%
% Simple Prolog client of the APE webservice.
% Note: this program is specific to SWI-Prolog.
% Also: http-libraries have to be installed.
%
% This example uses APE to parse a pseudo-Estonian sentence using a small user lexicon.
%
% @author Kaarel Kaljurand
% @version 2009-05-20
%

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

host('attempto.ifi.uzh.ch').

port(80).

path('/ws/ape/apews.perl').

ulextext('
	adv(kiiresti, eng:fast).
	tv_finsg(kirjutab, sym:\'✍\').
	adj_itr(kontrollitud, eng:controlled).
	noun_sg(keel, eng:language, neutr).
	noun_sg(robot, eng:robot, neutr).
').

%% acetext_drs(+ACE:atom, -DRS:term, -Tokens:list) is det.
%
% Usage:
%
%==
% ?- acetext_drs('A robot kirjutab a kontrollitud keel kiiresti. It kirjutab a keel.', DRS, Tokens).
%
% DRS = drs([_G6490, _G6493, _G6496, _G6499, _G6502], [object(_G6490, eng:robot, countable, na, eq, 1)-1, object(_G6493, ...
% Tokens = [['A', robot, kirjutab, a, kontrollitud, keel, kiiresti, '.'], ['It', kirjutab, a, keel, '.']] ;
%
% No
%==
%
% @param ACE is an ACE text
% @param DRS is an Attempto DRS
% @param Tokens is a list of sentences of tokens in the input text
%
acetext_drs(ACE, DRS, Tokens) :-
	host(Host),
	port(Port),
	path(Path),
	ulextext(Ulextext),
	parse_url(URL, [
		protocol(http),
		host(Host),
		port(Port),
		path(Path),
		search([
			ulextext = Ulextext,
			cdrs = on,
			ctokens = on,
			text = ACE
		])
	]),
	http_open(URL, Stream, []),
	load_xml_file(Stream, XML),
	close(Stream),
	extract_term(XML, drs, DRS),
	extract_term(XML, tokens, Tokens).


%% extract_term(+XML:term, -ElementName:atom, -TextContentAsTerm:term) is det.
%
% Parses the XML to extract the textual content of an element and convert it to a Prolog term.
% Only the first matching element is processed.
%
% @param XML is SWI Prolog's represenation of XML data
% @param ElementName is the name of the element to be extracted
% @param TextContentAsTerm is the text content of the element (converted to Prolog term)
%
extract_term(XML, ElementName, TextContentAsTerm) :-
	get_element_textcontent(XML, ElementName, TextContentAsAtom),
	!,
	atom_to_term(TextContentAsAtom, TextContentAsTerm, _).


%% get_element_textcontent(+XML:term, +ElementName:atom, -TextContent:atom) is nondet.
%
% @param XML is SWI Prolog's represenation of XML data
% @param ElementName is the name of the element whose text content we want to retrieve
% @param TextContent is the text content of the element (as an atom)
%
get_element_textcontent([element(ElementName, _Attributes, [TextContent]) | _], ElementName, TextContent).

get_element_textcontent([element(_, _Attributes, Children) | _], ElementName, TextContent) :-
	get_element_textcontent(Children, ElementName, TextContent).

get_element_textcontent([_ | Elements], ElementName, TextContent) :-
	get_element_textcontent(Elements, ElementName, TextContent).