From cc8c48adb0bea22be8d82344c8b8420a97054e40 Mon Sep 17 00:00:00 2001
From: Bartosz Broda <bartosz.broda@gmail.com>
Date: Mon, 13 Jun 2011 23:27:58 +0200
Subject: [PATCH] add add_lexicalunit to index, small improvement in creation
of mwe in sax parser
---
libmwereader/mwe.cpp | 56 +++++++++++++++++++++++++++---------
libmwereader/mwe.h | 58 ++++++++++++++++++++++++++++++++------
libmwereader/mweparser.cpp | 16 ++++++++---
3 files changed, 104 insertions(+), 26 deletions(-)
diff --git a/libmwereader/mwe.cpp b/libmwereader/mwe.cpp
index e76aaf4..897c80e 100644
--- a/libmwereader/mwe.cpp
+++ b/libmwereader/mwe.cpp
@@ -4,27 +4,21 @@
namespace Corpus2{
LexicalUnit::LexicalUnit(const std::string &base,
- boost::shared_ptr<Wccl::Operator<Wccl::Bool> > condition,
- boost::shared_ptr<Wccl::Operator<Wccl::Bool> > head_cond,
- std::map<std::string, std::string> variables)
+ LexicalUnit::BoolOpPtr condition,
+ LexicalUnit::BoolOpPtr head_cond,
+ LexicalUnit::strmap variables)
: condition_(condition),
head_cond_(head_cond),
variables_(variables),
base_(base),
nowhere_(Wccl::Position())
{
- // noop
+ for(strmap::iterator iter = variables_.begin();
+ iter != variables_.end(); ++iter)
+ potential_bases_.insert(iter->second);
}
-FixedLU::FixedLU(const std::string &base,
- boost::shared_ptr<Wccl::Operator<Wccl::Bool> > condition,
- boost::shared_ptr<Wccl::Operator<Wccl::Bool> > head_cond,
- std::map<std::string, std::string> variables)
- : LexicalUnit(base, condition, head_cond, variables)
-{
-}
-
-bool FixedLU::IsHere(const Wccl::SentenceContext &sc,
+bool LexicalUnit::IsHere(const Wccl::SentenceContext &sc,
std::set<size_t> &out_position)
{
boost::shared_ptr<const Wccl::Bool> pResult = condition_->apply(sc);
@@ -45,4 +39,40 @@ bool FixedLU::IsHere(const Wccl::SentenceContext &sc,
return true;
}
+
+FixedLU::FixedLU(const std::string &base,
+ boost::shared_ptr<Wccl::Operator<Wccl::Bool> > condition,
+ boost::shared_ptr<Wccl::Operator<Wccl::Bool> > head_cond,
+ std::map<std::string, std::string> variables)
+ : LexicalUnit(base, condition, head_cond, variables)
+{
+}
+
+FlexLU::FlexLU(const std::string &base,
+ boost::shared_ptr<Wccl::Operator<Wccl::Bool> > condition,
+ boost::shared_ptr<Wccl::Operator<Wccl::Bool> > head_cond,
+ std::map<std::string, std::string> variables)
+ : LexicalUnit(base, condition, head_cond, variables)
+{
+}
+
+MWEIndex::MWEIndex()
+{
+
+}
+
+void MWEIndex::add_lexicalunit(LexicalUnitPtr lu)
+{
+ foreach(const std::string& base, lu->get_potential_bases()){
+ value_type::iterator find = index_.find(base);
+ if(find == index_.end()){ // not found -> create new one
+ luvec v;
+ v.push_back(lu);
+ index_.insert( std::make_pair(base, v));
+ }else{// already exists -> add lu
+ (find->second).push_back(lu);
+ }
+ }
+}
+
}//ns Corpus2
diff --git a/libmwereader/mwe.h b/libmwereader/mwe.h
index f62f78f..0676ad0 100644
--- a/libmwereader/mwe.h
+++ b/libmwereader/mwe.h
@@ -1,6 +1,8 @@
#ifndef LIBMWEREADER_MWE_H
#define LIBMWEREADER_MWE_H
+#include <boost/unordered_map.hpp>
+
#include <libcorpus2/io/reader.h>
#include <libwccl/ops/operator.h>
@@ -11,24 +13,48 @@ namespace Corpus2 {
class LexicalUnit
{
public:
- LexicalUnit(const std::string &base,
- boost::shared_ptr<Wccl::Operator<Wccl::Bool> > condition,
- boost::shared_ptr<Wccl::Operator<Wccl::Bool> > head_cond,
- std::map<std::string, std::string> variables
+ typedef std::map<std::string, std::string> strmap;
+ typedef std::set<std::string> strset;
+ typedef boost::shared_ptr<Wccl::Operator<Wccl::Bool> > BoolOpPtr;
+
+ LexicalUnit(const std::string &base, BoolOpPtr condition,
+ BoolOpPtr head_cond, strmap variables
);
+ /**
+ * \param sc SentenceContext with position set to value which
+ * will be checked
+ * \param out_positions will contain absolute position in
+ * SentenceContext (called with sc->get_abs_position) only if
+ * the main condition of this LexicalUnit will return true in current
+ * sentence context
+ * \returns true if this lexical unit was found here
+ */
virtual bool IsHere(const Wccl::SentenceContext& sc,
- std::set<size_t> &out_position) = 0;
+ std::set<size_t> &out_positions) ;
+
+ const std::string & get_base() const{ return base_;}
+ const strmap & get_variables() const{ return variables_;}
+ const strset& get_potential_bases() const{ return potential_bases_;}
+
+
protected:
+
boost::shared_ptr<Wccl::Operator<Wccl::Bool> > condition_;
boost::shared_ptr<Wccl::Operator<Wccl::Bool> > head_cond_;
- std::map<std::string, std::string> variables_;
+ strmap variables_;
std::string base_;
+ strset potential_bases_;
+
const Wccl::Position nowhere_;
};
+typedef boost::shared_ptr<LexicalUnit> LexicalUnitPtr;
+
+
+// TODO: czy bedzie potrzebny podzial na fix/flex w kodzie?
class FixedLU : public LexicalUnit
{
public:
@@ -37,18 +63,32 @@ public:
boost::shared_ptr<Wccl::Operator<Wccl::Bool> > head_cond,
std::map<std::string, std::string> variables
);
- virtual bool IsHere(const Wccl::SentenceContext& sc,
- std::set<size_t> &out_position);
-
};
class FlexLU : public LexicalUnit
{
+public:
+ FlexLU(const std::string &base,
+ boost::shared_ptr<Wccl::Operator<Wccl::Bool> > condition,
+ boost::shared_ptr<Wccl::Operator<Wccl::Bool> > head_cond,
+ std::map<std::string, std::string> variables
+ );
};
class MWEIndex // lub base -> vector<LexicalUnit>
{
+public:
+ MWEIndex();
+
+ void add_lexicalunit(LexicalUnitPtr lu);
+
+protected:
+ typedef std::vector<LexicalUnitPtr> luvec;
+ typedef boost::unordered_map<std::string,luvec> value_type;
+
+
+ value_type index_;
};
}// ns Corpus2
diff --git a/libmwereader/mweparser.cpp b/libmwereader/mweparser.cpp
index 5f4e88d..c8fd91e 100644
--- a/libmwereader/mweparser.cpp
+++ b/libmwereader/mweparser.cpp
@@ -60,10 +60,17 @@ namespace Corpus2 {
MWEBuilder::BoolOpPtr head = mwe_builder_->get_head_condition(
head_cond_);
- //foreach(const std::string&varname, main->valid_variable_names())
- //if(boost::algorithm::starts_with(varname, "Pos"))
- //std::cout << "Pozycja: " << varname << std::endl;
-
+ LexicalUnitPtr lu;
+ if(group_type_ == "fix"){ // group_name_ -> lower case
+ lu = LexicalUnitPtr(new FixedLU(mwe_base_, main, head,
+ variables_));
+ } else if(group_type_ == "flex"){
+ lu = LexicalUnitPtr(new FlexLU(mwe_base_, main, head,
+ variables_));
+ } else {
+ throw Wccl::WcclError("Unknown type of lexical unit:"
+ + group_type_);
+ }
}
std::string MWEParser::get_attribute(const AttributeList& attributes,
@@ -85,6 +92,7 @@ namespace Corpus2 {
group_name_ = a.value;
} else if(a.name == "type"){
group_type_ = a.value;
+ boost::algorithm::to_lower(group_type_);
} else if(a.name == "class"){
group_class_ = a.value;
}
--
GitLab