seaweedfs/weed/s3api/s3tables/utils.go

package s3tables

import (
	"crypto/rand"
	"encoding/hex"
	"fmt"
	"net/url"
	"path"
	"regexp"
	"strings"
	"time"

	"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
)

const (
	bucketNamePatternStr     = `[a-z0-9-]+`
	tableNamespacePatternStr = `[a-z0-9_.-]+`
	tableNamePatternStr      = `[a-z0-9_-]+`
)

const (
	tableObjectRootDirName = ".objects"
)

var (
	bucketARNPattern = regexp.MustCompile(`^arn:aws:s3tables:[^:]*:[^:]*:bucket/(` + bucketNamePatternStr + `)$`)
	tableARNPattern  = regexp.MustCompile(`^arn:aws:s3tables:[^:]*:[^:]*:bucket/(` + bucketNamePatternStr + `)/table/(` + tableNamespacePatternStr + `)/(` + tableNamePatternStr + `)$`)
	tagPattern       = regexp.MustCompile(`^([\p{L}\p{Z}\p{N}_.:/=+\-@]*)$`)
)

// ARN parsing functions

// parseBucketNameFromARN extracts bucket name from table bucket ARN
// ARN format: arn:aws:s3tables:{region}:{account}:bucket/{bucket-name}
func parseBucketNameFromARN(arn string) (string, error) {
	matches := bucketARNPattern.FindStringSubmatch(arn)
	if len(matches) != 2 {
		return "", fmt.Errorf("invalid bucket ARN: %s", arn)
	}
	bucketName := matches[1]
	if !isValidBucketName(bucketName) {
		return "", fmt.Errorf("invalid bucket name in ARN: %s", bucketName)
	}
	return bucketName, nil
}

// ParseBucketNameFromARN is a wrapper to validate bucket ARN for other packages.
func ParseBucketNameFromARN(arn string) (string, error) {
	return parseBucketNameFromARN(arn)
}

// parseTableFromARN extracts bucket name, namespace, and table name from ARN
// ARN format: arn:aws:s3tables:{region}:{account}:bucket/{bucket-name}/table/{namespace}/{table-name}
func parseTableFromARN(arn string) (bucketName, namespace, tableName string, err error) {
	matches := tableARNPattern.FindStringSubmatch(arn)
	if len(matches) != 4 {
		return "", "", "", fmt.Errorf("invalid table ARN: %s", arn)
	}

	// Validate bucket name
	bucketName = matches[1]
	if err := validateBucketName(bucketName); err != nil {
		return "", "", "", fmt.Errorf("invalid bucket name in ARN: %v", err)
	}

	namespace, err = validateNamespace([]string{matches[2]})
	if err != nil {
		return "", "", "", fmt.Errorf("invalid namespace in ARN: %v", err)
	}

	// URL decode and validate the table name from the ARN path component
	tableNameUnescaped, err := url.PathUnescape(matches[3])
	if err != nil {
		return "", "", "", fmt.Errorf("invalid table name encoding in ARN: %v", err)
	}
	if _, err := validateTableName(tableNameUnescaped); err != nil {
		return "", "", "", fmt.Errorf("invalid table name in ARN: %v", err)
	}
	return bucketName, namespace, tableNameUnescaped, nil
}

// Path helpers

// GetTableBucketPath returns the filer path for a table bucket
func GetTableBucketPath(bucketName string) string {
	return path.Join(TablesPath, bucketName)
}

// GetNamespacePath returns the filer path for a namespace
func GetNamespacePath(bucketName, namespace string) string {
	return path.Join(TablesPath, bucketName, namespace)
}

// GetTablePath returns the filer path for a table
func GetTablePath(bucketName, namespace, tableName string) string {
	return path.Join(TablesPath, bucketName, namespace, tableName)
}

// TableDataDirFromMetadataLocation maps a table's s3:// metadata location to the
// filer directory holding its data. A renamed table is catalog-only, so its data
// stays at the original location while its catalog entry moves; this lets a drop
// purge the real data instead of the now-empty catalog path.
func TableDataDirFromMetadataLocation(metadataLocation string) string {
	loc := strings.TrimSuffix(metadataLocation, "/")
	if idx := strings.LastIndex(loc, "/metadata/"); idx != -1 {
		loc = loc[:idx]
	}
	loc = strings.TrimPrefix(loc, "s3://")
	if loc == "" {
		return ""
	}
	return path.Join(TablesPath, loc)
}

// GetTableObjectRootDir returns the root path for table bucket object storage
func GetTableObjectRootDir() string {
	return path.Join(TablesPath, tableObjectRootDirName)
}

// GetTableObjectBucketPath returns the filer path for table bucket object storage
func GetTableObjectBucketPath(bucketName string) string {
	return path.Join(GetTableObjectRootDir(), bucketName)
}

// Metadata structures

type tableBucketMetadata struct {
	Name           string    `json:"name"`
	CreatedAt      time.Time `json:"createdAt"`
	OwnerAccountID string    `json:"ownerAccountId"`
}

// namespaceMetadata stores metadata for a namespace
type namespaceMetadata struct {
	Namespace      []string          `json:"namespace"`
	CreatedAt      time.Time         `json:"createdAt"`
	OwnerAccountID string            `json:"ownerAccountId"`
	Properties     map[string]string `json:"properties,omitempty"`
}

// tableMetadataInternal stores metadata for a table
type tableMetadataInternal struct {
	Name             string         `json:"name"`
	Namespace        string         `json:"namespace"`
	Format           string         `json:"format"`
	CreatedAt        time.Time      `json:"createdAt"`
	ModifiedAt       time.Time      `json:"modifiedAt"`
	OwnerAccountID   string         `json:"ownerAccountId"`
	VersionToken     string         `json:"versionToken"`
	MetadataVersion  int            `json:"metadataVersion"`
	MetadataLocation string         `json:"metadataLocation,omitempty"`
	Metadata         *TableMetadata `json:"metadata,omitempty"`
}

// IsTableBucketEntry returns true when the entry is marked as a table bucket.
func IsTableBucketEntry(entry *filer_pb.Entry) bool {
	if entry == nil || entry.Extended == nil {
		return false
	}
	_, ok := entry.Extended[ExtendedKeyTableBucket]
	return ok
}

// entryType returns the entry-type marker for a catalog entry. Tables and views
// share the same on-disk layout; the marker distinguishes them. An absent marker
// means table for back-compat.
func entryType(extended map[string][]byte) string {
	if extended == nil {
		return EntryTypeTable
	}
	if v, ok := extended[ExtendedKeyEntryType]; ok && len(v) > 0 {
		return string(v)
	}
	return EntryTypeTable
}

// Utility functions

// validateBucketName validates bucket name and returns an error if invalid.
// Bucket names must contain only lowercase letters, numbers, and hyphens.
// Length must be between 3 and 63 characters.
// Must start and end with a letter or digit.
// Reserved prefixes/suffixes are rejected.
func validateBucketName(name string) error {
	if name == "" {
		return fmt.Errorf("bucket name is required")
	}

	if len(name) < 3 || len(name) > 63 {
		return fmt.Errorf("bucket name must be between 3 and 63 characters")
	}

	// Must start and end with a letter or digit
	start := name[0]
	end := name[len(name)-1]
	if !((start >= 'a' && start <= 'z') || (start >= '0' && start <= '9')) {
		return fmt.Errorf("bucket name must start with a letter or digit")
	}
	if !((end >= 'a' && end <= 'z') || (end >= '0' && end <= '9')) {
		return fmt.Errorf("bucket name must end with a letter or digit")
	}

	// Allowed characters: a-z, 0-9, -
	for i := 0; i < len(name); i++ {
		ch := name[i]
		if (ch >= 'a' && ch <= 'z') || (ch >= '0' && ch <= '9') || ch == '-' {
			continue
		}
		return fmt.Errorf("bucket name can only contain lowercase letters, numbers, and hyphens")
	}

	// Reserved prefixes
	reservedPrefixes := []string{"xn--", "sthree-", "amzn-s3-demo-", "aws"}
	for _, p := range reservedPrefixes {
		if strings.HasPrefix(name, p) {
			return fmt.Errorf("bucket name cannot start with reserved prefix: %s", p)
		}
	}

	// Reserved suffixes
	reservedSuffixes := []string{"-s3alias", "--ol-s3", "--x-s3", "--table-s3"}
	for _, s := range reservedSuffixes {
		if strings.HasSuffix(name, s) {
			return fmt.Errorf("bucket name cannot end with reserved suffix: %s", s)
		}
	}

	return nil
}

// BuildBucketARN builds a bucket ARN with the provided region and account ID.
// If region is empty, the ARN will omit the region field.
func BuildBucketARN(region, accountID, bucketName string) (string, error) {
	if bucketName == "" {
		return "", fmt.Errorf("bucket name is required")
	}
	if err := validateBucketName(bucketName); err != nil {
		return "", err
	}
	if accountID == "" {
		accountID = DefaultAccountID
	}
	return buildARN(region, accountID, fmt.Sprintf("bucket/%s", bucketName)), nil
}

// BuildTableARN builds a table ARN with the provided region and account ID.
func BuildTableARN(region, accountID, bucketName, namespace, tableName string) (string, error) {
	if bucketName == "" {
		return "", fmt.Errorf("bucket name is required")
	}
	if err := validateBucketName(bucketName); err != nil {
		return "", err
	}
	if namespace == "" {
		return "", fmt.Errorf("namespace is required")
	}
	normalizedNamespace, err := validateNamespace([]string{namespace})
	if err != nil {
		return "", err
	}
	if tableName == "" {
		return "", fmt.Errorf("table name is required")
	}
	normalizedTable, err := validateTableName(tableName)
	if err != nil {
		return "", err
	}
	if accountID == "" {
		accountID = DefaultAccountID
	}
	return buildARN(region, accountID, fmt.Sprintf("bucket/%s/table/%s/%s", bucketName, normalizedNamespace, normalizedTable)), nil
}

func buildARN(region, accountID, resourcePath string) string {
	return fmt.Sprintf("arn:aws:s3tables:%s:%s:%s", region, accountID, resourcePath)
}

// ValidateTags validates tags for S3 Tables.
func ValidateTags(tags map[string]string) error {
	if len(tags) > 10 {
		return fmt.Errorf("validate tags: %d tags more than 10", len(tags))
	}
	for k, v := range tags {
		if len(k) > 128 {
			return fmt.Errorf("validate tags: tag key longer than 128")
		}
		if !tagPattern.MatchString(k) {
			return fmt.Errorf("validate tags key %s error, incorrect key", k)
		}
		if len(v) > 256 {
			return fmt.Errorf("validate tags: tag value longer than 256")
		}
		if !tagPattern.MatchString(v) {
			return fmt.Errorf("validate tags value %s error, incorrect value", v)
		}
	}
	return nil
}

// isValidBucketName validates bucket name characters (kept for compatibility)
// Deprecated: use validateBucketName instead
func isValidBucketName(name string) bool {
	return validateBucketName(name) == nil
}

// generateVersionToken generates a unique, unpredictable version token
func generateVersionToken() string {
	b := make([]byte, 16)
	if _, err := rand.Read(b); err != nil {
		// Fallback to timestamp if crypto/rand fails
		return fmt.Sprintf("%x", time.Now().UnixNano())
	}
	return hex.EncodeToString(b)
}

// splitPath splits a path into directory and name components using stdlib
func splitPath(p string) (dir, name string) {
	dir = path.Dir(p)
	name = path.Base(p)
	return
}

func validateNamespacePart(name string) error {
	if len(name) < 1 || len(name) > 255 {
		return fmt.Errorf("namespace name must be between 1 and 255 characters")
	}

	// Prevent path traversal and multi-segment paths
	if name == "." || name == ".." {
		return fmt.Errorf("namespace name cannot be '.' or '..'")
	}
	if strings.Contains(name, "/") {
		return fmt.Errorf("namespace name cannot contain '/'")
	}

	// Must start and end with a letter or digit
	start := name[0]
	end := name[len(name)-1]
	if !((start >= 'a' && start <= 'z') || (start >= '0' && start <= '9')) {
		return fmt.Errorf("namespace name must start with a letter or digit")
	}
	if !((end >= 'a' && end <= 'z') || (end >= '0' && end <= '9')) {
		return fmt.Errorf("namespace name must end with a letter or digit")
	}

	// Allowed characters: a-z, 0-9, _, - (hyphen interior; start/end checked above)
	for _, ch := range name {
		if (ch >= 'a' && ch <= 'z') || (ch >= '0' && ch <= '9') || ch == '_' || ch == '-' {
			continue
		}
		return fmt.Errorf("invalid namespace name: only 'a-z', '0-9', '_', and '-' are allowed")
	}

	// Reserved prefix
	if strings.HasPrefix(name, "aws") {
		return fmt.Errorf("namespace name cannot start with reserved prefix 'aws'")
	}

	return nil
}

func normalizeNamespace(namespace []string) ([]string, error) {
	if len(namespace) == 0 {
		return nil, fmt.Errorf("namespace is required")
	}

	parts := namespace
	if len(namespace) == 1 {
		parts = strings.Split(namespace[0], ".")
	}

	normalized := make([]string, 0, len(parts))
	for _, part := range parts {
		if err := validateNamespacePart(part); err != nil {
			return nil, err
		}
		normalized = append(normalized, part)
	}
	return normalized, nil
}

// validateNamespace validates namespace identifiers and returns an internal namespace key.
// A single dotted namespace value is interpreted as multi-level namespace for compatibility
// with path-style APIs, for example "analytics.daily" => ["analytics", "daily"].
func validateNamespace(namespace []string) (string, error) {
	parts, err := normalizeNamespace(namespace)
	if err != nil {
		return "", err
	}
	return flattenNamespace(parts), nil
}

// ParseNamespace parses a namespace string into namespace parts.
func ParseNamespace(namespace string) ([]string, error) {
	return normalizeNamespace([]string{namespace})
}

// validateTableName validates a table name
func validateTableName(name string) (string, error) {
	if len(name) < 1 || len(name) > 255 {
		return "", fmt.Errorf("table name must be between 1 and 255 characters")
	}
	if name == "." || name == ".." || strings.Contains(name, "/") {
		return "", fmt.Errorf("invalid table name: cannot be '.', '..' or contain '/'")
	}

	// First character must be a letter or digit
	start := name[0]
	if !((start >= 'a' && start <= 'z') || (start >= '0' && start <= '9')) {
		return "", fmt.Errorf("table name must start with a letter or digit")
	}

	// Allowed characters: a-z, 0-9, _, - (start checked above)
	for _, ch := range name {
		if (ch >= 'a' && ch <= 'z') || (ch >= '0' && ch <= '9') || ch == '_' || ch == '-' {
			continue
		}
		return "", fmt.Errorf("invalid table name: only 'a-z', '0-9', '_', and '-' are allowed")
	}
	return name, nil
}

// ValidateTableName is a wrapper to validate table name for other packages.
func ValidateTableName(name string) (string, error) {
	return validateTableName(name)
}

// flattenNamespace joins namespace elements into a single string (using dots as per AWS S3 Tables)
func flattenNamespace(namespace []string) string {
	if len(namespace) == 0 {
		return ""
	}
	return strings.Join(namespace, ".")
}

func expandNamespace(namespace string) []string {
	if namespace == "" {
		return nil
	}
	parts, err := ParseNamespace(namespace)
	if err != nil {
		return []string{namespace}
	}
	return parts
}