Fix Content-Disposition in media repository (#4176)

3 files changed, 110 insertions, 90 deletions

diff --git a/synapse/rest/media/v1/_base.py b/synapse/rest/media/v1/_base.py
index 76e479afa3..efe42a429d 100644
--- a/synapse/rest/media/v1/_base.py
+++ b/synapse/rest/media/v1/_base.py
@@ -16,6 +16,7 @@
 import logging
 import os
 
+from six import PY3
 from six.moves import urllib
 
 from twisted.internet import defer
@@ -48,26 +49,21 @@ def parse_media_id(request):
         return server_name, media_id, file_name
     except Exception:
         raise SynapseError(
-            404,
-            "Invalid media id token %r" % (request.postpath,),
-            Codes.UNKNOWN,
+            404, "Invalid media id token %r" % (request.postpath,), Codes.UNKNOWN
         )
 
 
 def respond_404(request):
     respond_with_json(
-        request, 404,
-        cs_error(
-            "Not found %r" % (request.postpath,),
-            code=Codes.NOT_FOUND,
-        ),
-        send_cors=True
+        request,
+        404,
+        cs_error("Not found %r" % (request.postpath,), code=Codes.NOT_FOUND),
+        send_cors=True,
     )
 
 
 @defer.inlineCallbacks
-def respond_with_file(request, media_type, file_path,
-                      file_size=None, upload_name=None):
+def respond_with_file(request, media_type, file_path, file_size=None, upload_name=None):
     logger.debug("Responding with %r", file_path)
 
     if os.path.isfile(file_path):
@@ -97,31 +93,26 @@ def add_file_headers(request, media_type, file_size, upload_name):
         file_size (int): Size in bytes of the media, if known.
         upload_name (str): The name of the requested file, if any.
     """
+
     def _quote(x):
         return urllib.parse.quote(x.encode("utf-8"))
 
     request.setHeader(b"Content-Type", media_type.encode("UTF-8"))
     if upload_name:
         if is_ascii(upload_name):
-            disposition = ("inline; filename=%s" % (_quote(upload_name),)).encode("ascii")
+            disposition = "inline; filename=%s" % (_quote(upload_name),)
         else:
-            disposition = (
-                "inline; filename*=utf-8''%s" % (_quote(upload_name),)).encode("ascii")
+            disposition = "inline; filename*=utf-8''%s" % (_quote(upload_name),)
 
-        request.setHeader(b"Content-Disposition", disposition)
+        request.setHeader(b"Content-Disposition", disposition.encode('ascii'))
 
     # cache for at least a day.
     # XXX: we might want to turn this off for data we don't want to
     # recommend caching as it's sensitive or private - or at least
     # select private. don't bother setting Expires as all our
     # clients are smart enough to be happy with Cache-Control
-    request.setHeader(
-        b"Cache-Control", b"public,max-age=86400,s-maxage=86400"
-    )
-
-    request.setHeader(
-        b"Content-Length", b"%d" % (file_size,)
-    )
+    request.setHeader(b"Cache-Control", b"public,max-age=86400,s-maxage=86400")
+    request.setHeader(b"Content-Length", b"%d" % (file_size,))
 
 
 @defer.inlineCallbacks
@@ -153,6 +144,7 @@ class Responder(object):
     Responder is a context manager which *must* be used, so that any resources
     held can be cleaned up.
     """
+
     def write_to_consumer(self, consumer):
         """Stream response into consumer
 
@@ -186,9 +178,18 @@ class FileInfo(object):
         thumbnail_method (str)
         thumbnail_type (str): Content type of thumbnail, e.g. image/png
     """
-    def __init__(self, server_name, file_id, url_cache=False,
-                 thumbnail=False, thumbnail_width=None, thumbnail_height=None,
-                 thumbnail_method=None, thumbnail_type=None):
+
+    def __init__(
+        self,
+        server_name,
+        file_id,
+        url_cache=False,
+        thumbnail=False,
+        thumbnail_width=None,
+        thumbnail_height=None,
+        thumbnail_method=None,
+        thumbnail_type=None,
+    ):
         self.server_name = server_name
         self.file_id = file_id
         self.url_cache = url_cache
@@ -197,3 +198,74 @@ class FileInfo(object):
         self.thumbnail_height = thumbnail_height
         self.thumbnail_method = thumbnail_method
         self.thumbnail_type = thumbnail_type
+
+
+def get_filename_from_headers(headers):
+    """
+    Get the filename of the downloaded file by inspecting the
+    Content-Disposition HTTP header.
+
+    Args:
+        headers (twisted.web.http_headers.Headers): The HTTP
+            request headers.
+
+    Returns:
+        A Unicode string of the filename, or None.
+    """
+    content_disposition = headers.get(b"Content-Disposition", [b''])
+
+    # No header, bail out.
+    if not content_disposition[0]:
+        return
+
+    # dict of unicode: bytes, corresponding to the key value sections of the
+    # Content-Disposition header.
+    params = {}
+    parts = content_disposition[0].split(b";")
+    for i in parts:
+        # Split into key-value pairs, if able
+        # We don't care about things like `inline`, so throw it out
+        if b"=" not in i:
+            continue
+
+        key, value = i.strip().split(b"=")
+        params[key.decode('ascii')] = value
+
+    upload_name = None
+
+    # First check if there is a valid UTF-8 filename
+    upload_name_utf8 = params.get("filename*", None)
+    if upload_name_utf8:
+        if upload_name_utf8.lower().startswith(b"utf-8''"):
+            upload_name_utf8 = upload_name_utf8[7:]
+            # We have a filename*= section. This MUST be ASCII, and any UTF-8
+            # bytes are %-quoted.
+            if PY3:
+                try:
+                    # Once it is decoded, we can then unquote the %-encoded
+                    # parts strictly into a unicode string.
+                    upload_name = urllib.parse.unquote(
+                        upload_name_utf8.decode('ascii'), errors="strict"
+                    )
+                except UnicodeDecodeError:
+                    # Incorrect UTF-8.
+                    pass
+            else:
+                # On Python 2, we first unquote the %-encoded parts and then
+                # decode it strictly using UTF-8.
+                try:
+                    upload_name = urllib.parse.unquote(upload_name_utf8).decode('utf8')
+                except UnicodeDecodeError:
+                    pass
+
+    # If there isn't check for an ascii name.
+    if not upload_name:
+        upload_name_ascii = params.get("filename", None)
+        if upload_name_ascii and is_ascii(upload_name_ascii):
+            # Make sure there's no %-quoted bytes. If there is, reject it as
+            # non-valid ASCII.
+            if b"%" not in upload_name_ascii:
+                upload_name = upload_name_ascii.decode('ascii')
+
+    # This may be None here, indicating we did not find a matching name.
+    return upload_name
diff --git a/synapse/rest/media/v1/media_repository.py b/synapse/rest/media/v1/media_repository.py
index d6c5f07af0..e117836e9a 100644
--- a/synapse/rest/media/v1/media_repository.py
+++ b/synapse/rest/media/v1/media_repository.py
@@ -14,14 +14,12 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import cgi
 import errno
 import logging
 import os
 import shutil
 
-from six import PY3, iteritems
-from six.moves.urllib import parse as urlparse
+from six import iteritems
 
 import twisted.internet.error
 import twisted.web.http
@@ -34,14 +32,18 @@ from synapse.api.errors import (
     NotFoundError,
     SynapseError,
 )
-from synapse.http.matrixfederationclient import MatrixFederationHttpClient
 from synapse.metrics.background_process_metrics import run_as_background_process
 from synapse.util import logcontext
 from synapse.util.async_helpers import Linearizer
 from synapse.util.retryutils import NotRetryingDestination
-from synapse.util.stringutils import is_ascii, random_string
+from synapse.util.stringutils import random_string
 
-from ._base import FileInfo, respond_404, respond_with_responder
+from ._base import (
+    FileInfo,
+    get_filename_from_headers,
+    respond_404,
+    respond_with_responder,
+)
 from .config_resource import MediaConfigResource
 from .download_resource import DownloadResource
 from .filepath import MediaFilePaths
@@ -62,7 +64,7 @@ class MediaRepository(object):
     def __init__(self, hs):
         self.hs = hs
         self.auth = hs.get_auth()
-        self.client = MatrixFederationHttpClient(hs)
+        self.client = hs.get_http_client()
         self.clock = hs.get_clock()
         self.server_name = hs.hostname
         self.store = hs.get_datastore()
@@ -397,39 +399,9 @@ class MediaRepository(object):
             yield finish()
 
         media_type = headers[b"Content-Type"][0].decode('ascii')
-
+        upload_name = get_filename_from_headers(headers)
         time_now_ms = self.clock.time_msec()
 
-        content_disposition = headers.get(b"Content-Disposition", None)
-        if content_disposition:
-            _, params = cgi.parse_header(content_disposition[0].decode('ascii'),)
-            upload_name = None
-
-            # First check if there is a valid UTF-8 filename
-            upload_name_utf8 = params.get("filename*", None)
-            if upload_name_utf8:
-                if upload_name_utf8.lower().startswith("utf-8''"):
-                    upload_name = upload_name_utf8[7:]
-
-            # If there isn't check for an ascii name.
-            if not upload_name:
-                upload_name_ascii = params.get("filename", None)
-                if upload_name_ascii and is_ascii(upload_name_ascii):
-                    upload_name = upload_name_ascii
-
-            if upload_name:
-                if PY3:
-                    upload_name = urlparse.unquote(upload_name)
-                else:
-                    upload_name = urlparse.unquote(upload_name.encode('ascii'))
-                try:
-                    if isinstance(upload_name, bytes):
-                        upload_name = upload_name.decode("utf-8")
-                except UnicodeDecodeError:
-                    upload_name = None
-        else:
-            upload_name = None
-
         logger.info("Stored remote media in file %r", fname)
 
         yield self.store.store_cached_remote_media(
diff --git a/synapse/rest/media/v1/preview_url_resource.py b/synapse/rest/media/v1/preview_url_resource.py
index 9b15699e4d..d0ecf241b6 100644
--- a/synapse/rest/media/v1/preview_url_resource.py
+++ b/synapse/rest/media/v1/preview_url_resource.py
@@ -13,7 +13,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import cgi
 import datetime
 import errno
 import fnmatch
@@ -44,10 +43,11 @@ from synapse.http.server import (
 )
 from synapse.http.servlet import parse_integer, parse_string
 from synapse.metrics.background_process_metrics import run_as_background_process
+from synapse.rest.media.v1._base import get_filename_from_headers
 from synapse.util.async_helpers import ObservableDeferred
 from synapse.util.caches.expiringcache import ExpiringCache
 from synapse.util.logcontext import make_deferred_yieldable, run_in_background
-from synapse.util.stringutils import is_ascii, random_string
+from synapse.util.stringutils import random_string
 
 from ._base import FileInfo
 
@@ -336,31 +336,7 @@ class PreviewUrlResource(Resource):
                 media_type = "application/octet-stream"
             time_now_ms = self.clock.time_msec()
 
-            content_disposition = headers.get(b"Content-Disposition", None)
-            if content_disposition:
-                _, params = cgi.parse_header(content_disposition[0],)
-                download_name = None
-
-                # First check if there is a valid UTF-8 filename
-                download_name_utf8 = params.get("filename*", None)
-                if download_name_utf8:
-                    if download_name_utf8.lower().startswith("utf-8''"):
-                        download_name = download_name_utf8[7:]
-
-                # If there isn't check for an ascii name.
-                if not download_name:
-                    download_name_ascii = params.get("filename", None)
-                    if download_name_ascii and is_ascii(download_name_ascii):
-                        download_name = download_name_ascii
-
-                if download_name:
-                    download_name = urlparse.unquote(download_name)
-                    try:
-                        download_name = download_name.decode("utf-8")
-                    except UnicodeDecodeError:
-                        download_name = None
-            else:
-                download_name = None
+            download_name = get_filename_from_headers(headers)
 
             yield self.store.store_local_media(
                 media_id=file_id,