diff --git a/libwccl/CMakeLists.txt b/libwccl/CMakeLists.txt
index a1d869794ff87c66b71bab02fafa6fa4cc38dbb2..c969bf7866504908bfc591e752162d76419d1d12 100644
--- a/libwccl/CMakeLists.txt
+++ b/libwccl/CMakeLists.txt
@@ -4,7 +4,7 @@ PROJECT(wccl)
include_directories( ${CMAKE_CURRENT_BINARY_DIR}/include/ )
-find_package(Corpus2 0.1.2 REQUIRED)
+find_package(Corpus2 1.0.2 REQUIRED)
set(LIBS ${LIBS} ${Corpus2_LIBRARY})
find_package(PwrUtils 0.0.3 REQUIRED)
@@ -40,6 +40,7 @@ SET(libwccl_STAT_SRC
ops/functions/strset/getorth.cpp
ops/functions/strset/tolower.cpp
ops/functions/strset/toupper.cpp
+ ops/functions/tset/getsymbols.cpp
parser/grammar.g
parser/Parser.cpp
parser/ParserException.cpp
diff --git a/libwccl/ops/function.h b/libwccl/ops/function.h
index f2e4842d3f7d0541a4f674ea697ce14ce047167e..3aa008325f83f49f4371353afb150028d4a76559 100644
--- a/libwccl/ops/function.h
+++ b/libwccl/ops/function.h
@@ -21,7 +21,7 @@ public:
/**
* @returns Name of the function. By default it is same as raw name.
*/
- virtual const std::string name(const Corpus2::Tagset&) const {
+ virtual std::string name(const Corpus2::Tagset&) const {
return raw_name();
}
/**
diff --git a/libwccl/ops/functions/tset/getsymbols.cpp b/libwccl/ops/functions/tset/getsymbols.cpp
new file mode 100644
index 0000000000000000000000000000000000000000..fe5d1fa35e48ac2390c00ea7f742d36db918d694
--- /dev/null
+++ b/libwccl/ops/functions/tset/getsymbols.cpp
@@ -0,0 +1,42 @@
+#include <libwccl/ops/functions/tset/getsymbols.h>
+#include <libwccl/ops/formatters.h>
+#include <libwccl/ops/functions/constant.h>
+
+namespace Wccl {
+
+std::string GetSymbols::to_string(const Corpus2::Tagset& tagset) const
+{
+ return UnaryFunctionFormatter::to_string(tagset, *this, *pos_expr_, "[", "]");
+}
+
+std::string GetSymbols::to_raw_string() const {
+ return UnaryFunctionFormatter::to_raw_string(*this, *pos_expr_, "[", "]");
+}
+
+std::string GetSymbols::name(const Corpus2::Tagset &tagset) const
+{
+ return tagset.get_attribute_name(mask_.get_values());
+}
+
+std::string GetSymbols::raw_name() const
+{
+ return mask_.raw_dump();
+}
+
+GetSymbols::BaseRetValPtr GetSymbols::apply_internal(const FunExecContext& context) const
+{
+ const boost::shared_ptr<const Position>& pos = pos_expr_->apply(context);
+ const SentenceContext& sc = context.sentence_context();
+ if(pos->is_outside(sc) || !sc.is_current_inside()) {
+ return detail::DefaultFunction<TSet>()->apply(context);
+ }
+
+ boost::shared_ptr<TSet> tset = boost::make_shared<TSet>();
+ const Corpus2::Token* token = sc.at(*pos);
+ foreach (const Corpus2::Lexeme& lexeme, token->lexemes()) {
+ tset->combine_with(lexeme.tag());
+ }
+ return tset;
+}
+
+} /* end ns Wccl */
diff --git a/libwccl/ops/functions/tset/getsymbols.h b/libwccl/ops/functions/tset/getsymbols.h
new file mode 100644
index 0000000000000000000000000000000000000000..2c492016924d8024d39fb52d8a413de8d43c8272
--- /dev/null
+++ b/libwccl/ops/functions/tset/getsymbols.h
@@ -0,0 +1,56 @@
+#ifndef LIBWCCL_OPS_FUNCTIONS_TSET_GETSYMBOLS_H
+#define LIBWCCL_OPS_FUNCTIONS_TSET_GETSYMBOLS_H
+
+#include <libwccl/values/tset.h>
+#include <libwccl/values/position.h>
+#include <libwccl/ops/function.h>
+
+namespace Wccl {
+
+class GetSymbols : public Function<TSet> {
+public:
+ typedef boost::shared_ptr<Function<Position> > PosFunctionPtr;
+
+ GetSymbols(const PosFunctionPtr& pos_expr, const Corpus2::Tag& mask)
+ : pos_expr_(pos_expr), mask_(mask)
+ {
+ BOOST_ASSERT(pos_expr_);
+ }
+
+ /**
+ * @returns String representation of the function in the form of:
+ * "attribute_name[pos_expr_string]"
+ */
+ std::string to_string(const Corpus2::Tagset& tagset) const;
+
+ /**
+ * @returns String representation of the function in the form of:
+ * "attribute_name[pos_expr_string]"
+ * @note This version does not require tagset, but will be inclomplete
+ * and/or contain internal info.
+ */
+ std::string to_raw_string() const;
+
+ std::string raw_name() const;
+
+ std::string name(const Corpus2::Tagset& tagset) const;
+
+protected:
+ const PosFunctionPtr pos_expr_;
+
+ Corpus2::Tag mask_;
+
+ /**
+ * Gets a position from the argument expression, then gets the
+ * word at that position from the Sentence in the SentenceContext,
+ * then gets the tagset symbols matching the mask of the word.
+ *
+ * @returns A tagset symbol set of the word if position pointed to
+ * lies within boundaries of the Sentence. Empty Tset otherwise.
+ */
+ BaseRetValPtr apply_internal(const FunExecContext& context) const;
+};
+
+} /* end ns Wccl */
+
+#endif // LIBWCCL_OPS_FUNCTIONS_TSET_GETSYMBOLS_H
diff --git a/libwccl/sentencecontext.h b/libwccl/sentencecontext.h
index 6ff68a3351077c988cc6a73d7f2fd357fddee7ca..7451d78c17bacf179c454ae02af0906a1a5028f5 100644
--- a/libwccl/sentencecontext.h
+++ b/libwccl/sentencecontext.h
@@ -75,7 +75,7 @@ public:
/// Token access convenience function - const.
/// Will return NULL if the passed position is not valid in this Sentence
const Corpus2::Token* at(int abs_pos) const {
- if (is_current_inside()) {
+ if (is_inside(abs_pos)) {
return get_sentence()[abs_pos];
} else {
return NULL;
@@ -90,7 +90,7 @@ public:
/// Token access convenience function.
/// Will return NULL if the passed position is not valid in this Sentence
Corpus2::Token* at(int abs_pos) {
- if (is_current_inside()) {
+ if (is_inside(abs_pos)) {
return get_sentence()[abs_pos];
} else {
return NULL;
diff --git a/libwccl/values/tset.h b/libwccl/values/tset.h
index 0378628128aae708f119a46790c0a8c4de790e48..47533f126b335ddc9917daf766d76f5cd1b4c8dc 100644
--- a/libwccl/values/tset.h
+++ b/libwccl/values/tset.h
@@ -61,6 +61,15 @@ public:
return tag_ == other.tag_;
}
+ void combine_with(const Corpus2::Tag& other) {
+ tag_.combine_with(other);
+ }
+
+ void combine_with(const TSet& other) {
+ tag_.combine_with(other.get_value());
+ }
+
+
std::string to_string(const Corpus2::Tagset &) const;
std::string to_raw_string() const;
diff --git a/tests/values.cpp b/tests/values.cpp
index 6f9e5c391c56d3a61c65d43ab741d7748aa0fce8..d5f02c842f5fa3a531c7a2a39a254bbc01cb269c 100644
--- a/tests/values.cpp
+++ b/tests/values.cpp
@@ -123,8 +123,8 @@ BOOST_AUTO_TEST_CASE(position_ops)
v.push_back(minus_one);
v.push_back(minus_two);
- for(int i = 0; i < v.size(); ++i) {
- for(int j = 0; j < v.size(); ++j) {
+ for(size_t i = 0; i < v.size(); ++i) {
+ for(size_t j = 0; j < v.size(); ++j) {
BOOST_CHECK_EQUAL(i == j, v[i].equals(v[j]));
BOOST_CHECK_EQUAL(i == j, v[j].equals(v[i]));
if(i >= 2 && j >= 2) { //nowhere, zero, one, minus_one, minus_two