Merge branch 'master' of nlp.pwr.wroc.pl:wccl

libwccl/ops/functions/bool/iterations/leftlook.cpp

@@ -8,7 +8,7 @@ bool LeftLook::iterate(
 	Wccl::Position &p,
 	const Wccl::FunExecContext &context) const
 {
-	for(int i = left; i <= right; ++i) {
+	for(int i = right; i >= left; --i) {
 		p.set_value(i);
 		if(evaluating_expr_->apply(context)->get_value()) {
 			return true;

libwccl/ops/functions/bool/iterations/leftlook.h

@@ -6,8 +6,9 @@
 namespace Wccl {
 
 /**
- * Iterative operator "llook", which looks for a first
- * position from left that makes evaluating expression return true.
+ * Iterative operator "llook", which looks for the rightmost position
+ * in a range that makes the evaluating expression return true, i.e.
+ * it "looks to the left".
  */
 class LeftLook : public Iteration
 {
@@ -30,12 +31,12 @@ public:
 
 protected:
 	/**
-	 * @returns True if, when scanning from left,
-	 * a position within range is found that makes
+	 * @returns True if, when scanning right-to-left,
+	 * a position within the range is found that makes
 	 * the evaluating function return true. False
 	 * otherwise.
 	 * @note Upon success, the iteration variable is
-	 * set to the first position in range from left
+	 * set to the rightmost position in range
 	 * that has made evaluation function return true
 	 * (that is the position that made the stop condition
 	 * true in this case).

libwccl/ops/functions/bool/iterations/rightlook.cpp

@@ -8,7 +8,7 @@ bool RightLook::iterate(
 	Wccl::Position &p,
 	const Wccl::FunExecContext &context) const
 {
-	for(int i = right; i >= left; --i) {
+	for(int i = left; i <= right; ++i) {
 		p.set_value(i);
 		if(evaluating_expr_->apply(context)->get_value()) {
 			return true;

libwccl/ops/functions/bool/iterations/rightlook.h

@@ -6,8 +6,9 @@
 namespace Wccl {
 
 /**
- * Iterative operator "rlook", which looks for a first
- * position from right that makes evaluating expression return true.
+ * Iterative operator "rlook", which looks for the leftmost position
+ * in a range that makes the evaluating expression return true, i.e.
+ * it "looks to the right".
  */
 class RightLook : public Iteration
 {
@@ -30,12 +31,12 @@ public:
 
 protected:
 	/**
-	 * @returns True if, when scanning from right,
-	 * a position within range is found that makes
+	 * @returns True if, when scanning left-to-right,
+	 * a position within the range is found that makes
 	 * the evaluating function return true. False
 	 * otherwise.
 	 * @note Upon success, the iteration variable is
-	 * set to the first position in range from right
+	 * set to the leftmost position in range
 	 * that has made evaluation function return true
 	 * (that is the position that made the stop condition
 	 * true in this case).

libwccl/ops/functions/bool/predicates/pointagreement.h

@@ -19,10 +19,12 @@ public:
 	PointAgreement(
 		const PosFunctionPtr& pos1_expr,
 		const PosFunctionPtr& pos2_expr,
-		const TSetFunctionPtr& attribs_expr)
+		const TSetFunctionPtr& attribs_expr,
+		const Corpus2::Tagset& tagset)
 		: pos1_expr_(pos1_expr),
 		  pos2_expr_(pos2_expr),
-		  attribs_expr_(attribs_expr)
+		  attribs_expr_(attribs_expr),
+		  tagset_(tagset)
 	{
 		BOOST_ASSERT(pos1_expr_);
 		BOOST_ASSERT(pos2_expr_);
@@ -45,6 +47,7 @@ protected:
 	const PosFunctionPtr pos1_expr_;
 	const PosFunctionPtr pos2_expr_;
 	const TSetFunctionPtr attribs_expr_;
+	const Corpus2::Tagset& tagset_;
 
 	/**
 	 * Gets two positions from arguments. If either of them

libwccl/ops/functions/bool/predicates/strongagreement.h

@@ -18,10 +18,12 @@ public:
 	StrongAgreement(
 		const PosFunctionPtr& left_pos_expr,
 		const PosFunctionPtr& right_pos_expr,
-		const TSetFunctionPtr& attribs_expr)
+		const TSetFunctionPtr& attribs_expr,
+		const Corpus2::Tagset& tagset)
 		: left_pos_expr_(left_pos_expr),
 		  right_pos_expr_(right_pos_expr),
-		  attribs_expr_(attribs_expr)
+		  attribs_expr_(attribs_expr),
+		  tagset_(tagset)
 	{
 		BOOST_ASSERT(left_pos_expr_);
 		BOOST_ASSERT(right_pos_expr_);
@@ -45,6 +47,7 @@ protected:
 	const PosFunctionPtr left_pos_expr_;
 	const PosFunctionPtr right_pos_expr_;
 	const TSetFunctionPtr attribs_expr_;
+	const Corpus2::Tagset& tagset_;
 
 	/**
 	 * Gets start and end positions for range from arguments.

libwccl/ops/functions/bool/predicates/weakagreement.h

@@ -18,10 +18,12 @@ public:
 	WeakAgreement(
 		const PosFunctionPtr& left_pos_expr,
 		const PosFunctionPtr& right_pos_expr,
-		const TSetFunctionPtr& attribs_expr)
+		const TSetFunctionPtr& attribs_expr,
+		const Corpus2::Tagset& tagset)
 		: left_pos_expr_(left_pos_expr),
 		  right_pos_expr_(right_pos_expr),
-		  attribs_expr_(attribs_expr)
+		  attribs_expr_(attribs_expr),
+		  tagset_(tagset)
 	{
 		BOOST_ASSERT(left_pos_expr_);
 		BOOST_ASSERT(right_pos_expr_);
@@ -45,6 +47,7 @@ protected:
 	const PosFunctionPtr left_pos_expr_;
 	const PosFunctionPtr right_pos_expr_;
 	const TSetFunctionPtr attribs_expr_;
+	const Corpus2::Tagset& tagset_;
 
 	/**
 	 * Gets start and end positions for range from arguments.

libwccl/ops/functions/tset/agrfilter.h

@@ -22,11 +22,13 @@ public:
 			const PosFunctionPtr& left_pos_expr,
 			const PosFunctionPtr& right_pos_expr,
 			const TSetFunctionPtr& attribs_expr,
-			const TSetFunctionPtr& mask_expr)
+			const TSetFunctionPtr& mask_expr,
+			const Corpus2::Tagset& tagset)
 		: left_pos_expr_(left_pos_expr),
 		  right_pos_expr_(right_pos_expr),
 		  attribs_expr_(attribs_expr),
-		  mask_expr_(mask_expr)
+		  mask_expr_(mask_expr),
+		  tagset_(tagset)
 	{
 		BOOST_ASSERT(left_pos_expr_);
 		BOOST_ASSERT(right_pos_expr_);
@@ -54,7 +56,7 @@ protected:
 	const TSetFunctionPtr attribs_expr_;
 	const TSetFunctionPtr mask_expr_;
 
-
+	const Corpus2::Tagset& tagset_;
 
 	/**
 	 * Gets range specified via Position expressions. Trims the range

libwccl/parser/grammar.g

@@ -1098,9 +1098,9 @@ bool_iteration
 		RPAREN {
 			ret.reset(new AtLeast(lpos, rpos, *pacc, expr, min_match));
 		}
-	| "llook" LPAREN
-			lpos = position_operator [tagset, vars] COMMA 
+	| "llook" LPAREN //note inverted rpos/lpos order
 			rpos = position_operator [tagset, vars] COMMA
+			lpos = position_operator [tagset, vars] COMMA
 			pacc = position_variable_acc [vars]     COMMA
 			expr = bool_operator     [tagset, vars] 
 		RPAREN {

libwccl/values/value.h

@@ -6,7 +6,10 @@
 #define WCCL_VALUE_PREAMBLE \
 static const char* type_name; \
 const char* get_type_name() const { return type_name; } \
-static std::string var_repr(const std::string &var_name);
+static std::string var_repr(const std::string &var_name); \
+std::string make_var_repr(const std::string &var_name) const { \
+	return var_repr(var_name); \
+}
 
 
 namespace Wccl {
@@ -35,6 +38,8 @@ public:
 
 	virtual ~Value() {}
 
+	virtual std::string make_var_repr(const std::string&) const = 0;
+
 	/**
 	 * String representation function, in general, a tagset is required,
 	 * but some classes might not need that, so by default just forward

tests/data/example.ccl

@@ -4,6 +4,7 @@ After the --- line comes the first test -- the operator string, followed by an e
 If the first operator line is "position=XXX", position is parsed as an int and that is used as the current position in the sentence for this and any following tests. Default position is 0.
 Tests are separated from one another by a --- line.
 This test loads no sentence, but can change the position anyway. Tagset is kipi by default.
+Variable state can be checked by a name=value lines after the expected output.
 tagset=ikipi
 ---
 equal(["aaa"], "aaa")
@@ -13,3 +14,10 @@ True
 lower(["A"])
 
 ["a"]
+---
+setvar($b:A, setvar($t:B, subst))
+
+True
+A=True
+B={subst}
+

tests/data/iterations.ccl

 tagset=kipi
+sentence=t01.xml
 ---
 rlook(begin, end, $Test, True)
 
 True
 ---
+llook(0, end, $V, False)
+
+False
+V=nowhere
+---
+position=0
+rlook(0,end,$P,
+	inter(class[$P], {subst})
+)
+
+True
+P=3
+---
+position=3
+rlook(0,end,$P,
+	inter(class[$P], {subst})
+)
+
+True
+P=0
+---
+position=4
+rlook(0,end,$P,
+	inter(class[$P], {subst})
+)
+
+True
+P=8
+---
+position=19
+rlook(0,end,$P,inter(class[$P], {subst}))
+
+False
+P=nowhere
+---
+position=26
+llook(0,begin,$NomAcc, equal(cas[$NomAcc],{nom,acc}))
+
+True
+NomAcc=-1
+---
+position=12
+llook(0,begin,$NomAcc, equal(cas[$NomAcc],{nom,acc}))
+
+True
+NomAcc=-4
+---
+position=2
+llook(0,begin,$NomAcc, equal(cas[$NomAcc],{nom,acc}))
+
+False
+NomAcc=nowhere
+---
+position=0
+only(10,12,$Loc, in(loc, cas[$Loc]))
+
+True
+Loc=12
+---
+position=0
+only(10,13,$Loc, in(loc, cas[$Loc]))
+
+False
+Loc=nowhere
+---
+position=0
+atleast(9,13,$Loc, in(loc, cas[$Loc]), 3)
+
+True
+Loc=12
+---
+position=0
+atleast(9,13,$Loc, in(loc, cas[$Loc]), 4)
+
+False
+Loc=nowhere
+---
+position=-1
+llook(0,begin,$NomAcc, equal(cas[$NomAcc],{nom,acc}))
+
+False
+NomAcc=nowhere
+---

tests/datadriven.cpp

@@ -98,7 +98,24 @@ void test_one_item_actual(const compare_test& c)
 			expected_output = "";
 			operator_string = "";
 			++line_no;
-			std::getline(ifs_in, line);
+			while (ifs_in.good() && line != "---" && line != "") {
+				std::getline(ifs_in, line);
+				std::vector<std::string> fields;
+				boost::algorithm::split(fields, line, boost::is_any_of(separators));
+				if (fields.size() == 2) {
+					boost::shared_ptr<Wccl::Value> v;
+					v = fu.variables()->get<Wccl::Value>(fields[0]);
+					if (!v) {
+						BOOST_ERROR("Invalid variable name in test: "
+							<< fields[0] << " on line " << line_no);
+					} else if (v->to_string(tagset) != fields[1]) {
+						BOOST_ERROR("Variable " << fields[0]
+							<< " value mismatch on line "
+							<< line_no << "\n: expected " << fields[1]
+							<< " got " << v->to_string(tagset));
+					}
+				}
+			}
 			BOOST_REQUIRE(line == "---" || line == "");
 		} else {
 			if (operator_string.empty() && line.substr(0, 9) == "position=") {

wcclparser/main.cpp

@@ -90,8 +90,17 @@ void libedit_read_loop(boost::function<bool (const std::string&)>& line_cb)
 }
 #endif
 
+void dumpvariables(const Wccl::Variables& vars, const Corpus2::Tagset& tagset)
+{
+	typedef std::pair<std::string, boost::shared_ptr<Wccl::Value> > v_t;
+	foreach (const v_t& v, vars.get_all<Wccl::Value>()) {
+		std::cerr << v.second->make_var_repr(v.first) << "="
+			<< v.second->to_string(tagset) << "\n";
+	}
+}
+
 bool process_line(const std::string& line, Wccl::Parser& parser,
-	Wccl::SentenceContext& sc, bool all_positions)
+	Wccl::SentenceContext& sc, bool all_positions, bool dump_variables)
 {
 	if (line.empty() || line == "exit" || line == "quit") {
 		return true;
@@ -123,6 +132,9 @@ bool process_line(const std::string& line, Wccl::Parser& parser,
 						<< "Parsed expression: "
 						<< retVal->to_string(parser.tagset())
 						<< std::endl;
+					if (dump_variables) {
+						dumpvariables(*cx.variables(), parser.tagset());
+					}
 				} else {
 					std::cerr << "Problem while parsing -- "
 						<< "retVal is NULL!" << std::endl;
@@ -156,6 +168,7 @@ int main(int argc, char** argv)
 	std::string query_load = "";
 	std::string position = "0";
 	bool quiet = false;
+	bool dump_variables = false;
 	using boost::program_options::value;
 
 	boost::program_options::options_description desc("Allowed options");
@@ -172,6 +185,8 @@ int main(int argc, char** argv)
 			 "Query to run (disables interactive mode)\n")
 			("quiet,q", value(&quiet)->zero_tokens(),
 			 "Suppress messages\n")
+			("variables,V", value(&dump_variables)->zero_tokens(),
+			 "Dump variables after running each query\n")
 			("help,h", "Show help")
 			;
 	boost::program_options::variables_map vm;
@@ -214,14 +229,14 @@ int main(int argc, char** argv)
 		sc.set_position(pos);
 		Wccl::Parser parser(tagset);
 		if (!query.empty()) {
-			process_line(query, parser, sc, position == "all");
+			process_line(query, parser, sc, position == "all", dump_variables);
 			return 0;
 		} else if (!query_load.empty()) {
 			std::ifstream qf(query_load.c_str());
 			if (qf.good()) {
 				std::stringstream ss;
 				ss << qf.rdbuf();
-				process_line(ss.str(), parser, sc, position == "all");
+				process_line(ss.str(), parser, sc, position == "all", dump_variables);
 				return 0;
 			} else {
 				throw Wccl::FileNotFound(sentence_load, "", "Query file");
@@ -229,7 +244,7 @@ int main(int argc, char** argv)
 		}
 		boost::function<bool (const std::string&)> f;
 		f = boost::bind(&process_line, _1, boost::ref(parser), boost::ref(sc),
-			position == "all");
+			position == "all", dump_variables);
 #ifdef HAVE_LIBEDIT
 		libedit_read_loop(f);
 #else