From cee3b953c92809f06897d6d41e192163b91efef2 Mon Sep 17 00:00:00 2001
From: Bartosz Broda <bartosz.broda@gmail.com>
Date: Mon, 13 Jun 2011 22:29:55 +0200
Subject: [PATCH] add preliminary impl. of IsHere for fixed LU
---
libmwereader/mwe.cpp | 51 +++++++++++++++++++++++++++++++-------
libmwereader/mwe.h | 14 +++++++++++
libmwereader/mweparser.cpp | 8 +++---
3 files changed, 59 insertions(+), 14 deletions(-)
diff --git a/libmwereader/mwe.cpp b/libmwereader/mwe.cpp
index 046b644..e76aaf4 100644
--- a/libmwereader/mwe.cpp
+++ b/libmwereader/mwe.cpp
@@ -1,15 +1,48 @@
#include "mwe.h"
+#include <boost/algorithm/string.hpp>
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)
- : condition_(condition),
- head_cond_(head_cond),
- variables_(variables),
- base_(base)
- {
+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)
+ : condition_(condition),
+ head_cond_(head_cond),
+ variables_(variables),
+ base_(base),
+ nowhere_(Wccl::Position())
+{
+ // noop
+}
+
+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,
+ std::set<size_t> &out_position)
+{
+ boost::shared_ptr<const Wccl::Bool> pResult = condition_->apply(sc);
+ if(pResult->get_value() == false)
+ return false;
+
+ foreach(const std::string&varname, condition_->valid_variable_names()){
+ if(boost::algorithm::starts_with(varname, "Pos")){
+ Wccl::Position pos = condition_->get<Wccl::Position>(varname);
+ if(pos.equals(nowhere_)){
+ std::string errmsg("Position for found MWE cannot be zero.");
+ errmsg += " Offending unit: " + base_;
+ throw Wccl::WcclError(errmsg);
+ }
+ out_position.insert( sc.get_abs_position(pos) );
+ }
}
+ return true;
+}
+
}//ns Corpus2
diff --git a/libmwereader/mwe.h b/libmwereader/mwe.h
index 7508b6b..f62f78f 100644
--- a/libmwereader/mwe.h
+++ b/libmwereader/mwe.h
@@ -16,15 +16,29 @@ 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) = 0;
+
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_;
std::string base_;
+
+ const Wccl::Position nowhere_;
};
class FixedLU : public LexicalUnit
{
+public:
+ 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
+ );
+ virtual bool IsHere(const Wccl::SentenceContext& sc,
+ std::set<size_t> &out_position);
};
diff --git a/libmwereader/mweparser.cpp b/libmwereader/mweparser.cpp
index e0f1c5f..5f4e88d 100644
--- a/libmwereader/mweparser.cpp
+++ b/libmwereader/mweparser.cpp
@@ -60,11 +60,9 @@ namespace Corpus2 {
MWEBuilder::BoolOpPtr head = mwe_builder_->get_head_condition(
head_cond_);
- //const Corpus2::Tagset& tagset = Corpus2::get_named_tagset(tagset_);
-
- foreach(const std::string&varname, main->valid_variable_names())
- if(boost::algorithm::starts_with(varname, "Pos"))
- std::cout << "Pozycja: " << varname << std::endl;
+ //foreach(const std::string&varname, main->valid_variable_names())
+ //if(boost::algorithm::starts_with(varname, "Pos"))
+ //std::cout << "Pozycja: " << varname << std::endl;
}
--
GitLab