Added module libcorpus2_whole

libcorpus2/CMakeLists.txt

@@ -41,10 +41,8 @@ SET(libcorpus2_STAT_SRC
 	ann/iob.cpp 
 	ann/view.cpp
 	chunk.cpp
-	document.cpp
 	exception.cpp
 	lexeme.cpp
-	relation.cpp
 	sentence.cpp
 	tag.cpp
 	tagging.cpp
@@ -55,7 +53,6 @@ SET(libcorpus2_STAT_SRC
 	tokenmetadata.cpp
 	io/cclreader.cpp
 	io/cclwriter.cpp
-	io/docreader.cpp
 	io/helpers.cpp
 	io/fastxces.cpp
 	io/iob-chan.cpp
@@ -66,7 +63,6 @@ SET(libcorpus2_STAT_SRC
 	io/plainwriter.cpp
 	io/premorphwriter.cpp
 	io/reader.cpp
-	io/relreader.cpp
 	io/rft.cpp
 	io/sax.cpp
 	io/statwriter.cpp

libcorpus2_whole/CMakeLists.txt

+PROJECT(corpus2_whole)
+
+find_package(LibXML++ REQUIRED QUIET)
+include_directories(${LibXML++_INCLUDE_DIRS})
+link_directories(${LibXML++_LIBRARY_DIRS})
+set(LIBS ${LIBS} ${LibXML++_LIBRARIES})
+
+SET(libcorpus2_whole_SRC
+	corpus.cpp
+	corpusreader.cpp
+	docreader.cpp
+	document.cpp
+	poliqarpdocumentreader.cpp
+	relation.cpp
+	relreader.cpp
+	docreaderi.h
+)
+
+file(GLOB_RECURSE INCS "*.h")
+
+if(WIN32)
+	add_library(corpus2_whole STATIC ${libcorpus2_whole_SRC} ${INCS})
+else(WIN32)
+	add_library(corpus2_whole SHARED ${libcorpus2_whole_SRC} ${INCS})
+endif(WIN32)
+target_link_libraries(corpus2_whole corpus2 ${LIBS})
+
+if(UNIX)
+	install(TARGETS corpus2_whole
+		LIBRARY DESTINATION lib)
+
+	install(
+		DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}
+		DESTINATION include
+		FILES_MATCHING PATTERN "*.h"
+		PATTERN ".svn" EXCLUDE
+		PATTERN "bin" EXCLUDE
+		PATTERN "build" EXCLUDE
+		PATTERN "CMake*" EXCLUDE
+	)
+	install(
+		DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}/include/
+		DESTINATION include
+		FILES_MATCHING PATTERN "version.h"
+		PATTERN "config_d.h"
+	)
+	install(
+		DIRECTORY ${LIBCORPUS2_SRC_DATA_DIR}/
+		DESTINATION ${LIBCORPUS2_INSTALL_DATA_DIR}
+		FILES_MATCHING PATTERN "*.ini"
+		PATTERN "*.tagset"
+		PATTERN ".svn" EXCLUDE
+		PATTERN "local/*" EXCLUDE
+	)
+endif(UNIX)
+

libcorpus2_whole/corpus.cpp

+/*
+	Copyright (C) 2010 Tomasz Śniatowski, Adam Radziszewski, Paweł Kędzia
+	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 <libcorpus2_whole/corpus.h>
+
+namespace Corpus2 {
+namespace whole{
+
+Corpus::Corpus(const std::string name) : name_(name), documents_()
+{
+	this->current_document_ = documents_.begin();
+}
+
+} // whole ns
+} // Corpus2 ns

libcorpus2_whole/corpus.h

+/*
+	Copyright (C) 2010 Tomasz Śniatowski, Adam Radziszewski, Paweł Kędzia
+	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_CORPUS_H
+#define LIBCORPUS2_WHOLE_CORPUS_H
+
+#include <vector>
+#include <boost/shared_ptr.hpp>
+#include <libcorpus2_whole/document.h>
+
+namespace Corpus2 {
+namespace whole {
+
+/**
+ * Represents Corpus
+ */
+class Corpus
+{
+public:
+	/**
+	 * Corpus constructor takes one parameter:
+	 * @arg name Name of the corpus - default is empty
+	 */
+	Corpus(const std::string name = "");
+
+	/**
+	 * Adds one (readed) document to corpus
+	 */
+	void add_document(boost::shared_ptr<Document> document) {
+		this->documents_.push_back(document);
+	}
+
+	/// Returns list of the documents from corpus
+	const std::vector<boost::shared_ptr<Document> > documents() const {
+		return this->documents_;
+	}
+
+	/// Next document in corpus
+	/// @todo I don't know if it'll be working... It should be tested!
+	boost::shared_ptr<Document> next_document() {
+		return *(current_document_++);
+	}
+
+private:
+	/// Corpus name
+	const std::string name_;
+
+	/// List of the documents in corpus
+	std::vector<boost::shared_ptr<Document> > documents_;
+
+	/// Current document
+	std::vector<boost::shared_ptr<Document> >::iterator current_document_;
+};
+
+} // whole ns
+} // Corpus2 ns
+
+#endif // LIBCORPUS2_WHOLE_CORPUS_H

libcorpus2_whole/corpusreader.cpp

+#include <fstream>
+#include <boost/algorithm/string.hpp>
+
+#include <libcorpus2/exception.h>
+#include <libcorpus2_whole/corpusreader.h>
+
+namespace Corpus2 {
+namespace whole {
+
+CorpusReader::CorpusReader(const Tagset& tagset, const std::string& corpus_type)
+	: tagset_(tagset), corpus_type_(corpus_type)
+{
+	//
+}
+
+boost::shared_ptr<Corpus> CorpusReader::read(const std::string& corpus_file_path)
+{
+	std::string line;
+	std::ifstream corpus_file(corpus_file_path.c_str());
+	if (!corpus_file) {
+		throw Corpus2Error(corpus_file_path + " file not found!");
+	}
+
+	boost::shared_ptr<Corpus> corpus = boost::make_shared<Corpus>(corpus_file_path);
+
+	while(getline(corpus_file, line)) {
+		boost::shared_ptr<DocumentReaderI> reader;
+		std::string ann_path, rel_path;
+		// split line by semicolon
+		std::vector<std::string> splitted_line;
+		boost::split(splitted_line, line, boost::is_any_of(";"));
+		if (splitted_line.empty()) {
+			continue;
+		}
+		else if (splitted_line.size() == 1) {
+			ann_path = splitted_line[0];
+			rel_path = "";
+		}
+		else {
+			ann_path = splitted_line[0];
+			rel_path = splitted_line[1];
+		}
+		reader = this->get_reader_by_type(this->corpus_type_, ann_path, rel_path);
+		boost::shared_ptr<Document> document = reader->read();
+		corpus->add_document(document);
+	}
+
+	return corpus;
+}
+
+//
+boost::shared_ptr<DocumentReaderI> CorpusReader::get_reader_by_type(
+	const std::string &type,
+	const std::string &ann_path,
+	const std::string &rel_path)
+{
+	/*if (type == "poliqarp") {
+		static boost::shared_ptr<PoliqarpDocumentReader> pq_doc_reader;
+		if (!pq_doc_reader) {
+			pq_doc_reader = boost::shared_ptr<PoliqarpDocumentReader>(
+					new PoliqarpDocumentReader(this->tagset_, ann_path));
+		}
+		return pq_doc_reader;
+	} else */if (type == "document") {
+		return boost::shared_ptr<DocumentReader>(
+				new DocumentReader(this->tagset_, ann_path, rel_path));
+	}
+	throw Corpus2Error(type + " is unknown reader type!");
+}
+
+} // whole ns
+} // Corpus2 ns

libcorpus2_whole/corpusreader.h

+#ifndef LIBCORPUS2_WHOLE_CORPUSREADER_H
+#define LIBCORPUS2_WHOLE_CORPUSREADER_H
+
+#include <string>
+#include <boost/shared_ptr.hpp>
+
+#include <libcorpus2_whole/docreaderi.h>
+#include <libcorpus2_whole/corpus.h>
+#include <libcorpus2_whole/docreader.h>
+
+namespace Corpus2 {
+namespace whole {
+
+/**
+ * CorpusReader is a corpus-like reader
+ */
+class CorpusReader
+{
+public:
+	/**
+	 * @arg corpus_type may be:
+	 *  - document (contains relations)
+	 *  - poliqarp
+	 */
+	CorpusReader(const Tagset& tagset, const std::string& corpus_type);
+
+	/**
+	 * Reads corpus from given path
+	 * @arg corpus_file Path to file contains paths to corpus files.
+	 *      Depend on corpus type, each line in this file should contains only
+	 *      path to one document from corpus or path to (in particular DocReader)
+	 *      relations and annotatons (in one line, first is path to annotations
+	 *      and second are relations -- these paths, should be separated by semicolon)
+	 * @return Readed corpus
+	 */
+	boost::shared_ptr<Corpus> read(const std::string& corpus_file);
+
+protected:
+	/// Tagset to use, sets only onece in constructor
+	const Tagset& tagset_;
+
+	/// Type of corpus, sets only once in constructor
+	const std::string& corpus_type_;
+
+private:
+	/// Returns reader based on corpus type (poliqarp/document)
+	boost::shared_ptr<DocumentReaderI> get_reader_by_type(
+		const std::string &type,
+		const std::string &ann_path,
+		const std::string &rel_path = "");
+};
+
+} // whole ns
+} // Corpus2 ns
+
+#endif // LIBCORPUS2_WHOLE_CORPUSREADER_H

libcorpus2/io/docreader.cpp → libcorpus2_whole/docreader.cpp

@@ -15,11 +15,12 @@ or FITNESS FOR A PARTICULAR PURPOSE.
 */
 
 #include <boost/make_shared.hpp>
-#include <libcorpus2/io/docreader.h>
+#include <libcorpus2_whole/docreader.h>
 
 namespace Corpus2 {
 	DocumentReader::DocumentReader(const Tagset& tagset,
 		const std::string &annot_path, const std::string &rela_path)
+			: DocumentReaderI("document")
 	{
 		make_readers(tagset, annot_path, rela_path);
 	}

libcorpus2/io/docreader.h → libcorpus2_whole/docreader.h

@@ -14,13 +14,14 @@ or FITNESS FOR A PARTICULAR PURPOSE.
 	See the LICENSE and COPYING files for more details.
 */
 
-#ifndef LIBCORPUS2_DOCREADER_H
-#define LIBCORPUS2_DOCREADER_H
+#ifndef LIBCORPUS2_WHOLE__DOCREADER_H
+#define LIBCORPUS2_WHOLE__DOCREADER_H
 
-#include <libcorpus2/relation.h>
-#include <libcorpus2/document.h>
+#include <libcorpus2_whole/relation.h>
+#include <libcorpus2_whole/document.h>
+#include <libcorpus2_whole/docreaderi.h>
 #include <libcorpus2/io/cclreader.h>
-#include <libcorpus2/io/relreader.h>
+#include <libcorpus2_whole/relreader.h>
 
 #include <boost/shared_ptr.hpp>
 
@@ -30,7 +31,7 @@ namespace Corpus2 {
  * A reader for whole documents. Note that a whole document is read into memory
  * before any processing may take place.
  */
-class DocumentReader {
+class DocumentReader : public DocumentReaderI {
 public:
 	/**
 	 * Reads a whole document, using the two given path: the morphosyntax and
@@ -85,4 +86,4 @@ private:
 };
 } /* end ns Corpus2 */
 
-#endif // LIBCORPUS2_DOCREADER_H
+#endif // LIBCORPUS2_WHOLE_DOCREADER_H

libcorpus2_whole/docreaderi.h

+#ifndef READERI_H
+#define READERI_H
+
+#include <string>
+
+#include <libcorpus2_whole/document.h>
+#include <boost/shared_ptr.hpp>
+
+namespace Corpus2 {
+
+/**
+ * Reader interface,
+ * Contains method to get reader type (such as document, poliqarp).
+ * Method for reading read must by implemented!
+ */
+class DocumentReaderI
+{
+public:
+	DocumentReaderI(const std::string type) : type_(type) {}
+
+	const std::string& type() const { return type_; }
+
+	virtual boost::shared_ptr<Document> read() = 0;
+
+protected:
+	const std::string type_;
+};
+
+}
+
+#endif // READERI_H

libcorpus2/document.cpp → libcorpus2_whole/document.cpp

@@ -14,7 +14,7 @@ or FITNESS FOR A PARTICULAR PURPOSE.
 	See the LICENSE and COPYING files for more details.
 */
 
-#include <libcorpus2/document.h>
+#include <libcorpus2_whole/document.h>
 #include <boost/make_shared.hpp>
 
 namespace Corpus2 {

libcorpus2/document.h → libcorpus2_whole/document.h

 /*
-	Copyright (C) 2010 Tomasz Śniatowski, Adam Radziszewski
+	Copyright (C) 2010 Tomasz Śniatowski, Adam Radziszewski, Paweł Kędzia
 	Part of the libcorpus2 project
 
 	This program is free software; you can redistribute it and/or modify it
@@ -14,11 +14,11 @@ or FITNESS FOR A PARTICULAR PURPOSE.
 	See the LICENSE and COPYING files for more details.
 */
 
-#ifndef LIBCORPUS2_DOCUMENT_H
-#define LIBCORPUS2_DOCUMENT_H
+#ifndef LIBCORPUS2_WHOLE_DOCUMENT_H
+#define LIBCORPUS2_WHOLE_DOCUMENT_H
 
 #include <libcorpus2/chunk.h>
-#include <libcorpus2/relation.h>
+#include <libcorpus2_whole/relation.h>
 #include <boost/shared_ptr.hpp>
 
 namespace Corpus2 {
@@ -65,4 +65,4 @@ protected:
 
 } /* end ns Corpus2 */
 
-#endif // LIBCORPUS2_DOCUMENT_H
+#endif // LIBCORPUS2_WHOLE_DOCUMENT_H

libcorpus2_whole/poliqarpdocumentreader.cpp

+#include <libcorpus2_whole/poliqarpdocumentreader.h>
+
+namespace Corpus2 {
+namespace whole {
+
+PoliqarpDocumentReader::PoliqarpDocumentReader(const Tagset& tagset, const std::string& corpus_path)
+	: DocumentReaderI("poliqarp"), corpus_path_(corpus_path)
+{
+	this->pqr_ = boost::shared_ptr<PoliqarpReader> (new PoliqarpReader(tagset, corpus_path));
+}
+
+boost::shared_ptr<Document> PoliqarpDocumentReader::read()
+{
+	boost::shared_ptr<Document> document = boost::make_shared<Document>();
+	// boost::shared_ptr<Chunk> chunk = this->pqr_->pq_->get_next_sentence();
+//	if (chunk) {
+//		document->add_paragraph(chunk);
+//	}
+	return document;
+}
+
+} // whole ns
+} // Corpus2 ns

libcorpus2_whole/poliqarpdocumentreader.h

+#ifndef LIBCORPUS2_WHOLE_POLIQARPDOCUMENTREADER_H
+#define LIBCORPUS2_WHOLE_POLIQARPDOCUMENTREADER_H
+
+#include <poliqarp/pqreader.h>
+#include <libcorpus2_whole/docreaderi.h>
+#include <libcorpus2_whole/document.h>
+
+namespace Corpus2 {
+namespace whole {
+
+class PoliqarpDocumentReader : public DocumentReaderI
+{
+public:
+	PoliqarpDocumentReader(const Tagset& tagset, const std::string& corpus_path);
+
+	/// semantic of this methd is like get_next_document
+	boost::shared_ptr<Document> read();
+
+private:
+	const std::string corpus_path_;
+	boost::shared_ptr<PoliqarpReader> pqr_;
+};
+
+} // whole ns
+} // Corpus2 ns
+
+#endif // LIBCORPUS2_WHOLE_POLIQARPDOCUMENTREADER_H

libcorpus2/relation.cpp → libcorpus2_whole/relation.cpp

@@ -15,7 +15,7 @@ or FITNESS FOR A PARTICULAR PURPOSE.
 */
 
 #include <boost/make_shared.hpp>
-#include <libcorpus2/relation.h>
+#include <libcorpus2_whole/relation.h>
 
 namespace Corpus2 {
 

libcorpus2/io/relreader.cpp → libcorpus2_whole/relreader.cpp

@@ -16,7 +16,7 @@ or FITNESS FOR A PARTICULAR PURPOSE.
 
 #include <libpwrutils/foreach.h>
 #include <libcorpus2/exception.h>
-#include <libcorpus2/io/relreader.h>
+#include <libcorpus2_whole/relreader.h>
 
 #include <fstream>
 #include <boost/make_shared.hpp>

libcorpus2/io/relreader.h → libcorpus2_whole/relreader.h

@@ -14,15 +14,15 @@ or FITNESS FOR A PARTICULAR PURPOSE.
 	See the LICENSE and COPYING files for more details.
 */
 
-#ifndef LIBCORPUS2_RELREADER_H
-#define LIBCORPUS2_RELREADER_H
+#ifndef LIBCORPUS2_WHOLE_RELREADER_H
+#define LIBCORPUS2_WHOLE_RELREADER_H
 
 #include <vector>
 #include <boost/shared_ptr.hpp>
 #include <boost/scoped_ptr.hpp>
 #include <libxml++/parsers/saxparser.h>
 
-#include <libcorpus2/relation.h>
+#include <libcorpus2_whole/relation.h>
 
 #include <iostream>
 
@@ -120,4 +120,4 @@ private:
 };
 } /* end ns Corpus2 */
 
-#endif // LIBCORPUS2_RELREADER_H
+#endif // LIBCORPUS2_WHOLE_RELREADER_H

poliqarp/CMakeLists.txt

-PROJECT(Corpus2Poliqarp)
+PROJECT(corpus2_poliqarpreader)
 cmake_minimum_required(VERSION 2.8.0)
 set(c2pq_ver_major 1)
 set(c2pq_ver_minor 0)

swig/CMakeLists.txt

@@ -2,7 +2,7 @@
 
 PROJECT(corpus2SwigWrap)
 
-set(LIBS "corpus2" "pwrutils")
+set(LIBS "corpus2" "corpus2_whole" "corpus2_poliqarpreader" "pwrutils")
 
 include_directories (${corpus2_SOURCE_DIR})
 include_directories (${pwrutils_SOURCE_DIR})

swig/document.i

@@ -3,7 +3,7 @@
 
 %module libcorpusdocument
 %{
-  #include <libcorpus2/document.h>
+  #include <libcorpus2_whole/document.h>
 %}
 
 %include "std_defs.i"