From e750b2aa3698d5bd7f4a9d23f568031e34c6ba2a Mon Sep 17 00:00:00 2001 From: Lars Thorsen Date: Thu, 16 Jan 2014 12:21:20 +0100 Subject: [xmerl] Fix problem with header of next XML document is in the buffer when using xmerl_sax_parser:stream/2 function. --- lib/xmerl/src/xmerl_sax_parser.erl | 32 +++++++++++++++++--------------- 1 file changed, 17 insertions(+), 15 deletions(-) (limited to 'lib/xmerl/src/xmerl_sax_parser.erl') diff --git a/lib/xmerl/src/xmerl_sax_parser.erl b/lib/xmerl/src/xmerl_sax_parser.erl index 5c006aada2..ad71072d95 100644 --- a/lib/xmerl/src/xmerl_sax_parser.erl +++ b/lib/xmerl/src/xmerl_sax_parser.erl @@ -74,7 +74,8 @@ file(Name,Options) -> CL = filename:absname(Dir), File = filename:basename(Name), ContinuationFun = fun default_continuation_cb/1, - Res = stream(<<>>, [{continuation_fun, ContinuationFun}, + Res = stream(<<>>, + [{continuation_fun, ContinuationFun}, {continuation_state, FD}, {current_location, CL}, {entity, File} @@ -98,9 +99,13 @@ stream(Xml, Options) when is_list(Xml), is_list(Options) -> State = parse_options(Options, initial_state()), case State#xmerl_sax_parser_state.file_type of dtd -> - xmerl_sax_parser_list:parse_dtd(Xml, State#xmerl_sax_parser_state{encoding = list}); + xmerl_sax_parser_list:parse_dtd(Xml, + State#xmerl_sax_parser_state{encoding = list, + input_type = stream}); normal -> - xmerl_sax_parser_list:parse(Xml, State#xmerl_sax_parser_state{encoding = list}) + xmerl_sax_parser_list:parse(Xml, + State#xmerl_sax_parser_state{encoding = list, + input_type = stream}) end; stream(Xml, Options) when is_binary(Xml), is_list(Options) -> case parse_options(Options, initial_state()) of @@ -124,17 +129,14 @@ stream(Xml, Options) when is_binary(Xml), is_list(Options) -> [], State#xmerl_sax_parser_state.event_state}; {Xml1, State1} -> - parse(Xml1, State1, ParseFunction) + parse_binary(Xml1, + State1#xmerl_sax_parser_state{input_type = stream}, + ParseFunction) end end. - -%%====================================================================== -%% Internal functions -%%====================================================================== - %%---------------------------------------------------------------------- -%% Function: parse(Encoding, Xml, State, F) -> Result +%% Function: parse_binary(Encoding, Xml, State, F) -> Result %% Input: Encoding = atom() %% Xml = [integer()] | binary() %% State = #xmerl_sax_parser_state @@ -144,15 +146,15 @@ stream(Xml, Options) when is_binary(Xml), is_list(Options) -> %% EventState = term() %% Description: Chooses the correct parser depending on the encoding. %%---------------------------------------------------------------------- -parse(Xml, #xmerl_sax_parser_state{encoding=utf8}=State, F) -> +parse_binary(Xml, #xmerl_sax_parser_state{encoding=utf8}=State, F) -> xmerl_sax_parser_utf8:F(Xml, State); -parse(Xml, #xmerl_sax_parser_state{encoding={utf16,little}}=State, F) -> +parse_binary(Xml, #xmerl_sax_parser_state{encoding={utf16,little}}=State, F) -> xmerl_sax_parser_utf16le:F(Xml, State); -parse(Xml, #xmerl_sax_parser_state{encoding={utf16,big}}=State, F) -> +parse_binary(Xml, #xmerl_sax_parser_state{encoding={utf16,big}}=State, F) -> xmerl_sax_parser_utf16be:F(Xml, State); -parse(Xml, #xmerl_sax_parser_state{encoding=latin1}=State, F) -> +parse_binary(Xml, #xmerl_sax_parser_state{encoding=latin1}=State, F) -> xmerl_sax_parser_latin1:F(Xml, State); -parse(_, #xmerl_sax_parser_state{encoding=Enc}, _) -> +parse_binary(_, #xmerl_sax_parser_state{encoding=Enc}, _) -> {error, lists:flatten(io_lib:format("Charcter set ~p not supported", [Enc]))}. %%---------------------------------------------------------------------- -- cgit v1.2.3