sax.erl

-module(sax).
-compile(export_all).

-record(node, {name, attributes = [], content = [], children = [], depth}).
-record(state, {depth = 0, tree = []}).

%% xmerl_sax_parser:file("simple.xml", [{event_fun, fun sax:parser/3}, {event_state, sax:new_state()}]).

new_state() ->
    #state{ depth = 0, tree = []}.

parser({startElement, _NS, Name, _, Attributes} = Elem, _, #state{ tree = Tree } = State) ->
    %% io:format("'~p'Starting ~p: ~p~n", [State#state.depth, Name, Elem]),
    State#state{
        tree = [#node{
            name = Name,
            attributes = [{AttribName, AttribVal} || {_, _, AttribName, AttribVal} <- Attributes],
            depth = State#state.depth
        } | Tree],
        depth = State#state.depth + 1
    };

parser({characters, MoreData}, _, #state{ tree = [This | Tree]} = State) ->
    State#state{
        tree = [ This#node{ content = [MoreData | This#node.content]} | Tree]
    };

parser({endElement, _, Name, _} = Elem, _, State) ->
    %% io:format("'~p'Ending ~p: ~p~n", [State#state.depth, Name, Elem]),
    State#state{
        depth = State#state.depth - 1
    };

parser(endDocument, _, State) ->
    {[Parent | Children], []} = compose_tree(0, lists:reverse(State#state.tree), []),
    Parent#node{ children = Children};

parser({ignorableWhitespace, _}, _, State) ->
    State;

parser(Other, _, State) -> 
    %% io:format("'~p' Other ~p~n", [State#state.depth, Other]),
    State.

compose_tree(_I, [], Acc) -> {lists:reverse(Acc), []};

compose_tree(Indent, [#node{ depth = Indent } = Node | Rest], Acc) ->
    io:format("A '~p' - '~p' ~p~n", [Indent, Node#node.name, length(Acc)]),
    io:format("Rest ~p~n", [Rest]),
    compose_tree(Indent, Rest, [ Node | Acc ]);

compose_tree(Indent, [ #node{ depth = Ind } = Node | Rest], Acc) when Ind > Indent ->
    io:format("B '~p' - '~p' ~p~n", [Ind, Node#node.name, length(Acc)]),
    {Tree, Remains} = compose_tree(Ind, Rest, [Node]),
    compose_tree(Indent, Remains, [Tree | Acc]);

compose_tree(Indent, [  #node{ depth = Ind } = Node | Rest], Acc) when Ind < Indent ->
    io:format("C '~p' - '~p' ~p~n", [Ind, Node#node.name, length(Acc)]),
    {lists:reverse(Acc), [Node | Rest]}.

simple.xml

<bodyparts>
    <heads/>
    <arms>
        <hands>
            <fingers/>
        </hands>
    </arms>
    <shoulders/>
    <knees>
        <toes/>
    </knees>
</bodyparts>