mirror of
https://github.com/versity/versitygw.git
synced 2026-05-22 20:01:27 +00:00
Fixes #2123 Fixes #2120 Fixes #2116 Fixes #2111 Fixes #2108 Fixes #2086 Fixes #2085 Fixes #2083 Fixes #2081 Fixes #2080 Fixes #2073 Fixes #2072 Fixes #2071 Fixes #2069 Fixes #2044 Fixes #2043 Fixes #2042 Fixes #2041 Fixes #2040 Fixes #2039 Fixes #2036 Fixes #2035 Fixes #2034 Fixes #2028 Fixes #2020 Fixes #1842 Fixes #1810 Fixes #1780 Fixes #1775 Fixes #1736 Fixes #1705 Fixes #1663 Fixes #1645 Fixes #1583 Fixes #1526 Fixes #1514 Fixes #1493 Fixes #1487 Fixes #959 Fixes #779 Closes #823 Closes #85 Refactor global S3 error handling around structured error types and centralized XML response generation. All S3 errors now share the common APIError base for the fields every error has: Code, HTTP status code, and Message. Non-traditional errors that need AWS-compatible XML fields now have dedicated typed errors in the s3err package. Each typed error implements the shared S3Error behavior so controllers and middleware can handle errors consistently while still emitting error-specific XML fields. Add a dedicated InvalidArgumentError type because InvalidArgument is used widely across request validation, auth, copy source handling, object lock validation, multipart validation, and header parsing. The new InvalidArgument path uses explicit InvalidArgErrorCode constants with predefined descriptions and ArgumentName values, keeping call sites readable while preserving the correct InvalidArgument XML shape and optional ArgumentValue. New structured errors added in s3err: - `AccessForbiddenError`: Method, ResourceType - `BadDigestError`: CalculatedDigest, ExpectedDigest - `BucketError`: BucketName - `ContentSHA256MismatchError`: ClientComputedContentSHA256, S3ComputedContentSHA256 - `EntityTooLargeError`: ProposedSize, MaxSizeAllowed - `EntityTooSmallError`: ProposedSize, MinSizeAllowed - `ExpiredPresignedURLError`: ServerTime, XAmzExpires, Expires - `InvalidAccessKeyIdError`: AWSAccessKeyId - `InvalidArgumentError`: Description, ArgumentName, ArgumentValue - `InvalidChunkSizeError`: Chunk, BadChunkSize - `InvalidDigestError`: ContentMD5 - `InvalidLocationConstraintError`: LocationConstraint - `InvalidPartError`: UploadId, PartNumber, ETag - `InvalidRangeError`: RangeRequested, ActualObjectSize - `InvalidTagError`: TagKey, TagValue - `KeyTooLongError`: Size, MaxSizeAllowed - `MetadataTooLargeError`: Size, MaxSizeAllowed - `MethodNotAllowedError`: Method, ResourceType, AllowedMethods - `NoSuchUploadError`: UploadId - `NoSuchVersionError`: Key, VersionId - `NotImplementedError`: Header, AdditionalMessage - `PreconditionFailedError`: Condition - `RequestTimeTooSkewedError`: RequestTime, ServerTime, MaxAllowedSkewMilliseconds - `SignatureDoesNotMatchError`: AWSAccessKeyId, StringToSign, SignatureProvided, StringToSignBytes, CanonicalRequest, CanonicalRequestBytes Fix CompleteMultipartUpload validation in the Azure backend so missing or empty `ETag` values return the appropriate S3 error instead of allowing a gateway panic. Fix presigned authentication expiration validation to compare server time in `UTC`, matching the `UTC` timestamp used by presigned URL signing. Add request ID and host ID support across S3 requests. Each request now receives AWS S3-like identifiers, returned in response headers as `x-amz-request-id` and `x-amz-id-2` and included in all XML error responses as RequestId and HostId. The generated ID structure is designed to resemble AWS S3 request IDs and host IDs. The request signature calculation/validation for streaming uploads was previously delayed until the request body was fully read, both for Authorization header authentication and presigned URLs. Now, the signature is validated immediately in the authorization middlewares without reading the request body, since the signature calculation itself does not depend on the request body. Instead, only the `x-amz-content-sha256` SHA-256 hash calculation is delayed.
619 lines
20 KiB
Go
619 lines
20 KiB
Go
// Copyright 2024 Versity Software
|
|
// This file is 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 utils
|
|
|
|
import (
|
|
"bufio"
|
|
"bytes"
|
|
"crypto/hmac"
|
|
"crypto/sha256"
|
|
"encoding/base64"
|
|
"encoding/hex"
|
|
"errors"
|
|
"fmt"
|
|
"hash"
|
|
"io"
|
|
"math"
|
|
"strconv"
|
|
"strings"
|
|
"time"
|
|
|
|
"github.com/aws/aws-sdk-go-v2/service/s3/types"
|
|
"github.com/versity/versitygw/debuglogger"
|
|
"github.com/versity/versitygw/s3err"
|
|
)
|
|
|
|
// chunked uploads described in:
|
|
// https://docs.aws.amazon.com/AmazonS3/latest/API/sigv4-streaming.html
|
|
|
|
const (
|
|
zeroLenSig = "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
|
|
awsV4 = "AWS4"
|
|
awsS3Service = "s3"
|
|
awsV4Request = "aws4_request"
|
|
trailerSignatureHeader = "x-amz-trailer-signature:"
|
|
streamPayloadAlgo = "AWS4-HMAC-SHA256-PAYLOAD"
|
|
streamPayloadTrailerAlgo = "AWS4-HMAC-SHA256-TRAILER"
|
|
|
|
maxHeaderSize = 1024
|
|
)
|
|
|
|
var (
|
|
errskipHeader = errors.New("skip to next header")
|
|
delimiter = []byte{'\r', '\n'}
|
|
)
|
|
|
|
// ChunkReader reads from chunked upload request body, and returns
|
|
// object data stream
|
|
type ChunkReader struct {
|
|
r io.Reader
|
|
signingKey []byte
|
|
prevSig string
|
|
parsedSig string
|
|
canonicalString string
|
|
accessKey string
|
|
chunkDataLeft int64
|
|
trailer checksumType
|
|
trailerSig string
|
|
parsedChecksum string
|
|
stash []byte
|
|
chunkHash hash.Hash
|
|
checksumHash hash.Hash
|
|
isEOF bool
|
|
isFirstHeader bool
|
|
region string
|
|
date time.Time
|
|
requireTrailer bool
|
|
chunkSizes []int64
|
|
cLength int64
|
|
dataRead int64
|
|
}
|
|
|
|
// NewChunkReader reads from request body io.Reader and parses out the
|
|
// chunk metadata in stream. The headers are validated for proper signatures.
|
|
// Reading from the chunk reader will read only the object data stream
|
|
// without the chunk headers/trailers.
|
|
func NewSignedChunkReader(r io.Reader, authdata AuthData, canonicalString, secret string, date time.Time, chType checksumType, requireTrailer bool, cLength int64) (io.Reader, error) {
|
|
chRdr := &ChunkReader{
|
|
r: r,
|
|
signingKey: getSigningKey(secret, authdata.Region, date),
|
|
// the authdata.Signature is validated in the auth-reader,
|
|
// so we can use that here without any other checks
|
|
prevSig: authdata.Signature,
|
|
canonicalString: canonicalString,
|
|
accessKey: authdata.Access,
|
|
chunkHash: sha256.New(),
|
|
isFirstHeader: true,
|
|
date: date,
|
|
region: authdata.Region,
|
|
trailer: chType,
|
|
requireTrailer: requireTrailer,
|
|
chunkSizes: []int64{},
|
|
cLength: cLength,
|
|
}
|
|
|
|
if chType != "" {
|
|
checksumHasher, err := getHasher(chType)
|
|
if err != nil {
|
|
debuglogger.Logf("failed to initialize hash calculator: %v", err)
|
|
return nil, err
|
|
}
|
|
|
|
chRdr.checksumHash = checksumHasher
|
|
}
|
|
if !requireTrailer {
|
|
debuglogger.Infof("initializing signed chunk reader")
|
|
} else {
|
|
debuglogger.Infof("initializing signed chunk reader with '%v' trailing checksum", chType)
|
|
}
|
|
return chRdr, nil
|
|
}
|
|
|
|
// Read satisfies the io.Reader for this type
|
|
func (cr *ChunkReader) Read(p []byte) (int, error) {
|
|
n, err := cr.r.Read(p)
|
|
if err != nil && err != io.EOF {
|
|
return 0, err
|
|
}
|
|
|
|
cr.isEOF = err == io.EOF
|
|
|
|
if cr.chunkDataLeft < int64(n) {
|
|
chunkSize := cr.chunkDataLeft
|
|
if chunkSize > 0 {
|
|
cr.chunkHash.Write(p[:chunkSize])
|
|
if cr.checksumHash != nil {
|
|
cr.checksumHash.Write(p[:chunkSize])
|
|
}
|
|
}
|
|
n, err := cr.parseAndRemoveChunkInfo(p[chunkSize:n])
|
|
if err != nil && err != io.EOF {
|
|
return 0, err
|
|
}
|
|
n += int(chunkSize)
|
|
cr.dataRead += int64(n)
|
|
if cr.isEOF {
|
|
if cr.cLength != cr.dataRead {
|
|
debuglogger.Logf("number of bytes expected: (%v), number of bytes read: (%v)", cr.cLength, cr.dataRead)
|
|
return 0, s3err.GetAPIError(s3err.ErrContentLengthMismatch)
|
|
}
|
|
}
|
|
return n, err
|
|
}
|
|
|
|
cr.chunkDataLeft -= int64(n)
|
|
cr.chunkHash.Write(p[:n])
|
|
if cr.checksumHash != nil {
|
|
cr.checksumHash.Write(p[:n])
|
|
}
|
|
cr.dataRead += int64(n)
|
|
if cr.isEOF {
|
|
if cr.cLength != cr.dataRead {
|
|
debuglogger.Logf("number of bytes expected: (%v), number of bytes read: (%v)", cr.cLength, cr.dataRead)
|
|
return 0, s3err.GetAPIError(s3err.ErrContentLengthMismatch)
|
|
}
|
|
}
|
|
return n, err
|
|
}
|
|
|
|
// https://docs.aws.amazon.com/AmazonS3/latest/API/sigv4-streaming.html#sigv4-chunked-body-definition
|
|
// This part is the same for all chunks,
|
|
// only the previous signature and hash of current chunk changes
|
|
func (cr *ChunkReader) getStringToSignPrefix(algo string) string {
|
|
credentialScope := fmt.Sprintf("%s/%s/%s/%s",
|
|
cr.date.Format("20060102"),
|
|
cr.region,
|
|
awsS3Service,
|
|
awsV4Request)
|
|
|
|
return fmt.Sprintf("%s\n%s\n%s",
|
|
algo,
|
|
cr.date.Format("20060102T150405Z"),
|
|
credentialScope)
|
|
}
|
|
|
|
// https://docs.aws.amazon.com/AmazonS3/latest/API/sigv4-streaming.html#sigv4-chunked-body-definition
|
|
// signature For each chunk, you calculate the signature using the following
|
|
// string to sign. For the first chunk, you use the seed-signature as the
|
|
// previous signature.
|
|
func (cr *ChunkReader) getChunkStringToSign() string {
|
|
prefix := cr.getStringToSignPrefix(streamPayloadAlgo)
|
|
chunkHash := cr.chunkHash.Sum(nil)
|
|
strToSign := fmt.Sprintf("%s\n%s\n%s\n%s",
|
|
prefix,
|
|
cr.prevSig,
|
|
zeroLenSig,
|
|
hex.EncodeToString(chunkHash))
|
|
debuglogger.PrintInsideHorizontalBorders(debuglogger.Purple, "STRING TO SIGN", strToSign, 64)
|
|
return strToSign
|
|
}
|
|
|
|
// https://docs.aws.amazon.com/AmazonS3/latest/API/sigv4-streaming-trailers.html#example-signature-calculations-trailing-header
|
|
// Builds the final chunk trailing signature string to sign
|
|
func (cr *ChunkReader) getTrailerChunkStringToSign() string {
|
|
trailer := fmt.Sprintf("%v:%v\n", cr.trailer, cr.parsedChecksum)
|
|
hsh := sha256.Sum256([]byte(trailer))
|
|
sig := hex.EncodeToString(hsh[:])
|
|
|
|
prefix := cr.getStringToSignPrefix(streamPayloadTrailerAlgo)
|
|
|
|
strToSign := fmt.Sprintf("%s\n%s\n%s",
|
|
prefix,
|
|
cr.prevSig,
|
|
sig,
|
|
)
|
|
|
|
debuglogger.PrintInsideHorizontalBorders(debuglogger.Purple, "TRAILER STRING TO SIGN", strToSign, 64)
|
|
|
|
return strToSign
|
|
}
|
|
|
|
// https://docs.aws.amazon.com/AmazonS3/latest/API/sigv4-streaming-trailers.html#example-signature-calculations-trailing-header
|
|
// Calculates and validates the final chunk trailer signature
|
|
func (cr *ChunkReader) verifyTrailerSignature() error {
|
|
strToSign := cr.getTrailerChunkStringToSign()
|
|
sig := hex.EncodeToString(hmac256(cr.signingKey, []byte(strToSign)))
|
|
|
|
if sig != cr.trailerSig {
|
|
debuglogger.Logf("incorrect trailing signature: (calculated): %v, (got): %v", sig, cr.trailerSig)
|
|
return s3err.GetSignatureDoesNotMatchErr(cr.accessKey, strToSign, cr.trailerSig, HexBytes(strToSign), cr.canonicalString, HexBytes(cr.canonicalString))
|
|
}
|
|
|
|
return nil
|
|
}
|
|
|
|
// Verifies the object checksum
|
|
func (cr *ChunkReader) verifyChecksum() error {
|
|
checksumHash := cr.checksumHash.Sum(nil)
|
|
checksum := base64.StdEncoding.EncodeToString(checksumHash)
|
|
if checksum != cr.parsedChecksum {
|
|
algo := types.ChecksumAlgorithm(strings.ToUpper(strings.TrimPrefix(string(cr.trailer), "x-amz-checksum-")))
|
|
debuglogger.Logf("incorrect trailing checksum: (calculated): %v, (got): %v", checksum, cr.parsedChecksum)
|
|
return s3err.GetChecksumBadDigestErr(algo)
|
|
}
|
|
|
|
return nil
|
|
}
|
|
|
|
// Calculates and verifies the chunk signature
|
|
func (cr *ChunkReader) checkSignature() error {
|
|
sigstr := cr.getChunkStringToSign()
|
|
cr.chunkHash.Reset()
|
|
cr.prevSig = hex.EncodeToString(hmac256(cr.signingKey, []byte(sigstr)))
|
|
|
|
if cr.prevSig != cr.parsedSig {
|
|
debuglogger.Logf("incorrect signature: (calculated): %v, (got) %v", cr.prevSig, cr.parsedSig)
|
|
return s3err.GetSignatureDoesNotMatchErr(cr.accessKey, sigstr, cr.parsedSig, HexBytes(sigstr), cr.canonicalString, HexBytes(cr.canonicalString))
|
|
}
|
|
cr.parsedSig = ""
|
|
return nil
|
|
}
|
|
|
|
// The provided p should have all of the previous chunk data and trailer
|
|
// consumed already. The positioning here is expected that p[0] starts the
|
|
// new chunk size with the ";chunk-signature=" following. The only exception
|
|
// is if we started consuming the trailer, but hit the end of the read buffer.
|
|
// In this case, parseAndRemoveChunkInfo is called with skipcheck=true to
|
|
// finish consuming the final trailer bytes.
|
|
// This parses the chunk metadata in situ without allocating an extra buffer.
|
|
// It will just read and validate the chunk metadata and then move the
|
|
// following chunk data to overwrite the metadata in the provided buffer.
|
|
func (cr *ChunkReader) parseAndRemoveChunkInfo(p []byte) (int, error) {
|
|
n := len(p)
|
|
|
|
if cr.parsedSig != "" {
|
|
err := cr.checkSignature()
|
|
if err != nil {
|
|
return 0, err
|
|
}
|
|
}
|
|
|
|
chunkSize, sig, bufOffset, err := cr.parseChunkHeaderBytes(p[:n])
|
|
if err == errskipHeader {
|
|
cr.chunkDataLeft = 0
|
|
return 0, nil
|
|
}
|
|
if err != nil {
|
|
debuglogger.Logf("failed to parse chunk headers: %v", err)
|
|
return 0, err
|
|
}
|
|
cr.parsedSig = sig
|
|
// If we hit the final chunk, calculate and validate the final
|
|
// chunk signature and finish reading
|
|
if chunkSize == 0 {
|
|
debuglogger.Infof("final chunk parsed:\nchunk size: %v\nsignature: %v\nbuffer offset: %v", chunkSize, sig, bufOffset)
|
|
cr.chunkHash.Reset()
|
|
err := cr.checkSignature()
|
|
if err != nil {
|
|
return 0, err
|
|
}
|
|
|
|
if cr.trailer != "" {
|
|
debuglogger.Infof("final chunk trailers parsed:\nchecksum: %v\ntrailing signature: %v", cr.parsedChecksum, cr.trailerSig)
|
|
err := cr.verifyChecksum()
|
|
if err != nil {
|
|
return 0, err
|
|
}
|
|
err = cr.verifyTrailerSignature()
|
|
if err != nil {
|
|
return 0, err
|
|
}
|
|
}
|
|
|
|
return 0, io.EOF
|
|
}
|
|
debuglogger.Infof("chunk headers parsed:\nchunk size: %v\nsignature: %v\nbuffer offset: %v", chunkSize, sig, bufOffset)
|
|
|
|
// move data up to remove chunk header
|
|
copy(p, p[bufOffset:n])
|
|
n -= bufOffset
|
|
|
|
// if remaining buffer larger than chunk data,
|
|
// parse next header in buffer
|
|
if int64(n) > chunkSize {
|
|
cr.chunkDataLeft = 0
|
|
cr.chunkHash.Write(p[:chunkSize])
|
|
if cr.checksumHash != nil {
|
|
cr.checksumHash.Write(p[:chunkSize])
|
|
}
|
|
n, err := cr.parseAndRemoveChunkInfo(p[chunkSize:n])
|
|
if (chunkSize + int64(n)) > math.MaxInt {
|
|
debuglogger.Logf("exceeding the limit of maximum integer allowed: (value): %v, (limit): %v", chunkSize+int64(n), math.MaxInt)
|
|
return 0, s3err.GetAPIError(s3err.ErrIncompleteBody)
|
|
}
|
|
return n + int(chunkSize), err
|
|
}
|
|
|
|
cr.chunkDataLeft = chunkSize - int64(n)
|
|
cr.chunkHash.Write(p[:n])
|
|
if cr.checksumHash != nil {
|
|
cr.checksumHash.Write(p[:n])
|
|
}
|
|
|
|
return n, nil
|
|
}
|
|
|
|
// https://docs.aws.amazon.com/AmazonS3/latest/API/sig-v4-header-based-auth.html
|
|
// Task 3: Calculate Signature
|
|
// https://docs.aws.amazon.com/AmazonS3/latest/API/sig-v4-authenticating-requests.html#signing-request-intro
|
|
func getSigningKey(secret, region string, date time.Time) []byte {
|
|
dateKey := hmac256([]byte(awsV4+secret), []byte(date.Format(yyyymmdd)))
|
|
dateRegionKey := hmac256(dateKey, []byte(region))
|
|
dateRegionServiceKey := hmac256(dateRegionKey, []byte(awsS3Service))
|
|
signingKey := hmac256(dateRegionServiceKey, []byte(awsV4Request))
|
|
debuglogger.Infof("signing key: %s", hex.EncodeToString(signingKey))
|
|
return signingKey
|
|
}
|
|
|
|
func hmac256(key []byte, data []byte) []byte {
|
|
hash := hmac.New(sha256.New, key)
|
|
hash.Write(data)
|
|
return hash.Sum(nil)
|
|
}
|
|
|
|
// This returns the chunk payload size, signature, data start offset, and
|
|
// error if any. See the AWS documentation for the chunk header format. The
|
|
// header[0] byte is expected to be the first byte of the chunk size here.
|
|
func (cr *ChunkReader) parseChunkHeaderBytes(header []byte) (int64, string, int, error) {
|
|
stashLen := len(cr.stash)
|
|
if stashLen > maxHeaderSize {
|
|
debuglogger.Logf("the stash length exceeds the maximum allowed chunk header size: (stash len): %v, (header limit): %v", stashLen, maxHeaderSize)
|
|
return 0, "", 0, s3err.GetAPIError(s3err.ErrIncompleteBody)
|
|
}
|
|
if cr.stash != nil {
|
|
debuglogger.Logf("recovering the stash: (stash len): %v", stashLen)
|
|
tmp := make([]byte, stashLen+len(header))
|
|
copy(tmp, cr.stash)
|
|
copy(tmp[len(cr.stash):], header)
|
|
header = tmp
|
|
cr.stash = nil
|
|
}
|
|
|
|
rdr := bufio.NewReader(bytes.NewReader(header))
|
|
|
|
// After the first chunk each chunk header should start
|
|
// with "\n\r\n"
|
|
if !cr.isFirstHeader {
|
|
err := readAndSkip(rdr, delimiter...)
|
|
if err != nil {
|
|
debuglogger.Logf("failed to read chunk header first 2 bytes: (should be): \\r\\n, (got): %q", header[:min(2, len(header))])
|
|
return cr.handleRdrErr(err, header)
|
|
}
|
|
}
|
|
|
|
chunkSize, err := cr.parseChunkSize(rdr, header)
|
|
if err != nil {
|
|
return 0, "", 0, err
|
|
}
|
|
|
|
// read the chunk signature
|
|
err = readAndSkip(rdr, []byte("chunk-signature=")...)
|
|
if err != nil {
|
|
debuglogger.Logf("failed to read 'chunk-signature=': %v", err)
|
|
return cr.handleRdrErr(err, header)
|
|
}
|
|
sig, err := readBytes(rdr, 64)
|
|
if err != nil {
|
|
debuglogger.Logf("failed to read the chunk signature: %v", err)
|
|
return cr.handleRdrErr(err, header)
|
|
}
|
|
|
|
err = readAndSkip(rdr, delimiter...)
|
|
if err != nil {
|
|
debuglogger.Logf("failed to read '\\r\\n' after chunk signature")
|
|
return cr.handleRdrErr(err, header)
|
|
}
|
|
|
|
// read and parse the final chunk trailer and checksum
|
|
if chunkSize == 0 {
|
|
if cr.requireTrailer {
|
|
// parse and validate the trailing header
|
|
trailer, err := readAndTrim(rdr, ':')
|
|
if err != nil {
|
|
debuglogger.Logf("failed to read trailer prefix: %v", err)
|
|
return cr.handleRdrErr(err, header)
|
|
}
|
|
if trailer != string(cr.trailer) {
|
|
debuglogger.Logf("incorrect trailer prefix: (expected): %v, (got): %v", cr.trailer, trailer)
|
|
return 0, "", 0, s3err.GetAPIError(s3err.ErrMalformedTrailer)
|
|
}
|
|
|
|
algo := types.ChecksumAlgorithm(strings.ToUpper(strings.TrimPrefix(trailer, "x-amz-checksum-")))
|
|
|
|
// parse the checksum
|
|
checksum, err := readAndTrim(rdr, '\r')
|
|
if err != nil {
|
|
debuglogger.Logf("failed to read checksum value: %v", err)
|
|
return cr.handleRdrErr(err, header)
|
|
}
|
|
|
|
err = readAndSkip(rdr, '\n')
|
|
if err != nil {
|
|
debuglogger.Logf("failed to read \\n after checksum: %v", err)
|
|
return cr.handleRdrErr(err, header)
|
|
}
|
|
|
|
if !IsValidChecksum(checksum, algo) {
|
|
debuglogger.Logf("invalid checksum value: %v", checksum)
|
|
return 0, "", 0, s3err.GetInvalidTrailingChecksumHeaderErr(trailer)
|
|
}
|
|
|
|
// parse the trailing signature
|
|
trailerSigPrefix, err := readBytes(rdr, 24)
|
|
if err != nil {
|
|
debuglogger.Logf("failed to read trailing signature prefix: %v", err)
|
|
return cr.handleRdrErr(err, header)
|
|
}
|
|
|
|
if trailerSigPrefix != trailerSignatureHeader {
|
|
debuglogger.Logf("invalid trailing signature prefix: (expected): %v, (got): %v", trailerSignatureHeader, trailerSigPrefix)
|
|
return 0, "", 0, s3err.GetAPIError(s3err.ErrIncompleteBody)
|
|
}
|
|
|
|
trailerSig, err := readBytes(rdr, 64)
|
|
if err != nil {
|
|
debuglogger.Logf("failed to read trailing signature: %v", err)
|
|
return cr.handleRdrErr(err, header)
|
|
}
|
|
|
|
err = readAndSkip(rdr, delimiter...)
|
|
if err != nil {
|
|
debuglogger.Logf("failed to read '\\r\\n' after last chunk signature")
|
|
return cr.handleRdrErr(err, header)
|
|
}
|
|
|
|
cr.trailerSig = trailerSig
|
|
cr.parsedChecksum = checksum
|
|
}
|
|
|
|
// "\r\n\r\n" is followed after the last chunk
|
|
err = readAndSkip(rdr, delimiter...)
|
|
if err != nil {
|
|
debuglogger.Logf("failed to read \\r\\n at the end of chunk header: %v", err)
|
|
return cr.handleRdrErr(err, header)
|
|
}
|
|
|
|
return 0, sig, 0, nil
|
|
}
|
|
|
|
// add the chunk size at the end of header parsing
|
|
// to avoid duplication because of header stashing
|
|
cr.addChunkSize(chunkSize)
|
|
|
|
// find the index of chunk ending: '\r\n'
|
|
// skip the first 2 bytes as it is the starting '\r\n'
|
|
// the first chunk doesn't contain the starting '\r\n', but
|
|
// anyway, trimming the first 2 bytes doesn't pollute the logic.
|
|
ind := bytes.Index(header[2:], delimiter)
|
|
cr.isFirstHeader = false
|
|
|
|
// the offset is the found index + 4 - the stash length
|
|
// where:
|
|
// ind is the index of '\r\n'
|
|
// 4 specifies the trimmed 2 bytes plus 2 to shift the index at the end of '\r\n'
|
|
offset := ind + 4 - stashLen
|
|
return chunkSize, sig, offset, nil
|
|
}
|
|
|
|
// Stashes the header in cr.stash and returns "errskipHeader"
|
|
func (cr *ChunkReader) stashAndSkipHeader(header []byte) (int64, string, int, error) {
|
|
cr.stash = make([]byte, len(header))
|
|
copy(cr.stash, header)
|
|
debuglogger.Logf("stashing the header: (header length): %v", len(header))
|
|
return 0, "", 0, errskipHeader
|
|
}
|
|
|
|
// Returns "errInvalidChunkFormat" if the passed err is "io.EOF" and cr.rdr EOF is reached
|
|
// calls "cr.stashAndSkipHeader" if the passed err is "io.EOF" and cr.isEOF is false
|
|
// Returns the error otherwise
|
|
func (cr *ChunkReader) handleRdrErr(err error, header []byte) (int64, string, int, error) {
|
|
if err == io.EOF || err == io.ErrUnexpectedEOF {
|
|
if cr.isEOF {
|
|
debuglogger.Logf("incomplete chunk encoding, EOF reached")
|
|
return 0, "", 0, s3err.GetAPIError(s3err.ErrIncompleteBody)
|
|
}
|
|
return cr.stashAndSkipHeader(header)
|
|
}
|
|
return 0, "", 0, s3err.GetAPIError(s3err.ErrIncompleteBody)
|
|
}
|
|
|
|
// parseChunkSize parses and validates the chunk size
|
|
func (cr *ChunkReader) parseChunkSize(rdr *bufio.Reader, header []byte) (int64, error) {
|
|
// read and parse the chunk size
|
|
chunkSizeStr, err := readAndTrim(rdr, ';')
|
|
if err != nil {
|
|
debuglogger.Logf("failed to read chunk size: %v", err)
|
|
_, _, _, err := cr.handleRdrErr(err, header)
|
|
return 0, err
|
|
}
|
|
chunkSize, err := strconv.ParseInt(chunkSizeStr, 16, 64)
|
|
if err != nil {
|
|
debuglogger.Logf("failed to parse chunk size: (size): %v, (err): %v", chunkSizeStr, err)
|
|
return 0, s3err.GetAPIError(s3err.ErrIncompleteBody)
|
|
}
|
|
|
|
if !cr.isValidChunkSize(chunkSize) {
|
|
return 0, s3err.GetInvalidChunkSizeErr(len(cr.chunkSizes)+1, chunkSize)
|
|
}
|
|
|
|
return chunkSize, nil
|
|
}
|
|
|
|
// addChunkSize adds the input chunk size to chunkSizes slice
|
|
func (cr *ChunkReader) addChunkSize(size int64) {
|
|
cr.chunkSizes = append(cr.chunkSizes, size)
|
|
}
|
|
|
|
// isValidChunkSize checks if the parsed chunk size is valid
|
|
// they follow one rule: all chunk sizes except for the last one
|
|
// should be greater than 8192
|
|
func (cr *ChunkReader) isValidChunkSize(size int64) bool {
|
|
if len(cr.chunkSizes) == 0 {
|
|
// any valid number is valid as a first chunk size
|
|
return true
|
|
}
|
|
|
|
lastChunkSize := cr.chunkSizes[len(cr.chunkSizes)-1]
|
|
// any chunk size, except the last one should be greater than 8192
|
|
if size != 0 && lastChunkSize < minChunkSize {
|
|
debuglogger.Logf("invalid chunk size %v", lastChunkSize)
|
|
return false
|
|
}
|
|
|
|
return true
|
|
}
|
|
|
|
// Algorithm returns the checksum algorithm
|
|
func (cr *ChunkReader) Algorithm() string {
|
|
return strings.TrimPrefix(string(cr.trailer), "x-amz-checksum-")
|
|
}
|
|
|
|
// Checksum returns the parsed trailing checksum
|
|
func (cr *ChunkReader) Checksum() string {
|
|
return cr.parsedChecksum
|
|
}
|
|
|
|
// reads data from the "rdr" and validates the passed data bytes
|
|
func readAndSkip(rdr *bufio.Reader, expected ...byte) error {
|
|
buf := make([]byte, len(expected))
|
|
_, err := io.ReadFull(rdr, buf)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
|
|
if bytes.Equal(buf, expected) {
|
|
return nil
|
|
}
|
|
|
|
return s3err.GetAPIError(s3err.ErrIncompleteBody)
|
|
}
|
|
|
|
// reads string by "delim" and trims the delimiter at the end
|
|
func readAndTrim(r *bufio.Reader, delim byte) (string, error) {
|
|
str, err := r.ReadString(delim)
|
|
if err != nil {
|
|
return "", err
|
|
}
|
|
|
|
return strings.TrimSuffix(str, string(delim)), nil
|
|
}
|
|
|
|
func readBytes(r *bufio.Reader, count int) (string, error) {
|
|
buf := make([]byte, count)
|
|
_, err := io.ReadFull(r, buf)
|
|
|
|
return string(buf), err
|
|
}
|