Files
seaweedfs/weed/s3api/s3tables/iceberg_layout.go
T
qzhelloandChris Lu 69c84801e4 fix(s3tables/iceberg): make metadata spec-compliant and accept real-world manifest names (#9703)
* fix(s3tables/iceberg): make metadata spec-compliant and accept real-world manifest names

Two related issues prevent SeaweedFS S3 Tables from interoperating with
strict Iceberg clients (Java/Spark/Flink/Trino):

1. iceberg-go v0.5.0 serializes empty TableMetadata state by dropping
   keys via `omitempty` on optional pointer/slice fields. The Iceberg
   table spec, however, requires `current-snapshot-id`, `snapshots`,
   `snapshot-log`, `metadata-log`, and `refs` to be present even when
   empty (`current-snapshot-id` must be -1 for a table with no
   snapshots). Java's TableMetadataParser uses JsonUtil.getLong on
   `current-snapshot-id` and throws "Cannot parse missing long
   current-snapshot-id" against responses produced by this server.

2. The Iceberg layout validator only accepts manifest filenames that
   match Iceberg's internal naming (`{uuid}-m{n}.avro`,
   `snap-{n}-{n}-{uuid}.avro`). Real writers — notably Flink's sink —
   emit manifests like
   `{flink-job-id}-{checkpoint}-{operator-id}-{n}.avro`, which the
   validator rejects with 403, breaking INSERT commits.

Fixes:

* Add ensureMetadataSpecCompliance helper that backfills the five
  spec-required empty-state fields when iceberg-go omits them or emits
  explicit JSON null. Apply it on every code path that writes
  v*.metadata.json to S3 or returns metadata to clients
  (handlers_table create-table, handlers_commit, commit_helpers
  create-on-commit, plus MarshalJSON on LoadTableResult and
  CommitTableResponse). Real values from non-empty tables are never
  overwritten.

* Add catch-all regex entries to metadataFilePatterns accepting any
  *.avro / *.metadata.json filename composed of [A-Za-z0-9._-]. The
  Iceberg spec does not mandate filename format; the strict patterns
  remain for documentation. Metadata-directory subdirectory rejection
  and the data-file path validation are unchanged.

No upstream dependencies are forked: iceberg-go stays at v0.5.0 and
go.mod is untouched. The compliance layer can be removed once upstream
emits spec-compliant output.

Tests (all pass under `go test -race`):
- metadata_compliance_test.go: 5 cases covering missing fields,
  preserved real values, explicit null, invalid JSON, empty input.
- iceberg_layout_test.go: 3 groups (16 subtests) covering real-world
  manifest names from Flink/Spark/Iceberg, security boundary
  (subdirectories, bad extensions), and data-file regression.

* fix(s3tables/iceberg): preserve metadata key order and keep config field stable

Two small follow-ups on the spec-compliance fix:

* ensureMetadataSpecCompliance now splices missing keys in at the byte
  level just before the closing brace, so iceberg-go's struct-declared
  key order survives the backfill. The previous unmarshal/remarshal
  through map[string]json.RawMessage silently alphabetized every key in
  the document, which is spec-legal but breaks byte-equality fixtures
  and any downstream hashing of the persisted metadata. The slower
  remarshal path is kept for the rare explicit-null replacement case.

* LoadTableResult.MarshalJSON now serializes Config without omitempty,
  matching the struct field tag. The custom marshaler had silently
  flipped the tag to ,omitempty, which made the "config" key disappear
  from the response whenever s3Endpoint was unset (since
  buildFileIOConfig returned an empty but non-nil Properties map).

Tests:
- PreservesOriginalKeyOrder pins the byte-level output against
  iceberg-go's emitted shape; would have caught the alphabetization
  regression.
- EmptyObjectBackfilled covers the {} -> sentinels-only case (no
  leading comma).
- AllPresentReturnsSameBytes confirms the no-op path returns input
  bytes unchanged, with whitespace intact.
- iceberg_layout_test pins the catch-all $ anchor: metadata/file.avro.txt
  must still be rejected.

* fix(s3tables/iceberg): guard ensureMetadataSpecCompliance against top-level null

json.Unmarshal of a JSON `null` literal succeeds but leaves the map nil.
The current byte-append path no-ops gracefully on this input, but the
slow remarshal path would panic with "assignment to entry in nil map"
if the input ever combined `null` with the explicit-null detection. Add
an explicit nil-map short-circuit so the safety property is obvious
from the source, and a test that pins the contract.

* test(s3tables/iceberg): assert full byte equality in AllPresentReturnsSameBytes

The prefix check only caught a missing "{\n  " opener, so the test
would have passed even if the function silently reordered keys or
collapsed whitespace later in the document. Switch to a full string
comparison so any future regression in the no-op path is loud.

---------

Co-authored-by: Chris Lu <chris.lu@gmail.com>
2026-05-27 13:05:41 -07:00

321 lines
11 KiB
Go

package s3tables
import (
pathpkg "path"
"regexp"
"strings"
)
// Iceberg file layout validation
// Apache Iceberg tables follow a specific file layout structure:
// - metadata/ directory containing metadata files (*.json, *.avro)
// - data/ directory containing data files (*.parquet, *.orc, *.avro)
//
// Valid file patterns include:
// - metadata/v*.metadata.json (table metadata)
// - metadata/snap-*.avro (snapshot manifest lists)
// - metadata/*.avro (manifest files)
// - data/*.parquet, data/*.orc, data/*.avro (data files)
const uuidPattern = `[a-f0-9]{8}-[a-f0-9]{4}-[a-f0-9]{4}-[a-f0-9]{4}-[a-f0-9]{12}`
var (
// Allowed directories in an Iceberg table
icebergAllowedDirs = map[string]bool{
"metadata": true,
"data": true,
}
// Patterns for valid metadata files.
//
// Note: Iceberg engines (Flink/Spark/Trino, plus different Iceberg versions)
// use a variety of manifest/snapshot naming schemes that the strict patterns
// below don't cover - e.g. Flink emits manifests like
// "{flink-job-id}-{checkpoint}-{operator-id}-{n}.avro". The Iceberg spec
// itself doesn't mandate a specific filename; engines just need a stable
// unique name. So in addition to the strict patterns we keep below for
// documentation, the catch-all entries at the bottom accept any
// *.avro and *.metadata.json filename composed of safe characters
// ([A-Za-z0-9._-]). The catch-all deliberately does NOT cover arbitrary
// *.json — only the *.metadata.json suffix — to avoid letting random
// JSON drop into the metadata/ directory.
metadataFilePatterns = []*regexp.Regexp{
regexp.MustCompile(`^v\d+\.metadata\.json$`), // Table metadata: v1.metadata.json, v2.metadata.json
regexp.MustCompile(`^snap-\d+-\d+-` + uuidPattern + `\.avro$`), // Snapshot manifests: snap-123-1-uuid.avro
regexp.MustCompile(`^` + uuidPattern + `-m\d+\.avro$`), // Manifest files: uuid-m0.avro
regexp.MustCompile(`^` + uuidPattern + `\.avro$`), // General manifest files
regexp.MustCompile(`^version-hint\.text$`), // Version hint file
regexp.MustCompile(`^` + uuidPattern + `\.metadata\.json$`), // UUID-named metadata
regexp.MustCompile(`^[^/]+\.stats$`), // Trino/Iceberg stats files
// Catch-all for Iceberg writer-generated manifest / metadata files
// whose naming we can't anticipate across engines and versions.
regexp.MustCompile(`^[A-Za-z0-9._-]+\.avro$`),
regexp.MustCompile(`^[A-Za-z0-9._-]+\.metadata\.json$`),
}
// Patterns for valid data files
dataFilePatterns = []*regexp.Regexp{
regexp.MustCompile(`^[^/]+\.parquet$`), // Parquet files
regexp.MustCompile(`^[^/]+\.orc$`), // ORC files
regexp.MustCompile(`^[^/]+\.avro$`), // Avro files
}
// Data file partition path pattern (e.g., year=2024/month=01/)
partitionPathPattern = regexp.MustCompile(`^[a-zA-Z_][a-zA-Z0-9_]*=[^/]+$`)
// Pattern for valid subdirectory names (alphanumeric, underscore, hyphen, and UUID-style directories)
validSubdirectoryPattern = regexp.MustCompile(`^[a-zA-Z0-9_-]+$`)
)
// IcebergLayoutValidator validates that files conform to Iceberg table layout
type IcebergLayoutValidator struct{}
// NewIcebergLayoutValidator creates a new Iceberg layout validator
func NewIcebergLayoutValidator() *IcebergLayoutValidator {
return &IcebergLayoutValidator{}
}
// ValidateFilePath validates that a file path conforms to Iceberg layout
// The path should be relative to the table root (e.g., "metadata/v1.metadata.json" or "data/file.parquet")
func (v *IcebergLayoutValidator) ValidateFilePath(relativePath string) error {
// Normalize path separators
relativePath = strings.TrimPrefix(relativePath, "/")
if relativePath == "" {
return &IcebergLayoutError{
Code: ErrCodeInvalidIcebergLayout,
Message: "empty file path",
}
}
parts := strings.SplitN(relativePath, "/", 2)
topDir := parts[0]
// Check if top-level directory is allowed
if !icebergAllowedDirs[topDir] {
return &IcebergLayoutError{
Code: ErrCodeInvalidIcebergLayout,
Message: "files must be placed in 'metadata/' or 'data/' directories",
}
}
// If it's just a bare top-level key (no trailing slash and no subpath), reject it
if len(parts) == 1 {
return &IcebergLayoutError{
Code: ErrCodeInvalidIcebergLayout,
Message: "must be a directory (use trailing slash) or contain a subpath",
}
}
remainingPath := parts[1]
if remainingPath == "" {
return nil // allow paths like "data/" or "metadata/"
}
switch topDir {
case "metadata":
return v.validateMetadataFile(remainingPath)
case "data":
return v.validateDataFile(remainingPath)
}
return nil
}
// validateDirectoryPath validates intermediate subdirectories in a path
// isMetadata indicates if we're in the metadata directory (true) or data directory (false)
func validateDirectoryPath(normalizedPath string, isMetadata bool) error {
if isMetadata {
// For metadata, reject any subdirectories (enforce flat structure under metadata/)
return &IcebergLayoutError{
Code: ErrCodeInvalidIcebergLayout,
Message: "metadata directory does not support subdirectories",
}
}
// For data, validate each partition or subdirectory segment
subdirs := strings.Split(normalizedPath, "/")
for _, subdir := range subdirs {
if subdir == "" {
return &IcebergLayoutError{
Code: ErrCodeInvalidIcebergLayout,
Message: "invalid partition or subdirectory format in data path: empty segment",
}
}
// For data, allow both partitions and valid subdirectories
if !partitionPathPattern.MatchString(subdir) && !isValidSubdirectory(subdir) {
return &IcebergLayoutError{
Code: ErrCodeInvalidIcebergLayout,
Message: "invalid partition or subdirectory format in data path",
}
}
}
return nil
}
// validateFilePatterns validates a filename against allowed patterns
// isMetadata indicates if we're validating metadata files (true) or data files (false)
func validateFilePatterns(filename string, isMetadata bool) error {
var patterns []*regexp.Regexp
var errorMsg string
if isMetadata {
patterns = metadataFilePatterns
errorMsg = "invalid metadata file format: must be a valid Iceberg metadata, manifest, or snapshot file"
} else {
patterns = dataFilePatterns
errorMsg = "invalid data file format: must be .parquet, .orc, or .avro"
}
for _, pattern := range patterns {
if pattern.MatchString(filename) {
return nil
}
}
return &IcebergLayoutError{
Code: ErrCodeInvalidIcebergLayout,
Message: errorMsg,
}
}
// validateFile validates files with a unified logic for metadata and data directories
// isMetadata indicates whether we're validating metadata files (true) or data files (false)
// The logic is:
// 1. If path ends with "/", it's a directory - validate all parts and return nil
// 2. Otherwise, validate intermediate parts, then check the filename against patterns
func (v *IcebergLayoutValidator) validateFile(path string, isMetadata bool) error {
// Detect if it's a directory (path ends with "/")
if strings.HasSuffix(path, "/") {
// Normalize by removing trailing slash
normalizedPath := strings.TrimSuffix(path, "/")
return validateDirectoryPath(normalizedPath, isMetadata)
}
filename := pathpkg.Base(path)
// Validate intermediate subdirectories if present
// Find if there are intermediate directories by looking for the last slash
lastSlash := strings.LastIndex(path, "/")
if lastSlash != -1 {
dir := path[:lastSlash]
if err := validateDirectoryPath(dir, isMetadata); err != nil {
return err
}
}
// Check against allowed file patterns
err := validateFilePatterns(filename, isMetadata)
if err == nil {
return nil
}
// Path could be for a directory without a trailing slash, e.g., "data/year=2024"
if !isMetadata {
if partitionPathPattern.MatchString(filename) || isValidSubdirectory(filename) {
return nil
}
}
return err
}
// validateMetadataFile validates files in the metadata/ directory
// This is a thin wrapper that calls validateFile with isMetadata=true
func (v *IcebergLayoutValidator) validateMetadataFile(path string) error {
return v.validateFile(path, true)
}
// validateDataFile validates files in the data/ directory
// This is a thin wrapper that calls validateFile with isMetadata=false
func (v *IcebergLayoutValidator) validateDataFile(path string) error {
return v.validateFile(path, false)
}
// isValidSubdirectory checks if a path component is a valid subdirectory name
func isValidSubdirectory(name string) bool {
// Allow alphanumeric, underscore, hyphen, and UUID-style directories
return validSubdirectoryPattern.MatchString(name)
}
// IcebergLayoutError represents an Iceberg layout validation error
type IcebergLayoutError struct {
Code string
Message string
}
func (e *IcebergLayoutError) Error() string {
return e.Message
}
// Error code for Iceberg layout violations
const (
ErrCodeInvalidIcebergLayout = "InvalidIcebergLayout"
)
// TableBucketFileValidator validates file uploads to table buckets
type TableBucketFileValidator struct {
layoutValidator *IcebergLayoutValidator
}
// NewTableBucketFileValidator creates a new table bucket file validator
func NewTableBucketFileValidator() *TableBucketFileValidator {
return &TableBucketFileValidator{
layoutValidator: NewIcebergLayoutValidator(),
}
}
// ValidateTableBucketUpload checks if a file upload to a table bucket conforms to Iceberg layout
// fullPath is the complete filer path (e.g., /buckets/mybucket/mynamespace/mytable/data/file.parquet)
// Returns nil if the path is not a table bucket path or if validation passes
// Returns an error if the file doesn't conform to Iceberg layout
func (v *TableBucketFileValidator) ValidateTableBucketUpload(fullPath string) error {
// Check if this is a table bucket path
if !strings.HasPrefix(fullPath, TablesPath+"/") {
return nil // Not a table bucket, no validation needed
}
// Extract the path relative to table bucket root
// Format: /buckets/{bucket}/{namespace}/{table}/{relative-path}
relativePath := strings.TrimPrefix(fullPath, TablesPath+"/")
parts := strings.SplitN(relativePath, "/", 4)
// Need at least bucket/namespace/table/file
if len(parts) < 4 {
// Creating bucket, namespace, or table directories - allow only if preceding parts are non-empty
for i := 0; i < len(parts); i++ {
if parts[i] == "" {
return &IcebergLayoutError{
Code: ErrCodeInvalidIcebergLayout,
Message: "bucket, namespace, and table segments cannot be empty",
}
}
}
return nil
}
// For full paths, also verify bucket, namespace, and table segments are non-empty
if parts[0] == "" || parts[1] == "" || parts[2] == "" {
return &IcebergLayoutError{
Code: ErrCodeInvalidIcebergLayout,
Message: "bucket, namespace, and table segments cannot be empty",
}
}
// The last part is the path within the table (data/file.parquet or metadata/v1.json)
tableRelativePath := parts[3]
if tableRelativePath == "" {
return nil
}
// Reject paths with empty segments (double slashes) within the table path
if strings.HasPrefix(tableRelativePath, "/") || strings.Contains(tableRelativePath, "//") {
return &IcebergLayoutError{
Code: ErrCodeInvalidIcebergLayout,
Message: "bucket, namespace, and table segments cannot be empty",
}
}
return v.layoutValidator.ValidateFilePath(tableRelativePath)
}