From 9916f226f110a714d3698367b975e2f7bd8e84e2 Mon Sep 17 00:00:00 2001
From: lburdka <lburdka@wojtek-desktop.(none)>
Date: Wed, 2 May 2012 10:37:14 +0200
Subject: [PATCH] Extended corpus2_whole to fit anaphora requirements
---
libcorpus2_whole/CMakeLists.txt | 1 +
libcorpus2_whole/io/relwriter.cpp | 105 ++++++++++++++++++++++++++++++
libcorpus2_whole/io/relwriter.h | 85 ++++++++++++++++++++++++
libcorpus2_whole/relation.cpp | 22 +++++++
libcorpus2_whole/relation.h | 21 ++++--
swig/relation.i | 4 ++
swig/relationwriter.i | 29 +++++++++
7 files changed, 263 insertions(+), 4 deletions(-)
mode change 100644 => 100755 libcorpus2_whole/CMakeLists.txt
create mode 100755 libcorpus2_whole/io/relwriter.cpp
create mode 100755 libcorpus2_whole/io/relwriter.h
mode change 100644 => 100755 libcorpus2_whole/relation.cpp
mode change 100644 => 100755 libcorpus2_whole/relation.h
mode change 100644 => 100755 swig/relation.i
create mode 100755 swig/relationwriter.i
diff --git a/libcorpus2_whole/CMakeLists.txt b/libcorpus2_whole/CMakeLists.txt
old mode 100644
new mode 100755
index 0e07fa9..9606df2
--- a/libcorpus2_whole/CMakeLists.txt
+++ b/libcorpus2_whole/CMakeLists.txt
@@ -11,6 +11,7 @@ SET(libcorpus2_whole_SRC
relation.cpp
io/reader_i.h
io/relreader.cpp
+ io/relwriter.cpp
io/cclrelreader.cpp
io/documentcorpusreader.cpp
io/corpusreader.cpp
diff --git a/libcorpus2_whole/io/relwriter.cpp b/libcorpus2_whole/io/relwriter.cpp
new file mode 100755
index 0000000..6255719
--- /dev/null
+++ b/libcorpus2_whole/io/relwriter.cpp
@@ -0,0 +1,105 @@
+/*
+ Copyright (C) 2010 Tomasz ¦niatowski, Adam Radziszewski, Pawe³ Kêdzia
+ Part of the libcorpus2 project
+
+ This program is free software; you can redistribute it and/or modify it
+under the terms of the GNU General Public License as published by the Free
+Software Foundation; either version 3 of the License, or (at your option)
+any later version.
+
+ This program is distributed in the hope that it will be useful, but
+WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+or FITNESS FOR A PARTICULAR PURPOSE.
+
+ See the LICENSE and COPYING files for more details.
+*/
+
+#include <libpwrutils/foreach.h>
+#include <libcorpus2/exception.h>
+#include <libcorpus2_whole/io/relwriter.h>
+
+#include <fstream>
+#include <sstream>
+#include <boost/make_shared.hpp>
+
+namespace Corpus2 {
+namespace whole {
+
+RelationWriter::RelationWriter(const std::string &rela_path)
+{
+ rela_path_= rela_path;
+}
+
+
+void RelationWriter::write(const std::vector< boost::shared_ptr<Relation> >& relations)
+{
+ file_.reset(new std::ofstream(rela_path_.c_str()));
+
+ //header
+ file_->write("<relations>\n",12);
+
+ foreach (const boost::shared_ptr<Relation>& r, relations){
+ file_->write(" <rel name=\"",14);
+ std::streamsize size = r->name().size();
+ file_->write(r->name().c_str(),size);
+ //Maybe additional info about set is needed in Relation class
+ //file_->write("\" set=\"Anaphora\">\n",18);
+ file_->write("\">\n",3);
+ file_->write(" <from sent=\"",18);
+ size = r->from()->sentence_id().size();
+ file_->write(r->from()->sentence_id().c_str(),size);
+ file_->write("\" chan=\"",8);
+ size = r->from()->channel_name().size();
+ file_->write(r->from()->channel_name().c_str(),size);
+ file_->write("\">",2);
+ int an_num = r->from()->annotation_number();
+ std::ostringstream ss;
+ ss << an_num;
+ std::string str = ss.str();
+ size=str.size();
+ file_->write(str.c_str(),size);
+ file_->write("</from>\n",8);
+ file_->write(" <to sent=\"",16);
+ size = r->to()->sentence_id().size();
+ file_->write(r->to()->sentence_id().c_str(),size);
+ file_->write("\" chan=\"",8);
+ size = r->to()->channel_name().size();
+ file_->write(r->to()->channel_name().c_str(),size);
+ file_->write("\">",2);
+ an_num = r->to()->annotation_number();
+ std::ostringstream ss1;
+ ss1 << an_num;
+ str = ss1.str();
+ size=str.size();
+ file_->write(str.c_str(),size);
+ file_->write("</to>\n",6);
+ //file_->write("\n",1);
+ file_->write(" </rel>",9);
+ file_->write("\n",1);
+ }
+
+
+
+ //footer
+ file_->write("</relations>",12);
+ file_->close();
+}
+
+
+void write_relation(const boost::shared_ptr<Relation>& r)
+{
+ //std::streamsize size = r->from()->sentence_id().size();
+ //f->write(r->from()->sentence_id().c_str(),size);
+
+}
+
+void do_header(){
+}
+
+void do_footer(){
+}
+
+
+} // whole ns
+} // Corpus2 ns
+
diff --git a/libcorpus2_whole/io/relwriter.h b/libcorpus2_whole/io/relwriter.h
new file mode 100755
index 0000000..7ce5469
--- /dev/null
+++ b/libcorpus2_whole/io/relwriter.h
@@ -0,0 +1,85 @@
+/*
+ Copyright (C) 2010 Tomasz ¦niatowski, Adam Radziszewski, Pawe³ Kêdzia
+ Part of the libcorpus2 project
+
+ This program is free software; you can redistribute it and/or modify it
+under the terms of the GNU General Public License as published by the Free
+Software Foundation; either version 3 of the License, or (at your option)
+any later version.
+
+ This program is distributed in the hope that it will be useful, but
+WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+or FITNESS FOR A PARTICULAR PURPOSE.
+
+ See the LICENSE and COPYING files for more details.
+*/
+
+#ifndef LIBCORPUS2_WHOLE_RELWRITER_H
+#define LIBCORPUS2_WHOLE_RELWRITER_H
+
+#include <vector>
+#include <boost/shared_ptr.hpp>
+#include <boost/scoped_ptr.hpp>
+
+#include <libcorpus2_whole/relation.h>
+
+#include <libcorpus2/io/xmlwriter.h>
+
+#include <iostream>
+
+namespace Corpus2 {
+namespace whole {
+
+
+/**
+ * A reader for realtion documents. Note that document is read into memory
+ * before any processing may take place.
+ */
+class RelationWriter {
+public:
+ /**
+ * Reads a document with relations
+ * @param rela_path path to file with relations
+ */
+ RelationWriter(const std::string &rela_path);
+
+ /**
+ * Lazy relations accessor.
+ * If relations are not readed then read relations and returns list of them.
+ * @return List of readed relations
+ */
+ void write(const std::vector< boost::shared_ptr<Relation> >& relations);
+
+
+private:
+
+ void write_relation(const boost::shared_ptr<Relation>& r);
+
+ void do_header();
+
+ void do_footer();
+
+
+ // -------------------------------------------------------------------------
+ /// List of the relations in given relation file
+ std::vector< boost::shared_ptr<Relation> > relations_;
+
+ /// Path to file with relations
+ std::string rela_path_;
+
+ /// File pointer
+ boost::shared_ptr<std::ofstream> file_;
+
+ // -------------------------------------------------------------------------
+ // Temporary information of actual parsing relation
+ std::string rel_name_;
+ std::string ann_number_;
+ boost::shared_ptr<DirectionPoint> rel_from_;
+ boost::shared_ptr<DirectionPoint> rel_to_;
+};
+
+} // whole ns
+} // Corpus2 ns
+
+#endif // LIBCORPUS2_WHOLE_RELREADER_H
+
diff --git a/libcorpus2_whole/relation.cpp b/libcorpus2_whole/relation.cpp
old mode 100644
new mode 100755
index d1df54f..a069a02
--- a/libcorpus2_whole/relation.cpp
+++ b/libcorpus2_whole/relation.cpp
@@ -16,6 +16,7 @@ or FITNESS FOR A PARTICULAR PURPOSE.
#include <boost/make_shared.hpp>
#include <libcorpus2_whole/relation.h>
+#include <stdio.h>
namespace Corpus2 {
namespace whole {
@@ -27,6 +28,27 @@ Relation::Relation(const std::string& name,
{
}
+boost::shared_ptr<Relation> Relation::rel_pt(){
+ relation_=boost::shared_ptr<Relation>(new Relation(name_,from_,to_));
+ return relation_;
+}
+
+void Relation::set_to(const DirectionPoint& dp){
+ boost::shared_ptr<const DirectionPoint> temp(new DirectionPoint(dp.sentence_id(),dp.channel_name(),dp.annotation_number()));
+ to_=temp;
+
+}
+
+void Relation::set_from(const DirectionPoint& dp){
+ boost::shared_ptr<const DirectionPoint> temp(new DirectionPoint(dp.sentence_id(),dp.channel_name(),dp.annotation_number()));
+ from_=temp;
+
+}
+
+void Relation::set_name(const std::string& s){
+ name_=s;
+}
+
Relation::Relation(const std::string& name,
const DirectionPoint& from,
const DirectionPoint& to)
diff --git a/libcorpus2_whole/relation.h b/libcorpus2_whole/relation.h
old mode 100644
new mode 100755
index 8e8ed1d..136f5ca
--- a/libcorpus2_whole/relation.h
+++ b/libcorpus2_whole/relation.h
@@ -94,6 +94,15 @@ public:
~Relation();
+ //Setter of "form" direction point
+ void set_from(const DirectionPoint& dp);
+
+ //Setter of "to" direction point
+ void set_to(const DirectionPoint& dp);
+
+ //Setterf of name
+ void set_name(const std::string& s);
+
/// Accessor to "from" direction point
const boost::shared_ptr<const DirectionPoint>& from() const {
return from_;
@@ -109,13 +118,17 @@ public:
return name_;
}
+ boost::shared_ptr<Relation> rel_pt();
+
private:
/// Direction name
- const std::string name_;
-
+ std::string name_;
+
+ ///Pointer to relation
+ boost::shared_ptr<Relation> relation_;
/// Direction points: from and to
- const boost::shared_ptr<const DirectionPoint> from_;
- const boost::shared_ptr<const DirectionPoint> to_;
+ boost::shared_ptr<const DirectionPoint> from_;
+ boost::shared_ptr<const DirectionPoint> to_;
};
} // whole ns
diff --git a/swig/relation.i b/swig/relation.i
old mode 100644
new mode 100755
index 507e77e..ad4b110
--- a/swig/relation.i
+++ b/swig/relation.i
@@ -40,7 +40,11 @@ namespace whole {
const DirectionPoint& to);
~Relation();
+ void set_from(const DirectionPoint& dp);
+ void set_to(const DirectionPoint& dp);
+ void set_name(const std::string& s);
+ boost::shared_ptr<Relation> rel_pt();
/* It must be renamed because "from" is python keyword */
%rename(rel_from) from() const;
const boost::shared_ptr<const DirectionPoint>& from() const;
diff --git a/swig/relationwriter.i b/swig/relationwriter.i
new file mode 100755
index 0000000..41f23f4
--- /dev/null
+++ b/swig/relationwriter.i
@@ -0,0 +1,29 @@
+#ifndef SWIG_LIBCORPUS2_RELATIONWRITER_I
+#define SWIG_LIBCORPUS2_RELATIONWRITER_I
+
+%module libcorpusrelationwriter
+%{
+ #include <libcorpus2_whole/io/relwriter.h>
+%}
+
+
+namespace Corpus2 {
+namespace whole {
+ class RelationWriter {
+ public:
+ RelationWriter(const std::string &rela_path);
+ void write(const std::vector< boost::shared_ptr<Relation> > relations);
+ private:
+ void do_header();
+ void do_footer();
+ //void write_relation(const boost::shared_ptr<Relation>& r);
+};
+} // whole ns
+} // Corpus2 ns
+
+using namespace std;
+using namespace Corpus2;
+using namespace Corpus2::whole;
+
+#endif /* SWIG_LIBCORPUS2_RELATIONWRITER_I */
+
--
GitLab