath.Clean(cfg.WorkingDirectory)
	if binary == "" {
		return fmt.Errorf("pipeline.notarius.binary is required when pipeline.notarius.enabled is true")
	}
	if strings.TrimSpace(cfg.ConfigPath) == "" {
		return fmt.Errorf("pipeline.notarius.config_path is required when pipeline.notarius.enabled is true")
	}
	if pipelineID == "" {
		return fmt.Errorf("pipeline.notarius.pipeline_id is required when pipeline.notarius.enabled is true")
	}
	if timeout == "" {
		return fmt.Errorf("pipeline.notarius.timeout must be a valid duration")
	}
	if err := validateDuration("pipeline.notarius.timeout", timeout); err != nil {
		return err
	}
	if strings.TrimSpace(cfg.WorkingDirectory) == "" {
		return fmt.Errorf("pipeline.notarius.working_directory is required when pipeline.notarius.enabled is true")
	}
	if len(cfg.Outputs) == 0 {
		return fmt.Errorf("pipeline.notarius.outputs must contain at least one output when pipeline.notarius.enabled is true")
	}
	if len(cfg.References) > MaxNotariusReferenceBindings {
		return fmt.Errorf("pipeline.notarius.references must contain at most %d bindings", MaxNotariusReferenceBindings)
	}

	references, err := normalizeNotariusReferences(cfg.References)
	if err != nil {
		return err
	}
	outputs, err := normalizeNotariusOutputs(cfg.Outputs, scriptorium)
	if err != nil {
		return err
	}

	cfg.Binary = binary
	cfg.ConfigPath = configPath
	cfg.PipelineID = pipelineID
	cfg.Timeout = timeout
	cfg.WorkingDirectory = workingDirectory
	cfg.References = references
	cfg.Outputs = outputs
	return nil
}

func normalizeNotariusReferences(references map[string]string) (map[string]string, error) {
	selectors := make([]string, 0, len(references))
	for selector := range references {
		selectors = append(selectors, selector)
	}
	sort.Strings(selectors)

	var normalized map[string]string
	if references != nil {
		normalized = make(map[string]string, len(references))
	}
	owners := make(map[string]string, len(references))
	for _, rawSelector := range selectors {
		selector, err := notariusref.NormalizeSelector(rawSelector)
		if err != nil {
			return nil, fmt.Errorf("pipeline.notarius.references selector %q is invalid: %w", rawSelector, err)
		}
		if previous, ok := owners[selector]; ok {
			return nil, fmt.Errorf("pipeline.notarius.references selectors %q and %q normalize to %q", previous, rawSelector, selector)
		}
		owners[selector] = rawSelector

		source := strings.TrimSpace(references[rawSelector])
		if source == "" {
			return nil, fmt.Errorf("pipeline.notarius.references.%s source is required", selector)
		}
		if _, ok := artifactpolicy.DescribePreparedInputSource(source); !ok {
			return nil, fmt.Errorf("pipeline.notarius.references.%s source %q is not a supported prepared input source", selector, source)
		}
		normalized[selector] = source
	}
	return normalized, nil
}

func normalizeNotariusOutputs(outputs map[string]NotariusOutputConfig, scriptorium *ScriptoriumConfig) (map[string]NotariusOutputConfig, error) {
	reservedSources := notariusReservedSources(scriptorium)
	keys := make([]string, 0, len(outputs))
	for key := range outputs {
		keys = append(keys, key)
	}
	sort.Strings(keys)

	normalized := make(map[string]NotariusOutputConfig, len(outputs))
	sourceOwners := make(map[string]string, len(outputs))
	laneOwners := make(map[string]string, len(outputs))
	for _, rawKey := range keys {
		output := outputs[rawKey]
		key := strings.TrimSpace(rawKey)
		if !artifactpolicy.IsConfiguredKey(key) {
			return nil, fmt.Errorf("pipeline.notarius.outputs keys must match ^[a-z][a-z0-9_]*$")
		}
		sourceID := artifactpolicy.ExtractionSourceID(key)
		if previousKey, ok := sourceOwners[sourceID]; ok {
			return nil, fmt.Errorf("pipeline.notarius.outputs keys %q and %q normalize to duplicate source %q", previousKey, rawKey, sourceID)
		}
		if owner, ok := reservedSources[sourceID]; ok {
			return nil, fmt.Errorf("pipeline.notarius.outputs.%s source %q collides with %s", key, sourceID, owner)
		}
		sourceOwners[sourceID] = rawKey

		output.LaneID = strings.TrimSpace(output.LaneID)
		output.MediaType = strings.TrimSpace(output.MediaType)
		output.SchemaID = strings.TrimSpace(output.SchemaID)
		output.SchemaVersion = strings.TrimSpace(output.SchemaVersion)
		output.ModuleKey = strings.TrimSpace(output.ModuleKey)
		prefix := "pipeline.notarius.outputs." + key
		if output.LaneID == "" {
			return nil, fmt.Errorf("%s.lane_id is required", prefix)
		}
		if previousKey, ok := laneOwners[output.LaneID]; ok {
			return nil, fmt.Errorf("%s.lane_id %q duplicates pipeline.notarius.outputs.%s.lane_id", prefix, output.LaneID, previousKey)
		}
		laneOwners[output.LaneID] = key
		if output.MediaType == "" {
			return nil, fmt.Errorf("%s.media_type is required", prefix)
		}
		if output.SchemaID == "" {
			return nil, fmt.Errorf("%s.schema_id is required", prefix)
		}
		if output.SchemaVersion == "" {
			return nil, fmt.Errorf("%s.schema_version is required", prefix)
		}
		normalized[key] = output
	}
	return normalized, nil
}

func notariusReservedSources(scriptorium *ScriptoriumConfig) map[string]string {
	reserved := make(map[string]string)
	for _, spec := range artifactmodel.RuntimeTranscriptArtifacts() {
		reserved[spec.SourceID] = "built-in source"
	}
	reserved[artifactpolicy.SourceBoundsSession] = "built-in source"
	if scriptorium != nil {
		for key := range scriptorium.Artifacts {
			normalizedKey := strings.TrimSpace(key)
			reserved[artifactpolicy.ConfiguredSourceID(normalizedKey)] = "configured Scriptorium source"
			reserved[artifactpolicy.PreviousSessionSourceID(normalizedKey)] = "previous-session source"
		}
	}
	return reserved
}

func notariusOutputKeySet(notarius *NotariusConfig) map[string]struct{} {
	out := map[string]struct{}{}
	if notarius == nil {
		return out
	}
	for key := range notarius.Outputs {
		trimmed := strings.TrimSpace(key)
		if artifactpolicy.IsConfiguredKey(trimmed) {
			out[trimmed] = struct{}{}
		}
	}
	return out
}