From a3897913ea0ac318cc9d80c07514942da827e5f4 Mon Sep 17 00:00:00 2001
From: Adam Wardynski <award@.(win7-laptop)>
Date: Tue, 19 Apr 2011 09:15:04 +0200
Subject: [PATCH] Redesign Match to act more like other Value types.
MatchVector, AnnotationMatch, TokenMatch are now MatchData MatchData is
value_type of Match which now has get_value/set_value Match has also copy
construction working properly now Default value for Match holds MatchVector
so adding $m:_M var in grammar works as intended.
---
.../ops/match/conditions/tokencondition.cpp | 2 +-
libwccl/ops/match/matchresult.h | 28 +++++++++
libwccl/parser/grammar.g | 15 ++++-
libwccl/values/annotationmatch.cpp | 7 ---
libwccl/values/annotationmatch.h | 19 +++---
libwccl/values/match.cpp | 5 --
libwccl/values/match.h | 59 +++++++++++++++++--
libwccl/values/matchdata.h | 46 +++++++++++++++
libwccl/values/matchvector.cpp | 28 ++++++---
libwccl/values/matchvector.h | 28 ++++++---
libwccl/values/tokenmatch.cpp | 7 ---
libwccl/values/tokenmatch.h | 19 +++---
tests/match.cpp | 2 +-
13 files changed, 209 insertions(+), 56 deletions(-)
create mode 100644 libwccl/values/matchdata.h
diff --git a/libwccl/ops/match/conditions/tokencondition.cpp b/libwccl/ops/match/conditions/tokencondition.cpp
index e7f670f..533a2ab 100644
--- a/libwccl/ops/match/conditions/tokencondition.cpp
+++ b/libwccl/ops/match/conditions/tokencondition.cpp
@@ -23,7 +23,7 @@ MatchResult TokenCondition::apply(const ActionExecContext& context) const
{
int orig_iter = context.sentence_context().get_position();
if (_predicate->apply(context)->get_value()) {
- boost::shared_ptr<Match> match(new TokenMatch(context.sentence_context().get_position()));
+ boost::shared_ptr<TokenMatch> match(new TokenMatch(context.sentence_context().get_position()));
// increase current sentence position by one after successful token match
context.sentence_context().set_position(orig_iter + 1);
return MatchResult(match);
diff --git a/libwccl/ops/match/matchresult.h b/libwccl/ops/match/matchresult.h
index 84cdea5..a8b0745 100644
--- a/libwccl/ops/match/matchresult.h
+++ b/libwccl/ops/match/matchresult.h
@@ -13,6 +13,10 @@ class MatchResult
{
public:
MatchResult(const boost::shared_ptr<Match>& match);
+ MatchResult(const boost::shared_ptr<MatchData>& match);
+ MatchResult(const boost::shared_ptr<MatchVector>& match);
+ MatchResult(const boost::shared_ptr<TokenMatch>& match);
+ MatchResult(const boost::shared_ptr<AnnotationMatch>& match);
MatchResult();
boost::shared_ptr<Match> get_match() const;
bool matched() const;
@@ -35,6 +39,30 @@ MatchResult::MatchResult(const boost::shared_ptr<Match>& match)
_match(match) {
}
+inline
+MatchResult::MatchResult(const boost::shared_ptr<MatchData>& match)
+ : _matched(true),
+ _match(new Match(match)) {
+}
+
+inline
+MatchResult::MatchResult(const boost::shared_ptr<TokenMatch>& match)
+ : _matched(true),
+ _match(new Match(match)) {
+}
+
+inline
+MatchResult::MatchResult(const boost::shared_ptr<MatchVector>& match)
+ : _matched(true),
+ _match(new Match(match)) {
+}
+
+inline
+MatchResult::MatchResult(const boost::shared_ptr<AnnotationMatch>& match)
+ : _matched(true),
+ _match(new Match(match)) {
+}
+
inline
bool MatchResult::matched() const {
return _matched;
diff --git a/libwccl/parser/grammar.g b/libwccl/parser/grammar.g
index 4892a65..f6c5c4d 100644
--- a/libwccl/parser/grammar.g
+++ b/libwccl/parser/grammar.g
@@ -108,7 +108,7 @@ options {
private:
//
const UnicodeString token_ref_to_ustring(antlr::RefToken& rstr) const {
- return UnicodeString::fromUTF8(((antlr::Token*)rstr)->getText()).unescape();
+ return UnicodeString::fromUTF8(((antlr::Token*)rstr)->getText().c_str()).unescape();
}
/*
const UnicodeString str_token_ref_to_ustring(antlr::RefToken& rstr) const {
@@ -409,6 +409,19 @@ position_value
// Returns boost::shared_ptr<Match>
match_value
returns [boost::shared_ptr<Match> val]
+{
+ boost::shared_ptr<MatchData> m;
+}
+ : m = match_data_value {
+ val.reset(new Match(m));
+ }
+;
+
+// ----------------------------------------------------------------------------
+// Value used into match operator such as TOK[position] and ANN[position, name]
+// Returns boost::shared_ptr<MatchData>
+match_data_value
+ returns [boost::shared_ptr<MatchData> val]
: val = token_match_value
| val = ann_match_value
| val = match_vector_value
diff --git a/libwccl/values/annotationmatch.cpp b/libwccl/values/annotationmatch.cpp
index adca5db..54fb777 100644
--- a/libwccl/values/annotationmatch.cpp
+++ b/libwccl/values/annotationmatch.cpp
@@ -2,18 +2,11 @@
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 boost::shared_ptr<Corpus2::AnnotatedSentence>& s) const
{
size_t fpos = position_.get_value();
diff --git a/libwccl/values/annotationmatch.h b/libwccl/values/annotationmatch.h
index 43c8885..90015fe 100644
--- a/libwccl/values/annotationmatch.h
+++ b/libwccl/values/annotationmatch.h
@@ -1,14 +1,13 @@
#ifndef LIBWCCL_VALUES_ANNOTATIONMATCH_H
#define LIBWCCL_VALUES_ANNOTATIONMATCH_H
-#include <libwccl/values/match.h>
+#include <libwccl/values/matchdata.h>
namespace Wccl {
-class AnnotationMatch : public Match
+class AnnotationMatch : public MatchData
{
public:
- WCCL_VALUE_PREAMBLE
explicit AnnotationMatch(Position position, const std::string channel)
: position_(position), channel_(channel)
@@ -22,20 +21,26 @@ public:
BOOST_ASSERT(position_.get_value() != Position::Nowhere);
}
- /// Match override.
+ /// MatchData override.
bool empty() const {
return false;
}
- /// Match override.
+ /// MatchData override.
Position first_token(const boost::shared_ptr<Corpus2::AnnotatedSentence>& s) const;
- /// Match override.
+ /// MatchData override.
Position last_token(const boost::shared_ptr<Corpus2::AnnotatedSentence>& s) const;
- /// Value override
+ /// MatchData override
std::string to_raw_string() const;
+protected:
+ /// MatchData override
+ AnnotationMatch* clone_internal() const {
+ return new AnnotationMatch(*this);
+ }
+
private:
Position position_;
std::string channel_;
diff --git a/libwccl/values/match.cpp b/libwccl/values/match.cpp
index a35cf8f..a0f5cbc 100644
--- a/libwccl/values/match.cpp
+++ b/libwccl/values/match.cpp
@@ -5,11 +5,6 @@ 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;
diff --git a/libwccl/values/match.h b/libwccl/values/match.h
index 0b05c44..d2ac17f 100644
--- a/libwccl/values/match.h
+++ b/libwccl/values/match.h
@@ -3,6 +3,9 @@
#include <libwccl/values/position.h>
#include <libcorpus2/ann/annotatedsentence.h>
+#include <libwccl/values/matchvector.h>
+#include <libwccl/values/annotationmatch.h>
+#include <libwccl/values/tokenmatch.h>
namespace Wccl {
@@ -11,36 +14,80 @@ class Match : public Value
public:
WCCL_VALUE_PREAMBLE
+ typedef MatchData value_type;
+
+ /**
+ * The default data held is an empty MatchVector
+ */
Match()
+ : match_(new MatchVector())
+ {
+ }
+
+ Match(const boost::shared_ptr<MatchData>& data)
+ : match_(data)
+ {
+ }
+
+ Match(const boost::shared_ptr<TokenMatch>& data)
+ : match_(data)
+ {
+ }
+
+ Match(const boost::shared_ptr<AnnotationMatch>& data)
+ : match_(data)
{
}
+ Match(const boost::shared_ptr<MatchVector>& data)
+ : match_(data)
+ {
+ }
+
+ Match(const MatchData& data)
+ : match_(data.clone())
+ {
+ }
+
+ const MatchData& get_value() const {
+ return *match_;
+ }
+
+ void set_value(const MatchData& m) {
+ match_ = m.clone();
+ }
+
/**
* 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;
+ return match_->empty();
}
/**
* Getter for the first token matched. If the match is empty, must return
* Nowhere.
*/
- virtual Position first_token(const boost::shared_ptr<Corpus2::AnnotatedSentence>&) const {
- return Position(Position::Nowhere);
+ virtual Position first_token(const boost::shared_ptr<Corpus2::AnnotatedSentence>& s) const {
+ return match_->first_token(s);
}
/**
* Getter for the last token matched. If the match is empty, must return
* Nowhere.
*/
- virtual Position last_token(const boost::shared_ptr<Corpus2::AnnotatedSentence>&) const {
- return Position(Position::Nowhere);
+ virtual Position last_token(const boost::shared_ptr<Corpus2::AnnotatedSentence>& s) const {
+ return match_->last_token(s);
}
/// Value override
- std::string to_raw_string() const;
+ std::string to_raw_string() const {
+ return match_->to_raw_string();
+ }
+
+private:
+ boost::shared_ptr<MatchData> match_;
};
} /* end ns Wccl */
diff --git a/libwccl/values/matchdata.h b/libwccl/values/matchdata.h
new file mode 100644
index 0000000..c8bd50d
--- /dev/null
+++ b/libwccl/values/matchdata.h
@@ -0,0 +1,46 @@
+#ifndef LIBWCCL_VALUES_MATCHDATA_H
+#define LIBWCCL_VALUES_MATCHDATA_H
+
+#include <libwccl/values/position.h>
+#include <libcorpus2/ann/annotatedsentence.h>
+
+namespace Wccl {
+
+/**
+ * Base abstract class for data held by a Match Value
+ * - VectorMatch, TokenMatch or AnnotationMatch.
+ * (empty VectorMatch should be default option)
+ */
+class MatchData// : boost::noncopyable
+{
+public:
+
+ /**
+ * 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 = 0;
+ /**
+ * Getter for the first token matched. If the match is empty, must return
+ * Nowhere.
+ */
+ virtual Position first_token(const boost::shared_ptr<Corpus2::AnnotatedSentence>&) const = 0;
+
+ /**
+ * Getter for the last token matched. If the match is empty, must return
+ * Nowhere.
+ */
+ virtual Position last_token(const boost::shared_ptr<Corpus2::AnnotatedSentence>&) const = 0;
+
+ boost::shared_ptr<MatchData> clone() const {
+ return boost::shared_ptr<MatchData>(clone_internal());
+ }
+
+ virtual std::string to_raw_string() const = 0;
+
+protected:
+ virtual MatchData* clone_internal() const = 0;
+};
+
+}
+#endif // LIBWCCL_VALUES_MATCHDATA_H
diff --git a/libwccl/values/matchvector.cpp b/libwccl/values/matchvector.cpp
index faf4459..6ad6096 100644
--- a/libwccl/values/matchvector.cpp
+++ b/libwccl/values/matchvector.cpp
@@ -1,12 +1,11 @@
#include <libwccl/values/matchvector.h>
+#include <libwccl/values/match.h>
#include <libpwrutils/foreach.h>
#include <sstream>
#include <libwccl/exception.h>
namespace Wccl {
-const char* MatchVector::type_name = "MatchVector";
-
std::string MatchVector::to_raw_string() const
{
std::stringstream ss;
@@ -23,11 +22,6 @@ std::string MatchVector::to_raw_string() const
return ss.str();
}
-std::string MatchVector::var_repr(const std::string &var_name)
-{
- return Match::var_repr(var_name);
-}
-
Position MatchVector::first_token(const boost::shared_ptr<Corpus2::AnnotatedSentence>& s) const
{
if (matches_.empty()) {
@@ -75,6 +69,26 @@ void MatchVector::append(const boost::shared_ptr<Match> &m)
matches_.push_back(m);
}
+void MatchVector::append(const boost::shared_ptr<MatchVector> &m)
+{
+ matches_.push_back(boost::shared_ptr<Match>(new Match(m)));
+}
+
+void MatchVector::append(const boost::shared_ptr<TokenMatch> &m)
+{
+ matches_.push_back(boost::shared_ptr<Match>(new Match(m)));
+}
+
+void MatchVector::append(const boost::shared_ptr<AnnotationMatch> &m)
+{
+ matches_.push_back(boost::shared_ptr<Match>(new Match(m)));
+}
+
+void MatchVector::append(const boost::shared_ptr<MatchData> &m)
+{
+ matches_.push_back(boost::shared_ptr<Match>(new Match(m)));
+}
+
const boost::shared_ptr<Match>& MatchVector::submatch(size_t idx)
{
if (idx < matches_.size()) {
diff --git a/libwccl/values/matchvector.h b/libwccl/values/matchvector.h
index d25f799..795ef86 100644
--- a/libwccl/values/matchvector.h
+++ b/libwccl/values/matchvector.h
@@ -1,36 +1,45 @@
#ifndef LIBWCCL_VALUES_MATCHVECTOR_H
#define LIBWCCL_VALUES_MATCHVECTOR_H
-#include <libwccl/values/match.h>
+#include <libwccl/values/matchdata.h>
#include <boost/shared_ptr.hpp>
#include <vector>
namespace Wccl {
-class MatchVector : public Match
+class Match;
+class MatchData;
+class MatchVector;
+class TokenMatch;
+class AnnotationMatch;
+
+class MatchVector : public MatchData
{
public:
- WCCL_VALUE_PREAMBLE
MatchVector()
{
}
- /// Match override. A MatchVector is empty if it contains no sub-matches,
+ /// MatchData override. A MatchVector is empty if it contains no sub-matches,
/// or if they are all empty.
bool empty() const;
- /// Match override.
+ /// MatchData override.
Position first_token(const boost::shared_ptr<Corpus2::AnnotatedSentence>& s) const;
- /// Match override.
+ /// MatchData override.
Position last_token(const boost::shared_ptr<Corpus2::AnnotatedSentence>& s) const;
- /// Value override
+ /// MatchData override
std::string to_raw_string() const;
/// Append a sub-match
void append(const boost::shared_ptr<Match>& m);
+ void append(const boost::shared_ptr<MatchData>& m);
+ void append(const boost::shared_ptr<MatchVector>& m);
+ void append(const boost::shared_ptr<TokenMatch>& m);
+ void append(const boost::shared_ptr<AnnotationMatch>& m);
/// Size (number of direct sub-matches)
size_t size() const {
@@ -54,6 +63,11 @@ public:
matches_.clear();
}
+protected:
+ MatchVector* clone_internal() const {
+ return new MatchVector(*this);
+ }
+
private:
std::vector< boost::shared_ptr<Match> > matches_;
};
diff --git a/libwccl/values/tokenmatch.cpp b/libwccl/values/tokenmatch.cpp
index 1af5e1a..0852ad9 100644
--- a/libwccl/values/tokenmatch.cpp
+++ b/libwccl/values/tokenmatch.cpp
@@ -2,16 +2,9 @@
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
index f220c2d..46e085c 100644
--- a/libwccl/values/tokenmatch.h
+++ b/libwccl/values/tokenmatch.h
@@ -1,14 +1,13 @@
#ifndef LIBWCCL_VALUES_TOKENMATCH_H
#define LIBWCCL_VALUES_TOKENMATCH_H
-#include <libwccl/values/match.h>
+#include <libwccl/values/matchdata.h>
namespace Wccl {
-class TokenMatch : public Match
+class TokenMatch : public MatchData
{
public:
- WCCL_VALUE_PREAMBLE
explicit TokenMatch(Position position)
: position_(position)
@@ -22,24 +21,30 @@ public:
BOOST_ASSERT(position_.get_value() != Position::Nowhere);
}
- /// Match override.
+ /// MatchData override.
bool empty() const {
return false;
}
- /// Match override.
+ /// MatchData override.
Position first_token(const boost::shared_ptr<Corpus2::AnnotatedSentence>&) const {
return position_;
}
- /// Match override.
+ /// MatchData override.
Position last_token(const boost::shared_ptr<Corpus2::AnnotatedSentence>&) const {
return position_;
}
- /// Value override
+ /// MatchData override
std::string to_raw_string() const;
+protected:
+ /// MatchData override
+ TokenMatch* clone_internal() const {
+ return new TokenMatch(*this);
+ }
+
private:
Position position_;
};
diff --git a/tests/match.cpp b/tests/match.cpp
index 15b0790..72ea47d 100644
--- a/tests/match.cpp
+++ b/tests/match.cpp
@@ -104,7 +104,7 @@ BOOST_AUTO_TEST_CASE(varmatch)
{
boost::shared_ptr<Corpus2::AnnotatedSentence> ptr;
Wccl::Variables v;
- v.put<Wccl::Match>("a", new Wccl::TokenMatch(1));
+ v.put<Wccl::Match>("a", new Wccl::Match(boost::shared_ptr<MatchData>(new Wccl::TokenMatch(1))));
BOOST_CHECK_EQUAL(v.get<Wccl::Match>("a")->first_token(ptr).get_value(), 1);
BOOST_CHECK(v.get_put<Wccl::Match>("b")->empty());
BOOST_CHECK_EQUAL(v.get_put<Wccl::Match>("b")->first_token(ptr).get_value(),
--
GitLab