2023-03-12 18:00:57 +03:00
|
|
|
// GoToSocial
|
|
|
|
// Copyright (C) GoToSocial Authors admin@gotosocial.org
|
|
|
|
// SPDX-License-Identifier: AGPL-3.0-or-later
|
|
|
|
//
|
|
|
|
// This program is free software: you can redistribute it and/or modify
|
|
|
|
// it under the terms of the GNU Affero General Public License as published by
|
|
|
|
// the Free Software Foundation, either version 3 of the License, or
|
|
|
|
// (at your option) any later version.
|
|
|
|
//
|
|
|
|
// This program is distributed in the hope that it will be useful,
|
|
|
|
// but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
|
// GNU Affero General Public License for more details.
|
|
|
|
//
|
|
|
|
// You should have received a copy of the GNU Affero General Public License
|
|
|
|
// along with this program. If not, see <http://www.gnu.org/licenses/>.
|
2021-05-08 15:25:55 +03:00
|
|
|
|
|
|
|
package fileserver
|
|
|
|
|
|
|
|
import (
|
2022-06-08 21:38:03 +03:00
|
|
|
"fmt"
|
2022-12-21 13:17:43 +03:00
|
|
|
"io"
|
2021-05-08 15:25:55 +03:00
|
|
|
"net/http"
|
2022-07-30 15:42:47 +03:00
|
|
|
"strconv"
|
2023-02-06 11:50:16 +03:00
|
|
|
"strings"
|
2023-02-16 16:18:53 +03:00
|
|
|
"time"
|
2021-05-08 15:25:55 +03:00
|
|
|
|
2023-02-06 11:50:16 +03:00
|
|
|
"codeberg.org/gruf/go-fastcopy"
|
2021-05-08 15:25:55 +03:00
|
|
|
"github.com/gin-gonic/gin"
|
2023-01-02 15:10:50 +03:00
|
|
|
apimodel "github.com/superseriousbusiness/gotosocial/internal/api/model"
|
|
|
|
apiutil "github.com/superseriousbusiness/gotosocial/internal/api/util"
|
2022-06-08 21:38:03 +03:00
|
|
|
"github.com/superseriousbusiness/gotosocial/internal/gtserror"
|
2022-07-19 11:47:55 +03:00
|
|
|
"github.com/superseriousbusiness/gotosocial/internal/log"
|
2021-05-08 15:25:55 +03:00
|
|
|
"github.com/superseriousbusiness/gotosocial/internal/oauth"
|
|
|
|
)
|
|
|
|
|
|
|
|
// ServeFile is for serving attachments, headers, and avatars to the requester from instance storage.
|
|
|
|
//
|
|
|
|
// Note: to mitigate scraping attempts, no information should be given out on a bad request except "404 page not found".
|
|
|
|
// Don't give away account ids or media ids or anything like that; callers shouldn't be able to infer anything.
|
2023-01-02 15:10:50 +03:00
|
|
|
func (m *Module) ServeFile(c *gin.Context) {
|
2021-05-08 15:25:55 +03:00
|
|
|
authed, err := oauth.Authed(c, false, false, false, false)
|
|
|
|
if err != nil {
|
2023-02-02 16:08:13 +03:00
|
|
|
apiutil.ErrorHandler(c, gtserror.NewErrorNotFound(err), m.processor.InstanceGetV1)
|
2021-05-08 15:25:55 +03:00
|
|
|
return
|
|
|
|
}
|
|
|
|
|
|
|
|
// We use request params to check what to pull out of the database/storage so check everything. A request URL should be formatted as follows:
|
|
|
|
// "https://example.org/fileserver/[ACCOUNT_ID]/[MEDIA_TYPE]/[MEDIA_SIZE]/[FILE_NAME]"
|
|
|
|
// "FILE_NAME" consists of two parts, the attachment's database id, a period, and the file extension.
|
|
|
|
accountID := c.Param(AccountIDKey)
|
|
|
|
if accountID == "" {
|
2022-06-08 21:38:03 +03:00
|
|
|
err := fmt.Errorf("missing %s from request", AccountIDKey)
|
2023-02-02 16:08:13 +03:00
|
|
|
apiutil.ErrorHandler(c, gtserror.NewErrorNotFound(err), m.processor.InstanceGetV1)
|
2021-05-08 15:25:55 +03:00
|
|
|
return
|
|
|
|
}
|
|
|
|
|
|
|
|
mediaType := c.Param(MediaTypeKey)
|
|
|
|
if mediaType == "" {
|
2022-06-08 21:38:03 +03:00
|
|
|
err := fmt.Errorf("missing %s from request", MediaTypeKey)
|
2023-02-02 16:08:13 +03:00
|
|
|
apiutil.ErrorHandler(c, gtserror.NewErrorNotFound(err), m.processor.InstanceGetV1)
|
2021-05-08 15:25:55 +03:00
|
|
|
return
|
|
|
|
}
|
|
|
|
|
|
|
|
mediaSize := c.Param(MediaSizeKey)
|
|
|
|
if mediaSize == "" {
|
2022-06-08 21:38:03 +03:00
|
|
|
err := fmt.Errorf("missing %s from request", MediaSizeKey)
|
2023-02-02 16:08:13 +03:00
|
|
|
apiutil.ErrorHandler(c, gtserror.NewErrorNotFound(err), m.processor.InstanceGetV1)
|
2021-05-08 15:25:55 +03:00
|
|
|
return
|
|
|
|
}
|
|
|
|
|
|
|
|
fileName := c.Param(FileNameKey)
|
|
|
|
if fileName == "" {
|
2022-06-08 21:38:03 +03:00
|
|
|
err := fmt.Errorf("missing %s from request", FileNameKey)
|
2023-02-02 16:08:13 +03:00
|
|
|
apiutil.ErrorHandler(c, gtserror.NewErrorNotFound(err), m.processor.InstanceGetV1)
|
2021-05-08 15:25:55 +03:00
|
|
|
return
|
|
|
|
}
|
|
|
|
|
2023-02-17 14:02:29 +03:00
|
|
|
// Acquire context from gin request.
|
|
|
|
ctx := c.Request.Context()
|
|
|
|
|
2023-02-22 18:05:26 +03:00
|
|
|
content, errWithCode := m.processor.Media().GetFile(ctx, authed.Account, &apimodel.GetContentRequestForm{
|
2021-05-08 15:25:55 +03:00
|
|
|
AccountID: accountID,
|
|
|
|
MediaType: mediaType,
|
|
|
|
MediaSize: mediaSize,
|
|
|
|
FileName: fileName,
|
|
|
|
})
|
2022-03-07 13:08:26 +03:00
|
|
|
if errWithCode != nil {
|
2023-02-02 16:08:13 +03:00
|
|
|
apiutil.ErrorHandler(c, errWithCode, m.processor.InstanceGetV1)
|
2021-05-08 15:25:55 +03:00
|
|
|
return
|
|
|
|
}
|
|
|
|
|
2022-09-19 14:43:22 +03:00
|
|
|
if content.URL != nil {
|
2023-02-13 14:58:22 +03:00
|
|
|
// This is a non-local, non-proxied S3 file we're redirecting to.
|
2023-02-16 16:18:53 +03:00
|
|
|
// Derive the max-age value from how long the link has left until
|
|
|
|
// it expires.
|
|
|
|
maxAge := int(time.Until(content.URL.Expiry).Seconds())
|
|
|
|
c.Header("Cache-Control", "private,max-age="+strconv.Itoa(maxAge))
|
2022-09-19 14:43:22 +03:00
|
|
|
c.Redirect(http.StatusFound, content.URL.String())
|
|
|
|
return
|
|
|
|
}
|
|
|
|
|
2023-02-06 11:50:16 +03:00
|
|
|
defer func() {
|
|
|
|
// Close content when we're done, catch errors.
|
|
|
|
if err := content.Content.Close(); err != nil {
|
2023-02-17 14:02:29 +03:00
|
|
|
log.Errorf(ctx, "ServeFile: error closing readcloser: %s", err)
|
2023-02-06 11:50:16 +03:00
|
|
|
}
|
|
|
|
}()
|
|
|
|
|
2021-12-11 19:50:00 +03:00
|
|
|
// TODO: if the requester only accepts text/html we should try to serve them *something*.
|
2021-05-10 17:29:05 +03:00
|
|
|
// This is mostly needed because when sharing a link to a gts-hosted file on something like mastodon, the masto servers will
|
|
|
|
// attempt to look up the content to provide a preview of the link, and they ask for text/html.
|
2023-01-02 15:10:50 +03:00
|
|
|
format, err := apiutil.NegotiateAccept(c, apiutil.MIME(content.ContentType))
|
2022-06-08 21:38:03 +03:00
|
|
|
if err != nil {
|
2023-02-02 16:08:13 +03:00
|
|
|
apiutil.ErrorHandler(c, gtserror.NewErrorNotAcceptable(err, err.Error()), m.processor.InstanceGetV1)
|
2021-05-10 17:29:05 +03:00
|
|
|
return
|
|
|
|
}
|
|
|
|
|
2023-01-02 15:10:50 +03:00
|
|
|
// if this is a head request, just return info + throw the reader away
|
2022-07-30 15:42:47 +03:00
|
|
|
if c.Request.Method == http.MethodHead {
|
|
|
|
c.Header("Content-Type", format)
|
|
|
|
c.Header("Content-Length", strconv.FormatInt(content.ContentLength, 10))
|
|
|
|
c.Status(http.StatusOK)
|
|
|
|
return
|
|
|
|
}
|
|
|
|
|
2023-02-06 11:50:16 +03:00
|
|
|
// Look for a provided range header.
|
|
|
|
rng := c.GetHeader("Range")
|
|
|
|
if rng == "" {
|
|
|
|
// This is a simple query for the whole file, so do a read from whole reader.
|
|
|
|
c.DataFromReader(http.StatusOK, content.ContentLength, format, content.Content, nil)
|
2022-12-21 13:17:43 +03:00
|
|
|
return
|
|
|
|
}
|
|
|
|
|
2023-02-17 14:02:29 +03:00
|
|
|
// Set known content-type and serve range.
|
2023-02-06 11:50:16 +03:00
|
|
|
c.Header("Content-Type", format)
|
2023-02-17 14:02:29 +03:00
|
|
|
serveFileRange(
|
|
|
|
c.Writer,
|
|
|
|
c.Request,
|
|
|
|
content.Content,
|
|
|
|
rng,
|
|
|
|
content.ContentLength,
|
|
|
|
)
|
2023-02-06 11:50:16 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
// serveFileRange serves the range of a file from a given source reader, without the
|
|
|
|
// need for implementation of io.Seeker. Instead we read the first 'start' many bytes
|
|
|
|
// into a discard reader. Code is adapted from https://codeberg.org/gruf/simplehttp.
|
2023-02-17 14:02:29 +03:00
|
|
|
func serveFileRange(rw http.ResponseWriter, r *http.Request, src io.Reader, rng string, size int64) {
|
2023-02-06 11:50:16 +03:00
|
|
|
var i int
|
2023-01-16 18:19:17 +03:00
|
|
|
|
2023-02-06 11:50:16 +03:00
|
|
|
if i = strings.IndexByte(rng, '='); i < 0 {
|
|
|
|
// Range must include a separating '=' to indicate start
|
|
|
|
http.Error(rw, "Bad Range Header", http.StatusBadRequest)
|
2023-01-16 18:19:17 +03:00
|
|
|
return
|
|
|
|
}
|
|
|
|
|
2023-02-06 11:50:16 +03:00
|
|
|
if rng[:i] != "bytes" {
|
|
|
|
// We only support byte ranges in our implementation
|
|
|
|
http.Error(rw, "Unsupported Range Unit", http.StatusBadRequest)
|
2023-01-16 18:19:17 +03:00
|
|
|
return
|
|
|
|
}
|
2023-02-06 11:50:16 +03:00
|
|
|
|
|
|
|
// Reslice past '='
|
|
|
|
rng = rng[i+1:]
|
|
|
|
|
|
|
|
if i = strings.IndexByte(rng, '-'); i < 0 {
|
|
|
|
// Range header must contain a beginning and end separated by '-'
|
|
|
|
http.Error(rw, "Bad Range Header", http.StatusBadRequest)
|
|
|
|
return
|
|
|
|
}
|
|
|
|
|
|
|
|
var (
|
|
|
|
err error
|
|
|
|
|
|
|
|
// default start + end ranges
|
|
|
|
start, end = int64(0), size - 1
|
|
|
|
|
|
|
|
// start + end range strings
|
|
|
|
startRng, endRng string
|
|
|
|
)
|
|
|
|
|
|
|
|
if startRng = rng[:i]; len(startRng) > 0 {
|
|
|
|
// Parse the start of this byte range
|
|
|
|
start, err = strconv.ParseInt(startRng, 10, 64)
|
|
|
|
if err != nil {
|
|
|
|
http.Error(rw, "Bad Range Header", http.StatusBadRequest)
|
|
|
|
return
|
2023-01-16 18:19:17 +03:00
|
|
|
}
|
|
|
|
|
2023-02-06 11:50:16 +03:00
|
|
|
if start < 0 {
|
|
|
|
// This range starts *before* the file start, why did they send this lol
|
|
|
|
rw.Header().Set("Content-Range", "bytes *"+strconv.FormatInt(size, 10))
|
|
|
|
http.Error(rw, "Unsatisfiable Range", http.StatusRequestedRangeNotSatisfiable)
|
|
|
|
return
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
// No start supplied, implying file start
|
|
|
|
startRng = "0"
|
|
|
|
}
|
|
|
|
|
|
|
|
if endRng = rng[i+1:]; len(endRng) > 0 {
|
|
|
|
// Parse the end of this byte range
|
|
|
|
end, err = strconv.ParseInt(endRng, 10, 64)
|
|
|
|
if err != nil {
|
|
|
|
http.Error(rw, "Bad Range Header", http.StatusBadRequest)
|
|
|
|
return
|
|
|
|
}
|
2023-01-16 18:19:17 +03:00
|
|
|
|
2023-02-06 11:50:16 +03:00
|
|
|
if end > size {
|
|
|
|
// This range exceeds length of the file, therefore unsatisfiable
|
|
|
|
rw.Header().Set("Content-Range", "bytes *"+strconv.FormatInt(size, 10))
|
|
|
|
http.Error(rw, "Unsatisfiable Range", http.StatusRequestedRangeNotSatisfiable)
|
|
|
|
return
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
// No end supplied, implying file end
|
|
|
|
endRng = strconv.FormatInt(end, 10)
|
|
|
|
}
|
|
|
|
|
|
|
|
if start >= end {
|
|
|
|
// This range starts _after_ their range end, unsatisfiable and nonsense!
|
|
|
|
rw.Header().Set("Content-Range", "bytes *"+strconv.FormatInt(size, 10))
|
|
|
|
http.Error(rw, "Unsatisfiable Range", http.StatusRequestedRangeNotSatisfiable)
|
|
|
|
return
|
|
|
|
}
|
|
|
|
|
|
|
|
// Dump the first 'start' many bytes into the void...
|
|
|
|
if _, err := fastcopy.CopyN(io.Discard, src, start); err != nil {
|
2023-02-17 14:02:29 +03:00
|
|
|
log.Errorf(r.Context(), "error reading from source: %v", err)
|
2023-02-06 11:50:16 +03:00
|
|
|
return
|
|
|
|
}
|
|
|
|
|
2023-02-07 11:51:15 +03:00
|
|
|
// Determine new content length
|
|
|
|
// after slicing to given range.
|
|
|
|
length := end - start + 1
|
2023-02-06 11:50:16 +03:00
|
|
|
|
|
|
|
if end < size-1 {
|
|
|
|
// Range end < file end, limit the reader
|
|
|
|
src = io.LimitReader(src, length)
|
|
|
|
}
|
|
|
|
|
|
|
|
// Write the necessary length and range headers
|
|
|
|
rw.Header().Set("Content-Range", "bytes "+startRng+"-"+endRng+"/"+strconv.FormatInt(size, 10))
|
|
|
|
rw.Header().Set("Content-Length", strconv.FormatInt(length, 10))
|
|
|
|
rw.WriteHeader(http.StatusPartialContent)
|
|
|
|
|
|
|
|
// Read the "seeked" source reader into destination writer.
|
|
|
|
if _, err := fastcopy.Copy(rw, src); err != nil {
|
2023-02-17 14:02:29 +03:00
|
|
|
log.Errorf(r.Context(), "error reading from source: %v", err)
|
2023-02-06 11:50:16 +03:00
|
|
|
return
|
|
|
|
}
|
2021-05-08 15:25:55 +03:00
|
|
|
}
|