Make Scriptorium validation deterministic

This commit is contained in:
2026-09-05 00:29:05 +00:00
parent fddf78dc80
commit abd405015b
3 changed files with 130 additions and 33 deletions

View File

@@ -310,7 +310,7 @@ configuration only after every check succeeds.
## Stage 4 — Decompose Deterministic Scriptorium Validation ## Stage 4 — Decompose Deterministic Scriptorium Validation
**Status:** Pending **Status:** Completed
### Goal ### Goal

View File

@@ -574,6 +574,28 @@ func TestScriptoriumLoadAndValidate(t *testing.T) {
} }
} }
func TestScriptoriumDiagnosticsAreIndependentOfMapInsertionOrder(t *testing.T) {
invalidArtifact := ScriptoriumArtifactConfig{Enabled: true}
firstArtifacts := make(map[string]ScriptoriumArtifactConfig)
firstArtifacts["beta"] = invalidArtifact
firstArtifacts["alpha"] = invalidArtifact
secondArtifacts := make(map[string]ScriptoriumArtifactConfig)
secondArtifacts["alpha"] = invalidArtifact
secondArtifacts["beta"] = invalidArtifact
firstErr := validateScriptorium(&ScriptoriumConfig{Binary: "scriptorium", Artifacts: firstArtifacts}, nil)
secondErr := validateScriptorium(&ScriptoriumConfig{Binary: "scriptorium", Artifacts: secondArtifacts}, nil)
if firstErr == nil || secondErr == nil {
t.Fatalf("validateScriptorium() errors = %v / %v, want validation failures", firstErr, secondErr)
}
if firstErr.Error() != secondErr.Error() {
t.Fatalf("equivalent maps produced different diagnostics:\nfirst: %v\nsecond: %v", firstErr, secondErr)
}
if !strings.Contains(firstErr.Error(), "pipeline.scriptorium.artifacts.") || !strings.Contains(firstErr.Error(), "prompt_id") {
t.Fatalf("validation error = %v, want artifact field and semantic category", firstErr)
}
}
func TestScriptoriumLegacyTranscriptSourcesRejected(t *testing.T) { func TestScriptoriumLegacyTranscriptSourcesRejected(t *testing.T) {
legacyTranscriptSources := []string{ legacyTranscriptSources := []string{
"narratio.transcript." + "merged", "narratio.transcript." + "merged",

View File

@@ -23,17 +23,56 @@ func validateScriptorium(cfg *ScriptoriumConfig, notarius *NotariusConfig) error
return err return err
} }
configuredArtifacts := make(map[string]struct{}, len(cfg.Artifacts)) artifactNames, configuredArtifacts, err := validateScriptoriumArtifactKeys(cfg.Artifacts)
referencedArtifacts := make(map[string]struct{}) if err != nil {
outputOwners := make(map[string]string, len(cfg.Artifacts)) return err
for artifactName := range cfg.Artifacts { }
if !artifactpolicy.IsConfiguredKey(artifactName) { if err := validateScriptoriumDeclarations(cfg.Artifacts, artifactNames); err != nil {
return fmt.Errorf("pipeline.scriptorium.artifacts keys must match ^[a-z][a-z0-9_]*$") return err
} }
configuredArtifacts[artifactName] = struct{}{} referencedArtifacts, err := validateScriptoriumDependenciesAndInputs(
cfg.Artifacts,
artifactNames,
configuredArtifacts,
notariusOutputKeySet(notarius),
)
if err != nil {
return err
}
if err := validateScriptoriumVariables(cfg.Artifacts, artifactNames); err != nil {
return err
}
if err := validateScriptoriumReferencedOutputs(cfg.Artifacts, referencedArtifacts); err != nil {
return err
}
if err := ValidateScriptoriumArtifactDependencies(cfg.Artifacts); err != nil {
return err
} }
for artifactName, artifactCfg := range cfg.Artifacts { return nil
}
func validateScriptoriumArtifactKeys(artifacts map[string]ScriptoriumArtifactConfig) ([]string, map[string]struct{}, error) {
names := make([]string, 0, len(artifacts))
for name := range artifacts {
names = append(names, name)
}
sort.Strings(names)
configured := make(map[string]struct{}, len(names))
for _, name := range names {
if !artifactpolicy.IsConfiguredKey(name) {
return nil, nil, fmt.Errorf("pipeline.scriptorium.artifacts keys must match ^[a-z][a-z0-9_]*$")
}
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) == "" { if artifactCfg.Enabled && strings.TrimSpace(artifactCfg.PromptID) == "" {
return fmt.Errorf("pipeline.scriptorium.artifacts.%s.prompt_id is required when enabled", artifactName) return fmt.Errorf("pipeline.scriptorium.artifacts.%s.prompt_id is required when enabled", artifactName)
} }
@@ -49,48 +88,66 @@ func validateScriptorium(cfg *ScriptoriumConfig, notarius *NotariusConfig) error
return err return err
} }
if previous, exists := outputOwners[artifactCfg.OutputPath]; exists { if previous, exists := outputOwners[artifactCfg.OutputPath]; exists {
return fmt.Errorf("%s.output_path %q duplicates artifact %q", pathField, artifactCfg.OutputPath, previous) return fmt.Errorf("%s %q duplicates artifact %q", pathField, artifactCfg.OutputPath, previous)
} }
outputOwners[artifactCfg.OutputPath] = artifactName outputOwners[artifactCfg.OutputPath] = artifactName
} }
if err := validateDuration("pipeline.scriptorium.artifacts."+artifactName+".timeout", artifactCfg.Timeout); err != nil { if err := validateDuration("pipeline.scriptorium.artifacts."+artifactName+".timeout", artifactCfg.Timeout); err != nil {
return err 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)) depSet := make(map[string]struct{}, len(artifactCfg.DependsOn))
for i, depName := range artifactCfg.DependsOn { for i, depName := range artifactCfg.DependsOn {
trimmedDep := strings.TrimSpace(depName) trimmedDep := strings.TrimSpace(depName)
field := fmt.Sprintf("pipeline.scriptorium.artifacts.%s.depends_on[%d]", artifactName, i) field := fmt.Sprintf("pipeline.scriptorium.artifacts.%s.depends_on[%d]", artifactName, i)
if trimmedDep == "" { if trimmedDep == "" {
return fmt.Errorf("%s must be non-empty", field) return nil, fmt.Errorf("%s must be non-empty", field)
} }
if _, ok := configuredArtifacts[trimmedDep]; !ok { if _, ok := configuredArtifacts[trimmedDep]; !ok {
return fmt.Errorf("%s %q is not a configured artifact key", field, depName) return nil, fmt.Errorf("%s %q is not a configured artifact key", field, depName)
} }
if trimmedDep == artifactName { if trimmedDep == artifactName {
return fmt.Errorf("pipeline.scriptorium.artifacts.%s.depends_on must not include itself", artifactName) return nil, fmt.Errorf("pipeline.scriptorium.artifacts.%s.depends_on must not include itself", artifactName)
} }
depSet[trimmedDep] = struct{}{} depSet[trimmedDep] = struct{}{}
referencedArtifacts[trimmedDep] = struct{}{} referencedArtifacts[trimmedDep] = struct{}{}
} }
for inputName, inputCfg := range artifactCfg.Inputs { 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) trimmedInputName := strings.TrimSpace(inputName)
if trimmedInputName == "" { if trimmedInputName == "" {
return fmt.Errorf("pipeline.scriptorium.artifacts.%s.inputs keys must be non-empty", artifactName) return nil, fmt.Errorf("pipeline.scriptorium.artifacts.%s.inputs keys must be non-empty", artifactName)
} }
source := strings.TrimSpace(inputCfg.Source) source := strings.TrimSpace(inputCfg.Source)
if source == "" { if source == "" {
return fmt.Errorf("pipeline.scriptorium.artifacts.%s.inputs.%s.source is required", artifactName, inputName) return nil, fmt.Errorf("pipeline.scriptorium.artifacts.%s.inputs.%s.source is required", artifactName, inputName)
} }
referencedArtifact, err := validateScriptoriumInputSource(artifactName, inputName, source, configuredArtifacts, notariusOutputKeySet(notarius)) referencedArtifact, err := validateScriptoriumInputSource(artifactName, inputName, source, configuredArtifacts, extractionOutputs)
if err != nil { if err != nil {
return err return nil, err
} }
if referencedArtifact != "" { if referencedArtifact != "" {
if _, ok := depSet[referencedArtifact]; !ok { if _, ok := depSet[referencedArtifact]; !ok {
return fmt.Errorf( return nil, fmt.Errorf(
"pipeline.scriptorium.artifacts.%s.inputs.%s.source %q requires depends_on entry %q", "pipeline.scriptorium.artifacts.%s.inputs.%s.source %q requires depends_on entry %q",
artifactName, artifactName,
inputName, inputName,
@@ -101,7 +158,20 @@ func validateScriptorium(cfg *ScriptoriumConfig, notarius *NotariusConfig) error
referencedArtifacts[referencedArtifact] = struct{}{} referencedArtifacts[referencedArtifact] = struct{}{}
} }
} }
for varName, varValue := range artifactCfg.Vars { }
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) == "" { if strings.TrimSpace(varName) == "" {
return fmt.Errorf("pipeline.scriptorium.artifacts.%s.vars keys must be non-empty", artifactName) return fmt.Errorf("pipeline.scriptorium.artifacts.%s.vars keys must be non-empty", artifactName)
} }
@@ -112,18 +182,21 @@ func validateScriptorium(cfg *ScriptoriumConfig, notarius *NotariusConfig) error
} }
} }
} }
return nil
}
func validateScriptoriumReferencedOutputs(artifacts map[string]ScriptoriumArtifactConfig, referencedArtifacts map[string]struct{}) error {
artifactNames := make([]string, 0, len(referencedArtifacts))
for artifactName := range referencedArtifacts { for artifactName := range referencedArtifacts {
artifactCfg := cfg.Artifacts[artifactName] artifactNames = append(artifactNames, artifactName)
}
sort.Strings(artifactNames)
for _, artifactName := range artifactNames {
artifactCfg := artifacts[artifactName]
if strings.TrimSpace(artifactCfg.OutputPath) == "" { if strings.TrimSpace(artifactCfg.OutputPath) == "" {
return fmt.Errorf("pipeline.scriptorium.artifacts.%s.output_path is required when artifact is referenced", artifactName) return fmt.Errorf("pipeline.scriptorium.artifacts.%s.output_path is required when artifact is referenced", artifactName)
} }
} }
if err := ValidateScriptoriumArtifactDependencies(cfg.Artifacts); err != nil {
return err
}
return nil return nil
} }
@@ -192,8 +265,15 @@ func ValidateScriptoriumArtifactDependencies(artifacts map[string]ScriptoriumArt
return nil 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)) graph := make(map[string][]string, len(artifacts))
for name, cfg := range artifacts { for _, name := range nodes {
cfg := artifacts[name]
if !artifactpolicy.IsConfiguredKey(name) { if !artifactpolicy.IsConfiguredKey(name) {
return fmt.Errorf("pipeline.scriptorium.artifacts keys must match ^[a-z][a-z0-9_]*$") return fmt.Errorf("pipeline.scriptorium.artifacts keys must match ^[a-z][a-z0-9_]*$")
} }
@@ -240,11 +320,6 @@ func ValidateScriptoriumArtifactDependencies(artifacts map[string]ScriptoriumArt
return nil return nil
} }
nodes := make([]string, 0, len(artifacts))
for node := range artifacts {
nodes = append(nodes, node)
}
sort.Strings(nodes)
for _, node := range nodes { for _, node := range nodes {
if err := visit(node); err != nil { if err := visit(node); err != nil {
return err return err