Making Parsers with yecc
Yecc is an erlang version of yacc/bison.
We have a BNF(Backus-Naur_form) grammar in a source file ending in .yrl, yrl means yecc rule list. We can parse a simple xhtml file using yecc. Actually, we will apply yecc to html.yrl to create a parser called html_parser.erl. Next we use the html_parser to parse some xhtml, voila.
yecc:yecc("html.yrl","html_parser.erl").
c(html_parser).
f(B), {_,B,_} =
erl_scan:string(
"<html><head></head><body>hello_world</body></html>").
html_parser:parse(B).
All tags in the xhtml code must have matching open and close. (Of course a more powerful way to parse an xml file in erlang is to use xmerl).
html.yrl source:
Nonterminals tag elements element start_tag end_tag . Terminals 'atom' '<' '>' '/'. Rootsymbol tag. tag -> start_tag tag end_tag : ['$1', '$2', '$3']. tag -> start_tag tag tag end_tag : ['$1', '$2', '$3', '$4']. tag -> start_tag elements end_tag : ['$1', {'contents','$2'}, '$3']. tag -> start_tag end_tag : ['$1','$2']. start_tag -> '<' 'atom' '>' : {'open','$2'}. end_tag -> '<' '/' 'atom' '>' : {'close','$3'}. elements -> element : ['$1']. elements -> element elements : ['$1', '$2']. element -> atom : '$1'. % yecc:yecc("html.yrl","html_parser.erl"). % c(html_parser). % f(B), {_,B,_} = % erl_scan:string( % "<html><head></head><body>hello_world</body></html>"). % html_parser:parse(B).
It can be a pain to build and run a parser each time we edit the source yrl file. To speed things up, we can use a program to build and run the parser for us. We compile and run the test program which builds the parser and tests it for us on some document.
-module(html_test).
-compile(export_all).
start() ->
yecc:yecc("html.yrl","html_parser.erl"),
cover:compile(html_parser),
{_,List_of_symbols,_}=erl_scan:string(
"<html><head><title>greeting</title></head>
<body>
hello there world what is up
</body>
</html>"),
{ok,L} = html_parser:parse(List_of_symbols),
register(do_event, spawn(html_test,event_loop,[])),
Events = lists:flatten(L),
send_events(Events),
Events.
send_events([]) -> do_event ! {exit};
send_events([H|T]) ->
do_event ! H,
%io:format(" ~w ~n",[H]),
send_events(T).
event_loop() ->
receive
{open,{atom,_Line_Number,html}} ->
io:format("~n start scan ~n", []),
event_loop();
{contents,List} ->
Contents = get_contents(List,[]),
io:format("~n contents: ~w ~n", [Contents]);
{exit} -> exit(normal)
end,
event_loop().
get_contents([],Items) -> Items;
get_contents([H|T],Items)->
if
length(T) > 0 ->
NT = hd(T);
true ->
NT = T
end,
{atom,_N,Item} = H,
NItems = Items++[Item],
% io:format(" ~w ",[Item]),
get_contents(NT,NItems).
% 6> c(html_test).
% {ok,html_test}
% 7> html_test:start().
% [greeting]
% [hello,there,world,what,is,up]
% and events.