From 95637630b9ede5a1a0b619fcef0e49464f4ae05c Mon Sep 17 00:00:00 2001
From: Pavel Pogodaev <p.pogodaev@yadro.com>
Date: Sun, 8 Dec 2024 15:02:31 +0300
Subject: [PATCH] [#147] Add Kludge profiles

Signed-off-by: Pavel Pogodaev <p.pogodaev@yadro.com>
---
 api/handler/api.go              |  9 ++++++--
 api/handler/cors.go             |  1 +
 api/handler/delete.go           |  2 +-
 api/handler/handlers_test.go    |  4 ++--
 api/handler/lifecycle.go        |  2 +-
 api/handler/locking.go          |  6 +++---
 api/handler/multipart_upload.go |  2 +-
 api/handler/put.go              |  8 ++++---
 api/handler/versioning.go       |  2 +-
 api/layer/cors.go               |  2 +-
 api/layer/layer.go              |  3 ++-
 api/middleware/policy.go        |  4 ++--
 api/router_mock_test.go         |  2 +-
 cmd/s3-gw/app.go                | 37 ++++++++++++++++++++++++++++-----
 cmd/s3-gw/app_settings.go       | 33 +++++++++++++++++++++++++++++
 cmd/s3-gw/decoder_test.go       |  2 +-
 config/config.env               |  4 ++++
 config/config.yaml              |  7 +++++++
 docs/configuration.md           |  9 +++++++-
 19 files changed, 113 insertions(+), 26 deletions(-)

diff --git a/api/handler/api.go b/api/handler/api.go
index 559977a..100bae8 100644
--- a/api/handler/api.go
+++ b/api/handler/api.go
@@ -32,11 +32,11 @@ type (
 		PlacementPolicy(namespace, constraint string) (netmap.PlacementPolicy, bool)
 		CopiesNumbers(namespace, constraint string) ([]uint32, bool)
 		DefaultCopiesNumbers(namespace string) []uint32
-		NewXMLDecoder(io.Reader) *xml.Decoder
+		NewXMLDecoder(reader io.Reader, agent string) *xml.Decoder
 		DefaultMaxAge() int
 		ResolveZoneList() []string
 		IsResolveListAllow() bool
-		BypassContentEncodingInChunks() bool
+		BypassContentEncodingInChunks(agent string) bool
 		MD5Enabled() bool
 		RetryMaxAttempts() int
 		RetryMaxBackoff() time.Duration
@@ -59,6 +59,11 @@ type (
 
 type RetryStrategy string
 
+type KludgeParams struct {
+	IsSet bool
+	Value bool
+}
+
 const (
 	RetryStrategyExponential = "exponential"
 	RetryStrategyConstant    = "constant"
diff --git a/api/handler/cors.go b/api/handler/cors.go
index ddbc829..6a671a5 100644
--- a/api/handler/cors.go
+++ b/api/handler/cors.go
@@ -55,6 +55,7 @@ func (h *handler) PutBucketCorsHandler(w http.ResponseWriter, r *http.Request) {
 		BktInfo:    bktInfo,
 		Reader:     r.Body,
 		NewDecoder: h.cfg.NewXMLDecoder,
+		UserAgent:  r.UserAgent(),
 	}
 
 	p.CopiesNumbers, err = h.pickCopiesNumbers(parseMetadata(r), reqInfo.Namespace, bktInfo.LocationConstraint)
diff --git a/api/handler/delete.go b/api/handler/delete.go
index 8ac9fa9..67b025d 100644
--- a/api/handler/delete.go
+++ b/api/handler/delete.go
@@ -147,7 +147,7 @@ func (h *handler) DeleteMultipleObjectsHandler(w http.ResponseWriter, r *http.Re
 
 	// Unmarshal list of keys to be deleted.
 	requested := &DeleteObjectsRequest{}
-	if err := h.cfg.NewXMLDecoder(r.Body).Decode(requested); err != nil {
+	if err := h.cfg.NewXMLDecoder(r.Body, r.UserAgent()).Decode(requested); err != nil {
 		h.logAndSendError(ctx, w, "couldn't decode body", reqInfo, fmt.Errorf("%w: %s", errors.GetAPIError(errors.ErrMalformedXML), err.Error()))
 		return
 	}
diff --git a/api/handler/handlers_test.go b/api/handler/handlers_test.go
index e5a224b..57f9741 100644
--- a/api/handler/handlers_test.go
+++ b/api/handler/handlers_test.go
@@ -97,11 +97,11 @@ func (c *configMock) DefaultCopiesNumbers(_ string) []uint32 {
 	return c.defaultCopiesNumbers
 }
 
-func (c *configMock) NewXMLDecoder(r io.Reader) *xml.Decoder {
+func (c *configMock) NewXMLDecoder(r io.Reader, _ string) *xml.Decoder {
 	return xml.NewDecoder(r)
 }
 
-func (c *configMock) BypassContentEncodingInChunks() bool {
+func (c *configMock) BypassContentEncodingInChunks(_ string) bool {
 	return c.bypassContentEncodingInChunks
 }
 
diff --git a/api/handler/lifecycle.go b/api/handler/lifecycle.go
index 3aa3ccc..4ebdb2e 100644
--- a/api/handler/lifecycle.go
+++ b/api/handler/lifecycle.go
@@ -69,7 +69,7 @@ func (h *handler) PutBucketLifecycleHandler(w http.ResponseWriter, r *http.Reque
 	}
 
 	cfg := new(data.LifecycleConfiguration)
-	if err = h.cfg.NewXMLDecoder(tee).Decode(cfg); err != nil {
+	if err = h.cfg.NewXMLDecoder(tee, r.UserAgent()).Decode(cfg); err != nil {
 		h.logAndSendError(ctx, w, "could not decode body", reqInfo, fmt.Errorf("%w: %s", apierr.GetAPIError(apierr.ErrMalformedXML), err.Error()))
 		return
 	}
diff --git a/api/handler/locking.go b/api/handler/locking.go
index bc5813d..e9a0a23 100644
--- a/api/handler/locking.go
+++ b/api/handler/locking.go
@@ -42,7 +42,7 @@ func (h *handler) PutBucketObjectLockConfigHandler(w http.ResponseWriter, r *htt
 	}
 
 	lockingConf := &data.ObjectLockConfiguration{}
-	if err = h.cfg.NewXMLDecoder(r.Body).Decode(lockingConf); err != nil {
+	if err = h.cfg.NewXMLDecoder(r.Body, r.UserAgent()).Decode(lockingConf); err != nil {
 		h.logAndSendError(ctx, w, "couldn't parse locking configuration", reqInfo, err)
 		return
 	}
@@ -124,7 +124,7 @@ func (h *handler) PutObjectLegalHoldHandler(w http.ResponseWriter, r *http.Reque
 	}
 
 	legalHold := &data.LegalHold{}
-	if err = h.cfg.NewXMLDecoder(r.Body).Decode(legalHold); err != nil {
+	if err = h.cfg.NewXMLDecoder(r.Body, r.UserAgent()).Decode(legalHold); err != nil {
 		h.logAndSendError(ctx, w, "couldn't parse legal hold configuration", reqInfo, err)
 		return
 	}
@@ -214,7 +214,7 @@ func (h *handler) PutObjectRetentionHandler(w http.ResponseWriter, r *http.Reque
 	}
 
 	retention := &data.Retention{}
-	if err = h.cfg.NewXMLDecoder(r.Body).Decode(retention); err != nil {
+	if err = h.cfg.NewXMLDecoder(r.Body, r.UserAgent()).Decode(retention); err != nil {
 		h.logAndSendError(ctx, w, "couldn't parse object retention", reqInfo, err)
 		return
 	}
diff --git a/api/handler/multipart_upload.go b/api/handler/multipart_upload.go
index bb10927..725052d 100644
--- a/api/handler/multipart_upload.go
+++ b/api/handler/multipart_upload.go
@@ -401,7 +401,7 @@ func (h *handler) CompleteMultipartUploadHandler(w http.ResponseWriter, r *http.
 	)
 
 	reqBody := new(CompleteMultipartUpload)
-	if err = h.cfg.NewXMLDecoder(r.Body).Decode(reqBody); err != nil {
+	if err = h.cfg.NewXMLDecoder(r.Body, r.UserAgent()).Decode(reqBody); err != nil {
 		h.logAndSendError(ctx, w, "could not read complete multipart upload xml", reqInfo,
 			fmt.Errorf("%w: %s", errors.GetAPIError(errors.ErrMalformedXML), err.Error()), additional...)
 		return
diff --git a/api/handler/put.go b/api/handler/put.go
index d08389f..58518f6 100644
--- a/api/handler/put.go
+++ b/api/handler/put.go
@@ -331,7 +331,9 @@ func (h *handler) getBodyReader(r *http.Request) (io.ReadCloser, error) {
 	}
 	r.Header.Set(api.ContentEncoding, strings.Join(resultContentEncoding, ","))
 
-	if !chunkedEncoding && !h.cfg.BypassContentEncodingInChunks() {
+	defBypass := h.cfg.BypassContentEncodingInChunks(r.UserAgent())
+
+	if !chunkedEncoding && !defBypass {
 		return nil, fmt.Errorf("%w: request is not chunk encoded, encodings '%s'",
 			apierr.GetAPIError(apierr.ErrInvalidEncodingMethod), strings.Join(encodings, ","))
 	}
@@ -442,7 +444,7 @@ func (h *handler) PostObject(w http.ResponseWriter, r *http.Request) {
 	if tagging := auth.MultipartFormValue(r, "tagging"); tagging != "" {
 		buffer := bytes.NewBufferString(tagging)
 		tags := new(data.Tagging)
-		if err = h.cfg.NewXMLDecoder(buffer).Decode(tags); err != nil {
+		if err = h.cfg.NewXMLDecoder(buffer, r.UserAgent()).Decode(tags); err != nil {
 			h.logAndSendError(ctx, w, "could not decode tag set", reqInfo,
 				fmt.Errorf("%w: %s", apierr.GetAPIError(apierr.ErrMalformedXML), err.Error()))
 			return
@@ -1033,7 +1035,7 @@ func (h *handler) parseLocationConstraint(r *http.Request) (*createBucketParams,
 	}
 
 	params := new(createBucketParams)
-	if err := h.cfg.NewXMLDecoder(r.Body).Decode(params); err != nil {
+	if err := h.cfg.NewXMLDecoder(r.Body, r.UserAgent()).Decode(params); err != nil {
 		return nil, fmt.Errorf("%w: %s", apierr.GetAPIError(apierr.ErrMalformedXML), err.Error())
 	}
 	return params, nil
diff --git a/api/handler/versioning.go b/api/handler/versioning.go
index d8b5bda..bf9c895 100644
--- a/api/handler/versioning.go
+++ b/api/handler/versioning.go
@@ -14,7 +14,7 @@ func (h *handler) PutBucketVersioningHandler(w http.ResponseWriter, r *http.Requ
 	reqInfo := middleware.GetReqInfo(ctx)
 
 	configuration := new(VersioningConfiguration)
-	if err := h.cfg.NewXMLDecoder(r.Body).Decode(configuration); err != nil {
+	if err := h.cfg.NewXMLDecoder(r.Body, r.UserAgent()).Decode(configuration); err != nil {
 		h.logAndSendError(ctx, w, "couldn't decode versioning configuration", reqInfo, errors.GetAPIError(errors.ErrIllegalVersioningConfigurationException))
 		return
 	}
diff --git a/api/layer/cors.go b/api/layer/cors.go
index 49b6cc4..ebf0a19 100644
--- a/api/layer/cors.go
+++ b/api/layer/cors.go
@@ -28,7 +28,7 @@ func (n *Layer) PutBucketCORS(ctx context.Context, p *PutCORSParams) error {
 		cors = &data.CORSConfiguration{}
 	)
 
-	if err := p.NewDecoder(tee).Decode(cors); err != nil {
+	if err := p.NewDecoder(tee, p.UserAgent).Decode(cors); err != nil {
 		return fmt.Errorf("xml decode cors: %w", err)
 	}
 
diff --git a/api/layer/layer.go b/api/layer/layer.go
index cafbb66..1589c7a 100644
--- a/api/layer/layer.go
+++ b/api/layer/layer.go
@@ -149,7 +149,8 @@ type (
 		BktInfo       *data.BucketInfo
 		Reader        io.Reader
 		CopiesNumbers []uint32
-		NewDecoder    func(io.Reader) *xml.Decoder
+		NewDecoder    func(io.Reader, string) *xml.Decoder
+		UserAgent     string
 	}
 
 	// CopyObjectParams stores object copy request parameters.
diff --git a/api/middleware/policy.go b/api/middleware/policy.go
index 9e36f0b..9faf069 100644
--- a/api/middleware/policy.go
+++ b/api/middleware/policy.go
@@ -61,7 +61,7 @@ type FrostFSIDInformer interface {
 }
 
 type XMLDecoder interface {
-	NewXMLDecoder(io.Reader) *xml.Decoder
+	NewXMLDecoder(io.Reader, string) *xml.Decoder
 }
 
 type ResourceTagging interface {
@@ -476,7 +476,7 @@ func determineRequestTags(r *http.Request, decoder XMLDecoder, op string) (map[s
 
 	if strings.HasSuffix(op, PutObjectTaggingOperation) || strings.HasSuffix(op, PutBucketTaggingOperation) {
 		tagging := new(data.Tagging)
-		if err := decoder.NewXMLDecoder(r.Body).Decode(tagging); err != nil {
+		if err := decoder.NewXMLDecoder(r.Body, r.UserAgent()).Decode(tagging); err != nil {
 			return nil, fmt.Errorf("%w: %s", apierr.GetAPIError(apierr.ErrMalformedXML), err.Error())
 		}
 		GetReqInfo(r.Context()).Tagging = tagging
diff --git a/api/router_mock_test.go b/api/router_mock_test.go
index 4f237f9..4c4bb82 100644
--- a/api/router_mock_test.go
+++ b/api/router_mock_test.go
@@ -151,7 +151,7 @@ func (f *frostFSIDMock) GetUserGroupIDsAndClaims(util.Uint160) ([]string, map[st
 type xmlMock struct {
 }
 
-func (m *xmlMock) NewXMLDecoder(r io.Reader) *xml.Decoder {
+func (m *xmlMock) NewXMLDecoder(r io.Reader, _ string) *xml.Decoder {
 	return xml.NewDecoder(r)
 }
 
diff --git a/cmd/s3-gw/app.go b/cmd/s3-gw/app.go
index dc5677b..31703b9 100644
--- a/cmd/s3-gw/app.go
+++ b/cmd/s3-gw/app.go
@@ -112,6 +112,7 @@ type (
 		namespaces                    Namespaces
 		defaultXMLNS                  bool
 		bypassContentEncodingInChunks bool
+		kludgeProfiles                map[string]*KludgeParams
 		clientCut                     bool
 		maxBufferSizeForPut           uint64
 		md5Enabled                    bool
@@ -297,6 +298,7 @@ func (s *appSettings) update(v *viper.Viper, log *zap.Logger) {
 	vhsNamespacesEnabled := s.prepareVHSNamespaces(v, log, defaultNamespaces)
 	defaultXMLNS := v.GetBool(cfgKludgeUseDefaultXMLNS)
 	bypassContentEncodingInChunks := v.GetBool(cfgKludgeBypassContentEncodingCheckInChunks)
+	kludgeProfiles := fetchKludgeProfiles(v)
 	clientCut := v.GetBool(cfgClientCut)
 	maxBufferSizeForPut := v.GetUint64(cfgBufferMaxSizeForPut)
 	md5Enabled := v.GetBool(cfgMD5Enabled)
@@ -332,6 +334,7 @@ func (s *appSettings) update(v *viper.Viper, log *zap.Logger) {
 	s.namespaces = nsConfig.Namespaces
 	s.defaultXMLNS = defaultXMLNS
 	s.bypassContentEncodingInChunks = bypassContentEncodingInChunks
+	s.kludgeProfiles = kludgeProfiles
 	s.clientCut = clientCut
 	s.maxBufferSizeForPut = maxBufferSizeForPut
 	s.md5Enabled = md5Enabled
@@ -392,10 +395,25 @@ func (s *appSettings) VHSNamespacesEnabled() map[string]bool {
 	return s.vhsNamespacesEnabled
 }
 
-func (s *appSettings) BypassContentEncodingInChunks() bool {
+func (s *appSettings) BypassContentEncodingInChunks(agent string) bool {
 	s.mu.RLock()
 	defer s.mu.RUnlock()
-	return s.bypassContentEncodingInChunks
+
+	var profile *KludgeParams
+	profiles := s.kludgeProfiles
+	for p := range profiles {
+		if strings.Contains(agent, p) {
+			profile = profiles[p]
+		}
+	}
+
+	return s.bypassContentEncodingInChunks || (profile != nil && profile.BypassContentEncodingCheckInChunks)
+}
+
+func (s *appSettings) KludgeProfiles() map[string]*KludgeParams {
+	s.mu.RLock()
+	defer s.mu.RUnlock()
+	return s.kludgeProfiles
 }
 
 func (s *appSettings) ClientCut() bool {
@@ -445,7 +463,7 @@ func (s *appSettings) LogHTTPConfig() s3middleware.LogHTTPConfig {
 	return s.httpLogging
 }
 
-func (s *appSettings) NewXMLDecoder(r io.Reader) *xml.Decoder {
+func (s *appSettings) NewXMLDecoder(r io.Reader, agent string) *xml.Decoder {
 	dec := xml.NewDecoder(r)
 	dec.CharsetReader = func(charset string, reader io.Reader) (io.Reader, error) {
 		enc, err := ianaindex.IANA.Encoding(charset)
@@ -456,10 +474,19 @@ func (s *appSettings) NewXMLDecoder(r io.Reader) *xml.Decoder {
 	}
 
 	s.mu.RLock()
-	if s.defaultXMLNS {
+	defer s.mu.RUnlock()
+
+	var profile *KludgeParams
+	for p := range s.kludgeProfiles {
+		if strings.Contains(agent, p) {
+			profile = s.kludgeProfiles[p]
+			break
+		}
+	}
+
+	if s.defaultXMLNS || (profile != nil && profile.UseDefaultXMLNS) {
 		dec.DefaultSpace = awsDefaultNamespace
 	}
-	s.mu.RUnlock()
 
 	return dec
 }
diff --git a/cmd/s3-gw/app_settings.go b/cmd/s3-gw/app_settings.go
index edf8769..06502b7 100644
--- a/cmd/s3-gw/app_settings.go
+++ b/cmd/s3-gw/app_settings.go
@@ -197,6 +197,7 @@ const ( // Settings.
 	cfgKludgeUseDefaultXMLNS                    = "kludge.use_default_xmlns"
 	cfgKludgeBypassContentEncodingCheckInChunks = "kludge.bypass_content_encoding_check_in_chunks"
 	cfgKludgeDefaultNamespaces                  = "kludge.default_namespaces"
+	cfgKludgeProfile                            = "kludge.profile"
 	// Web.
 	cfgWebReadTimeout       = "web.read_timeout"
 	cfgWebReadHeaderTimeout = "web.read_header_timeout"
@@ -563,6 +564,38 @@ func fetchDefaultCopiesNumbers(l *zap.Logger, v *viper.Viper) []uint32 {
 	return result
 }
 
+type KludgeParams struct {
+	UseDefaultXMLNS                    bool
+	BypassContentEncodingCheckInChunks bool
+}
+
+func fetchKludgeProfiles(v *viper.Viper) map[string]*KludgeParams {
+	kludgeProfiles := make(map[string]*KludgeParams)
+	for i := 0; ; i++ {
+		key := cfgKludgeProfile + "." + strconv.Itoa(i) + "."
+		userAgent := v.GetString(key + "user_agent")
+		if userAgent == "" {
+			break
+		}
+
+		kludgeParam := &KludgeParams{
+			UseDefaultXMLNS:                    v.GetBool(cfgKludgeUseDefaultXMLNS),
+			BypassContentEncodingCheckInChunks: v.GetBool(cfgKludgeBypassContentEncodingCheckInChunks),
+		}
+		if v.IsSet(key + "use_default_xmlns") {
+			kludgeParam.UseDefaultXMLNS = v.GetBool(key + "use_default_xmlns")
+		}
+
+		if v.IsSet(key + "bypass_content_encoding_check_in_chunks") {
+			kludgeParam.BypassContentEncodingCheckInChunks = v.GetBool(key + "bypass_content_encoding_check_in_chunks")
+		}
+
+		kludgeProfiles[userAgent] = kludgeParam
+	}
+
+	return kludgeProfiles
+}
+
 func fetchCopiesNumbers(l *zap.Logger, v *viper.Viper) map[string][]uint32 {
 	copiesNums := make(map[string][]uint32)
 	for i := 0; ; i++ {
diff --git a/cmd/s3-gw/decoder_test.go b/cmd/s3-gw/decoder_test.go
index 93827e6..50cd752 100644
--- a/cmd/s3-gw/decoder_test.go
+++ b/cmd/s3-gw/decoder_test.go
@@ -101,7 +101,7 @@ func TestDefaultNamespace(t *testing.T) {
 	} {
 		t.Run("", func(t *testing.T) {
 			model := new(handler.CompleteMultipartUpload)
-			err := tc.settings.NewXMLDecoder(bytes.NewBufferString(tc.input)).Decode(model)
+			err := tc.settings.NewXMLDecoder(bytes.NewBufferString(tc.input), "test").Decode(model)
 			if tc.err {
 				require.Error(t, err)
 			} else {
diff --git a/config/config.env b/config/config.env
index 44a8971..c556a52 100644
--- a/config/config.env
+++ b/config/config.env
@@ -186,6 +186,10 @@ S3_GW_KLUDGE_USE_DEFAULT_XMLNS=false
 S3_GW_KLUDGE_BYPASS_CONTENT_ENCODING_CHECK_IN_CHUNKS=false
 # Namespaces that should be handled as default
 S3_GW_KLUDGE_DEFAULT_NAMESPACES="" "root"
+# Kludge profiles
+S3_GW_KLUDGE_PROFILE_0_USER_AGENT=aws-cli
+S3_GW_KLUDGE_PROFILE_0_USE_DEFAULT_XMLNS=true
+S3_GW_KLUDGE_PROFILE_0_BYPASS_CONTENT_ENCODING_CHECK_IN_CHUNKS=true
 
 S3_GW_TRACING_ENABLED=false
 S3_GW_TRACING_ENDPOINT="localhost:4318"
diff --git a/config/config.yaml b/config/config.yaml
index 8f16946..98b706a 100644
--- a/config/config.yaml
+++ b/config/config.yaml
@@ -226,6 +226,13 @@ kludge:
   bypass_content_encoding_check_in_chunks: false
   # Namespaces that should be handled as default
   default_namespaces: [ "", "root" ]
+  # new profile section override defaults based on user agent
+  profile:
+    - user_agent: aws-cli
+      use_default_xmlns: false
+    - user_agent: aws-sdk-go
+      use_default_xmlns: true
+      bypass_content_encoding_check_in_chunks: false
 
 runtime:
   soft_memory_limit: 1gb
diff --git a/docs/configuration.md b/docs/configuration.md
index 0f98747..313c790 100644
--- a/docs/configuration.md
+++ b/docs/configuration.md
@@ -625,13 +625,19 @@ resolve_bucket:
 
 # `kludge` section
 
-Workarounds for non-standard use cases.
+Workarounds for non-standard use cases. In `profiles` subsection has the ability to override behavior for specific user agent.
 
 ```yaml
 kludge:
   use_default_xmlns: false
   bypass_content_encoding_check_in_chunks: false
   default_namespaces: [ "", "root" ]
+  profile:
+     - user_agent: aws-cli
+       use_default_xmlns: false
+     - user_agent: aws-sdk-go
+       use_default_xmlns: true
+       bypass_content_encoding_check_in_chunks: false
 ```
 
 | Parameter                                 | Type       | SIGHUP reload | Default value | Description                                                                                                                                                                                      |
@@ -639,6 +645,7 @@ kludge:
 | `use_default_xmlns`                       | `bool`     | yes           | `false`       | Enable using default xml namespace `http://s3.amazonaws.com/doc/2006-03-01/` when parse xml bodies.                                                                                              |
 | `bypass_content_encoding_check_in_chunks` | `bool`     | yes           | `false`       | Use this flag to be able to use [chunked upload approach](https://docs.aws.amazon.com/AmazonS3/latest/API/sigv4-streaming.html) without having `aws-chunked` value in `Content-Encoding` header. |
 | `default_namespaces`                      | `[]string` | yes           | `["","root"]` | Namespaces that should be handled as default.                                                                                                                                                    |
+| `profile.user_agent`                      | `string`   | yes           |               | Request UserAgent value.                                                                                                                                                                         |
 
 # `runtime` section
 Contains runtime parameters.