aboutsummaryrefslogtreecommitdiffstats
path: root/lib/xmerl/src
diff options
context:
space:
mode:
authorLars Thorsen <[email protected]>2017-03-07 11:43:44 +0100
committerLars Thorsen <[email protected]>2017-03-07 11:43:44 +0100
commit1982fbc8a8b25709b7c7b6831e786d7a9ce361f2 (patch)
tree406c3de5d0730c29ce61ca4ace783be61b44fbcc /lib/xmerl/src
parent3300b3059d13cd204f553f0c7d46d784adc5a8ab (diff)
parent266b7f805250acfc91d5242e5a73bd8dc51f8d47 (diff)
downloadotp-1982fbc8a8b25709b7c7b6831e786d7a9ce361f2.tar.gz
otp-1982fbc8a8b25709b7c7b6831e786d7a9ce361f2.tar.bz2
otp-1982fbc8a8b25709b7c7b6831e786d7a9ce361f2.zip
Merge branch 'maint'
Diffstat (limited to 'lib/xmerl/src')
-rw-r--r--lib/xmerl/src/xmerl_sax_parser_base.erlsrc60
-rw-r--r--lib/xmerl/src/xmerl_sax_parser_latin1.erlsrc38
-rw-r--r--lib/xmerl/src/xmerl_sax_parser_list.erlsrc19
-rw-r--r--lib/xmerl/src/xmerl_sax_parser_utf16be.erlsrc50
-rw-r--r--lib/xmerl/src/xmerl_sax_parser_utf16le.erlsrc50
-rw-r--r--lib/xmerl/src/xmerl_sax_parser_utf8.erlsrc50
6 files changed, 199 insertions, 68 deletions
diff --git a/lib/xmerl/src/xmerl_sax_parser_base.erlsrc b/lib/xmerl/src/xmerl_sax_parser_base.erlsrc
index 2b9b37b5f3..f3470b2809 100644
--- a/lib/xmerl/src/xmerl_sax_parser_base.erlsrc
+++ b/lib/xmerl/src/xmerl_sax_parser_base.erlsrc
@@ -145,10 +145,11 @@ parse_dtd(Xml, State) ->
%% [1] document ::= prolog element Misc*
%%----------------------------------------------------------------------
parse_document(Rest, State) when is_record(State, xmerl_sax_parser_state) ->
- {Rest1, State1} = parse_xml_decl(Rest, State),
+ {Rest1, State1} = parse_byte_order_mark(Rest, State),
{Rest2, State2} = parse_misc(Rest1, State1, true),
{ok, Rest2, State2}.
+?PARSE_BYTE_ORDER_MARK(Bytes, State).
%%----------------------------------------------------------------------
%% Function: parse_xml_decl(Rest, State) -> Result
@@ -159,15 +160,8 @@ parse_document(Rest, State) when is_record(State, xmerl_sax_parser_state) ->
%% [22] prolog ::= XMLDecl? Misc* (doctypedecl Misc*)?
%% [23] XMLDecl ::= '<?xml' VersionInfo EncodingDecl? SDDecl? S? '?>'
%%----------------------------------------------------------------------
--dialyzer({[no_fail_call, no_match], parse_xml_decl/2}).
parse_xml_decl(?STRING_EMPTY, State) ->
cf(?STRING_EMPTY, State, fun parse_xml_decl/2);
-parse_xml_decl(?BYTE_ORDER_MARK_1, State) ->
- cf(?BYTE_ORDER_MARK_1, State, fun parse_xml_decl/2);
-parse_xml_decl(?BYTE_ORDER_MARK_2, State) ->
- cf(?BYTE_ORDER_MARK_2, State, fun parse_xml_decl/2);
-parse_xml_decl(?BYTE_ORDER_MARK_REST(Rest), State) ->
- cf(Rest, State, fun parse_xml_decl/2);
parse_xml_decl(?STRING("<") = Bytes, State) ->
cf(Bytes, State, fun parse_xml_decl/2);
parse_xml_decl(?STRING("<?") = Bytes, State) ->
@@ -179,31 +173,19 @@ parse_xml_decl(?STRING("<?xm") = Bytes, State) ->
parse_xml_decl(?STRING("<?xml") = Bytes, State) ->
cf(Bytes, State, fun parse_xml_decl/2);
parse_xml_decl(?STRING_REST("<?xml", Rest1), State) ->
- parse_xml_decl_1(Rest1, State);
-parse_xml_decl(Bytes, #xmerl_sax_parser_state{encoding=Enc} = State) when is_binary(Bytes) ->
- case unicode:characters_to_list(Bytes, Enc) of
- {incomplete, _, _} ->
- cf(Bytes, State, fun parse_xml_decl/2);
- {error, _Encoded, _Rest} ->
- ?fatal_error(State, lists:flatten(io_lib:format("Bad character, not in ~p\n", [Enc])));
- _ ->
- parse_prolog(Bytes, State)
- end;
-parse_xml_decl(Bytes, State) ->
- parse_prolog(Bytes, State).
-
+ parse_xml_decl_rest(Rest1, State);
+?PARSE_XML_DECL(Bytes, State).
-parse_xml_decl_1(?STRING_UNBOUND_REST(C, Rest) = Bytes, State) ->
+parse_xml_decl_rest(?STRING_UNBOUND_REST(C, Rest) = Bytes, State) ->
if
?is_whitespace(C) ->
{_XmlAttributes, Rest1, State1} = parse_version_info(Rest, State, []),
- %State2 = event_callback({processingInstruction, "xml", XmlAttributes}, State1),% The XML decl. should not be reported as a PI
parse_prolog(Rest1, State1);
true ->
parse_prolog(?STRING_REST("<?xml", Bytes), State)
end;
-parse_xml_decl_1(Bytes, State) ->
- unicode_incomplete_check([Bytes, State, fun parse_xml_decl_1/2], undefined).
+parse_xml_decl_rest(Bytes, State) ->
+ unicode_incomplete_check([Bytes, State, fun parse_xml_decl_rest/2], undefined).
@@ -225,8 +207,6 @@ parse_prolog(?STRING_REST("<?", Rest), State) ->
parse_prolog(Rest1, State1);
{endDocument, Rest1, State1} ->
parse_prolog(Rest1, State1)
- % IValue = ?TO_INPUT_FORMAT("<?"),
- % {?APPEND_STRING(IValue, Rest1), State1}
end;
parse_prolog(?STRING_REST("<!", Rest), State) ->
parse_prolog_1(Rest, State);
@@ -239,7 +219,6 @@ parse_prolog(Bytes, State) ->
unicode_incomplete_check([Bytes, State, fun parse_prolog/2],
"expecting < or whitespace").
-
parse_prolog_1(?STRING_EMPTY, State) ->
cf(?STRING_EMPTY, State, fun parse_prolog_1/2);
parse_prolog_1(?STRING("D") = Bytes, State) ->
@@ -1232,7 +1211,6 @@ send_character_event(_, true, String, State) ->
%% Description: Parse whitespaces.
%% [3] S ::= (#x20 | #x9 | #xD | #xA)+
%%----------------------------------------------------------------------
--dialyzer({no_fail_call, whitespace/3}).
whitespace(?STRING_EMPTY, State, Acc) ->
case cf(?STRING_EMPTY, State, Acc, fun whitespace/3) of
{?STRING_EMPTY, State} ->
@@ -1258,16 +1236,7 @@ whitespace(?STRING_REST("\r", Rest), State, Acc) ->
whitespace(Rest, State#xmerl_sax_parser_state{line_no=N+1}, [?lf |Acc]);
whitespace(?STRING_UNBOUND_REST(C, Rest), State, Acc) when ?is_whitespace(C) ->
whitespace(Rest, State, [C|Acc]);
-whitespace(?STRING_UNBOUND_REST(_C, _) = Bytes, State, Acc) ->
- {lists:reverse(Acc), Bytes, State};
-whitespace(Bytes, #xmerl_sax_parser_state{encoding=Enc} = State, Acc) when is_binary(Bytes) ->
- case unicode:characters_to_list(Bytes, Enc) of
- {incomplete, _, _} ->
- cf(Bytes, State, Acc, fun whitespace/3);
- {error, _Encoded, _Rest} ->
- ?fatal_error(State, lists:flatten(io_lib:format("Bad character, not in ~p\n", [Enc])))
- end.
-
+?WHITESPACE(Bytes, State, Acc).
%%----------------------------------------------------------------------
%% Function: parse_reference(Rest, State, HaveToExist) -> Result
@@ -1390,7 +1359,6 @@ parse_pe_reference_1(Bytes, State, Name) ->
"missing ; after reference " ++ Name).
-
%%----------------------------------------------------------------------
%% Function: insert_reference(Reference, State) -> Result
%% Parameters: Reference = string()
@@ -1406,7 +1374,6 @@ insert_reference({Name, Type, Value}, Table) ->
end.
-
%%----------------------------------------------------------------------
%% Function: look_up_reference(Reference, State) -> Result
%% Parameters: Reference = string()
@@ -1721,7 +1688,7 @@ handle_external_entity({http, Url}, State) ->
++ file:format_error(Reason));
{ok, FD} ->
{?STRING_EMPTY, EntityState} =
- parse_external_entity_1(<<>>,
+ parse_external_entity_byte_order_mark(<<>>,
State#xmerl_sax_parser_state{continuation_state=FD,
current_location=filename:dirname(Url),
entity=filename:basename(Url),
@@ -1737,6 +1704,8 @@ handle_external_entity({http, Url}, State) ->
handle_external_entity({Tag, _Url}, State) ->
?fatal_error(State, "Unsupported URI type: " ++ atom_to_list(Tag)).
+?PARSE_EXTERNAL_ENTITY_BYTE_ORDER_MARK(Bytes, State).
+
%%----------------------------------------------------------------------
%% Function : parse_external_entity_1(Rest, State) -> Result
%% Parameters: Rest = string() | binary()
@@ -1744,7 +1713,6 @@ handle_external_entity({Tag, _Url}, State) ->
%% Result : {Rest, State}
%% Description: Parse the external entity.
%%----------------------------------------------------------------------
--dialyzer({[no_fail_call, no_match], parse_external_entity_1/2}).
parse_external_entity_1(?STRING_EMPTY, #xmerl_sax_parser_state{file_type=Type} = State) ->
case catch cf(?STRING_EMPTY, State, fun parse_external_entity_1/2) of
{Rest, State1} when is_record(State1, xmerl_sax_parser_state) ->
@@ -1754,12 +1722,6 @@ parse_external_entity_1(?STRING_EMPTY, #xmerl_sax_parser_state{file_type=Type} =
Other ->
throw(Other)
end;
-parse_external_entity_1(?BYTE_ORDER_MARK_1, State) ->
- cf(?BYTE_ORDER_MARK_1, State, fun parse_external_entity_1/2);
-parse_external_entity_1(?BYTE_ORDER_MARK_2, State) ->
- cf(?BYTE_ORDER_MARK_2, State, fun parse_external_entity_1/2);
-parse_external_entity_1(?BYTE_ORDER_MARK_REST(Rest), State) ->
- parse_external_entity_1(Rest, State);
parse_external_entity_1(?STRING("<") = Bytes, State) ->
cf(Bytes, State, fun parse_external_entity_1/2);
parse_external_entity_1(?STRING("<?") = Bytes, State) ->
diff --git a/lib/xmerl/src/xmerl_sax_parser_latin1.erlsrc b/lib/xmerl/src/xmerl_sax_parser_latin1.erlsrc
index 961806bf4c..6e59347fb8 100644
--- a/lib/xmerl/src/xmerl_sax_parser_latin1.erlsrc
+++ b/lib/xmerl/src/xmerl_sax_parser_latin1.erlsrc
@@ -2,7 +2,7 @@
%%--------------------------------------------------------------------
%% %CopyrightBegin%
%%
-%% Copyright Ericsson AB 2008-2016. All Rights Reserved.
+%% Copyright Ericsson AB 2008-2017. All Rights Reserved.
%%
%% Licensed under the Apache License, Version 2.0 (the "License");
%% you may not use this file except in compliance with the License.
@@ -34,8 +34,36 @@
-define(APPEND_STRING(Rest, New), <<Rest/binary, New/binary>>).
-define(TO_INPUT_FORMAT(Val), unicode:characters_to_binary(Val, unicode, latin1)).
-%% STRING_REST and STRING_UNBOUND_REST is only different in the list case
-define(STRING_UNBOUND_REST(MatchChar, Rest), <<MatchChar, Rest/binary>>).
--define(BYTE_ORDER_MARK_1, undefined_bom1).
--define(BYTE_ORDER_MARK_2, undefined_bom2).
--define(BYTE_ORDER_MARK_REST(Rest), <<undefined, Rest/binary>>).
+
+-define(PARSE_BYTE_ORDER_MARK(Bytes, State),
+ parse_byte_order_mark(Bytes, State) ->
+ parse_xml_decl(Bytes, State)).
+
+-define(PARSE_XML_DECL(Bytes, State),
+ parse_xml_decl(Bytes, #xmerl_sax_parser_state{encoding=Enc} = State) when is_binary(Bytes) ->
+ case unicode:characters_to_list(Bytes, Enc) of
+ {incomplete, _, _} ->
+ cf(Bytes, State, fun parse_xml_decl/2);
+ {error, _Encoded, _Rest} ->
+ ?fatal_error(State, lists:flatten(io_lib:format("Bad character, not in ~p\n", [Enc])));
+ _ ->
+ parse_prolog(Bytes, State)
+ end;
+ parse_xml_decl(Bytes, State) ->
+ parse_prolog(Bytes, State)).
+
+-define(WHITESPACE(Bytes, State, Acc),
+ whitespace(?STRING_UNBOUND_REST(_C, _) = Bytes, State, Acc) ->
+ {lists:reverse(Acc), Bytes, State};
+ whitespace(Bytes, #xmerl_sax_parser_state{encoding=Enc} = State, Acc) when is_binary(Bytes) ->
+ case unicode:characters_to_list(Bytes, Enc) of
+ {incomplete, _, _} ->
+ cf(Bytes, State, Acc, fun whitespace/3);
+ {error, _Encoded, _Rest} ->
+ ?fatal_error(State, lists:flatten(io_lib:format("Bad character, not in ~p\n", [Enc])))
+ end).
+
+-define(PARSE_EXTERNAL_ENTITY_BYTE_ORDER_MARK(Bytes, State),
+ parse_external_entity_byte_order_mark(Bytes, State) ->
+ parse_external_entity_1(Bytes, State)).
diff --git a/lib/xmerl/src/xmerl_sax_parser_list.erlsrc b/lib/xmerl/src/xmerl_sax_parser_list.erlsrc
index 624a621d92..6a4435b1d9 100644
--- a/lib/xmerl/src/xmerl_sax_parser_list.erlsrc
+++ b/lib/xmerl/src/xmerl_sax_parser_list.erlsrc
@@ -36,6 +36,19 @@
%% In the list case we can't use a '++' when matchin against an unbound variable
-define(STRING_UNBOUND_REST(MatchChar, Rest), [MatchChar | Rest]).
--define(BYTE_ORDER_MARK_1, undefined_bom1).
--define(BYTE_ORDER_MARK_2, undefined_bom2).
--define(BYTE_ORDER_MARK_REST(Rest), [undefined|Rest]).
+
+-define(PARSE_BYTE_ORDER_MARK(Bytes, State),
+ parse_byte_order_mark(Bytes, State) ->
+ parse_xml_decl(Bytes, State)).
+
+-define(PARSE_XML_DECL(Bytes, State),
+ parse_xml_decl(Bytes, State) ->
+ parse_prolog(Bytes, State)).
+
+-define(WHITESPACE(Bytes, State, Acc),
+ whitespace(?STRING_UNBOUND_REST(_C, _) = Bytes, State, Acc) ->
+ {lists:reverse(Acc), Bytes, State}).
+
+-define(PARSE_EXTERNAL_ENTITY_BYTE_ORDER_MARK(Bytes, State),
+ parse_external_entity_byte_order_mark(Bytes, State) ->
+ parse_external_entity_1(Bytes, State)).
diff --git a/lib/xmerl/src/xmerl_sax_parser_utf16be.erlsrc b/lib/xmerl/src/xmerl_sax_parser_utf16be.erlsrc
index ff84ece97a..ec89024729 100644
--- a/lib/xmerl/src/xmerl_sax_parser_utf16be.erlsrc
+++ b/lib/xmerl/src/xmerl_sax_parser_utf16be.erlsrc
@@ -2,7 +2,7 @@
%%--------------------------------------------------------------------
%% %CopyrightBegin%
%%
-%% Copyright Ericsson AB 2008-2016. All Rights Reserved.
+%% Copyright Ericsson AB 2008-2017. All Rights Reserved.
%%
%% Licensed under the Apache License, Version 2.0 (the "License");
%% you may not use this file except in compliance with the License.
@@ -34,8 +34,50 @@
-define(APPEND_STRING(Rest, New), <<Rest/binary, New/binary>>).
-define(TO_INPUT_FORMAT(Val), unicode:characters_to_binary(Val, unicode, {utf16, big})).
-%% STRING_REST and STRING_UNBOUND_REST is only different in the list case
-define(STRING_UNBOUND_REST(MatchChar, Rest), <<MatchChar/big-utf16, Rest/binary>>).
--define(BYTE_ORDER_MARK_1, undefined_bom1).
--define(BYTE_ORDER_MARK_2, <<16#FE>>).
+-define(BYTE_ORDER_MARK_1, <<16#FE>>).
-define(BYTE_ORDER_MARK_REST(Rest), <<16#FE, 16#FF, Rest/binary>>).
+
+-define(PARSE_BYTE_ORDER_MARK(Bytes, State),
+ parse_byte_order_mark(?STRING_EMPTY, State) ->
+ cf(?STRING_EMPTY, State, fun parse_byte_order_mark/2);
+ parse_byte_order_mark(?BYTE_ORDER_MARK_1, State) ->
+ cf(?BYTE_ORDER_MARK_1, State, fun parse_byte_order_mark/2);
+ parse_byte_order_mark(?BYTE_ORDER_MARK_REST(Rest), State) ->
+ parse_xml_decl(Rest, State);
+ parse_byte_order_mark(Bytes, State) ->
+ parse_xml_decl(Bytes, State)).
+
+-define(PARSE_XML_DECL(Bytes, State),
+ parse_xml_decl(Bytes, #xmerl_sax_parser_state{encoding=Enc} = State) when is_binary(Bytes) ->
+ case unicode:characters_to_list(Bytes, Enc) of
+ {incomplete, _, _} ->
+ cf(Bytes, State, fun parse_xml_decl/2);
+ {error, _Encoded, _Rest} ->
+ ?fatal_error(State, lists:flatten(io_lib:format("Bad character, not in ~p\n", [Enc])));
+ _ ->
+ parse_prolog(Bytes, State)
+ end;
+ parse_xml_decl(Bytes, State) ->
+ parse_prolog(Bytes, State)).
+
+-define(WHITESPACE(Bytes, State, Acc),
+ whitespace(?STRING_UNBOUND_REST(_C, _) = Bytes, State, Acc) ->
+ {lists:reverse(Acc), Bytes, State};
+ whitespace(Bytes, #xmerl_sax_parser_state{encoding=Enc} = State, Acc) when is_binary(Bytes) ->
+ case unicode:characters_to_list(Bytes, Enc) of
+ {incomplete, _, _} ->
+ cf(Bytes, State, Acc, fun whitespace/3);
+ {error, _Encoded, _Rest} ->
+ ?fatal_error(State, lists:flatten(io_lib:format("Bad character, not in ~p\n", [Enc])))
+ end).
+
+-define(PARSE_EXTERNAL_ENTITY_BYTE_ORDER_MARK(Bytes, State),
+ parse_external_entity_byte_order_mark(?STRING_EMPTY, State) ->
+ cf(?STRING_EMPTY, State, fun parse_external_entity_byte_order_mark/2);
+ parse_external_entity_byte_order_mark(?BYTE_ORDER_MARK_1, State) ->
+ cf(?BYTE_ORDER_MARK_1, State, fun parse_external_entity_byte_order_mark/2);
+ parse_external_entity_byte_order_mark(?BYTE_ORDER_MARK_REST(Rest), State) ->
+ parse_external_entity_1(Rest, State);
+ parse_external_entity_byte_order_mark(Bytes, State) ->
+ parse_external_entity_1(Bytes, State)).
diff --git a/lib/xmerl/src/xmerl_sax_parser_utf16le.erlsrc b/lib/xmerl/src/xmerl_sax_parser_utf16le.erlsrc
index a330fce8d0..566333a045 100644
--- a/lib/xmerl/src/xmerl_sax_parser_utf16le.erlsrc
+++ b/lib/xmerl/src/xmerl_sax_parser_utf16le.erlsrc
@@ -2,7 +2,7 @@
%%--------------------------------------------------------------------
%% %CopyrightBegin%
%%
-%% Copyright Ericsson AB 2008-2016. All Rights Reserved.
+%% Copyright Ericsson AB 2008-2017. All Rights Reserved.
%%
%% Licensed under the Apache License, Version 2.0 (the "License");
%% you may not use this file except in compliance with the License.
@@ -34,8 +34,50 @@
-define(APPEND_STRING(Rest, New), <<Rest/binary, New/binary>>).
-define(TO_INPUT_FORMAT(Val), unicode:characters_to_binary(Val, unicode, {utf16, little})).
-%% STRING_REST and STRING_UNBOUND_REST is only different in the list case
-define(STRING_UNBOUND_REST(MatchChar, Rest), <<MatchChar/little-utf16, Rest/binary>>).
--define(BYTE_ORDER_MARK_1, undefined_bom1).
--define(BYTE_ORDER_MARK_2, <<16#FF>>).
+-define(BYTE_ORDER_MARK_1, <<16#FF>>).
-define(BYTE_ORDER_MARK_REST(Rest), <<16#FF, 16#FE, Rest/binary>>).
+
+-define(PARSE_BYTE_ORDER_MARK(Bytes, State),
+ parse_byte_order_mark(?STRING_EMPTY, State) ->
+ cf(?STRING_EMPTY, State, fun parse_byte_order_mark/2);
+ parse_byte_order_mark(?BYTE_ORDER_MARK_1, State) ->
+ cf(?BYTE_ORDER_MARK_1, State, fun parse_byte_order_mark/2);
+ parse_byte_order_mark(?BYTE_ORDER_MARK_REST(Rest), State) ->
+ parse_xml_decl(Rest, State);
+ parse_byte_order_mark(Bytes, State) ->
+ parse_xml_decl(Bytes, State)).
+
+-define(PARSE_XML_DECL(Bytes, State),
+ parse_xml_decl(Bytes, #xmerl_sax_parser_state{encoding=Enc} = State) when is_binary(Bytes) ->
+ case unicode:characters_to_list(Bytes, Enc) of
+ {incomplete, _, _} ->
+ cf(Bytes, State, fun parse_xml_decl/2);
+ {error, _Encoded, _Rest} ->
+ ?fatal_error(State, lists:flatten(io_lib:format("Bad character, not in ~p\n", [Enc])));
+ _ ->
+ parse_prolog(Bytes, State)
+ end;
+ parse_xml_decl(Bytes, State) ->
+ parse_prolog(Bytes, State)).
+
+-define(WHITESPACE(Bytes, State, Acc),
+ whitespace(?STRING_UNBOUND_REST(_C, _) = Bytes, State, Acc) ->
+ {lists:reverse(Acc), Bytes, State};
+ whitespace(Bytes, #xmerl_sax_parser_state{encoding=Enc} = State, Acc) when is_binary(Bytes) ->
+ case unicode:characters_to_list(Bytes, Enc) of
+ {incomplete, _, _} ->
+ cf(Bytes, State, Acc, fun whitespace/3);
+ {error, _Encoded, _Rest} ->
+ ?fatal_error(State, lists:flatten(io_lib:format("Bad character, not in ~p\n", [Enc])))
+ end).
+
+-define(PARSE_EXTERNAL_ENTITY_BYTE_ORDER_MARK(Bytes, State),
+ parse_external_entity_byte_order_mark(?STRING_EMPTY, State) ->
+ cf(?STRING_EMPTY, State, fun parse_external_entity_byte_order_mark/2);
+ parse_external_entity_byte_order_mark(?BYTE_ORDER_MARK_1, State) ->
+ cf(?BYTE_ORDER_MARK_1, State, fun parse_external_entity_byte_order_mark/2);
+ parse_external_entity_byte_order_mark(?BYTE_ORDER_MARK_REST(Rest), State) ->
+ parse_external_entity_1(Rest, State);
+ parse_external_entity_byte_order_mark(Bytes, State) ->
+ parse_external_entity_1(Bytes, State)).
diff --git a/lib/xmerl/src/xmerl_sax_parser_utf8.erlsrc b/lib/xmerl/src/xmerl_sax_parser_utf8.erlsrc
index d46d60d237..f41d06d013 100644
--- a/lib/xmerl/src/xmerl_sax_parser_utf8.erlsrc
+++ b/lib/xmerl/src/xmerl_sax_parser_utf8.erlsrc
@@ -2,7 +2,7 @@
%%--------------------------------------------------------------------
%% %CopyrightBegin%
%%
-%% Copyright Ericsson AB 2008-2016. All Rights Reserved.
+%% Copyright Ericsson AB 2008-2017. All Rights Reserved.
%%
%% Licensed under the Apache License, Version 2.0 (the "License");
%% you may not use this file except in compliance with the License.
@@ -34,11 +34,55 @@
-define(APPEND_STRING(Rest, New), <<Rest/binary, New/binary>>).
-define(TO_INPUT_FORMAT(Val), unicode:characters_to_binary(Val, unicode, utf8)).
-
-%% STRING_REST and STRING_UNBOUND_REST is only different in the list case
-define(STRING_UNBOUND_REST(MatchChar, Rest), <<MatchChar/utf8, Rest/binary>>).
-define(BYTE_ORDER_MARK_1, <<16#EF>>).
-define(BYTE_ORDER_MARK_2, <<16#EF, 16#BB>>).
-define(BYTE_ORDER_MARK_REST(Rest), <<16#EF, 16#BB, 16#BF, Rest/binary>>).
+-define(PARSE_BYTE_ORDER_MARK(Bytes, State),
+ parse_byte_order_mark(?STRING_EMPTY, State) ->
+ cf(?STRING_EMPTY, State, fun parse_byte_order_mark/2);
+ parse_byte_order_mark(?BYTE_ORDER_MARK_1, State) ->
+ cf(?BYTE_ORDER_MARK_1, State, fun parse_byte_order_mark/2);
+ parse_byte_order_mark(?BYTE_ORDER_MARK_2, State) ->
+ cf(?BYTE_ORDER_MARK_2, State, fun parse_byte_order_mark/2);
+ parse_byte_order_mark(?BYTE_ORDER_MARK_REST(Rest), State) ->
+ parse_xml_decl(Rest, State);
+ parse_byte_order_mark(Bytes, State) ->
+ parse_xml_decl(Bytes, State)).
+
+-define(PARSE_XML_DECL(Bytes, State),
+ parse_xml_decl(Bytes, #xmerl_sax_parser_state{encoding=Enc} = State) when is_binary(Bytes) ->
+ case unicode:characters_to_list(Bytes, Enc) of
+ {incomplete, _, _} ->
+ cf(Bytes, State, fun parse_xml_decl/2);
+ {error, _Encoded, _Rest} ->
+ ?fatal_error(State, lists:flatten(io_lib:format("Bad character, not in ~p\n", [Enc])));
+ _ ->
+ parse_prolog(Bytes, State)
+ end;
+ parse_xml_decl(Bytes, State) ->
+ parse_prolog(Bytes, State)).
+
+-define(WHITESPACE(Bytes, State, Acc),
+ whitespace(?STRING_UNBOUND_REST(_C, _) = Bytes, State, Acc) ->
+ {lists:reverse(Acc), Bytes, State};
+ whitespace(Bytes, #xmerl_sax_parser_state{encoding=Enc} = State, Acc) when is_binary(Bytes) ->
+ case unicode:characters_to_list(Bytes, Enc) of
+ {incomplete, _, _} ->
+ cf(Bytes, State, Acc, fun whitespace/3);
+ {error, _Encoded, _Rest} ->
+ ?fatal_error(State, lists:flatten(io_lib:format("Bad character, not in ~p\n", [Enc])))
+ end).
+-define(PARSE_EXTERNAL_ENTITY_BYTE_ORDER_MARK(Bytes, State),
+ parse_external_entity_byte_order_mark(?STRING_EMPTY, State) ->
+ cf(?STRING_EMPTY, State, fun parse_external_entity_byte_order_mark/2);
+ parse_external_entity_byte_order_mark(?BYTE_ORDER_MARK_1, State) ->
+ cf(?BYTE_ORDER_MARK_1, State, fun parse_external_entity_byte_order_mark/2);
+ parse_external_entity_byte_order_mark(?BYTE_ORDER_MARK_2, State) ->
+ cf(?BYTE_ORDER_MARK_2, State, fun parse_external_entity_byte_order_mark/2);
+ parse_external_entity_byte_order_mark(?BYTE_ORDER_MARK_REST(Rest), State) ->
+ parse_external_entity_1(Rest, State);
+ parse_external_entity_byte_order_mark(Bytes, State) ->
+ parse_external_entity_1(Bytes, State)).