From 5769ec06ee4b3da2d6134550dbe7b6a3ed0bba61 Mon Sep 17 00:00:00 2001
From: ilor <kailoran@gmail.com>
Date: Thu, 3 Mar 2011 08:48:59 +0100
Subject: [PATCH] First draft of Match value type with some tests. MatchVector
does not yet properly implement first/last token, hence a test fails, and
AnnotationMatch is pretty much a stub.
---
libwccl/CMakeLists.txt | 4 ++
libwccl/values/annotationmatch.cpp | 33 ++++++++++++
libwccl/values/annotationmatch.h | 46 +++++++++++++++++
libwccl/values/match.cpp | 20 ++++++++
libwccl/values/match.h | 47 +++++++++++++++++
libwccl/values/matchvector.cpp | 60 ++++++++++++++++++++++
libwccl/values/matchvector.h | 41 +++++++++++++++
libwccl/values/tokenmatch.cpp | 17 +++++++
libwccl/values/tokenmatch.h | 49 ++++++++++++++++++
libwccl/variables.h | 6 ++-
tests/CMakeLists.txt | 1 +
tests/match.cpp | 82 ++++++++++++++++++++++++++++++
12 files changed, 404 insertions(+), 2 deletions(-)
create mode 100644 libwccl/values/annotationmatch.cpp
create mode 100644 libwccl/values/annotationmatch.h
create mode 100644 libwccl/values/match.cpp
create mode 100644 libwccl/values/match.h
create mode 100644 libwccl/values/matchvector.cpp
create mode 100644 libwccl/values/matchvector.h
create mode 100644 libwccl/values/tokenmatch.cpp
create mode 100644 libwccl/values/tokenmatch.h
create mode 100644 tests/match.cpp
diff --git a/libwccl/CMakeLists.txt b/libwccl/CMakeLists.txt
index f9dc761..f5f5796 100644
--- a/libwccl/CMakeLists.txt
+++ b/libwccl/CMakeLists.txt
@@ -65,9 +65,13 @@ SET(libwccl_STAT_SRC
parser/Parser.cpp
parser/ParserException.cpp
sentencecontext.cpp
+ values/annotationmatch.cpp
values/bool.cpp
+ values/match.cpp
+ values/matchvector.cpp
values/position.cpp
values/strset.cpp
+ values/tokenmatch.cpp
values/tset.cpp
values/value.cpp
variables.cpp
diff --git a/libwccl/values/annotationmatch.cpp b/libwccl/values/annotationmatch.cpp
new file mode 100644
index 0000000..a8419e3
--- /dev/null
+++ b/libwccl/values/annotationmatch.cpp
@@ -0,0 +1,33 @@
+#include <libwccl/values/annotationmatch.h>
+
+namespace Wccl {
+
+const char* AnnotationMatch::type_name = "AnnotationMatch";
+
+std::string AnnotationMatch::to_raw_string() const
+{
+ return "ANN[" + position_.to_raw_string() + "," + channel_ + "]";
+}
+
+std::string AnnotationMatch::var_repr(const std::string &var_name)
+{
+ return Match::var_repr(var_name);
+}
+
+
+Position AnnotationMatch::first_token() const
+{
+ Position p;
+ // TODO
+ return p;
+}
+
+Position AnnotationMatch::last_token() const
+{
+ Position p;
+ // TODO
+ return p;
+}
+
+
+} /* end ns Wccl */
diff --git a/libwccl/values/annotationmatch.h b/libwccl/values/annotationmatch.h
new file mode 100644
index 0000000..e0d85b8
--- /dev/null
+++ b/libwccl/values/annotationmatch.h
@@ -0,0 +1,46 @@
+#ifndef LIBWCCL_VALUES_ANNOTATIONMATCH_H
+#define LIBWCCL_VALUES_ANNOTATIONMATCH_H
+
+#include <libwccl/values/match.h>
+
+namespace Wccl {
+
+class AnnotationMatch : public Match
+{
+public:
+ WCCL_VALUE_PREAMBLE
+
+ explicit AnnotationMatch(Position position, const std::string channel)
+ : position_(position), channel_(channel)
+ {
+ BOOST_ASSERT(position_.get_value() != Position::Nowhere);
+ }
+
+ explicit AnnotationMatch(int pos,const std::string channel)
+ : position_(pos), channel_(channel)
+ {
+ BOOST_ASSERT(position_.get_value() != Position::Nowhere);
+ }
+
+ /// Match override.
+ bool empty() const {
+ return false;
+ }
+
+ /// Match override.
+ Position first_token() const;
+
+ /// Match override.
+ Position last_token() const;
+
+ /// Value override
+ std::string to_raw_string() const;
+
+private:
+ Position position_;
+ std::string channel_;
+};
+
+} /* end ns Wccl */
+
+#endif // LIBWCCL_VALUES_ANNOTATIONMATCH_H
diff --git a/libwccl/values/match.cpp b/libwccl/values/match.cpp
new file mode 100644
index 0000000..a35cf8f
--- /dev/null
+++ b/libwccl/values/match.cpp
@@ -0,0 +1,20 @@
+#include <libwccl/values/match.h>
+#include <sstream>
+
+namespace Wccl {
+
+const char* Match::type_name = "Match";
+
+std::string Match::to_raw_string() const
+{
+ return "MATCH()";
+}
+
+std::string Match::var_repr(const std::string &var_name)
+{
+ std::stringstream ss;
+ ss << "$m:" << var_name;
+ return ss.str();
+}
+
+} /* end ns Wccl */
diff --git a/libwccl/values/match.h b/libwccl/values/match.h
new file mode 100644
index 0000000..2b80b9f
--- /dev/null
+++ b/libwccl/values/match.h
@@ -0,0 +1,47 @@
+#ifndef LIBWCCL_VALUES_MATCH_H
+#define LIBWCCL_VALUES_MATCH_H
+
+#include <libwccl/values/position.h>
+
+namespace Wccl {
+
+class Match : public Value
+{
+public:
+ WCCL_VALUE_PREAMBLE
+
+ Match()
+ {
+ }
+
+ /**
+ * Check if the match is empty (matches nothing). Match objects themselves
+ * are by definition empty, child classes are sometimes or always non-empty.
+ */
+ virtual bool empty() const {
+ return true;
+ }
+
+ /**
+ * Getter for the first token matched. If the match is empty, must return
+ * Nowhere.
+ */
+ virtual Position first_token() const {
+ return Position(Position::Nowhere);
+ }
+
+ /**
+ * Getter for the last token matched. If the match is empty, must return
+ * Nowhere.
+ */
+ virtual Position last_token() const {
+ return Position(Position::Nowhere);
+ }
+
+ /// Value override
+ std::string to_raw_string() const;
+};
+
+} /* end ns Wccl */
+
+#endif // LIBWCCL_VALUES_MATCH_H
diff --git a/libwccl/values/matchvector.cpp b/libwccl/values/matchvector.cpp
new file mode 100644
index 0000000..a0f1b3b
--- /dev/null
+++ b/libwccl/values/matchvector.cpp
@@ -0,0 +1,60 @@
+#include <libwccl/values/matchvector.h>
+#include <libpwrutils/foreach.h>
+#include <sstream>
+
+namespace Wccl {
+
+const char* MatchVector::type_name = "MatchVector";
+
+std::string MatchVector::to_raw_string() const
+{
+ std::stringstream ss;
+ ss << "MATCH(";
+ bool comma = false;
+ foreach (const boost::shared_ptr<Match>& m, matches_) {
+ if (comma) {
+ ss << ",";
+ }
+ ss << m->to_raw_string();
+ comma = true;
+ }
+ ss << ")";
+ return ss.str();
+}
+
+std::string MatchVector::var_repr(const std::string &var_name)
+{
+ return Match::var_repr(var_name);
+}
+
+Position MatchVector::first_token() const
+{
+ Position p;
+ // TODO
+ return p;
+}
+
+Position MatchVector::last_token() const
+{
+ Position p;
+ // TODO
+ return p;
+}
+
+bool MatchVector::empty() const
+{
+ foreach (const boost::shared_ptr<Match>& m, matches_) {
+ if (!m->empty()) {
+ return false;
+ }
+ }
+ return true;
+}
+
+void MatchVector::append(const boost::shared_ptr<Match> &m)
+{
+ matches_.push_back(m);
+}
+
+
+} /* end ns Wccl */
diff --git a/libwccl/values/matchvector.h b/libwccl/values/matchvector.h
new file mode 100644
index 0000000..c4912de
--- /dev/null
+++ b/libwccl/values/matchvector.h
@@ -0,0 +1,41 @@
+#ifndef LIBWCCL_VALUES_MATCHVECTOR_H
+#define LIBWCCL_VALUES_MATCHVECTOR_H
+
+#include <libwccl/values/match.h>
+#include <boost/shared_ptr.hpp>
+#include <vector>
+
+namespace Wccl {
+
+class MatchVector : public Match
+{
+public:
+ WCCL_VALUE_PREAMBLE
+
+ MatchVector()
+ {
+ }
+
+ /// Match override. A MatchVector is empty if it contains no sub-matches,
+ /// or if they are all empty.
+ bool empty() const;
+
+ /// Match override.
+ Position first_token() const;
+
+ /// Match override.
+ Position last_token() const;
+
+ /// Value override
+ std::string to_raw_string() const;
+
+ /// Append a sub-match
+ void append(const boost::shared_ptr<Match>& m);
+
+private:
+ std::vector< boost::shared_ptr<Match> > matches_;
+};
+
+} /* end ns Wccl */
+
+#endif // LIBWCCL_VALUES_MATCHVECTOR_H
diff --git a/libwccl/values/tokenmatch.cpp b/libwccl/values/tokenmatch.cpp
new file mode 100644
index 0000000..1af5e1a
--- /dev/null
+++ b/libwccl/values/tokenmatch.cpp
@@ -0,0 +1,17 @@
+#include <libwccl/values/tokenmatch.h>
+
+namespace Wccl {
+
+const char* TokenMatch::type_name = "TokenMatch";
+
+std::string TokenMatch::to_raw_string() const
+{
+ return "TOK[" + position_.to_raw_string() + "]";
+}
+
+std::string TokenMatch::var_repr(const std::string &var_name)
+{
+ return Match::var_repr(var_name);
+}
+
+} /* end ns Wccl */
diff --git a/libwccl/values/tokenmatch.h b/libwccl/values/tokenmatch.h
new file mode 100644
index 0000000..f7b340d
--- /dev/null
+++ b/libwccl/values/tokenmatch.h
@@ -0,0 +1,49 @@
+#ifndef LIBWCCL_VALUES_TOKENMATCH_H
+#define LIBWCCL_VALUES_TOKENMATCH_H
+
+#include <libwccl/values/match.h>
+
+namespace Wccl {
+
+class TokenMatch : public Match
+{
+public:
+ WCCL_VALUE_PREAMBLE
+
+ explicit TokenMatch(Position position)
+ : position_(position)
+ {
+ BOOST_ASSERT(position_.get_value() != Position::Nowhere);
+ }
+
+ explicit TokenMatch(int pos)
+ : position_(pos)
+ {
+ BOOST_ASSERT(position_.get_value() != Position::Nowhere);
+ }
+
+ /// Match override.
+ bool empty() const {
+ return false;
+ }
+
+ /// Match override.
+ Position first_token() const {
+ return position_;
+ }
+
+ /// Match override.
+ Position last_token() const {
+ return position_;
+ }
+
+ /// Value override
+ std::string to_raw_string() const;
+
+private:
+ Position position_;
+};
+
+} /* end ns Wccl */
+
+#endif // LIBWCCL_VALUES_TOKENMATCH_H
diff --git a/libwccl/variables.h b/libwccl/variables.h
index 35a99d3..d836bab 100644
--- a/libwccl/variables.h
+++ b/libwccl/variables.h
@@ -1,8 +1,9 @@
#ifndef LIBWCCL_VARIABLES_H
#define LIBWCCL_VARIABLES_H
-#include <libwccl/values/bool.h>
#include <libwccl/exception.h>
+#include <libwccl/values/bool.h>
+#include <libwccl/values/match.h>
#include <libwccl/values/position.h>
#include <libwccl/values/strset.h>
#include <libwccl/values/tset.h>
@@ -152,13 +153,14 @@ class Variables : detail::Vmap<Value>
, detail::Vmap<Position>
, detail::Vmap<StrSet>
, detail::Vmap<TSet>
+ , detail::Vmap<Match>
, boost::noncopyable
{
public:
/// Valid value types, should match the inheritance.
/// the type Value must be first, order of other items is not important
typedef boost::mpl::list<Value,
- Bool, Position, StrSet, TSet> types;
+ Bool, Position, StrSet, TSet, Match> types;
/// Constructor, creates an empty instance.
Variables();
diff --git a/tests/CMakeLists.txt b/tests/CMakeLists.txt
index 55e7aa8..5ff0f38 100644
--- a/tests/CMakeLists.txt
+++ b/tests/CMakeLists.txt
@@ -27,6 +27,7 @@ add_executable(tests
getorth.cpp
logicalpredicates.cpp
main.cpp
+ match.cpp
position.cpp
positionpredicates.cpp
regex.cpp
diff --git a/tests/match.cpp b/tests/match.cpp
new file mode 100644
index 0000000..1cbb430
--- /dev/null
+++ b/tests/match.cpp
@@ -0,0 +1,82 @@
+#include <boost/test/unit_test.hpp>
+#include <boost/bind.hpp>
+#include <libcorpus2/tagsetmanager.h>
+
+#include <libwccl/sentencecontext.h>
+#include <libwccl/variables.h>
+
+#include <libwccl/values/tokenmatch.h>
+#include <libwccl/values/annotationmatch.h>
+#include <libwccl/values/matchvector.h>
+
+#include <iostream>
+
+using namespace Wccl;
+
+BOOST_AUTO_TEST_SUITE(match_value)
+
+BOOST_AUTO_TEST_CASE(empty)
+{
+ Wccl::Match m;
+ BOOST_CHECK(m.empty());
+ BOOST_CHECK_EQUAL(m.first_token().get_value(), Wccl::Position::Nowhere);
+ BOOST_CHECK_EQUAL(m.last_token().get_value(), Wccl::Position::Nowhere);
+ BOOST_CHECK_EQUAL(m.to_raw_string(), "MATCH()");
+}
+
+BOOST_AUTO_TEST_CASE(token)
+{
+ Wccl::TokenMatch m(1);
+ BOOST_CHECK(!m.empty());
+ BOOST_CHECK_EQUAL(m.first_token().get_value(), 1);
+ BOOST_CHECK_EQUAL(m.last_token().get_value(), 1);
+ BOOST_CHECK_EQUAL(m.to_raw_string(), "TOK[1]");
+}
+
+BOOST_AUTO_TEST_CASE(vector1)
+{
+ Wccl::MatchVector m;
+ BOOST_CHECK(m.empty());
+ BOOST_CHECK_EQUAL(m.first_token().get_value(), Wccl::Position::Nowhere);
+ BOOST_CHECK_EQUAL(m.last_token().get_value(), Wccl::Position::Nowhere);
+ BOOST_CHECK_EQUAL(m.to_raw_string(), "MATCH()");
+ m.append(boost::make_shared<Wccl::MatchVector>());
+ BOOST_CHECK(m.empty());
+ BOOST_CHECK_EQUAL(m.first_token().get_value(), Wccl::Position::Nowhere);
+ BOOST_CHECK_EQUAL(m.last_token().get_value(), Wccl::Position::Nowhere);
+ BOOST_CHECK_EQUAL(m.to_raw_string(), "MATCH(MATCH())");
+ m.append(boost::make_shared<Wccl::MatchVector>());
+ BOOST_CHECK(m.empty());
+ BOOST_CHECK_EQUAL(m.first_token().get_value(), Wccl::Position::Nowhere);
+ BOOST_CHECK_EQUAL(m.last_token().get_value(), Wccl::Position::Nowhere);
+ BOOST_CHECK_EQUAL(m.to_raw_string(), "MATCH(MATCH(),MATCH())");
+ Wccl::MatchVector m2;
+ m2.append(boost::make_shared<Wccl::MatchVector>(m));
+ BOOST_CHECK(m2.empty());
+ BOOST_CHECK_EQUAL(m2.first_token().get_value(), Wccl::Position::Nowhere);
+ BOOST_CHECK_EQUAL(m2.last_token().get_value(), Wccl::Position::Nowhere);
+ BOOST_CHECK_EQUAL(m2.to_raw_string(), "MATCH(MATCH(MATCH(),MATCH()))");
+ m2.append(boost::make_shared<Wccl::MatchVector>());
+ BOOST_CHECK(m2.empty());
+ BOOST_CHECK_EQUAL(m2.first_token().get_value(), Wccl::Position::Nowhere);
+ BOOST_CHECK_EQUAL(m2.last_token().get_value(), Wccl::Position::Nowhere);
+ BOOST_CHECK_EQUAL(m2.to_raw_string(), "MATCH(MATCH(MATCH(),MATCH()),MATCH())");
+}
+
+BOOST_AUTO_TEST_CASE(vector2)
+{
+ Wccl::MatchVector m;
+ m.append(boost::make_shared<TokenMatch>(5));
+ BOOST_CHECK_EQUAL(m.first_token().get_value(), 5);
+ BOOST_CHECK_EQUAL(m.last_token().get_value(), 5);
+ BOOST_CHECK_EQUAL(m.to_raw_string(), "MATCH(TOK[5])");
+ m.append(boost::make_shared<TokenMatch>(6));
+ BOOST_CHECK_EQUAL(m.first_token().get_value(), 5);
+ BOOST_CHECK_EQUAL(m.last_token().get_value(), 6);
+ BOOST_CHECK_EQUAL(m.to_raw_string(), "MATCH(TOK[5],TOK[6])");
+ m.append(boost::make_shared<TokenMatch>(4));
+ BOOST_CHECK_EQUAL(m.first_token().get_value(), 4);
+ BOOST_CHECK_EQUAL(m.last_token().get_value(), 6);
+ BOOST_CHECK_EQUAL(m.to_raw_string(), "MATCH(TOK[5],TOK[6],TOK[4])");
+}
+BOOST_AUTO_TEST_SUITE_END()
--
GitLab