Fix some mp3/mp4 media files show as Raw

modules/setting/mime_type_map.go

@@ -4,7 +4,12 @@
 
 package setting
 
-import "strings"
+import (
+	"mime"
+	"strings"
+
+	"code.gitea.io/gitea/modules/log"
+)
 
 // MimeTypeMap defines custom mime type mapping settings
 var MimeTypeMap = struct {
@@ -21,6 +26,10 @@ func newMimeTypeMap() {
 	m := make(map[string]string, len(keys))
 	for _, key := range keys {
 		m[strings.ToLower(key.Name())] = key.Value()
+		err := mime.AddExtensionType(key.Name(), key.Value())
+		if err != nil {
+			log.Warn("mime.AddExtensionType(%s,%s): %v", key.Name(), key.Value(), err)
+		}
 	}
 	MimeTypeMap.Map = m
 	if len(keys) > 0 {

modules/typesniffer/typesniffer.go

@@ -7,7 +7,9 @@ package typesniffer
 import (
 	"fmt"
 	"io"
+	"mime"
 	"net/http"
+	"path/filepath"
 	"regexp"
 	"strings"
 
@@ -32,32 +34,32 @@ type SniffedType struct {
 
 // IsText etects if content format is plain text.
 func (ct SniffedType) IsText() bool {
-	return strings.Contains(ct.contentType, "text/")
+	return strings.HasPrefix(ct.contentType, "text/")
 }
 
 // IsImage detects if data is an image format
 func (ct SniffedType) IsImage() bool {
-	return strings.Contains(ct.contentType, "image/")
+	return strings.HasPrefix(ct.contentType, "image/")
 }
 
 // IsSvgImage detects if data is an SVG image format
 func (ct SniffedType) IsSvgImage() bool {
-	return strings.Contains(ct.contentType, SvgMimeType)
+	return strings.HasPrefix(ct.contentType, SvgMimeType)
 }
 
 // IsPDF detects if data is a PDF format
 func (ct SniffedType) IsPDF() bool {
-	return strings.Contains(ct.contentType, "application/pdf")
+	return strings.HasPrefix(ct.contentType, "application/pdf")
 }
 
 // IsVideo detects if data is an video format
 func (ct SniffedType) IsVideo() bool {
-	return strings.Contains(ct.contentType, "video/")
+	return strings.HasPrefix(ct.contentType, "video/")
 }
 
 // IsAudio detects if data is an video format
 func (ct SniffedType) IsAudio() bool {
-	return strings.Contains(ct.contentType, "audio/")
+	return strings.HasPrefix(ct.contentType, "audio/")
 }
 
 // IsRepresentableAsText returns true if file content can be represented as
@@ -66,6 +68,11 @@ func (ct SniffedType) IsRepresentableAsText() bool {
 	return ct.IsText() || ct.IsSvgImage()
 }
 
+// Mime return the mime
+func (ct SniffedType) Mime() string {
+	return strings.Split(ct.contentType, ";")[0]
+}
+
 // DetectContentType extends http.DetectContentType with more content types. Defaults to text/unknown if input is empty.
 func DetectContentType(data []byte) SniffedType {
 	if len(data) == 0 {
@@ -78,15 +85,33 @@ func DetectContentType(data []byte) SniffedType {
 		data = data[:sniffLen]
 	}
 
-	if (strings.Contains(ct, "text/plain") || strings.Contains(ct, "text/html")) && svgTagRegex.Match(data) ||
-		strings.Contains(ct, "text/xml") && svgTagInXMLRegex.Match(data) {
+	if (strings.HasPrefix(ct, "text/plain") || strings.HasPrefix(ct, "text/html")) && svgTagRegex.Match(data) ||
+		strings.HasPrefix(ct, "text/xml") && svgTagInXMLRegex.Match(data) {
 		// SVG is unsupported. https://github.com/golang/go/issues/15888
 		ct = SvgMimeType
 	}
 
 	return SniffedType{ct}
 }
 
+// DetectContentTypeExtFirst
+// detect content type by `name` first, if not found, detect by `reader`
+// Note: you may need `reader.Seek(0, io.SeekStart)` to reset the offset
+func DetectContentTypeExtFirst(name string, bytesOrReader interface{}) (SniffedType, error) {
+	ct := mime.TypeByExtension(filepath.Ext(name))
+	if ct != "" && !strings.HasPrefix(ct, "text/") {
+		return SniffedType{ct}, nil
+	}
+	if r, ok := bytesOrReader.(io.Reader); ok {
+		st, err := DetectContentTypeFromReader(r)
+		if nil != err {
+			return SniffedType{}, err
+		}
+		return st, nil
+	}
+	return DetectContentType(bytesOrReader.([]byte)), nil
+}
+
 // DetectContentTypeFromReader guesses the content type contained in the reader.
 func DetectContentTypeFromReader(r io.Reader) (SniffedType, error) {
 	buf := make([]byte, sniffLen)

routers/common/repo.go

@@ -7,9 +7,10 @@ package common
 import (
 	"fmt"
 	"io"
-	"path"
-	"path/filepath"
+	"net/http"
+	"strconv"
 	"strings"
+	"time"
 
 	"code.gitea.io/gitea/modules/charset"
 	"code.gitea.io/gitea/modules/context"
@@ -21,7 +22,55 @@ import (
 	"code.gitea.io/gitea/modules/util"
 )
 
-// ServeBlob download a git.Blob
+func setCommonHeaders(ctx *context.Context, name string, data interface{}) error {
+	// Google Chrome dislike commas in filenames, so let's change it to a space
+	name = strings.ReplaceAll(name, ",", " ")
+	// FIXME: we may need a setting for global cache control
+	ctx.Resp.Header().Set("Cache-Control", "public,max-age=86400")
+
+	st, err := typesniffer.DetectContentTypeExtFirst(name, data)
+	if nil != err {
+		return err
+	}
+
+	// reset the offset to the start of served file
+	if seeker, ok := data.(io.ReadSeeker); ok {
+		_, _ = seeker.Seek(0, io.SeekStart)
+	}
+
+	if st.IsText() || ctx.FormBool("render") {
+		var cs string
+		var err error
+		if reader, ok := data.(io.ReadSeeker); ok {
+			cs, err = charset.DetectEncodingFromReader(reader)
+			_, _ = reader.Seek(0, io.SeekStart)
+		} else {
+			cs, err = charset.DetectEncoding(data.([]byte))
+		}
+		if err != nil {
+			log.Error("Detect raw file %s charset failed: %v, using by default utf-8", name, err)
+			cs = "utf-8"
+		}
+
+		// http.ServeContent has bug on detecting GBK charset
+		ctx.Resp.Header().Set("Content-Type", fmt.Sprintf("%s; charset=%s", st.Mime(), strings.ToLower(cs)))
+	} else if (st.IsImage() || st.IsPDF()) && (setting.UI.SVG.Enabled || !st.IsSvgImage()) {
+		ctx.Resp.Header().Set("Access-Control-Expose-Headers", "Content-Disposition")
+		ctx.Resp.Header().Set("Content-Disposition", fmt.Sprintf(`inline; filename="%s"`, name))
+		if st.IsSvgImage() {
+			ctx.Resp.Header().Set("Content-Security-Policy", "default-src 'none'; style-src 'unsafe-inline'; sandbox")
+			ctx.Resp.Header().Set("X-Content-Type-Options", "nosniff")
+			ctx.Resp.Header().Set("Content-Type", typesniffer.SvgMimeType)
+		}
+	} else {
+		ctx.Resp.Header().Set("Access-Control-Expose-Headers", "Content-Disposition")
+		ctx.Resp.Header().Set("Content-Disposition", fmt.Sprintf(`attachment; filename="%s"`, name))
+	}
+	return nil
+}
+
+// ServeBlob serve git.Blob which represents a normal(non-lfs) file stored in repositories
+// todo: implement io.Seeker for git.Blob.blobReader to support Range-Request
 func ServeBlob(ctx *context.Context, blob *git.Blob) error {
 	if httpcache.HandleGenericETagCache(ctx.Req, ctx.Resp, `"`+blob.ID.String()+`"`) {
 		return nil
@@ -37,70 +86,39 @@ func ServeBlob(ctx *context.Context, blob *git.Blob) error {
 		}
 	}()
 
-	return ServeData(ctx, ctx.Repo.TreePath, blob.Size(), dataRc)
-}
-
-// ServeData download file from io.Reader
-func ServeData(ctx *context.Context, name string, size int64, reader io.Reader) error {
 	buf := make([]byte, 1024)
-	n, err := util.ReadAtMost(reader, buf)
+	n, err := util.ReadAtMost(dataRc, buf)
 	if err != nil {
 		return err
 	}
 	if n >= 0 {
 		buf = buf[:n]
 	}
 
-	ctx.Resp.Header().Set("Cache-Control", "public,max-age=86400")
-
+	size := blob.Size()
 	if size >= 0 {
-		ctx.Resp.Header().Set("Content-Length", fmt.Sprintf("%d", size))
+		ctx.Resp.Header().Set("Content-Length", strconv.FormatInt(size, 10))
 	} else {
-		log.Error("ServeData called to serve data: %s with size < 0: %d", name, size)
+		log.Error("ServeData called to serve data: %s with size < 0: %d", ctx.Repo.TreePath, size)
 	}
-	name = path.Base(name)
-
-	// Google Chrome dislike commas in filenames, so let's change it to a space
-	name = strings.ReplaceAll(name, ",", " ")
 
-	st := typesniffer.DetectContentType(buf)
-
-	mappedMimeType := ""
-	if setting.MimeTypeMap.Enabled {
-		fileExtension := strings.ToLower(filepath.Ext(name))
-		mappedMimeType = setting.MimeTypeMap.Map[fileExtension]
-	}
-	if st.IsText() || ctx.FormBool("render") {
-		cs, err := charset.DetectEncoding(buf)
-		if err != nil {
-			log.Error("Detect raw file %s charset failed: %v, using by default utf-8", name, err)
-			cs = "utf-8"
-		}
-		if mappedMimeType == "" {
-			mappedMimeType = "text/plain"
-		}
-		ctx.Resp.Header().Set("Content-Type", mappedMimeType+"; charset="+strings.ToLower(cs))
-	} else {
-		ctx.Resp.Header().Set("Access-Control-Expose-Headers", "Content-Disposition")
-		if mappedMimeType != "" {
-			ctx.Resp.Header().Set("Content-Type", mappedMimeType)
-		}
-		if (st.IsImage() || st.IsPDF()) && (setting.UI.SVG.Enabled || !st.IsSvgImage()) {
-			ctx.Resp.Header().Set("Content-Disposition", fmt.Sprintf(`inline; filename="%s"`, name))
-			if st.IsSvgImage() {
-				ctx.Resp.Header().Set("Content-Security-Policy", "default-src 'none'; style-src 'unsafe-inline'; sandbox")
-				ctx.Resp.Header().Set("X-Content-Type-Options", "nosniff")
-				ctx.Resp.Header().Set("Content-Type", typesniffer.SvgMimeType)
-			}
-		} else {
-			ctx.Resp.Header().Set("Content-Disposition", fmt.Sprintf(`attachment; filename="%s"`, name))
-		}
+	if err := setCommonHeaders(ctx, ctx.Repo.TreePath, buf); err != nil {
+		return err
 	}
 
 	_, err = ctx.Resp.Write(buf)
 	if err != nil {
 		return err
 	}
-	_, err = io.Copy(ctx.Resp, reader)
+	_, err = io.Copy(ctx.Resp, dataRc)
 	return err
 }
+
+// ServeLargeFile Serve files stored with Git LFS and attachments uploaded on the Releases page
+func ServeLargeFile(ctx *context.Context, name string, time time.Time, reader io.ReadSeeker) error {
+	if err := setCommonHeaders(ctx, name, reader); err != nil {
+		return err
+	}
+	http.ServeContent(ctx.Resp, ctx.Req, name, time, reader)
+	return nil
+}

routers/web/repo/view.go

@@ -427,7 +427,7 @@ func renderFile(ctx *context.Context, entry *git.TreeEntry, treeLink, rawLink st
 				}
 				buf = buf[:n]
 
-				st = typesniffer.DetectContentType(buf)
+				st, _ = typesniffer.DetectContentTypeExtFirst(blob.Name(), buf)
 				isTextFile = st.IsText()
 
 				fileSize = meta.Size