Files
seaweedfs/weed/s3api/s3_constants/header.go
T
Chris Lu 863fec6c3f S3: let a key that is a prefix of other keys be an object (#10912)
* filer: keep the sentinel when CreateEntry reports an update failure

CreateEntry flattened the error UpdateEntry wraps, so errors.Is stopped
matching and ErrExistingIsDirectory and ErrExistingIsFile never reached
the S3 mapper, which answered a retryable 500 instead.

* s3: let a key that is a prefix of other keys be an object

S3 keys are flat, so "a/b" and "a/b/c" are independent objects that
coexist in either write order. The filer stores a key as a path, so one
of them has to live on the directory the other is nested under.

Writing the nested key first refused the prefix key outright. Writing it
second promoted the file to a directory, which kept its data but lost the
key: an empty object left nothing to recognise it by and disappeared, and
one with data listed under a trailing slash it never had.

Mark the directory that carries such a key, and write the object onto it
when the path is already a directory. The mark makes an empty prefix
object visible to listings and readable by GET and HEAD, keeps the empty
folder cleaner off it, and lists it under the key it was written with.
Deleting the key strips the mark back off along with the data.

* filer: keep a TTL off a directory that stands for an object

An expired entry is deleted a row at a time, so expiring a directory
removes it and leaves everything under it unreachable. Promoting a file
to a directory carried its TTL across, and a promoted file is exactly the
one that has keys nested under it.

Drop the TTL on promotion, and leave one an older build wrote alone. The
lifecycle worker still expires the object, through the delete that leaves
the directory behind.

* s3: delete the null version of a key other keys are nested under

The routed delete cannot remove an entry that other keys live under, and
answered a retryable 500 rather than falling back to the lock path the
unversioned delete already falls back to. That path then looked the entry
up under the bucket with the whole key as its name, so the demote wrote it
back one directory too high and failed as not found.

Fall back on any non-precondition error, and split the key before deleting
it. Trailing-slash directory markers with children reach the same delete.

* filer: keep the sentinel when MkFile and Mkdir report a create failure

Same flattening one layer out: every mkFile caller lost the sentinel, so
a CopyObject onto a key that other keys are nested under answered a
retryable 500 where a PutObject of the same key answers 409.

* s3: copy and rename a key that other keys are nested under

Such a key is stored on the directory those keys live in, and copy and
rename both refused it: the source lookup maps every directory entry to
NoSuchKey, so a key a plain GET serves could not be copied or moved, and
the destination side refused it as a directory conflict.

The source is read through a view of the entry as the object it names.
The destination is written the way a PutObject of that key writes it. A
rename at either end copies the object's own data across and strips it off
the source key rather than going through AtomicRenameEntry, which moves a
directory by moving everything under it - the nested keys are not part of
what is being renamed.
2026-08-24 15:10:34 -07:00

423 lines
17 KiB
Go

/*
* MinIO Cloud Storage, (C) 2019 MinIO, Inc.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package s3_constants
import (
"context"
"net/http"
"strings"
"sync/atomic"
"github.com/gorilla/mux"
)
// S3 XML namespace
const (
S3Namespace = "http://s3.amazonaws.com/doc/2006-03-01/"
)
// S3 object key limits
const (
MaxS3ObjectKeyLength = 1024
)
// Standard S3 HTTP request constants
const (
// S3 storage class
AmzStorageClass = "X-Amz-Storage-Class"
// S3 user-defined metadata
AmzUserMetaPrefix = "X-Amz-Meta-"
AmzUserMetaDirective = "X-Amz-Metadata-Directive"
AmzUserMetaMtime = "X-Amz-Meta-Mtime"
// S3 object tagging
AmzObjectTagging = "X-Amz-Tagging"
AmzObjectTaggingPrefix = "X-Amz-Tagging-"
AmzObjectTaggingDirective = "X-Amz-Tagging-Directive"
AmzTagCount = "x-amz-tagging-count"
// GetObjectAttributes headers
AmzObjectAttributes = "X-Amz-Object-Attributes"
AmzMaxParts = "X-Amz-Max-Parts"
AmzPartNumberMarker = "X-Amz-Part-Number-Marker"
AmzDeleteMarker = "X-Amz-Delete-Marker"
SeaweedFSUploadId = "X-Seaweedfs-Upload-Id"
SeaweedFSMultipartPartsCount = "X-Seaweedfs-Multipart-Parts-Count"
SeaweedFSMultipartPartBoundaries = "X-Seaweedfs-Multipart-Part-Boundaries" // JSON: [{part:1,start:0,end:2,etag:"abc"},{part:2,start:2,end:3,etag:"def"}]
SeaweedFSExpiresS3 = "X-Seaweedfs-Expires-S3"
AmzMpPartsCount = "x-amz-mp-parts-count"
// S3 ACL headers
AmzCannedAcl = "X-Amz-Acl"
AmzAclFullControl = "X-Amz-Grant-Full-Control"
AmzAclRead = "X-Amz-Grant-Read"
AmzAclWrite = "X-Amz-Grant-Write"
AmzAclReadAcp = "X-Amz-Grant-Read-Acp"
AmzAclWriteAcp = "X-Amz-Grant-Write-Acp"
// S3 Object Lock headers
AmzBucketObjectLockEnabled = "X-Amz-Bucket-Object-Lock-Enabled"
AmzObjectLockMode = "X-Amz-Object-Lock-Mode"
AmzObjectLockRetainUntilDate = "X-Amz-Object-Lock-Retain-Until-Date"
AmzObjectLockLegalHold = "X-Amz-Object-Lock-Legal-Hold"
// S3 checksum headers
AmzChecksumAlgorithm = "X-Amz-Checksum-Algorithm"
AmzChecksumCRC32 = "X-Amz-Checksum-Crc32"
AmzChecksumCRC32C = "X-Amz-Checksum-Crc32c"
AmzChecksumCRC64NVME = "X-Amz-Checksum-Crc64nvme"
AmzChecksumSHA1 = "X-Amz-Checksum-Sha1"
AmzChecksumSHA256 = "X-Amz-Checksum-Sha256"
AmzTrailer = "X-Amz-Trailer"
AmzSdkChecksumAlgorithm = "X-Amz-Sdk-Checksum-Algorithm"
AmzChecksumType = "X-Amz-Checksum-Type"
// S3 checksum type values (x-amz-checksum-type). A COMPOSITE checksum is a
// checksum-of-per-part-checksums ("base64-N"); a FULL_OBJECT checksum is the
// checksum of the whole object as if uploaded in a single request ("base64").
ChecksumTypeComposite = "COMPOSITE"
ChecksumTypeFullObject = "FULL_OBJECT"
// S3 conditional headers
IfMatch = "If-Match"
IfNoneMatch = "If-None-Match"
IfModifiedSince = "If-Modified-Since"
IfUnmodifiedSince = "If-Unmodified-Since"
// S3 conditional copy headers
AmzCopySourceIfMatch = "X-Amz-Copy-Source-If-Match"
AmzCopySourceIfNoneMatch = "X-Amz-Copy-Source-If-None-Match"
AmzCopySourceIfModifiedSince = "X-Amz-Copy-Source-If-Modified-Since"
AmzCopySourceIfUnmodifiedSince = "X-Amz-Copy-Source-If-Unmodified-Since"
// RenameObject
AmzRenameSource = "X-Amz-Rename-Source"
AmzRenameSourceIfMatch = "X-Amz-Rename-Source-If-Match"
AmzRenameSourceIfNoneMatch = "X-Amz-Rename-Source-If-None-Match"
AmzRenameSourceIfModifiedSince = "X-Amz-Rename-Source-If-Modified-Since"
AmzRenameSourceIfUnmodifiedSince = "X-Amz-Rename-Source-If-Unmodified-Since"
// S3 Server-Side Encryption with Customer-provided Keys (SSE-C)
AmzServerSideEncryptionCustomerAlgorithm = "X-Amz-Server-Side-Encryption-Customer-Algorithm"
AmzServerSideEncryptionCustomerKey = "X-Amz-Server-Side-Encryption-Customer-Key"
AmzServerSideEncryptionCustomerKeyMD5 = "X-Amz-Server-Side-Encryption-Customer-Key-MD5"
AmzServerSideEncryptionContext = "X-Amz-Server-Side-Encryption-Context"
// S3 Server-Side Encryption with KMS (SSE-KMS)
AmzServerSideEncryption = "X-Amz-Server-Side-Encryption"
AmzServerSideEncryptionAwsKmsKeyId = "X-Amz-Server-Side-Encryption-Aws-Kms-Key-Id"
AmzServerSideEncryptionBucketKeyEnabled = "X-Amz-Server-Side-Encryption-Bucket-Key-Enabled"
// S3 SSE-C copy source headers
AmzCopySourceServerSideEncryptionCustomerAlgorithm = "X-Amz-Copy-Source-Server-Side-Encryption-Customer-Algorithm"
AmzCopySourceServerSideEncryptionCustomerKey = "X-Amz-Copy-Source-Server-Side-Encryption-Customer-Key"
AmzCopySourceServerSideEncryptionCustomerKeyMD5 = "X-Amz-Copy-Source-Server-Side-Encryption-Customer-Key-MD5"
)
// Metadata keys for internal storage
const (
// SSE-KMS metadata keys
AmzEncryptedDataKey = "x-amz-encrypted-data-key"
AmzEncryptionContextMeta = "x-amz-encryption-context"
// SeaweedFS internal metadata prefix (used to filter internal headers from client responses)
SeaweedFSInternalPrefix = "x-seaweedfs-"
// SeaweedFSPrefixObject marks a directory entry that also holds the object named by
// its path without a trailing slash. S3 keys are flat, so "a/b" and "a/b/c" are
// independent keys, and the filer stores the first one on the directory the second
// one lives under.
SeaweedFSPrefixObject = "x-seaweedfs-prefix-object"
// SeaweedFS internal metadata keys for encryption (prefixed to avoid automatic HTTP header conversion)
SeaweedFSSSEKMSKey = "x-seaweedfs-sse-kms-key" // Key for storing serialized SSE-KMS metadata
SeaweedFSSSES3Key = "x-seaweedfs-sse-s3-key" // Key for storing serialized SSE-S3 metadata
SeaweedFSSSEIV = "x-seaweedfs-sse-c-iv" // Key for storing SSE-C IV
// Multipart upload metadata keys for SSE-KMS (consistent with internal metadata key pattern)
SeaweedFSSSEKMSKeyID = "x-seaweedfs-sse-kms-key-id" // Key ID for multipart upload SSE-KMS inheritance
SeaweedFSSSEKMSEncryption = "x-seaweedfs-sse-kms-encryption" // Encryption type for multipart upload SSE-KMS inheritance
SeaweedFSSSEKMSBucketKeyEnabled = "x-seaweedfs-sse-kms-bucket-key-enabled" // Bucket key setting for multipart upload SSE-KMS inheritance
SeaweedFSSSEKMSEncryptionContext = "x-seaweedfs-sse-kms-encryption-context" // Encryption context for multipart upload SSE-KMS inheritance
SeaweedFSSSEKMSBaseIV = "x-seaweedfs-sse-kms-base-iv" // Base IV for multipart upload SSE-KMS (for IV offset calculation)
// Multipart upload metadata keys for SSE-S3
SeaweedFSSSES3Encryption = "x-seaweedfs-sse-s3-encryption" // Encryption type for multipart upload SSE-S3 inheritance
SeaweedFSSSES3BaseIV = "x-seaweedfs-sse-s3-base-iv" // Base IV for multipart upload SSE-S3 (for IV offset calculation)
SeaweedFSSSES3KeyData = "x-seaweedfs-sse-s3-key-data" // Encrypted key data for multipart upload SSE-S3 inheritance
)
// SeaweedFS internal headers for filer communication
const (
SeaweedFSSSEKMSKeyHeader = "X-SeaweedFS-SSE-KMS-Key" // Header for passing SSE-KMS metadata to filer
SeaweedFSSSEIVHeader = "X-SeaweedFS-SSE-IV" // Header for passing SSE-C IV to filer (SSE-C only)
SeaweedFSSSEKMSBaseIVHeader = "X-SeaweedFS-SSE-KMS-Base-IV" // Header for passing base IV for multipart SSE-KMS
SeaweedFSSSES3BaseIVHeader = "X-SeaweedFS-SSE-S3-Base-IV" // Header for passing base IV for multipart SSE-S3
SeaweedFSSSES3KeyDataHeader = "X-SeaweedFS-SSE-S3-Key-Data" // Header for passing key data for multipart SSE-S3
// SeaweedFSPrincipalHeader and SeaweedFSSessionTokenHeader carry the
// server-derived principal ARN and session token from the auth layer to
// downstream authorization (IAM policy evaluation, S3Tables IAM checks).
// These headers are trusted: callers of AuthSignatureOnly /
// authenticateRequestInternal must scrub any client-supplied values at the
// very start of authentication to prevent privilege escalation via header
// injection. Prefer these constants over literals so that any future scrub
// or consumer cannot drift by typo.
SeaweedFSPrincipalHeader = "X-SeaweedFS-Principal"
SeaweedFSSessionTokenHeader = "X-SeaweedFS-Session-Token"
)
// Non-Standard S3 HTTP request constants
const (
AmzIdentityId = "s3-identity-id"
AmzAccountId = "s3-account-id"
AmzAuthType = "s3-auth-type"
)
func GetBucketAndObject(r *http.Request) (bucket, object string) {
vars := mux.Vars(r)
bucket = vars["bucket"]
object = NormalizeObjectKey(vars["object"])
return
}
// IsValidObjectKey rejects S3 object keys that — after normalization
// (backslash→slash, slash collapse) — contain a `.` or `..` path segment, or
// embed a NUL byte. Such keys are collapsed by filepath.Join inside the filer
// and would escape the bucket directory, so they must not reach the gRPC layer.
// Gorilla mux URL-decodes captured vars before this runs, so `%2e%2e` is
// already `..` here.
func IsValidObjectKey(object string) bool {
if object == "" {
return true
}
if strings.ContainsRune(object, '\x00') {
return false
}
object = strings.ReplaceAll(object, "\\", "/")
for _, seg := range strings.Split(object, "/") {
if seg == "." || seg == ".." {
return false
}
}
return true
}
// IsValidBucketName rejects bucket names captured from the URL path that are
// unsafe to use in filer path construction (`.`, `..`, contain `/` or `\`, or
// embed NUL). This is a path-safety check, not a full S3 naming-rule check.
func IsValidBucketName(bucket string) bool {
if bucket == "" {
return true
}
if bucket == "." || bucket == ".." {
return false
}
return !strings.ContainsAny(bucket, "/\\\x00")
}
// IsValidPathSegment reports whether value is safe to use as one filer path
// component. It intentionally does not enforce any application-specific
// format; callers can layer stricter checks on top.
func IsValidPathSegment(value string) bool {
if value == "" || value == "." || value == ".." {
return false
}
return !strings.ContainsAny(value, "/\\\x00")
}
// NormalizeObjectKey normalizes object keys by removing duplicate slashes and converting backslashes.
// This normalizes keys from various sources (URL path, form values, etc.) to a consistent format.
// It also converts Windows-style backslashes to forward slashes for cross-platform compatibility.
// Returns keys WITHOUT leading slash to match S3 API format (e.g., "foo/bar" not "/foo/bar").
// Preserves trailing slash if present (e.g., "foo/" stays "foo/").
func NormalizeObjectKey(object string) string {
// Preserve trailing slash if present
hasTrailingSlash := strings.HasSuffix(object, "/")
// Convert Windows-style backslashes to forward slashes
object = strings.ReplaceAll(object, "\\", "/")
object = removeDuplicateSlashes(object)
// Remove leading slash to match S3 API format
object = strings.TrimPrefix(object, "/")
// Restore trailing slash if it was present and result is not empty
if hasTrailingSlash && object != "" && !strings.HasSuffix(object, "/") {
object = object + "/"
}
return object
}
// removeDuplicateSlashes removes consecutive slashes from a path
func removeDuplicateSlashes(s string) string {
var result strings.Builder
result.Grow(len(s))
lastWasSlash := false
for _, r := range s {
if r == '/' {
if !lastWasSlash {
result.WriteRune(r)
}
lastWasSlash = true
} else {
result.WriteRune(r)
lastWasSlash = false
}
}
return result.String()
}
func GetPrefix(r *http.Request) string {
query := r.URL.Query()
prefix := query.Get("prefix")
prefix = removeDuplicateSlashes(prefix)
return prefix
}
var PassThroughHeaders = map[string]string{
"response-cache-control": "Cache-Control",
"response-content-disposition": "Content-Disposition",
"response-content-encoding": "Content-Encoding",
"response-content-language": "Content-Language",
"response-content-type": "Content-Type",
"response-expires": "Expires",
}
// IsSeaweedFSInternalHeader checks if a header key is a SeaweedFS internal header
// that should be filtered from client responses.
// Header names are case-insensitive in HTTP, so this function normalizes to lowercase.
func IsSeaweedFSInternalHeader(headerKey string) bool {
return strings.HasPrefix(strings.ToLower(headerKey), SeaweedFSInternalPrefix)
}
// Context keys for storing authenticated identity information
type contextKey string
const (
contextKeyIdentityName contextKey = "s3-identity-name"
contextKeyIdentityObject contextKey = "s3-identity-object"
contextKeyIdentityHolder contextKey = "s3-identity-holder"
contextKeyPrincipalArn contextKey = "s3-principal-arn"
)
// identityHolder is a mutable container for the authenticated identity name,
// stored by pointer in the request context. Authentication runs as an inner
// handler that records the identity with r.WithContext, which returns a request
// copy; that copy's new context values are invisible to outer middleware (e.g.
// the audit logger) holding an earlier copy of the request. A pointer to a
// holder installed before authentication is shared across all copies, so the
// name written by the inner handler is readable by the outer middleware.
type identityHolder struct {
name atomic.Pointer[string]
principalArn atomic.Pointer[string]
}
// EnsureIdentityHolder attaches a mutable identity holder to the request context
// if one is not already present, returning the (possibly new) request. Install
// it before authentication so the authenticated requester remains recoverable
// from the original request, e.g. when emitting an audit entry for a handler
// that does not log the requester itself.
func EnsureIdentityHolder(r *http.Request) *http.Request {
if r == nil {
return nil
}
if h, ok := r.Context().Value(contextKeyIdentityHolder).(*identityHolder); ok && h != nil {
return r
}
return r.WithContext(context.WithValue(r.Context(), contextKeyIdentityHolder, &identityHolder{}))
}
// SetIdentityNameInContext stores the authenticated identity name in the request context
// This is the secure way to propagate identity - headers can be spoofed, context cannot
func SetIdentityNameInContext(ctx context.Context, identityName string) context.Context {
if identityName == "" {
return ctx
}
// Also record into the mutable holder (if installed) so the name survives the
// request-context copy and stays visible to outer middleware such as the
// audit logger.
if h, ok := ctx.Value(contextKeyIdentityHolder).(*identityHolder); ok && h != nil {
h.name.Store(&identityName)
}
return context.WithValue(ctx, contextKeyIdentityName, identityName)
}
// GetIdentityNameFromContext retrieves the authenticated identity name from the request context
// Returns empty string if no identity is set (unauthenticated request)
// This is the secure way to retrieve identity - never read from headers directly
func GetIdentityNameFromContext(r *http.Request) string {
// Prefer the per-request context value; fall back to the mutable holder for
// callers (e.g. the audit middleware) that hold a copy of the request taken
// before authentication set the identity.
if name, ok := r.Context().Value(contextKeyIdentityName).(string); ok && name != "" {
return name
}
if h, ok := r.Context().Value(contextKeyIdentityHolder).(*identityHolder); ok && h != nil {
if name := h.name.Load(); name != nil {
return *name
}
}
return ""
}
// SetPrincipalArnInContext stores the authenticated principal ARN in the request
// context. For an STS session the identity name is an opaque session subject, so
// the ARN is the only place the assumed role and session name survive to the
// audit log.
func SetPrincipalArnInContext(ctx context.Context, principalArn string) context.Context {
if principalArn == "" {
return ctx
}
if h, ok := ctx.Value(contextKeyIdentityHolder).(*identityHolder); ok && h != nil {
h.principalArn.Store(&principalArn)
}
return context.WithValue(ctx, contextKeyPrincipalArn, principalArn)
}
// GetPrincipalArnFromContext retrieves the authenticated principal ARN from the
// request context, or "" when the request is unauthenticated.
func GetPrincipalArnFromContext(r *http.Request) string {
if arn, ok := r.Context().Value(contextKeyPrincipalArn).(string); ok && arn != "" {
return arn
}
if h, ok := r.Context().Value(contextKeyIdentityHolder).(*identityHolder); ok && h != nil {
if arn := h.principalArn.Load(); arn != nil {
return *arn
}
}
return ""
}
// SetIdentityInContext stores the full authenticated identity object in the request context
// This is used to pass the full identity (including for JWT users) to handlers
func SetIdentityInContext(ctx context.Context, identity interface{}) context.Context {
if identity != nil {
return context.WithValue(ctx, contextKeyIdentityObject, identity)
}
return ctx
}
// GetIdentityFromContext retrieves the full identity object from the request context
// Returns nil if no identity is set (unauthenticated request)
func GetIdentityFromContext(r *http.Request) interface{} {
return r.Context().Value(contextKeyIdentityObject)
}