Bucket name rewriting

cmd/gofakes3/main.go

@@ -31,6 +31,7 @@ type fakeS3Flags struct {
 	initialBucket string
 	fixedTimeStr  string
 	noIntegrity   bool
+	hostBucket    bool
 
 	boltDb         string
 	directFsPath   string
@@ -45,6 +46,7 @@ func (f *fakeS3Flags) attach(flagSet *flag.FlagSet) {
 	flagSet.StringVar(&f.fixedTimeStr, "time", "", "RFC3339 format. If passed, the server's clock will always see this time; does not affect existing stored dates.")
 	flagSet.StringVar(&f.initialBucket, "initialbucket", "", "If passed, this bucket will be created on startup if it does not already exist.")
 	flagSet.BoolVar(&f.noIntegrity, "no-integrity", false, "Pass this flag to disable Content-MD5 validation when uploading.")
+	flagSet.BoolVar(&f.hostBucket, "hostbucket", false, "If passed, the bucket name will be extracted from the first segment of the hostname, rather than the first part of the URL path.")
 
 	// Backend specific:
 	flagSet.StringVar(&f.backendKind, "backend", "", "Backend to use to store data (memory, bolt, directfs, fs)")
@@ -181,6 +183,7 @@ func run() error {
 		gofakes3.WithTimeSkewLimit(timeSkewLimit),
 		gofakes3.WithTimeSource(timeSource),
 		gofakes3.WithLogger(gofakes3.GlobalLog()),
+		gofakes3.WithHostBucket(hostBucket),
 	)
 
 	return listenAndServe(values.host, faker.Server())

cors.go

@@ -2,7 +2,6 @@ package gofakes3
 
 import (
 	"net/http"
-	"regexp"
 	"strings"
 )
 
@@ -22,8 +21,6 @@ var (
 		"x-amz-meta-to",
 	}
 	corsHeadersString = strings.Join(corsHeaders, ", ")
-
-	bucketRewritePattern = regexp.MustCompile("(127.0.0.1:\\d{1,7})|(.localhost:\\d{1,7})|(localhost:\\d{1,7})")
 )
 
 type withCORS struct {
@@ -40,18 +37,5 @@ func (s *withCORS) ServeHTTP(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 
-	// Bucket name rewriting
-	// this is due to some inconsistencies in the AWS SDKs
-	bucket := bucketRewritePattern.ReplaceAllString(r.Host, "")
-	if len(bucket) > 0 {
-		s.log.Print(LogInfo, "rewrite bucket ->", bucket)
-		p := r.URL.Path
-		r.URL.Path = "/" + bucket
-		if p != "/" {
-			r.URL.Path += p
-		}
-	}
-	s.log.Print(LogInfo, "=>", r.URL)
-
 	s.r.ServeHTTP(w, r)
 }

gofakes3.go

@@ -60,6 +60,7 @@ type GoFakeS3 struct {
 	timeSkew          time.Duration
 	metadataSizeLimit int
 	integrityCheck    bool
+	hostBucket        bool
 	uploader          *uploader
 	requestID         uint64
 	log               Logger
@@ -96,12 +97,24 @@ func (g *GoFakeS3) nextRequestID() uint64 {
 
 // Create the AWS S3 API
 func (g *GoFakeS3) Server() http.Handler {
-	wc := &withCORS{r: http.HandlerFunc(g.routeBase), log: g.log}
+	var handler http.Handler = &withCORS{r: http.HandlerFunc(g.routeBase), log: g.log}
 
-	hf := func(w http.ResponseWriter, rq *http.Request) {
+	if g.timeSkew != 0 {
+		handler = g.timeSkewMiddleware(handler)
+	}
+
+	if g.hostBucket {
+		handler = g.hostBucketMiddleware(handler)
+	}
+
+	return handler
+}
+
+func (g *GoFakeS3) timeSkewMiddleware(handler http.Handler) http.Handler {
+	return http.HandlerFunc(func(w http.ResponseWriter, rq *http.Request) {
 		timeHdr := rq.Header.Get("x-amz-date")
 
-		if g.timeSkew > 0 && timeHdr != "" {
+		if timeHdr != "" {
 			rqTime, _ := time.Parse("20060102T150405Z", timeHdr)
 			at := g.timeSource.Now()
 			skew := at.Sub(rqTime)
@@ -112,10 +125,26 @@ func (g *GoFakeS3) Server() http.Handler {
 			}
 		}
 
-		wc.ServeHTTP(w, rq)
-	}
+		handler.ServeHTTP(w, rq)
+	})
+}
 
-	return http.HandlerFunc(hf)
+// hostBucketMiddleware forces the server to use VirtualHost-style bucket URLs:
+// https://docs.aws.amazon.com/AmazonS3/latest/dev/UsingBucket.html
+func (g *GoFakeS3) hostBucketMiddleware(handler http.Handler) http.Handler {
+	return http.HandlerFunc(func(w http.ResponseWriter, rq *http.Request) {
+		parts := strings.SplitN(rq.Host, ".", 2)
+		bucket := parts[0]
+
+		p := rq.URL.Path
+		rq.URL.Path = "/" + bucket
+		if p != "/" {
+			rq.URL.Path += p
+		}
+		g.log.Print(LogInfo, p, "=>", rq.URL)
+
+		handler.ServeHTTP(w, rq)
+	})
 }
 
 func (g *GoFakeS3) httpError(w http.ResponseWriter, r *http.Request, err error) {

gofakes3_internal_test.go

@@ -5,6 +5,7 @@ import (
 	"encoding/xml"
 	"fmt"
 	"log"
+	"net/http"
 	"net/http/httptest"
 	"testing"
 )
@@ -52,6 +53,35 @@ func TestHttpErrorWriteFailure(t *testing.T) {
 	}
 }
 
+func TestHostBucketMiddleware(t *testing.T) {
+	for _, tc := range []struct {
+		in   string
+		host string
+		out  string
+	}{
+		{"/", "foo", "/foo"},
+		{"/", "mybucket.localhost", "/mybucket"},
+		{"/object", "mybucket.localhost", "/mybucket/object"},
+	} {
+		t.Run("", func(t *testing.T) {
+			var g GoFakeS3
+			g.log = DiscardLog()
+
+			inner := http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+				if r.URL.Path != tc.out {
+					t.Fatal(r.URL.Path, "!=", tc.out)
+				}
+			})
+
+			handler := g.hostBucketMiddleware(inner)
+			rq := httptest.NewRequest("GET", tc.in, nil)
+			rq.Host = tc.host
+			rs := httptest.NewRecorder()
+			handler.ServeHTTP(rs, rq)
+		})
+	}
+}
+
 type failingResponseWriter struct {
 	*httptest.ResponseRecorder
 }

option.go

@@ -55,3 +55,13 @@ func WithGlobalLog() Option {
 func WithRequestID(id uint64) Option {
 	return func(g *GoFakeS3) { g.requestID = id }
 }
+
+// WithHostBucket enables or disables bucket rewriting in the router.
+// If active, the URL 'http://mybucket.localhost/object' will be routed
+// as if the URL path was '/mybucket/object'.
+//
+// See https://docs.aws.amazon.com/AmazonS3/latest/dev/UsingBucket.html
+// for details.
+func WithHostBucket(enabled bool) Option {
+	return func(g *GoFakeS3) { g.hostBucket = enabled }
+}