registry/internal/validators/validators.go
adam jones ef4607720d
Fix _meta field format for MCP spec compliance (#367)
## Summary

Updates the `_meta` field structure to be compliant with the MCP
specification by using proper `prefix/name` format for vendor
extensions.

### Changes

- **Registry metadata**: `io.modelcontextprotocol.registry` →
`io.modelcontextprotocol.registry/official`
- **Publisher metadata**: `publisher` →
`io.modelcontextprotocol.registry/publisher-provided`
- **Database migration**: Updates indexes to use new field paths
- **Updated all code**: Service layer, validators, tests, and seed data

### Format Before
```json
{
  "_meta": {
    "publisher": { "license": "MIT" },
    "io.modelcontextprotocol.registry": { "id": "...", "is_latest": true }
  }
}
```

### Format After
```json
{
  "_meta": {
    "io.modelcontextprotocol.registry/publisher-provided": { "license": "MIT" },
    "io.modelcontextprotocol.registry/official": { "id": "...", "is_latest": true }
  }
}
```

This resolves the naming collision with the planned top-level
`publisher` field and ensures compliance with MCP specification
requirements.

Fixes #356
2025-09-08 11:40:04 -07:00

373 lines
12 KiB
Go

package validators
import (
"context"
"encoding/json"
"fmt"
"net/url"
"slices"
"strings"
"github.com/modelcontextprotocol/registry/internal/config"
apiv0 "github.com/modelcontextprotocol/registry/pkg/api/v0"
"github.com/modelcontextprotocol/registry/pkg/model"
)
func ValidateServerJSON(serverJSON *apiv0.ServerJSON) error {
// Validate server name exists and format
if _, err := parseServerName(*serverJSON); err != nil {
return err
}
// Validate repository
if err := validateRepository(&serverJSON.Repository); err != nil {
return err
}
// Validate all packages (basic field validation)
// Detailed package validation (including registry checks) is done during publish
for _, pkg := range serverJSON.Packages {
if err := validatePackageField(&pkg); err != nil {
return err
}
}
// Validate all remotes
for _, remote := range serverJSON.Remotes {
if err := validateRemoteTransport(&remote); err != nil {
return err
}
}
// Validate reverse-DNS namespace matching for remote URLs
if err := validateRemoteNamespaceMatch(*serverJSON); err != nil {
return err
}
return nil
}
func validateRepository(obj *model.Repository) error {
// Skip validation for empty repository (optional field)
if obj.URL == "" && obj.Source == "" {
return nil
}
// validate the repository source
repoSource := RepositorySource(obj.Source)
if !IsValidRepositoryURL(repoSource, obj.URL) {
return fmt.Errorf("%w: %s", ErrInvalidRepositoryURL, obj.URL)
}
// validate subfolder if present
if obj.Subfolder != "" && !IsValidSubfolderPath(obj.Subfolder) {
return fmt.Errorf("%w: %s", ErrInvalidSubfolderPath, obj.Subfolder)
}
return nil
}
func validatePackageField(obj *model.Package) error {
if !HasNoSpaces(obj.Identifier) {
return ErrPackageNameHasSpaces
}
// Validate runtime arguments
for _, arg := range obj.RuntimeArguments {
if err := validateArgument(&arg); err != nil {
return fmt.Errorf("invalid runtime argument: %w", err)
}
}
// Validate package arguments
for _, arg := range obj.PackageArguments {
if err := validateArgument(&arg); err != nil {
return fmt.Errorf("invalid package argument: %w", err)
}
}
// Validate transport with template variable support
availableVariables := collectAvailableVariables(obj)
if err := validatePackageTransport(&obj.Transport, availableVariables); err != nil {
return fmt.Errorf("invalid transport: %w", err)
}
return nil
}
// validateArgument validates argument details
func validateArgument(obj *model.Argument) error {
if obj.Type == model.ArgumentTypeNamed {
// Validate named argument name format
if err := validateNamedArgumentName(obj.Name); err != nil {
return err
}
// Validate value and default don't start with the name
if err := validateArgumentValueFields(obj.Name, obj.Value, obj.Default); err != nil {
return err
}
}
return nil
}
func validateNamedArgumentName(name string) error {
// Check if name is required for named arguments
if name == "" {
return ErrNamedArgumentNameRequired
}
// Check for invalid characters that suggest embedded values or descriptions
// Valid: "--directory", "--port", "-v", "config", "verbose"
// Invalid: "--directory <absolute_path_to_adfin_mcp_folder>", "--port 8080"
if strings.Contains(name, "<") || strings.Contains(name, ">") ||
strings.Contains(name, " ") || strings.Contains(name, "$") {
return fmt.Errorf("%w: %s", ErrInvalidNamedArgumentName, name)
}
return nil
}
func validateArgumentValueFields(name, value, defaultValue string) error {
// Check if value starts with the argument name (using startsWith, not contains)
if value != "" && strings.HasPrefix(value, name) {
return fmt.Errorf("%w: value starts with argument name '%s': %s", ErrArgumentValueStartsWithName, name, value)
}
if defaultValue != "" && strings.HasPrefix(defaultValue, name) {
return fmt.Errorf("%w: default starts with argument name '%s': %s", ErrArgumentDefaultStartsWithName, name, defaultValue)
}
return nil
}
// collectAvailableVariables collects all available template variables from a package
func collectAvailableVariables(pkg *model.Package) []string {
var variables []string
// Add environment variable names
for _, env := range pkg.EnvironmentVariables {
variables = append(variables, env.Name)
}
// Add runtime argument names and value hints
for _, arg := range pkg.RuntimeArguments {
if arg.Name != "" {
variables = append(variables, arg.Name)
}
if arg.ValueHint != "" {
variables = append(variables, arg.ValueHint)
}
}
// Add package argument names and value hints
for _, arg := range pkg.PackageArguments {
if arg.Name != "" {
variables = append(variables, arg.Name)
}
if arg.ValueHint != "" {
variables = append(variables, arg.ValueHint)
}
}
return variables
}
// validatePackageTransport validates a package's transport with templating support
func validatePackageTransport(transport *model.Transport, availableVariables []string) error {
// Validate transport type is supported
switch transport.Type {
case model.TransportTypeStdio:
// Validate that URL is empty for stdio transport
if transport.URL != "" {
return fmt.Errorf("url must be empty for %s transport type, got: %s", transport.Type, transport.URL)
}
return nil
case model.TransportTypeStreamableHTTP, model.TransportTypeSSE:
// URL is required for streamable-http and sse
if transport.URL == "" {
return fmt.Errorf("url is required for %s transport type", transport.Type)
}
// Validate URL format with template variable support
if !IsValidTemplatedURL(transport.URL, availableVariables, true) {
// Check if it's a template variable issue or basic URL issue
templateVars := extractTemplateVariables(transport.URL)
if len(templateVars) > 0 {
return fmt.Errorf("%w: template variables in URL %s reference undefined variables. Available variables: %v",
ErrInvalidRemoteURL, transport.URL, availableVariables)
}
return fmt.Errorf("%w: %s", ErrInvalidRemoteURL, transport.URL)
}
return nil
default:
return fmt.Errorf("unsupported transport type: %s", transport.Type)
}
}
// validateRemoteTransport validates a remote transport (no templating allowed)
func validateRemoteTransport(obj *model.Transport) error {
// Validate transport type is supported - remotes only support streamable-http and sse
switch obj.Type {
case model.TransportTypeStreamableHTTP, model.TransportTypeSSE:
// URL is required for streamable-http and sse
if obj.URL == "" {
return fmt.Errorf("url is required for %s transport type", obj.Type)
}
// Validate URL format (no templates allowed for remotes, no localhost)
if !IsValidRemoteURL(obj.URL) {
return fmt.Errorf("%w: %s", ErrInvalidRemoteURL, obj.URL)
}
return nil
default:
return fmt.Errorf("unsupported transport type for remotes: %s (only streamable-http and sse are supported)", obj.Type)
}
}
// ValidatePublishRequest validates a complete publish request including extensions
func ValidatePublishRequest(req apiv0.ServerJSON, cfg *config.Config) error {
// Validate publisher extensions in _meta
if err := validatePublisherExtensions(req); err != nil {
return err
}
// Validate the server detail (includes all nested validation)
if err := ValidateServerJSON(&req); err != nil {
return err
}
// Validate registry ownership for all packages if validation is enabled and server is not deleted
if cfg.EnableRegistryValidation && req.Status != model.StatusDeleted {
ctx := context.Background()
for i, pkg := range req.Packages {
if err := ValidatePackage(ctx, pkg, req.Name); err != nil {
return fmt.Errorf("registry validation failed for package %d (%s): %w", i, pkg.Identifier, err)
}
}
}
return nil
}
func validatePublisherExtensions(req apiv0.ServerJSON) error {
const maxExtensionSize = 4 * 1024 // 4KB limit
// Check size limit for _meta publisher-provided extension
if req.Meta != nil && req.Meta.PublisherProvided != nil {
extensionsJSON, err := json.Marshal(req.Meta.PublisherProvided)
if err != nil {
return fmt.Errorf("failed to marshal _meta.io.modelcontextprotocol.registry/publisher-provided extension: %w", err)
}
if len(extensionsJSON) > maxExtensionSize {
return fmt.Errorf("_meta.io.modelcontextprotocol.registry/publisher-provided extension exceeds 4KB limit (%d bytes)", len(extensionsJSON))
}
}
if req.Meta != nil {
// Validate that only publisher-provided data is allowed in _meta during publish (no official registry metadata should be present)
if req.Meta.Official != nil {
return fmt.Errorf("official registry metadata '_meta.io.modelcontextprotocol.registry/official' is not allowed during publish")
}
}
return nil
}
func parseServerName(serverJSON apiv0.ServerJSON) (string, error) {
name := serverJSON.Name
if name == "" {
return "", fmt.Errorf("server name is required and must be a string")
}
// Validate format: dns-namespace/name
if !strings.Contains(name, "/") {
return "", fmt.Errorf("server name must be in format 'dns-namespace/name' (e.g., 'com.example.api/server')")
}
parts := strings.SplitN(name, "/", 2)
if len(parts) != 2 || parts[0] == "" || parts[1] == "" {
return "", fmt.Errorf("server name must be in format 'dns-namespace/name' with non-empty namespace and name parts")
}
return name, nil
}
// validateRemoteNamespaceMatch validates that remote URLs match the reverse-DNS namespace
func validateRemoteNamespaceMatch(serverJSON apiv0.ServerJSON) error {
namespace := serverJSON.Name
for _, remote := range serverJSON.Remotes {
if err := validateRemoteURLMatchesNamespace(remote.URL, namespace); err != nil {
return fmt.Errorf("remote URL %s does not match namespace %s: %w", remote.URL, namespace, err)
}
}
return nil
}
// validateRemoteURLMatchesNamespace checks if a remote URL's hostname matches the publisher domain from the namespace
func validateRemoteURLMatchesNamespace(remoteURL, namespace string) error {
// Parse the URL to extract the hostname
parsedURL, err := url.Parse(remoteURL)
if err != nil {
return fmt.Errorf("invalid URL format: %w", err)
}
hostname := parsedURL.Hostname()
if hostname == "" {
return fmt.Errorf("URL must have a valid hostname")
}
// Skip validation for localhost and local development URLs
if hostname == "localhost" || strings.HasSuffix(hostname, ".localhost") || hostname == "127.0.0.1" {
return nil
}
// Extract publisher domain from reverse-DNS namespace
publisherDomain := extractPublisherDomainFromNamespace(namespace)
if publisherDomain == "" {
return fmt.Errorf("invalid namespace format: cannot extract domain from %s", namespace)
}
// Check if the remote URL hostname matches the publisher domain or is a subdomain
if !isValidHostForDomain(hostname, publisherDomain) {
return fmt.Errorf("remote URL host %s does not match publisher domain %s", hostname, publisherDomain)
}
return nil
}
// extractPublisherDomainFromNamespace converts reverse-DNS namespace to normal domain format
// e.g., "com.example" -> "example.com"
func extractPublisherDomainFromNamespace(namespace string) string {
// Extract the namespace part before the first slash
namespacePart := namespace
if slashIdx := strings.Index(namespace, "/"); slashIdx != -1 {
namespacePart = namespace[:slashIdx]
}
// Split into parts and reverse them to get normal domain format
parts := strings.Split(namespacePart, ".")
if len(parts) < 2 {
return ""
}
// Reverse the parts to convert from reverse-DNS to normal domain
slices.Reverse(parts)
return strings.Join(parts, ".")
}
// isValidHostForDomain checks if a hostname is the domain or a subdomain of the publisher domain
func isValidHostForDomain(hostname, publisherDomain string) bool {
// Exact match
if hostname == publisherDomain {
return true
}
// Subdomain match - hostname should end with "." + publisherDomain
if strings.HasSuffix(hostname, "."+publisherDomain) {
return true
}
return false
}