Data Integrity

api/r0/upload_async.go

@@ -1,42 +1,81 @@
 package r0
 
 import (
+	"bytes"
 	"errors"
+	"io"
 	"net/http"
 	"path/filepath"
 
 	"github.com/getsentry/sentry-go"
+	"github.com/h2non/filetype"
 	"github.com/sirupsen/logrus"
 	"github.com/t2bot/matrix-media-repo/api/_apimeta"
 	"github.com/t2bot/matrix-media-repo/api/_responses"
 	"github.com/t2bot/matrix-media-repo/api/_routers"
 	"github.com/t2bot/matrix-media-repo/common"
 	"github.com/t2bot/matrix-media-repo/common/rcontext"
 	"github.com/t2bot/matrix-media-repo/pipelines/pipeline_upload"
+	"github.com/t2bot/matrix-media-repo/util"
 )
 
 func UploadMediaAsync(r *http.Request, rctx rcontext.RequestContext, user _apimeta.UserInfo) interface{} {
 	server := _routers.GetParam("server", r)
 	mediaId := _routers.GetParam("mediaId", r)
 	filename := filepath.Base(r.URL.Query().Get("filename"))
-
 	rctx = rctx.LogWithFields(logrus.Fields{
 		"mediaId":  mediaId,
 		"server":   server,
 		"filename": filename,
 	})
-
-	if r.Host != server {
+	// GK CUSTOMIZATION: Sanitize the filename
+	if len(filename) > rctx.Config.Uploads.MaxFilenameLength {
+		rctx.Log.Info("Filename too long")
 		return &_responses.ErrorResponse{
-			Code:         common.ErrCodeNotFound,
-			Message:      "Upload request is for another domain.",
-			InternalCode: common.ErrCodeForbidden,
+			Code:         common.ErrCodeBadRequest,
+			Message:      "Filename too long.",
+			InternalCode: common.ErrCodeBadRequest,
 		}
 	}
 
 	contentType := r.Header.Get("Content-Type")
 	if contentType == "" {
 		contentType = "application/octet-stream" // binary
+	} else {
+		// GK CUSTOMIZATION: Check if the file type is supported
+		buf, err := io.ReadAll(r.Body)
+		r.Body = io.NopCloser(bytes.NewBuffer(buf))
+		if err != nil {
+			return &_responses.ErrorResponse{
+				Code:         common.ErrCodeBadRequest,
+				Message:      "Error reading file.",
+				InternalCode: common.ErrCodeBadRequest,
+			}
+		}
+		kind, err := filetype.Match(buf)
+		if err != nil {
+			return &_responses.ErrorResponse{
+				Code:         common.ErrCodeBadRequest,
+				Message:      "Error matching file type.",
+				InternalCode: common.ErrCodeBadRequest,
+			}
+		}
+		if !util.IsSupportedFileType(kind.Extension, rctx.Config.Uploads.SupportedFileTypes) {
+			rctx.Log.Info("Unsupported file type: ", kind.Extension)
+			return &_responses.ErrorResponse{
+				Code:         common.ErrCodeBadRequest,
+				Message:      "Unsupported file type.",
+				InternalCode: common.ErrCodeBadRequest,
+			}
+		}
+	}
+
+	if r.Host != server {
+		return &_responses.ErrorResponse{
+			Code:         common.ErrCodeNotFound,
+			Message:      "Upload request is for another domain.",
+			InternalCode: common.ErrCodeForbidden,
+		}
 	}
 
 	// Early sizing constraints (reject requests which claim to be too large/small)

api/r0/upload_sync.go

@@ -1,12 +1,15 @@
 package r0
 
 import (
+	"bytes"
 	"errors"
+	"io"
 	"net/http"
 	"path/filepath"
 	"strconv"
 
 	"github.com/getsentry/sentry-go"
+	"github.com/h2non/filetype"
 	"github.com/sirupsen/logrus"
 	"github.com/t2bot/matrix-media-repo/api/_apimeta"
 	"github.com/t2bot/matrix-media-repo/api/_responses"
@@ -23,14 +26,50 @@ type MediaUploadedResponse struct {
 
 func UploadMediaSync(r *http.Request, rctx rcontext.RequestContext, user _apimeta.UserInfo) interface{} {
 	filename := filepath.Base(r.URL.Query().Get("filename"))
-
 	rctx = rctx.LogWithFields(logrus.Fields{
 		"filename": filename,
 	})
+	// GK CUSTOMIZATION: Sanitize the filename
+	if len(filename) > rctx.Config.Uploads.MaxFilenameLength {
+		rctx.Log.Info("Filename too long")
+		return &_responses.ErrorResponse{
+			Code:         common.ErrCodeBadRequest,
+			Message:      "Filename too long.",
+			InternalCode: common.ErrCodeBadRequest,
+		}
+	}
 
 	contentType := r.Header.Get("Content-Type")
 	if contentType == "" {
 		contentType = "application/octet-stream" // binary
+	} else {
+		// GK CUSTOMIZATION: Check if the file type is supported
+		buf, err := io.ReadAll(r.Body)
+		r.Body = io.NopCloser(bytes.NewBuffer(buf))
+		if err != nil {
+			return &_responses.ErrorResponse{
+				Code:         common.ErrCodeBadRequest,
+				Message:      "Error reading file.",
+				InternalCode: common.ErrCodeBadRequest,
+			}
+		}
+		kind, err := filetype.Match(buf)
+		if err != nil {
+			return &_responses.ErrorResponse{
+				Code:         common.ErrCodeBadRequest,
+				Message:      "Error matching file type.",
+				InternalCode: common.ErrCodeBadRequest,
+			}
+		}
+		if !util.IsSupportedFileType(kind.Extension, rctx.Config.Uploads.SupportedFileTypes) {
+			rctx.Log.Info("Unsupported file type: ", kind.Extension)
+			return &_responses.ErrorResponse{
+				Code:         common.ErrCodeBadRequest,
+				Message:      "Unsupported file type.",
+				InternalCode: common.ErrCodeBadRequest,
+			}
+		}
+		//
 	}
 
 	// Early sizing constraints (reject requests which claim to be too large/small)

api/routes.go

@@ -32,7 +32,7 @@ func buildRoutes() http.Handler {
 	// Standard (spec) features
 	register([]string{"PUT"}, PrefixMedia, "upload/:server/:mediaId", mxV3, router, makeRoute(_routers.RequireAccessToken(r0.UploadMediaAsync), "upload_async", counter))
 	register([]string{"POST"}, PrefixMedia, "upload", mxSpecV3Transition, router, makeRoute(_routers.RequireAccessToken(r0.UploadMediaSync), "upload", counter))
-	downloadRoute := makeRoute(_routers.OptionalAccessToken(r0.DownloadMedia), "download", counter)
+	downloadRoute := makeRoute(_routers.RequireAccessToken(r0.DownloadMedia), "download", counter)
 	register([]string{"GET"}, PrefixMedia, "download/:server/:mediaId/:filename", mxSpecV3Transition, router, downloadRoute)
 	register([]string{"GET"}, PrefixMedia, "download/:server/:mediaId", mxSpecV3Transition, router, downloadRoute)
 	register([]string{"GET"}, PrefixMedia, "thumbnail/:server/:mediaId", mxSpecV3Transition, router, makeRoute(_routers.OptionalAccessToken(r0.ThumbnailMedia), "thumbnail", counter))

common/config/conf_min_shared.go

@@ -20,6 +20,40 @@ func NewDefaultMinimumRepoConfig() MinimumRepoConfig {
 			TargetBytesPerPart: 209715200, // 200mb
 		},
 		Uploads: UploadsConfig{
+			MaxFilenameLength: 24,
+			SupportedFileTypes: []string{
+				"docx",
+				"doc",
+				"xlsx",
+				"xls",
+				"odt",
+				"ods",
+				"odp",
+				"csv",
+				"txt",
+				"pdf",
+				"ppr",
+				"pptx",
+				"ppt",
+				"jpg",
+				"png",
+				"gif",
+				"heic",
+				"bmp",
+				"webp",
+				"svg",
+				"mpeg4",
+				"mpeg",
+				"h264",
+				"webm",
+				"mkv",
+				"avi",
+				"mov",
+				"mp3",
+				"wav",
+				"flac",
+				"zip",
+			},
 			MaxSizeBytes:         104857600, // 100mb
 			MinSizeBytes:         100,
 			ReportedMaxSizeBytes: 0,

common/config/models_domain.go

@@ -19,6 +19,8 @@ type QuotasConfig struct {
 }
 
 type UploadsConfig struct {
+	MaxFilenameLength    int          `yaml:"maxFilenameLength"`
+	SupportedFileTypes   []string     `yaml:"supportedFileTypes"`
 	MaxSizeBytes         int64        `yaml:"maxBytes"`
 	MinSizeBytes         int64        `yaml:"minBytes"`
 	ReportedMaxSizeBytes int64        `yaml:"reportedMaxBytes"`

config.sample.yaml

@@ -232,6 +232,42 @@ archiving:
 
 # The file upload settings for the media repository
 uploads:
+  # GK CUSTOMIZATION
+  maxFilenameLength: 24
+  supportedFileTypes:
+  - "docx"
+  - "doc"
+  - "xlsx"
+  - "xls"
+  - "odt"
+  - "ods"
+  - "odp"
+  - "csv"
+  - "txt"
+  - "pdf"
+  - "ppr"
+  - "pptx"
+  - "ppt"
+  - "jpg"
+  - "png"
+  - "gif"
+  - "heic"
+  - "bmp"
+  - "webp"
+  - "svg"
+  - "mpeg4"
+  - "mpeg"
+  - "h264"
+  - "webm"
+  - "mkv"
+  - "avi"
+  - "mov"
+  - "mp3"
+  - "wav"
+  - "flac"
+  - "zip"
+  ################## 
+
   # The maximum individual file size a user can upload.
   maxBytes: 104857600 # 100MB default, 0 to disable
 

go.mod

@@ -151,6 +151,7 @@ require (
 	github.com/facebookgo/clock v0.0.0-20150410010913-600d898af40a // indirect
 	github.com/fastly/go-utils v0.0.0-20180712184237-d95a45783239 // indirect
 	github.com/go-sql-driver/mysql v1.5.0 // indirect
+	github.com/h2non/filetype v1.1.3
 	github.com/hashicorp/yamux v0.1.1 // indirect
 	github.com/jehiah/go-strftime v0.0.0-20171201141054-1d33003b3869 // indirect
 	github.com/lestrrat/go-envload v0.0.0-20180220120943-6ed08b54a570 // indirect