%
% Simple Prolog client of the APE webservice.
% Note: this program is specific to SWI-Prolog.
% Also: http-libraries have to be installed.
%
% @author Kaarel Kaljurand
% @version 2009-05-24
%

:- use_module(library('http/http_open')).


%% acetext_to_drs(+ACE:atom, -DRS:term, -Messages:list) is det.
%
% For a given ACE text, generates its DRS.
% In case parsing of the ACE text fails then the empty DRS =|drs([], [])|= is generated.
% Messages are generated in any case, but if there are no errors nor warnings
% then the message list is empty.
%
% Usage example:
%
%==
% ?- acetext_to_drs('Every man waits.', DRS, Messages).
%
% DRS = drs([], [=>(drs([_G2327], [object(_G2327, man, countable, na, eq, 1)-1/2]), ...
% Messages = [].
%
%
% ?- acetext_to_drs('Every manman waits.', DRS, Messages).
%
% DRS = drs([], []),
% Messages = [message(error, sentence, '1'-'2', 'Every <> manman waits.',
%            'This is the first sentence that was not ACE.
%		      The sign <> indicates the position where parsing failed.'),
%			  message(error, word, ''-'', manman, madman)].
%
%
% ?- acetext_to_drs('JohnJohn waits.', DRS, Messages).
%
% DRS = drs([_G2389], [predicate(_G2389, wait, named('JohnJohn'))-1/2]),
% Messages = [message(warning, word, '1'-'1', 'JohnJohn',
%            'Undefined word. Interpreted as a singular proper name.')].
%
%==
%
% @param Address is a triple in the form address(Host, Port, Path)
% @param ACE is an ACE text
% @param DRS is an Attempto DRS
% @param Messages is a list of APE error/warning messages
%
acetext_to_drs(ACE, DRS, Messages) :-
	acetext_to_drs(address('attempto.ifi.uzh.ch', 80, '/ws/ape/apews.perl'), ACE, DRS, Messages).

acetext_to_drs(address(Host, Port, Path), ACE, DRS, Messages) :-
	parse_url(URL, [
		protocol(http),
		host(Host),
		port(Port),
		path(Path),
		search([
			cdrs = on,
			text = ACE
		])
	]),
	http_open(URL, Stream, []),
	load_xml_file(Stream, XML),
	close(Stream),
	get_drs(XML, DRS),
	get_messages(XML, Messages).


%% get_drs(+XML:xml, -DRS:term) is det.
%
% Parses the XML to extract the DRS and convert it into a Prolog term.
%
% @param XML is SWI Prolog's representation of XML data
% @param DRS is an Attempto DRS
%
get_drs(XML, DRS) :-
	get_element_content(XML, drs, [DrsAtom]),
	!,
	atom_to_term(DrsAtom, DRS, _).


%% get_messages(+XML:xml, -Messages:list) is det.
%
% @param XML is SWI Prolog's representation of XML data
% @param Messages is a list of APE error/warning messages
%
get_messages(XML, Messages) :-
	get_element_content(XML, messages, MessagesXml),
	!,
	messagesxml_to_messages(MessagesXml, Messages).


%% messagesxml_to_messages(+MessagesXml:xml, -Messages:list) is det.
%
% Converts the APE messages in the SWI-Prolog XML notation into the native APE message-terms.
%
% @param MessagesXml is list of XML elements in SWI-Prolog notation
% @param Messages is list of messages
%
messagesxml_to_messages([], []).

% Note: we ignore textual childnodes.
% These are always whitespace, the byproduct of XML pretty printing.
messagesxml_to_messages([Atom | MessagesXml], Messages) :-
	atom(Atom),
	!,
	messagesxml_to_messages(MessagesXml, Messages).

messagesxml_to_messages(
	[element(message, [importance=Importance, type=Type, sentence=SentenceId, token=TokenId, value=Subject, repair=Description], []) | MessagesXml],
	[message(Importance, Type, SentenceId-TokenId, Subject, Description) | Messages]
	) :-
	messagesxml_to_messages(MessagesXml, Messages).


%% get_element_content(+XML:xml, +ElementName:atom, -Content:term) is nondet.
%
% @param XML is SWI Prolog's representation of XML data
% @param ElementName is the name of the element whose content we want to retrieve
% @param Content is the content of the element
%
get_element_content([element(ElementName, _Attributes, Content) | _], ElementName, Content).

get_element_content([element(_, _Attributes, Children) | _], ElementName, Content) :-
	get_element_content(Children, ElementName, Content).

get_element_content([_ | Elements], ElementName, Content) :-
	get_element_content(Elements, ElementName, Content).