feat(upload): auto-detect file content-type if not sent

openapi/Swarm.yaml

@@ -342,7 +342,7 @@ paths:
     post:
       summary: "Upload a file or collection of files"
       description:
-        "Upload single files or collections of files. To upload a collection, send a multipart request with files in the form data with appropriate headers. Tar files can be uploaded with the `swarm-collection` header to extract and upload the directory structure. Without the `swarm-collection` header, requests are treated as single file uploads. Multipart requests are always treated as collections; use the `swarm-index-document` header to specify a single file to serve."
+        "Upload single files or collections of files. For a single file, `Content-Type` is optional: when present it is stored as metadata as-is; when absent the server infers a type from the start of the body. To upload a collection, send a multipart request with files in the form data with appropriate headers. Tar files can be uploaded with the `swarm-collection` header to extract and upload the directory structure. Without the `swarm-collection` header, requests are treated as single file uploads. Multipart requests are always treated as collections; use the `swarm-index-document` header to specify a single file to serve."
       tags:
         - BZZ
       parameters:

openapi/SwarmCommon.yaml

@@ -1176,7 +1176,7 @@ components:
       name: Content-Type
       schema:
         type: string
-      description: The specified content-type is preserved for download of the asset
+      description: "Single file: trimmed Content-Type is stored as-is or, if omitted or empty, inferred from the first bytes without validating against the body; tar (`swarm-collection`) and multipart collection uploads still need a full-body Content-Type (e.g. `application/x-tar` or `multipart/form-data` with boundary) so the request can be parsed."
 
     SwarmIndexDocumentParameter:
       in: header

pkg/api/api.go

@@ -16,7 +16,6 @@ import (
 	"io"
 	"math"
 	"math/big"
-	"mime"
 	"net/http"
 	"reflect"
 	"strconv"
@@ -302,10 +301,6 @@ func New(
 	s.chainBackend = chainBackend
 	s.metricsRegistry = newDebugMetrics()
 	s.preMapHooks = map[string]func(v string) (string, error){
-		"mimeMediaType": func(v string) (string, error) {
-			typ, _, err := mime.ParseMediaType(v)
-			return typ, err
-		},
 		"decBase64url": func(v string) (string, error) {
 			buf, err := base64.URLEncoding.DecodeString(v)
 			return string(buf), err

pkg/api/bzz.go

@@ -5,10 +5,13 @@
 package api
 
 import (
+	"bytes"
 	"context"
 	"encoding/hex"
 	"errors"
 	"fmt"
+	"io"
+	"mime"
 	"net/http"
 	"path"
 	"path/filepath"
@@ -51,6 +54,8 @@ const (
 	largeFileBufferSize = 16 * 32 * 1024
 
 	largeBufferFilesizeThreshold = 10 * 1000000 // ten megs
+
+	contentTypeSniffLen = 512
 )
 
 func lookaheadBufferSize(size int64) int {
@@ -65,7 +70,7 @@ func (s *Service) bzzUploadHandler(w http.ResponseWriter, r *http.Request) {
 	defer span.Finish()
 
 	headers := struct {
-		ContentType    string           `map:"Content-Type,mimeMediaType" validate:"required"`
+		ContentType    string           `map:"Content-Type"`
 		BatchID        []byte           `map:"Swarm-Postage-Batch-Id" validate:"required"`
 		SwarmTag       uint64           `map:"Swarm-Tag"`
 		Pin            bool             `map:"Swarm-Pin"`
@@ -137,8 +142,18 @@ func (s *Service) bzzUploadHandler(w http.ResponseWriter, r *http.Request) {
 		logger:         logger,
 	}
 
-	if headers.IsDir || headers.ContentType == multiPartFormData {
-		s.dirUploadHandler(ctx, logger, span, ow, r, putter, r.Header.Get(ContentTypeHeader), headers.Encrypt, tag, headers.RLevel, headers.Act, headers.HistoryAddress)
+	contentTypeHdr := strings.TrimSpace(headers.ContentType)
+	r.Header.Set(ContentTypeHeader, contentTypeHdr)
+	mt, _, errParseCT := mime.ParseMediaType(contentTypeHdr)
+	isMultipart := errParseCT == nil && mt == multiPartFormData
+	if headers.IsDir || isMultipart {
+		if contentTypeHdr == "" {
+			logger.Debug("content-type required for directory upload")
+			logger.Error(nil, "content-type required for directory upload")
+			jsonhttp.BadRequest(w, errInvalidContentType)
+			return
+		}
+		s.dirUploadHandler(ctx, logger, span, ow, r, putter, headers.Encrypt, tag, headers.RLevel, headers.Act, headers.HistoryAddress)
 		return
 	}
 	s.fileUploadHandler(ctx, logger, span, ow, r, putter, headers.Encrypt, tag, headers.RLevel, headers.Act, headers.HistoryAddress)
@@ -174,8 +189,23 @@ func (s *Service) fileUploadHandler(
 
 	p := requestPipelineFn(putter, encrypt, rLevel)
 
+	sniffBuf := make([]byte, contentTypeSniffLen)
+	n, err := io.ReadFull(r.Body, sniffBuf)
+	sniffBuf = sniffBuf[:n]
+	if err != nil && !errors.Is(err, io.EOF) && !errors.Is(err, io.ErrUnexpectedEOF) {
+		logger.Debug("body read failed", "file_name", queries.FileName, "error", err)
+		logger.Error(nil, "body read failed", "file_name", queries.FileName)
+		jsonhttp.BadRequest(w, "failed to read request body")
+		return
+	}
+
+	if r.Header.Get(ContentTypeHeader) == "" {
+		r.Header.Set(ContentTypeHeader, http.DetectContentType(sniffBuf))
+	}
+	bodyForStore := io.MultiReader(bytes.NewReader(sniffBuf), r.Body)
+
 	// first store the file and get its reference
-	fr, err := p(ctx, r.Body)
+	fr, err := p(ctx, bodyForStore)
 	if err != nil {
 		logger.Debug("file store failed", "file_name", queries.FileName, "error", err)
 		logger.Error(nil, "file store failed", "file_name", queries.FileName)
@@ -240,7 +270,7 @@ func (s *Service) fileUploadHandler(
 	}
 
 	fileMtdt := map[string]string{
-		manifest.EntryMetadataContentTypeKey: r.Header.Get(ContentTypeHeader), // Content-Type has already been validated.
+		manifest.EntryMetadataContentTypeKey: r.Header.Get(ContentTypeHeader),
 		manifest.EntryMetadataFilenameKey:    queries.FileName,
 	}
 

pkg/api/bzz_test.go

@@ -465,6 +465,44 @@ func TestBzzFiles(t *testing.T) {
 		)
 	})
 
+	t.Run("omit-content-type-uses-sniff", func(t *testing.T) {
+		fileName := "plain.txt"
+		var resp api.BzzUploadResponse
+		jsonhttptest.Request(t, client, http.MethodPost, fileUploadResource+"?name="+fileName, http.StatusCreated,
+			jsonhttptest.WithRequestHeader(api.SwarmDeferredUploadHeader, "true"),
+			jsonhttptest.WithRequestHeader(api.SwarmPostageBatchIdHeader, batchOkStr),
+			jsonhttptest.WithRequestBody(bytes.NewReader(simpleData)),
+			jsonhttptest.WithUnmarshalJSONResponse(&resp),
+		)
+		rootHash := resp.Reference.String()
+		jsonhttptest.Request(t, client, http.MethodGet, fileDownloadResource(rootHash), http.StatusOK,
+			jsonhttptest.WithExpectedResponse(simpleData),
+			jsonhttptest.WithExpectedContentLength(len(simpleData)),
+			jsonhttptest.WithExpectedResponseHeader(api.ContentTypeHeader, "text/plain; charset=utf-8"),
+			jsonhttptest.WithExpectedResponseHeader(api.ContentDispositionHeader, fmt.Sprintf(`inline; filename="%s"`, fileName)),
+		)
+	})
+
+	t.Run("image-content-type-preserved", func(t *testing.T) {
+		ct := "image/png"
+		fileName := "test.txt"
+		var resp api.BzzUploadResponse
+		jsonhttptest.Request(t, client, http.MethodPost, fileUploadResource+"?name="+fileName, http.StatusCreated,
+			jsonhttptest.WithRequestHeader(api.SwarmDeferredUploadHeader, "true"),
+			jsonhttptest.WithRequestHeader(api.SwarmPostageBatchIdHeader, batchOkStr),
+			jsonhttptest.WithRequestHeader(api.ContentTypeHeader, ct),
+			jsonhttptest.WithRequestBody(bytes.NewReader(simpleData)),
+			jsonhttptest.WithUnmarshalJSONResponse(&resp),
+		)
+		rootHash := resp.Reference.String()
+		jsonhttptest.Request(t, client, http.MethodGet, fileDownloadResource(rootHash), http.StatusOK,
+			jsonhttptest.WithExpectedResponse(simpleData),
+			jsonhttptest.WithExpectedContentLength(len(simpleData)),
+			jsonhttptest.WithExpectedResponseHeader(api.ContentTypeHeader, ct),
+			jsonhttptest.WithExpectedResponseHeader(api.ContentDispositionHeader, fmt.Sprintf(`inline; filename="%s"`, fileName)),
+		)
+	})
+
 	t.Run("upload-then-download-and-check-data", func(t *testing.T) {
 		fileName := "sample.html"
 		rootHash := "36e6c1bbdfee6ac21485d5f970479fd1df458d36df9ef4e8179708ed46da557f"

pkg/api/dirs.go

@@ -44,7 +44,6 @@ func (s *Service) dirUploadHandler(
 	w http.ResponseWriter,
 	r *http.Request,
 	putter storer.PutterSession,
-	contentTypeString string,
 	encrypt bool,
 	tag uint64,
 	rLevel redundancy.Level,
@@ -58,7 +57,7 @@ func (s *Service) dirUploadHandler(
 	}
 
 	// The error is ignored because the header was already validated by the caller.
-	mediaType, params, _ := mime.ParseMediaType(contentTypeString)
+	mediaType, params, _ := mime.ParseMediaType(r.Header.Get(ContentTypeHeader))
 
 	var dReader dirReader
 	switch mediaType {