xmerl_scan.erl

来自「OTP是开放电信平台的简称」· ERL 代码 · 共 438 行 · 第 1/2 页

%%% @spec cont_state(ContinuationState, S::global_state()) -> global_state()
%%% @doc For controlling the ContinuationState, to be used in a continuation
%%% function, and called when the parser encounters the end of the byte stream.
%%% See <a href="xmerl_examples.html">tutorial</a> on customization functions.
cont_state(X, S=#xmerl_scanner{fun_states = FS}) ->
    FS1 = FS#xmerl_fun_states{cont = X},
    S#xmerl_scanner{fun_states = FS1}.


%% @spec file(Filename::string()) -> {xmlElement(),Rest}
%%   Rest = list()
%% @equiv file(Filename, [])
file(F) ->
    file(F, []).

%% @spec file(Filename::string(), Options::option_list()) -> {xmlElement(),Rest}
%%   Rest = list()
%%% @doc Parse file containing an XML document
file(F, Options) ->
    ExtCharset=case lists:keysearch(encoding,1,Options) of
		   {value,{_,Val}} -> Val;
		   false -> undefined
	       end,
    case int_file(F,Options,ExtCharset) of
	{Res, Tail,S=#xmerl_scanner{close_fun=Close}} ->
	    Close(S), % for side effects only - final state is dropped
	    {Res,Tail};
	{error, Reason} ->
	    {error, Reason}
    end.

int_file(F, Options,_ExtCharset) ->
     %%io:format("int_file F=~p~n",[F]),
    case file:read_file(F) of
	{ok, Bin} ->
	    int_string(binary_to_list(Bin), Options, filename:dirname(F),F);
	Error ->
	    Error
    end.

int_file_decl(F, Options,_ExtCharset) ->
%     io:format("int_file_decl F=~p~n",[F]),
    case file:read_file(F) of
	{ok, Bin} ->
	    int_string_decl(binary_to_list(Bin), Options, filename:dirname(F),F);
	Error ->
	    Error
    end.

%% @spec string(Text::list()) -> {xmlElement(),Rest}
%%   Rest = list()
%% @equiv string(Test, [])
string(Str) ->  
    string(Str, []).

%% @spec string(Text::list(),Options::option_list()) -> {xmlElement(),Rest}
%%   Rest = list()
%%% @doc Parse string containing an XML document
string(Str, Options) ->
     {Res, Tail, S=#xmerl_scanner{close_fun = Close}} =
	int_string(Str, Options,file_name_unknown),
    Close(S),    % for side effects only - final state is dropped
    {Res,Tail}.

int_string(Str, Options,FileName) ->
    {ok,  XMLBase} = file:get_cwd(),
    int_string(Str, Options, XMLBase, FileName).

int_string(Str, Options, XMLBase, FileName) ->
    S0=initial_state0(Options,XMLBase),
    S = S0#xmerl_scanner{filename=FileName},
    %%io:format("int_string1, calling xmerl_lib:detect_charset~n",[]),

    %% In case of no encoding attribute in document utf-8 is default, but
    %% another character set may be detected with help of Byte Order Marker or
    %% with help of the encoding of the first 4 bytes.
    case xmerl_lib:detect_charset(S#xmerl_scanner.encoding,Str) of
	{auto,'iso-10646-utf-1',Str2} ->
	    scan_document(Str2, S#xmerl_scanner{encoding="iso-10646-utf-1"});
	{external,'iso-10646-utf-1',Str2} ->
	    scan_document(Str2, S#xmerl_scanner{encoding="iso-10646-utf-1"});
	{undefined,undefined,Str2} -> %% no auto detection
	    scan_document(Str2, S);
	{external,ExtCharset,Str2} -> 
	    %% no auto detection, ExtCharset is an explicitly provided
	    %% 7 bit,8 bit or utf-8 encoding
	    scan_document(Str2, S#xmerl_scanner{encoding=atom_to_list(ExtCharset)})
    end.

int_string_decl(Str, Options, XMLBase, FileName) ->
    S0=initial_state0(Options,XMLBase),
    S = S0#xmerl_scanner{filename=FileName},
    case xmerl_lib:detect_charset(S#xmerl_scanner.encoding,Str) of
	{auto,'iso-10646-utf-1',Str2} ->
	    scan_decl(Str2, S#xmerl_scanner{encoding="iso-10646-utf-1"});
	{external,'iso-10646-utf-1',Str2} ->
	    scan_decl(Str2, S#xmerl_scanner{encoding="iso-10646-utf-1"});
	{undefined,undefined,Str2} ->
	    scan_decl(Str2, S);
	{external,ExtCharset,Str2} ->
	    scan_decl(Str2, S#xmerl_scanner{encoding=atom_to_list(ExtCharset)})
    end.
    


initial_state0(Options,XMLBase) ->
    initial_state(Options, #xmerl_scanner{
		    event_fun = fun event/2,
		    hook_fun = fun hook/2,
		    acc_fun = fun acc/3,
		    fetch_fun = fun fetch/2,
		    close_fun = fun close/1,
		    continuation_fun = fun cont/3,
		    rules_read_fun = fun rules_read/3,
		    rules_write_fun = fun rules_write/4,
		    rules_delete_fun= fun rules_delete/3,
		    xmlbase = XMLBase
		   }).

initial_state([{event_fun, F}|T], S) ->
    initial_state(T, S#xmerl_scanner{event_fun = F});
initial_state([{event_fun, F, ES}|T], S) ->
    S1 = event_state(ES, S#xmerl_scanner{event_fun = F}),
    initial_state(T, S1);
initial_state([{acc_fun, F}|T], S) ->
    initial_state(T, S#xmerl_scanner{acc_fun = F});
initial_state([{hook_fun, F}|T], S) ->
    initial_state(T, S#xmerl_scanner{hook_fun = F});
initial_state([{hook_fun, F, HS}|T], S) ->
    S1 = hook_state(HS, S#xmerl_scanner{hook_fun = F}),
    initial_state(T, S1);
initial_state([{close_fun, F}|T], S) ->
    initial_state(T, S#xmerl_scanner{close_fun = F});
initial_state([{fetch_fun, F}|T], S) ->
    initial_state(T, S#xmerl_scanner{fetch_fun = F});
initial_state([{fetch_fun, F, FS}|T], S) ->
    S1 = fetch_state(FS, S#xmerl_scanner{fetch_fun = F}),
    initial_state(T, S1);
initial_state([{fetch_path, P}|T], S) ->
    initial_state(T, S#xmerl_scanner{fetch_path = P});
initial_state([{continuation_fun, F}|T], S) ->
    initial_state(T, S#xmerl_scanner{continuation_fun = F});
initial_state([{continuation_fun, F, CS}|T], S) ->
    S1 = cont_state(CS, S#xmerl_scanner{continuation_fun = F}),
    initial_state(T, S1);
initial_state([{rules, R}|T], S) ->
    initial_state(T, S#xmerl_scanner{rules = R,
				     keep_rules = true});
initial_state([{rules, Read, Write, RS}|T], S) ->
    S1 = rules_state(RS, S#xmerl_scanner{rules_read_fun = Read,
					 rules_write_fun = Write,
					 keep_rules = true}),
    initial_state(T, S1);
initial_state([{user_state, F}|T], S) ->
    initial_state(T, S#xmerl_scanner{user_state = F});
initial_state([{space, L}|T], S) ->
    initial_state(T, S#xmerl_scanner{space = L});
initial_state([{line, L}|T], S) ->
    initial_state(T, S#xmerl_scanner{line = L});
initial_state([{namespace_conformant, F}|T], S) when F==true; F==false ->
    initial_state(T, S#xmerl_scanner{namespace_conformant = F});
initial_state([{validation, F}|T], S) 
  when F==off; F==dtd; F==schema; F==true; F==false ->
    initial_state(T, S#xmerl_scanner{validation = validation_value(F)});
initial_state([{schemaLocation, SL}|T], S) when is_list(SL) ->
    initial_state(T, S#xmerl_scanner{schemaLocation=SL});
initial_state([{quiet, F}|T], S) when F==true; F==false ->
    initial_state(T, S#xmerl_scanner{quiet = F});
initial_state([{doctype_DTD,DTD}|T], S) ->
    initial_state(T,S#xmerl_scanner{doctype_DTD = DTD});
initial_state([{text_decl,Bool}|T], S) ->
    initial_state(T,S#xmerl_scanner{text_decl=Bool});
initial_state([{environment,Env}|T], S) ->
    initial_state(T,S#xmerl_scanner{environment=Env});
initial_state([{xmlbase, D}|T], S) ->
    initial_state(T, S#xmerl_scanner{xmlbase = D});
initial_state([{encoding, Enc}|T], S) ->
    initial_state(T, S#xmerl_scanner{encoding = Enc});
initial_state([], S=#xmerl_scanner{rules = undefined}) ->
    Tab = ets:new(rules, [set, public]),
    S#xmerl_scanner{rules = Tab};
initial_state([], S) ->
    S.

validation_value(true) ->
    dtd;
validation_value(false) ->
    off;
validation_value(F) ->
    F.

%%% -----------------------------------------------------
%%% Default modifier functions

%%% Hooks:
%%% - {element, Line, Name, Attrs, Content}
%%% - {processing_instruction, Line, Data}

hook(X, State) ->
    {X, State}.

%%% Events:
%%%
%%% #xmerl_event{event : started | ended,
%%%              line  : integer(),
%%%		 col   : integer(),
%%%              data}
%%%
%%% Data		Events
%%% document		started, ended
%%% #xmlElement		started, ended
%%% #xmlAttribute	ended
%%% #xmlPI		ended
%%% #xmlComment		ended
%%% #xmlText		ended
event(_X, S) ->
    S.

%% The acc/3 function can return either {Acc