narratio/internal/config/validate_scriptorium.go

package config

import (
	"errors"
	"fmt"
	"sort"
	"strings"

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

func validateScriptorium(cfg *ScriptoriumConfig, notarius *NotariusConfig) error {
	if cfg == nil {
		return nil
	}
	if strings.TrimSpace(cfg.Binary) == "" {
		return fmt.Errorf("pipeline.scriptorium.binary is required when pipeline.scriptorium is configured")
	}
	if cfg.ConfigPath != "" && strings.TrimSpace(cfg.ConfigPath) == "" {
		return fmt.Errorf("pipeline.scriptorium.config_path must be non-empty when provided")
	}
	if err := validateDuration("pipeline.scriptorium.timeout", cfg.Timeout); err != nil {
		return err
	}

	artifactNames, configuredArtifacts, err := validateScriptoriumArtifactKeys(cfg.Artifacts)
	if err != nil {
		return err
	}
	if err := validateScriptoriumDeclarations(cfg.Artifacts, artifactNames); err != nil {
		return err
	}
	referencedArtifacts, err := validateScriptoriumDependenciesAndInputs(
		cfg.Artifacts,
		artifactNames,
		configuredArtifacts,
		notariusOutputKeySet(notarius),
	)
	if err != nil {
		return err
	}
	if err := validateScriptoriumVariables(cfg.Artifacts, artifactNames); err != nil {
		return err
	}
	if err := validateScriptoriumReferencedOutputs(cfg.Artifacts, referencedArtifacts); err != nil {
		return err
	}
	if err := ValidateScriptoriumArtifactDependencies(cfg.Artifacts); err != nil {
		return err
	}

	return nil
}

func validateScriptoriumArtifactKeys(artifacts map[string]ScriptoriumArtifactConfig) ([]string, map[string]struct{}, error) {
	names := make([]string, 0, len(artifacts))
	for name := range artifacts {
		names = append(names, name)
	}
	sort.Strings(names)

	configured := make(map[string]struct{}, len(names))
	for _, name := range names {
		if !artifactpolicy.IsConfiguredKey(name) {
			return nil, nil, fmt.Errorf("pipeline.scriptorium.artifacts keys must match ^[a-z][a-z0-9_]*$")
		}
		configured[name] = struct{}{}
	}
	return names, configured, nil
}

func validateScriptoriumDeclarations(artifacts map[string]ScriptoriumArtifactConfig, artifactNames []string) error {
	outputOwners := make(map[string]string, len(artifactNames))
	for _, artifactName := range artifactNames {
		artifactCfg := artifacts[artifactName]
		if artifactCfg.Enabled && strings.TrimSpace(artifactCfg.PromptID) == "" {
			return fmt.Errorf("pipeline.scriptorium.artifacts.%s.prompt_id is required when enabled", artifactName)
		}
		if artifactCfg.Enabled && strings.TrimSpace(artifactCfg.OutputPath) == "" {
			return fmt.Errorf("pipeline.scriptorium.artifacts.%s.output_path is required when enabled", artifactName)
		}
		if strings.TrimSpace(artifactCfg.OutputPath) != "" {
			pathField := "pipeline.scriptorium.artifacts." + artifactName + ".output_path"
			if err := validateRelativeSafePath(pathField, artifactCfg.OutputPath); err != nil {
				return err
			}
			if err := validatePathWithinRoot(pathField, artifactCfg.OutputPath, DefaultScriptoriumArtifactOutputRoot); err != nil {
				return err
			}
			if previous, exists := outputOwners[artifactCfg.OutputPath]; exists {
				return fmt.Errorf("%s %q duplicates artifact %q", pathField, artifactCfg.OutputPath, previous)
			}
			outputOwners[artifactCfg.OutputPath] = artifactName
		}
		if err := validateDuration("pipeline.scriptorium.artifacts."+artifactName+".timeout", artifactCfg.Timeout); err != nil {
			return err
		}
	}
	return nil
}

func validateScriptoriumDependenciesAndInputs(
	artifacts map[string]ScriptoriumArtifactConfig,
	artifactNames []string,
	configuredArtifacts map[string]struct{},
	extractionOutputs map[string]struct{},
) (map[string]struct{}, error) {
	referencedArtifacts := make(map[string]struct{})
	for _, artifactName := range artifactNames {
		artifactCfg := artifacts[artifactName]
		depSet := make(map[string]struct{}, len(artifactCfg.DependsOn))
		for i, depName := range artifactCfg.DependsOn {
			trimmedDep := strings.TrimSpace(depName)
			field := fmt.Sprintf("pipeline.scriptorium.artifacts.%s.depends_on[%d]", artifactName, i)
			if trimmedDep == "" {
				return nil, fmt.Errorf("%s must be non-empty", field)
			}
			if _, ok := configuredArtifacts[trimmedDep]; !ok {
				return nil, fmt.Errorf("%s %q is not a configured artifact key", field, depName)
			}
			if trimmedDep == artifactName {
				return nil, fmt.Errorf("pipeline.scriptorium.artifacts.%s.depends_on must not include itself", artifactName)
			}
			depSet[trimmedDep] = struct{}{}
			referencedArtifacts[trimmedDep] = struct{}{}
		}

		inputNames := make([]string, 0, len(artifactCfg.Inputs))
		for inputName := range artifactCfg.Inputs {
			inputNames = append(inputNames, inputName)
		}
		sort.Strings(inputNames)
		for _, inputName := range inputNames {
			inputCfg := artifactCfg.Inputs[inputName]
			trimmedInputName := strings.TrimSpace(inputName)
			if trimmedInputName == "" {
				return nil, fmt.Errorf("pipeline.scriptorium.artifacts.%s.inputs keys must be non-empty", artifactName)
			}
			source := strings.TrimSpace(inputCfg.Source)
			if source == "" {
				return nil, fmt.Errorf("pipeline.scriptorium.artifacts.%s.inputs.%s.source is required", artifactName, inputName)
			}

			referencedArtifact, err := validateScriptoriumInputSource(artifactName, inputName, source, configuredArtifacts, extractionOutputs)
			if err != nil {
				return nil, err
			}
			if referencedArtifact != "" {
				if _, ok := depSet[referencedArtifact]; !ok {
					return nil, fmt.Errorf(
						"pipeline.scriptorium.artifacts.%s.inputs.%s.source %q requires depends_on entry %q",
						artifactName,
						inputName,
						source,
						referencedArtifact,
					)
				}
				referencedArtifacts[referencedArtifact] = struct{}{}
			}
		}
	}
	return referencedArtifacts, nil
}

func validateScriptoriumVariables(artifacts map[string]ScriptoriumArtifactConfig, artifactNames []string) error {
	for _, artifactName := range artifactNames {
		artifactCfg := artifacts[artifactName]
		variableNames := make([]string, 0, len(artifactCfg.Vars))
		for varName := range artifactCfg.Vars {
			variableNames = append(variableNames, varName)
		}
		sort.Strings(variableNames)
		for _, varName := range variableNames {
			varValue := artifactCfg.Vars[varName]
			if strings.TrimSpace(varName) == "" {
				return fmt.Errorf("pipeline.scriptorium.artifacts.%s.vars keys must be non-empty", artifactName)
			}
			switch varValue.(type) {
			case bool, string:
			default:
				return fmt.Errorf("pipeline.scriptorium.artifacts.%s.vars.%s must be a string or boolean", artifactName, varName)
			}
		}
	}
	return nil
}

func validateScriptoriumReferencedOutputs(artifacts map[string]ScriptoriumArtifactConfig, referencedArtifacts map[string]struct{}) error {
	artifactNames := make([]string, 0, len(referencedArtifacts))
	for artifactName := range referencedArtifacts {
		artifactNames = append(artifactNames, artifactName)
	}
	sort.Strings(artifactNames)
	for _, artifactName := range artifactNames {
		artifactCfg := artifacts[artifactName]
		if strings.TrimSpace(artifactCfg.OutputPath) == "" {
			return fmt.Errorf("pipeline.scriptorium.artifacts.%s.output_path is required when artifact is referenced", artifactName)
		}
	}
	return nil
}

func validateScriptoriumInputSource(
	artifactName, inputName, source string,
	configuredArtifacts map[string]struct{},
	extractionOutputs map[string]struct{},
) (string, error) {
	trimmedSource := strings.TrimSpace(source)
	descriptor, err := artifactpolicy.DescribeScriptoriumInputSource(trimmedSource)
	if err != nil {
		if errors.Is(err, artifactpolicy.ErrInvalidPreviousSessionSource) {
			return "", fmt.Errorf(
				"pipeline.scriptorium.artifacts.%s.inputs.%s.source %q must reference configured artifact key matching ^[a-z][a-z0-9_]*$",
				artifactName,
				inputName,
				source,
			)
		}
		return "", fmt.Errorf(
			"pipeline.scriptorium.artifacts.%s.inputs.%s.source %q is unsupported",
			artifactName,
			inputName,
			source,
		)
	}
	if err := artifactpolicy.ValidateInputReference(descriptor, configuredArtifacts, extractionOutputs); err != nil {
		var unknownConfigured *artifactpolicy.UnknownConfiguredArtifactError
		if errors.As(err, &unknownConfigured) {
			return "", fmt.Errorf(
				"pipeline.scriptorium.artifacts.%s.inputs.%s.source %q references unknown artifact %q",
				artifactName,
				inputName,
				source,
				unknownConfigured.ConfiguredKey,
			)
		}
		var unknownExtraction *artifactpolicy.UnknownExtractionArtifactError
		if errors.As(err, &unknownExtraction) {
			return "", fmt.Errorf(
				"pipeline.scriptorium.artifacts.%s.inputs.%s.source %q references unknown extraction output %q",
				artifactName,
				inputName,
				source,
				unknownExtraction.ConfiguredKey,
			)
		}
		return "", fmt.Errorf(
			"pipeline.scriptorium.artifacts.%s.inputs.%s.source %q is unsupported",
			artifactName,
			inputName,
			source,
		)
	}
	if descriptor.Source.Kind == artifactpolicy.SourceKindConfiguredArtifact {
		return descriptor.Source.ConfiguredKey, nil
	}
	return "", nil
}

// ValidateScriptoriumArtifactDependencies validates the configured dependency
// graph independently of execution selection. Disabled artifacts remain valid
// prerequisites for explicit selections and therefore participate in cycles.
func ValidateScriptoriumArtifactDependencies(artifacts map[string]ScriptoriumArtifactConfig) error {
	if len(artifacts) == 0 {
		return nil
	}

	nodes := make([]string, 0, len(artifacts))
	for name := range artifacts {
		nodes = append(nodes, name)
	}
	sort.Strings(nodes)

	graph := make(map[string][]string, len(artifacts))
	for _, name := range nodes {
		cfg := artifacts[name]
		if !artifactpolicy.IsConfiguredKey(name) {
			return fmt.Errorf("pipeline.scriptorium.artifacts keys must match ^[a-z][a-z0-9_]*$")
		}
		seen := make(map[string]struct{}, len(cfg.DependsOn))
		for _, dep := range cfg.DependsOn {
			trimmedDep := strings.TrimSpace(dep)
			if trimmedDep == "" {
				return fmt.Errorf("pipeline.scriptorium.artifacts.%s.depends_on entries must be non-empty", name)
			}
			if _, ok := artifacts[trimmedDep]; !ok {
				return fmt.Errorf("pipeline.scriptorium.artifacts.%s dependency %q is not configured", name, dep)
			}
			if trimmedDep == name {
				return fmt.Errorf("pipeline.scriptorium.artifacts.%s.depends_on must not include itself", name)
			}
			if _, duplicate := seen[trimmedDep]; duplicate {
				continue
			}
			seen[trimmedDep] = struct{}{}
			graph[name] = append(graph[name], trimmedDep)
		}
		sort.Strings(graph[name])
	}

	visiting := make(map[string]bool, len(artifacts))
	visited := make(map[string]bool, len(artifacts))

	var visit func(node string) error
	visit = func(node string) error {
		if visiting[node] {
			return fmt.Errorf("pipeline.scriptorium.artifacts dependencies must not contain cycles")
		}
		if visited[node] {
			return nil
		}
		visiting[node] = true
		for _, dep := range graph[node] {
			if err := visit(dep); err != nil {
				return err
			}
		}
		visiting[node] = false
		visited[node] = true
		return nil
	}

	for _, node := range nodes {
		if err := visit(node); err != nil {
			return err
		}
	}
	return nil
}