Better total (body + attachments) size checking for documents

Use the already computed (conservative) body size.

Switch multipart length calcuation to accept body and boundary sizes.

Issue apache#1200
Issue apache#1253

src/couch/src/couch_doc.erl

@@ -133,10 +133,11 @@ from_json_obj_validate(EJson) ->
 from_json_obj_validate(EJson, DbName) ->
  MaxSize = config:get_integer("couchdb", "max_document_size", 4294967296),
  Doc = from_json_obj(EJson, DbName),
- case couch_ejson_size:encoded_size(Doc#doc.body) =< MaxSize of
+ BodySize = couch_ejson_size:encoded_size(Doc#doc.body),
+ case BodySize =< MaxSize of
  true ->
  validate_attachment_sizes(Doc#doc.atts),
- validate_total_document_size(Doc),
+ validate_total_document_size(Doc, BodySize),
  Doc;
  false ->
  throw({request_entity_too_large, Doc#doc.id})
@@ -145,12 +146,11 @@ from_json_obj_validate(EJson, DbName) ->
 
 % sum up the json body size + attachment body size and
 % make sure it is < max_http_request_size
-validate_total_document_size(#doc{id=DocId, body=Body, atts=Atts0}=Doc) ->
+validate_total_document_size(#doc{id=DocId, atts=Atts0}=Doc, BodySize) ->
  MaxReqSize = config:get_integer("httpd", "max_http_request_size", 4294967296), % 4 GB
- Boundary = <<"00000000000000000000000000000000">>,
  Atts = lists:map(fun couch_att:to_tuple/1, Atts0),
- {_, DocSum} = couch_httpd_multipart:length_multipart_stream(Boundary,
- ?JSON_ENCODE(Body), Atts),
+ {_, DocSum} = couch_httpd_multipart:length_multipart_stream(32, BodySize,
+ Atts),
  case DocSum =< MaxReqSize of
  true -> ok;
  false -> throw({request_entity_too_large, DocId})
@@ -436,7 +436,13 @@ merge_stubs(#doc{id=Id,atts=MemBins}=StubsDoc, #doc{atts=DiskBins}) ->
 len_doc_to_multi_part_stream(Boundary, JsonBytes, Atts, SendEncodedAtts) ->
  AttsToInclude = lists:filter(fun(Att) -> not couch_att:is_stub(Att) end, Atts),
  AttsDecoded = decode_attributes(AttsToInclude, SendEncodedAtts),
- couch_httpd_multipart:length_multipart_stream(Boundary, JsonBytes, AttsDecoded).
+ case couch_httpd_multipart:length_multipart_stream(byte_size(Boundary),
+ iolist_size(JsonBytes), AttsDecoded) of
+ {json, Len} ->
+ {<<"application/json">>, Len};
+ {multipart, Len} ->
+ {<<"multipart/related; boundary=\"", Boundary/binary, "\"">>, Len}
+ end.
 
 
 doc_to_multi_part_stream(Boundary, JsonBytes, Atts, WriteFun,

src/couch/src/couch_httpd_multipart.erl

@@ -262,14 +262,15 @@ atts_to_mp([{Att, Name, Len, Type, Encoding} | RestAtts], Boundary, WriteFun,
  WriteFun(<<"\r\n--", Boundary/binary>>),
  atts_to_mp(RestAtts, Boundary, WriteFun, AttFun).
 
-length_multipart_stream(Boundary, JsonBytes, Atts) ->
+length_multipart_stream(BoundarySize, JsonByteSize, Atts) when
+ is_integer(BoundarySize), is_integer(JsonByteSize) ->
  AttsSize = lists:foldl(fun({_Att, Name, Len, Type, Encoding}, AccAttsSize) ->
  AccAttsSize +
  4 + % "\r\n\r\n"
  length(integer_to_list(Len)) +
  Len +
  4 + % "\r\n--"
- size(Boundary) +
+ BoundarySize +
  % attachment headers
  % (the length of the Content-Length has already been set)
  size(Name) +
@@ -286,15 +287,15 @@ length_multipart_stream(Boundary, JsonBytes, Atts) ->
  end
  end, 0, Atts),
  if AttsSize == 0 ->
- {<<"application/json">>, iolist_size(JsonBytes)};
+ {json, JsonByteSize};
  true ->
- {<<"multipart/related; boundary=\"", Boundary/binary, "\"">>,
+ {multipart,
  2 + % "--"
- size(Boundary) +
+ BoundarySize +
  36 + % "\r\ncontent-type: application/json\r\n\r\n"
- iolist_size(JsonBytes) +
+ JsonByteSize +
  4 + % "\r\n--"
- size(Boundary) +
+ BoundarySize +
  + AttsSize +
  2 % "--"
  }