merge: quick/fase6-integrity — validacion de integridad al indexar

cmd/fn/main.go

@@ -94,8 +94,11 @@ func cmdIndex() {
 	}
 
 	fmt.Printf("Indexed %d functions, %d types\n", result.Functions, result.Types)
+	for _, e := range result.ValidationErrors {
+		fmt.Fprintf(os.Stderr, "  INVALID: %s\n", e)
+	}
 	for _, e := range result.Errors {
-		fmt.Fprintf(os.Stderr, "  warn: %s\n", e)
+		fmt.Fprintf(os.Stderr, "  ERROR: %s\n", e)
 	}
 }
 

registry/indexer.go

@@ -9,13 +9,16 @@ import (
 
 // IndexResult holds stats from an indexing run.
 type IndexResult struct {
-	Functions int
-	Types     int
-	Errors    []string
+	Functions        int
+	Types            int
+	ValidationErrors []string
+	Errors           []string
 }
 
-// Index walks the registry root, parses all .md files, and populates the database.
-// It purges existing data first to ensure a clean rebuild.
+// Index walks the registry root, parses all .md files, validates integrity,
+// and populates the database. It uses two passes:
+// 1. Parse all entries and collect known IDs
+// 2. Validate references against known IDs, then insert valid entries
 func Index(db *DB, root string) (*IndexResult, error) {
 	if err := db.Purge(); err != nil {
 		return nil, fmt.Errorf("purging database: %w", err)
@@ -23,64 +26,75 @@ func Index(db *DB, root string) (*IndexResult, error) {
 
 	result := &IndexResult{}
 
-	// Index functions
+	// Pass 1: parse everything
+	var functions []*Function
+	var types []*Type
+
 	functionsDir := filepath.Join(root, "functions")
 	if _, err := os.Stat(functionsDir); err == nil {
-		err := filepath.Walk(functionsDir, func(path string, info os.FileInfo, err error) error {
-			if err != nil {
+		filepath.Walk(functionsDir, func(path string, info os.FileInfo, err error) error {
+			if err != nil || info.IsDir() || !strings.HasSuffix(path, ".md") {
 				return nil
 			}
-			if info.IsDir() || !strings.HasSuffix(path, ".md") {
-				return nil
-			}
-
 			f, err := ParseFunctionMD(path)
 			if err != nil {
-				result.Errors = append(result.Errors, fmt.Sprintf("%s: %v", path, err))
+				result.Errors = append(result.Errors, fmt.Sprintf("parse %s: %v", path, err))
 				return nil
 			}
-
-			if err := db.InsertFunction(f); err != nil {
-				result.Errors = append(result.Errors, fmt.Sprintf("insert %s: %v", f.ID, err))
-				return nil
-			}
-
-			result.Functions++
+			functions = append(functions, f)
 			return nil
 		})
-		if err != nil {
-			return nil, fmt.Errorf("walking functions: %w", err)
-		}
 	}
 
-	// Index types
 	typesDir := filepath.Join(root, "types")
 	if _, err := os.Stat(typesDir); err == nil {
-		err := filepath.Walk(typesDir, func(path string, info os.FileInfo, err error) error {
-			if err != nil {
+		filepath.Walk(typesDir, func(path string, info os.FileInfo, err error) error {
+			if err != nil || info.IsDir() || !strings.HasSuffix(path, ".md") {
 				return nil
 			}
-			if info.IsDir() || !strings.HasSuffix(path, ".md") {
-				return nil
-			}
-
 			t, err := ParseTypeMD(path)
 			if err != nil {
-				result.Errors = append(result.Errors, fmt.Sprintf("%s: %v", path, err))
+				result.Errors = append(result.Errors, fmt.Sprintf("parse %s: %v", path, err))
 				return nil
 			}
-
-			if err := db.InsertType(t); err != nil {
-				result.Errors = append(result.Errors, fmt.Sprintf("insert %s: %v", t.ID, err))
-				return nil
-			}
-
-			result.Types++
+			types = append(types, t)
 			return nil
 		})
-		if err != nil {
-			return nil, fmt.Errorf("walking types: %w", err)
+	}
+
+	// Build known ID sets
+	knownFunctions := make(map[string]bool, len(functions))
+	for _, f := range functions {
+		knownFunctions[f.ID] = true
+	}
+	knownTypes := make(map[string]bool, len(types))
+	for _, t := range types {
+		knownTypes[t.ID] = true
+	}
+
+	// Pass 2: validate and insert
+	for _, t := range types {
+		if verr := ValidateType(t, knownTypes); verr != nil {
+			result.ValidationErrors = append(result.ValidationErrors, verr.Error())
+			continue
 		}
+		if err := db.InsertType(t); err != nil {
+			result.Errors = append(result.Errors, fmt.Sprintf("insert %s: %v", t.ID, err))
+			continue
+		}
+		result.Types++
+	}
+
+	for _, f := range functions {
+		if verr := ValidateFunction(f, knownFunctions, knownTypes); verr != nil {
+			result.ValidationErrors = append(result.ValidationErrors, verr.Error())
+			continue
+		}
+		if err := db.InsertFunction(f); err != nil {
+			result.Errors = append(result.Errors, fmt.Sprintf("insert %s: %v", f.ID, err))
+			continue
+		}
+		result.Functions++
 	}
 
 	return result, nil

registry/parser_test.go

@@ -209,6 +209,9 @@ func TestIndexFullCycle(t *testing.T) {
 	if len(result.Errors) != 0 {
 		t.Errorf("unexpected errors: %v", result.Errors)
 	}
+	if len(result.ValidationErrors) != 0 {
+		t.Errorf("unexpected validation errors: %v", result.ValidationErrors)
+	}
 
 	// Verify searchable
 	fns, err := db.SearchFunctions("filter", "", "", "", "")
@@ -236,3 +239,61 @@ func TestIndexFullCycle(t *testing.T) {
 		t.Error("re-index should produce same counts")
 	}
 }
+
+const invalidPipelineMD = `---
+name: bad_pipeline
+kind: pipeline
+lang: go
+domain: core
+version: "1.0.0"
+purity: pure
+description: "Pipeline puro sin uses_functions — debe fallar."
+tags: []
+uses_functions: []
+uses_types: []
+returns: []
+returns_optional: false
+error_type: ""
+imports: []
+tested: false
+tests: []
+test_file_path: ""
+file_path: "functions/pipelines/bad.go"
+---
+`
+
+func TestIndexRejectsInvalid(t *testing.T) {
+	root := t.TempDir()
+
+	// Valid function
+	writeTempFile(t, root, "functions/core/filter_slice.md", functionMD)
+	// Invalid pipeline (pure + empty uses_functions)
+	writeTempFile(t, root, "functions/pipelines/bad.md", invalidPipelineMD)
+	// Valid type
+	writeTempFile(t, root, "types/finance/ohlcv.md", typeMD)
+
+	dbPath := filepath.Join(root, "registry.db")
+	db, err := Open(dbPath)
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer db.Close()
+
+	result, err := Index(db, root)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	// Valid entries should be indexed
+	if result.Functions != 1 {
+		t.Errorf("functions: got %d, want 1 (only the valid one)", result.Functions)
+	}
+	if result.Types != 1 {
+		t.Errorf("types: got %d, want 1", result.Types)
+	}
+
+	// Invalid pipeline should produce validation error
+	if len(result.ValidationErrors) == 0 {
+		t.Error("expected validation errors for invalid pipeline")
+	}
+}

registry/validate.go

@@ -0,0 +1,167 @@
+package registry
+
+import (
+	"fmt"
+	"strings"
+)
+
+// ValidationError represents one or more integrity violations.
+type ValidationError struct {
+	ID     string
+	Errors []string
+}
+
+func (v *ValidationError) Error() string {
+	return fmt.Sprintf("%s: %s", v.ID, strings.Join(v.Errors, "; "))
+}
+
+// ValidateFunction checks integrity rules from docs/integrity.md.
+// knownFunctions and knownTypes are sets of IDs that exist in the registry
+// (including the current indexing batch).
+func ValidateFunction(f *Function, knownFunctions, knownTypes map[string]bool) *ValidationError {
+	var errs []string
+
+	// Required fields
+	if f.Name == "" {
+		errs = append(errs, "name is required")
+	}
+	if f.Kind == "" {
+		errs = append(errs, "kind is required")
+	}
+	if f.Lang == "" {
+		errs = append(errs, "lang is required")
+	}
+	if f.Domain == "" {
+		errs = append(errs, "domain is required")
+	}
+	if f.Description == "" {
+		errs = append(errs, "description is required")
+	}
+
+	// Pipeline rules
+	if f.Kind == KindPipeline {
+		if f.Purity != PurityImpure {
+			errs = append(errs, "pipeline must be impure")
+		}
+		if len(f.UsesFunctions) == 0 {
+			errs = append(errs, "pipeline uses_functions cannot be empty")
+		}
+	}
+
+	// Purity rules
+	if f.Purity == PurityPure {
+		if f.ReturnsOptional {
+			errs = append(errs, "pure function cannot have returns_optional: true (model as sum type)")
+		}
+		if f.ErrorType != "" {
+			errs = append(errs, "pure function cannot have error_type")
+		}
+	}
+	if f.Purity == PurityImpure && f.Kind != KindComponent {
+		if f.ErrorType == "" {
+			errs = append(errs, "impure function must declare error_type")
+		}
+	}
+
+	// Tested rules
+	if f.Tested {
+		if f.TestFilePath == "" {
+			errs = append(errs, "tested: true requires test_file_path")
+		}
+		if len(f.Tests) == 0 {
+			errs = append(errs, "tested: true requires non-empty tests")
+		}
+	} else {
+		if len(f.Tests) > 0 {
+			errs = append(errs, "tested: false but tests is not empty")
+		}
+		if f.TestFilePath != "" {
+			errs = append(errs, "tested: false but test_file_path is set")
+		}
+	}
+
+	// Component rules
+	if f.Kind == KindComponent {
+		if f.Framework == "" {
+			errs = append(errs, "component must declare framework")
+		}
+		if len(f.Returns) > 0 {
+			errs = append(errs, "component returns must be empty (use emits)")
+		}
+		if f.HasState != nil && *f.HasState && f.Purity != PurityImpure {
+			errs = append(errs, "component with has_state: true must be impure")
+		}
+	}
+
+	// File path must be relative
+	if f.FilePath != "" && strings.HasPrefix(f.FilePath, "/") {
+		errs = append(errs, "file_path must be relative to registry root")
+	}
+
+	// Reference validation
+	for _, ref := range f.UsesFunctions {
+		if !knownFunctions[ref] {
+			errs = append(errs, fmt.Sprintf("uses_functions references unknown function: %s", ref))
+		}
+	}
+	for _, ref := range f.UsesTypes {
+		if !knownTypes[ref] {
+			errs = append(errs, fmt.Sprintf("uses_types references unknown type: %s", ref))
+		}
+	}
+	for _, ref := range f.Returns {
+		if !knownTypes[ref] {
+			errs = append(errs, fmt.Sprintf("returns references unknown type: %s", ref))
+		}
+	}
+	if f.ErrorType != "" {
+		if !knownTypes[f.ErrorType] {
+			errs = append(errs, fmt.Sprintf("error_type references unknown type: %s", f.ErrorType))
+		}
+	}
+
+	if len(errs) > 0 {
+		return &ValidationError{ID: f.ID, Errors: errs}
+	}
+	return nil
+}
+
+// ValidateType checks integrity rules for types.
+func ValidateType(t *Type, knownTypes map[string]bool) *ValidationError {
+	var errs []string
+
+	if t.Name == "" {
+		errs = append(errs, "name is required")
+	}
+	if t.Lang == "" {
+		errs = append(errs, "lang is required")
+	}
+	if t.Domain == "" {
+		errs = append(errs, "domain is required")
+	}
+	if t.Description == "" {
+		errs = append(errs, "description is required")
+	}
+	if t.Algebraic != AlgebraicProduct && t.Algebraic != AlgebraicSum {
+		errs = append(errs, fmt.Sprintf("algebraic must be 'product' or 'sum', got %q", t.Algebraic))
+	}
+
+	if t.FilePath != "" && strings.HasPrefix(t.FilePath, "/") {
+		errs = append(errs, "file_path must be relative to registry root")
+	}
+
+	// Self-reference check
+	for _, ref := range t.UsesTypes {
+		if ref == t.ID {
+			errs = append(errs, "type cannot reference itself in uses_types")
+		}
+		if !knownTypes[ref] {
+			errs = append(errs, fmt.Sprintf("uses_types references unknown type: %s", ref))
+		}
+	}
+
+	if len(errs) > 0 {
+		return &ValidationError{ID: t.ID, Errors: errs}
+	}
+	return nil
+}

registry/validate_test.go

@@ -0,0 +1,258 @@
+package registry
+
+import (
+	"strings"
+	"testing"
+)
+
+func boolPtr(b bool) *bool { return &b }
+
+func knownFns(ids ...string) map[string]bool {
+	m := make(map[string]bool)
+	for _, id := range ids {
+		m[id] = true
+	}
+	return m
+}
+
+func knownTps(ids ...string) map[string]bool {
+	return knownFns(ids...)
+}
+
+func TestValidateFunction_Valid(t *testing.T) {
+	f := &Function{
+		ID: "filter_slice_go_core", Name: "filter_slice", Kind: KindFunction,
+		Lang: "go", Domain: "core", Purity: PurityPure,
+		Description: "Filtra un slice", Version: "1.0.0",
+	}
+	if err := ValidateFunction(f, knownFns(), knownTps()); err != nil {
+		t.Errorf("expected valid, got: %v", err)
+	}
+}
+
+func TestValidateFunction_PipelineMustBeImpure(t *testing.T) {
+	f := &Function{
+		ID: "p_go_core", Name: "p", Kind: KindPipeline,
+		Lang: "go", Domain: "core", Purity: PurityPure,
+		Description: "bad pipeline", Version: "1.0.0",
+		UsesFunctions: []string{"filter_slice_go_core"},
+	}
+	err := ValidateFunction(f, knownFns("filter_slice_go_core"), knownTps())
+	if err == nil {
+		t.Fatal("expected error")
+	}
+	if !strings.Contains(err.Error(), "pipeline must be impure") {
+		t.Errorf("unexpected error: %v", err)
+	}
+}
+
+func TestValidateFunction_PipelineNeedsUsesFunctions(t *testing.T) {
+	f := &Function{
+		ID: "p_go_core", Name: "p", Kind: KindPipeline,
+		Lang: "go", Domain: "core", Purity: PurityImpure,
+		Description: "bad pipeline", Version: "1.0.0",
+		ErrorType: "error_go_core",
+	}
+	err := ValidateFunction(f, knownFns(), knownTps("error_go_core"))
+	if err == nil {
+		t.Fatal("expected error")
+	}
+	if !strings.Contains(err.Error(), "uses_functions cannot be empty") {
+		t.Errorf("unexpected error: %v", err)
+	}
+}
+
+func TestValidateFunction_PureNoReturnsOptional(t *testing.T) {
+	f := &Function{
+		ID: "f_go_core", Name: "f", Kind: KindFunction,
+		Lang: "go", Domain: "core", Purity: PurityPure,
+		Description: "bad", Version: "1.0.0",
+		ReturnsOptional: true,
+	}
+	err := ValidateFunction(f, knownFns(), knownTps())
+	if err == nil {
+		t.Fatal("expected error")
+	}
+	if !strings.Contains(err.Error(), "returns_optional") {
+		t.Errorf("unexpected error: %v", err)
+	}
+}
+
+func TestValidateFunction_PureNoErrorType(t *testing.T) {
+	f := &Function{
+		ID: "f_go_core", Name: "f", Kind: KindFunction,
+		Lang: "go", Domain: "core", Purity: PurityPure,
+		Description: "bad", Version: "1.0.0",
+		ErrorType: "error_go_core",
+	}
+	err := ValidateFunction(f, knownFns(), knownTps("error_go_core"))
+	if err == nil {
+		t.Fatal("expected error")
+	}
+	if !strings.Contains(err.Error(), "pure function cannot have error_type") {
+		t.Errorf("unexpected error: %v", err)
+	}
+}
+
+func TestValidateFunction_ImpureNeedsErrorType(t *testing.T) {
+	f := &Function{
+		ID: "f_go_io", Name: "f", Kind: KindFunction,
+		Lang: "go", Domain: "io", Purity: PurityImpure,
+		Description: "bad", Version: "1.0.0",
+	}
+	err := ValidateFunction(f, knownFns(), knownTps())
+	if err == nil {
+		t.Fatal("expected error")
+	}
+	if !strings.Contains(err.Error(), "impure function must declare error_type") {
+		t.Errorf("unexpected error: %v", err)
+	}
+}
+
+func TestValidateFunction_TestedNeedsTestFile(t *testing.T) {
+	f := &Function{
+		ID: "f_go_core", Name: "f", Kind: KindFunction,
+		Lang: "go", Domain: "core", Purity: PurityPure,
+		Description: "test", Version: "1.0.0",
+		Tested: true,
+	}
+	err := ValidateFunction(f, knownFns(), knownTps())
+	if err == nil {
+		t.Fatal("expected error")
+	}
+	if !strings.Contains(err.Error(), "test_file_path") {
+		t.Errorf("unexpected error: %v", err)
+	}
+}
+
+func TestValidateFunction_NotTestedNoTests(t *testing.T) {
+	f := &Function{
+		ID: "f_go_core", Name: "f", Kind: KindFunction,
+		Lang: "go", Domain: "core", Purity: PurityPure,
+		Description: "test", Version: "1.0.0",
+		Tested: false, Tests: []string{"ghost test"}, TestFilePath: "test.go",
+	}
+	err := ValidateFunction(f, knownFns(), knownTps())
+	if err == nil {
+		t.Fatal("expected error")
+	}
+	if !strings.Contains(err.Error(), "tested: false but tests is not empty") {
+		t.Errorf("unexpected error: %v", err)
+	}
+}
+
+func TestValidateFunction_OrphanRefs(t *testing.T) {
+	f := &Function{
+		ID: "p_go_core", Name: "p", Kind: KindPipeline,
+		Lang: "go", Domain: "core", Purity: PurityImpure,
+		Description: "pipeline", Version: "1.0.0",
+		UsesFunctions: []string{"nonexistent_go_core"},
+		UsesTypes:     []string{"ghost_go_core"},
+		Returns:       []string{"phantom_go_core"},
+		ErrorType:     "missing_go_core",
+	}
+	err := ValidateFunction(f, knownFns(), knownTps())
+	if err == nil {
+		t.Fatal("expected error")
+	}
+	if len(err.Errors) < 4 {
+		t.Errorf("expected at least 4 errors, got %d: %v", len(err.Errors), err)
+	}
+}
+
+func TestValidateFunction_ComponentRules(t *testing.T) {
+	f := &Function{
+		ID: "dt_typescript_core", Name: "DataTable", Kind: KindComponent,
+		Lang: "typescript", Domain: "core", Purity: PurityImpure,
+		Description: "table", Version: "1.0.0",
+		HasState: boolPtr(true), Framework: "react",
+	}
+	if err := ValidateFunction(f, knownFns(), knownTps()); err != nil {
+		t.Errorf("expected valid, got: %v", err)
+	}
+
+	// Missing framework
+	f2 := *f
+	f2.Framework = ""
+	if err := ValidateFunction(&f2, knownFns(), knownTps()); err == nil {
+		t.Error("expected error for missing framework")
+	}
+
+	// Returns should be empty
+	f3 := *f
+	f3.Returns = []string{"some_go_core"}
+	if err := ValidateFunction(&f3, knownFns(), knownTps("some_go_core")); err == nil {
+		t.Error("expected error for non-empty returns on component")
+	}
+
+	// has_state: true but pure
+	f4 := *f
+	f4.Purity = PurityPure
+	if err := ValidateFunction(&f4, knownFns(), knownTps()); err == nil {
+		t.Error("expected error for stateful pure component")
+	}
+}
+
+func TestValidateFunction_AbsoluteFilePath(t *testing.T) {
+	f := &Function{
+		ID: "f_go_core", Name: "f", Kind: KindFunction,
+		Lang: "go", Domain: "core", Purity: PurityPure,
+		Description: "test", Version: "1.0.0",
+		FilePath: "/absolute/path.go",
+	}
+	err := ValidateFunction(f, knownFns(), knownTps())
+	if err == nil {
+		t.Fatal("expected error for absolute file_path")
+	}
+}
+
+func TestValidateType_Valid(t *testing.T) {
+	typ := &Type{
+		ID: "ohlcv_go_finance", Name: "ohlcv", Lang: "go", Domain: "finance",
+		Algebraic: AlgebraicProduct, Description: "candle", Version: "1.0.0",
+	}
+	if err := ValidateType(typ, knownTps("ohlcv_go_finance")); err != nil {
+		t.Errorf("expected valid, got: %v", err)
+	}
+}
+
+func TestValidateType_BadAlgebraic(t *testing.T) {
+	typ := &Type{
+		ID: "t_go_core", Name: "t", Lang: "go", Domain: "core",
+		Algebraic: "wrong", Description: "bad", Version: "1.0.0",
+	}
+	err := ValidateType(typ, knownTps("t_go_core"))
+	if err == nil {
+		t.Fatal("expected error")
+	}
+}
+
+func TestValidateType_SelfReference(t *testing.T) {
+	typ := &Type{
+		ID: "t_go_core", Name: "t", Lang: "go", Domain: "core",
+		Algebraic: AlgebraicProduct, Description: "self ref", Version: "1.0.0",
+		UsesTypes: []string{"t_go_core"},
+	}
+	err := ValidateType(typ, knownTps("t_go_core"))
+	if err == nil {
+		t.Fatal("expected error")
+	}
+	if !strings.Contains(err.Error(), "cannot reference itself") {
+		t.Errorf("unexpected error: %v", err)
+	}
+}
+
+func TestValidateType_OrphanRef(t *testing.T) {
+	typ := &Type{
+		ID: "t_go_core", Name: "t", Lang: "go", Domain: "core",
+		Algebraic: AlgebraicProduct, Description: "orphan ref", Version: "1.0.0",
+		UsesTypes: []string{"nonexistent_go_core"},
+	}
+	err := ValidateType(typ, knownTps("t_go_core"))
+	if err == nil {
+		t.Fatal("expected error")
+	}
+	if !strings.Contains(err.Error(), "unknown type") {
+		t.Errorf("unexpected error: %v", err)
+	}
+}