LZ77 compression and decompression

alib2algo/src/string/compress/LZ77.cpp

+/*
+ * LZ77.cpp
+ *
+ *  Created on: 1. 3. 2017
+ *      Author: Jan Parma
+ */
+
+#include "LZ77.h"
+
+namespace string {
+
+namespace compress {
+
+std::vector < std::tuple < unsigned int, unsigned int, DefaultSymbolType > > LZ77::compress ( const string::String & string, unsigned int searchBufferSize, unsigned int lookaheadBufferSize ) {
+	return dispatch ( string.getData ( ), searchBufferSize, lookaheadBufferSize );
+}
+
+auto LZ77Compress = LZ77::RegistratorWrapper < std::vector < std::tuple < unsigned int, unsigned int, DefaultSymbolType > >, string::LinearString < > > ( LZ77::compress );
+
+} /* namespace compress */
+
+} /* namespace string */

alib2algo/src/string/compress/LZ77.h

+/*
+ * LZ77.h
+ *
+ *  Created on: 1. 3. 2017
+ *      Author: Jan Parma
+ */
+
+#ifndef _LZ77_H_
+#define _LZ77_H_
+
+#include <core/multipleDispatch.hpp>
+#include <string/StringFeatures.h>
+#include <tuple>
+#include <vector>
+
+#include <string/LinearString.h>
+#include <string/String.h>
+
+#include <exception/CommonException.h>
+
+namespace string {
+
+namespace compress {
+
+class LZ77 : public std::SingleDispatch < LZ77, std::vector < std::tuple < unsigned int, unsigned int, DefaultSymbolType > >, const string::StringBase &, unsigned int, unsigned int > {
+public:
+	static std::vector < std::tuple < unsigned int, unsigned int, DefaultSymbolType > > compress ( const string::String & string, unsigned int searchBufferSize, unsigned int lookaheadBufferSize );
+
+	template < class SymbolType >
+	static std::vector < std::tuple < unsigned int, unsigned int, SymbolType > > compress ( const string::LinearString < SymbolType > & string, unsigned int searchBufferSize, unsigned int lookaheadBufferSize );
+
+private:
+	template < class SymbolType >
+	static int equal ( const string::LinearString < SymbolType > & string, unsigned int first1, unsigned int last1, unsigned int first2 );
+};
+
+// Main method that handle compress
+template < class SymbolType >
+std::vector < std::tuple < unsigned int, unsigned int, SymbolType > > LZ77::compress ( const string::LinearString < SymbolType > & string, unsigned int searchBufferSize, unsigned int lookaheadBufferSize ) {
+
+	if(searchBufferSize == 0)
+		throw exception::CommonException("LZ77: search buffer size must be greater than 0");
+
+	if(lookaheadBufferSize == 0)
+		throw exception::CommonException("LZ77: lookahead buffer size must be greater than 0");
+
+
+	size_t pointer = 0;
+	unsigned int match = 0;
+	unsigned int maxMatch = 0;
+	unsigned int sbPointer = 0;
+
+	std::vector < std::tuple < unsigned int, unsigned int, SymbolType > > output;
+
+	while ( pointer < string.getContent ( ).size ( ) ) {
+		if ( pointer + lookaheadBufferSize >= string.getContent ( ).size ( ) )
+			lookaheadBufferSize = string.getContent ( ).size ( ) - pointer - 1;
+
+		for ( int j = pointer - 1; j > ( int ) pointer - 1 - ( int ) searchBufferSize; j-- ) {
+			if ( j < 0 ) break;
+
+			match = equal ( string, pointer, pointer + lookaheadBufferSize - 1, j );
+
+			if ( maxMatch < match ) {
+				maxMatch = match;
+				sbPointer = j;
+			}
+		}
+
+		std::tuple < unsigned int, unsigned int, SymbolType > triple ( maxMatch == 0 ? 0 : pointer - sbPointer, maxMatch, string.getContent ( )[pointer + maxMatch] );
+		output.push_back ( triple );
+
+		pointer = pointer + maxMatch + 1;
+
+		maxMatch = 0;
+	}
+
+	return output;
+}
+
+// Method that return longest match of two substrings that are defined as incoming string and index of first staring letter
+template < class SymbolType >
+int LZ77::equal ( const string::LinearString < SymbolType > & string, unsigned int first1, unsigned int last1, unsigned int first2 ) {
+
+	int steps = 0;
+
+	while ( first1 <= last1 ) {
+		if ( string.getContent ( )[first1] != string.getContent ( )[first2] )
+			return steps;
+
+		first1++;
+		first2++;
+		steps++;
+	}
+
+	return steps;
+}
+
+} /* namespace compress */
+
+} /* namespace string */
+
+#endif /* _LZ77_H_ */

alib2algo/src/string/compress/LZ77Decompress.cpp

+/*
+ * LZ77Decompress.cpp
+ *
+ *  Created on: 1. 3. 2017
+ *      Author: Jan Parma
+ */
+
+#include "LZ77Decompress.h"
+
+namespace string {
+
+namespace compress {
+
+string::String LZ77Decompress::decompress ( const std::vector < std::tuple < unsigned int, unsigned int, DefaultSymbolType > > & input ) {
+	return dispatch ( input );
+}
+
+auto LZ77Decompress = LZ77Decompress::RegistratorWrapper < string::LinearString < >, std::vector < std::tuple < unsigned int, unsigned int, DefaultSymbolType > > > ( LZ77Decompress::decompress );
+
+} /* namespace compress */
+
+} /* namespace string */

alib2algo/src/string/compress/LZ77Decompress.h

+/*
+ * LZ77Decompress.h
+ *
+ *  Created on: 1. 3. 2017
+ *      Author: Jan Parma
+ */
+
+#ifndef _LZ77DECOMPRESS_H_
+#define _LZ77DECOMPRESS_H_
+
+#define SEARCH_BUFFER_LENGTH	   6
+#define LOOKAHEAD_BUFFER_LENGTH	   4
+
+#include <core/multipleDispatch.hpp>
+#include <string/StringFeatures.h>
+#include <tuple>
+#include <vector>
+
+#include <string/LinearString.h>
+#include <string/String.h>
+
+namespace string {
+
+namespace compress {
+
+class LZ77Decompress : public std::SingleDispatch < LZ77Decompress, string::String, const std::vector < std::tuple < unsigned int, unsigned int, DefaultSymbolType > > & > {
+public:
+	static string::String decompress ( const std::vector < std::tuple < unsigned int, unsigned int, DefaultSymbolType > > & input );
+
+	template < class SymbolType >
+	static string::LinearString < SymbolType > decompress ( const std::vector < std::tuple < unsigned int, unsigned int, SymbolType > > & input );
+};
+
+// Main method that handle decompress
+template < class SymbolType >
+string::LinearString < SymbolType > LZ77Decompress::decompress ( const std::vector < std::tuple < unsigned int, unsigned int, SymbolType > > & input ) {
+
+	string::LinearString < SymbolType > output;
+
+	for ( unsigned int i = 0; i < input.size ( ); i++ )
+	{
+		if ( ( std::get < 0 > ( input[i] ) == 0 ) && ( std::get < 1 > ( input[i] ) == 0 ) )
+		{
+			output.extendAlphabet ( std::set < SymbolType > { std::get < 2 > ( input[i] ) } );
+			output.appendSymbol ( std::get < 2 > ( input[i] ) );
+		} 
+		else 
+		{
+			for ( unsigned int j = 0; j < std::get < 1 > ( input[i] ); j++ )
+			{
+				output.extendAlphabet ( std::set < SymbolType > { output.getContent ( )[output.getContent ( ).size ( ) - std::get < 0 > ( input[i] )] } );
+				output.appendSymbol ( output.getContent ( )[output.getContent ( ).size ( ) - std::get < 0 > ( input[i] )] );
+			}
+
+			output.extendAlphabet ( std::set < SymbolType > { std::get < 2 > ( input[i] ) } );
+			output.appendSymbol ( std::get < 2 > ( input[i] ) );
+		}
+	}
+
+	return output;
+}
+
+} /* namespace compress */
+
+} /* namespace string */
+
+#endif /* _LZ77DECOMPRESS_H_ */

alib2algo/test-src/string/compress/LZ77DecompressTest.cpp

+#include "LZ77DecompressTest.h"
+
+#include <string/String.h>
+#include <string/compress/LZ77.h>
+#include <string/compress/LZ77Decompress.h>
+
+#define CPPUNIT_IMPLY( x, y )    CPPUNIT_ASSERT ( !( x ) || ( y ) )
+
+#include <iostream>
+#include <tuple>
+
+
+CPPUNIT_TEST_SUITE_NAMED_REGISTRATION ( LZ77DecompressTest, "stringology" );
+CPPUNIT_TEST_SUITE_REGISTRATION ( LZ77DecompressTest );
+
+void LZ77DecompressTest::setUp ( ) {
+}
+
+void LZ77DecompressTest::tearDown ( ) {
+}
+
+void LZ77DecompressTest::testLZ77Decompress1 ( ) {
+
+	std::vector < std::tuple < unsigned int, unsigned int, DefaultSymbolType > > input;
+
+	input.push_back(std::make_tuple(0, 0, DefaultSymbolType ( 'a' )));
+	input.push_back(std::make_tuple(1, 1, DefaultSymbolType ( 'b' )));
+	input.push_back(std::make_tuple(3, 2, DefaultSymbolType ( 'c' )));
+	input.push_back(std::make_tuple(3, 2, DefaultSymbolType ( 'a' )));
+	input.push_back(std::make_tuple(1, 3, DefaultSymbolType ( 'b' )));
+	input.push_back(std::make_tuple(2, 3, DefaultSymbolType ( 'b' )));
+
+	string::String res = string::compress::LZ77Decompress::decompress ( input );
+
+	std::string string = "aabaacaaaaaababab";
+	string::String expectedString = string::stringFrom ( string );
+
+	CPPUNIT_ASSERT ( res == expectedString );
+
+}
+
+void LZ77DecompressTest::testLZ77Decompress2 ( ) {
+
+	std::vector < std::tuple < unsigned int, unsigned int, DefaultSymbolType > > input;
+
+	input.push_back(std::make_tuple(0, 0, DefaultSymbolType ( 'a' )));
+	input.push_back(std::make_tuple(1, 1, DefaultSymbolType ( 'c' )));
+	input.push_back(std::make_tuple(3, 4, DefaultSymbolType ( 'b' )));
+	input.push_back(std::make_tuple(3, 3, DefaultSymbolType ( 'a' )));
+	input.push_back(std::make_tuple(1, 2, DefaultSymbolType ( 'c' )));
+
+	string::String res = string::compress::LZ77Decompress::decompress ( input );
+
+	std::string string = "aacaacabcabaaac";
+	string::String expectedString = string::stringFrom ( string );
+
+	CPPUNIT_ASSERT ( res == expectedString );
+
+}
+
+void LZ77DecompressTest::testLZ77Decompress3 ( ) {
+
+	std::string string = "";
+	string::String input = string::stringFrom ( string );
+
+	std::vector < std::tuple < unsigned int, unsigned int, DefaultSymbolType > > out = string::compress::LZ77::compress ( input, 6, 4 );
+
+	string::String res = string::compress::LZ77Decompress::decompress ( out );
+
+	CPPUNIT_ASSERT ( input == res );
+}

alib2algo/test-src/string/compress/LZ77DecompressTest.h

+#ifndef LZ77DECOMPRESS_TEST
+#define LZ77DECOMPRESS_TEST
+
+#include <cppunit/extensions/HelperMacros.h>
+
+class LZ77DecompressTest : public CppUnit::TestFixture {
+	CPPUNIT_TEST_SUITE ( LZ77DecompressTest );
+	CPPUNIT_TEST ( testLZ77Decompress1 );
+	CPPUNIT_TEST ( testLZ77Decompress2 );
+	CPPUNIT_TEST ( testLZ77Decompress3 );
+	CPPUNIT_TEST_SUITE_END ( );
+
+public:
+	void setUp ( );
+	void tearDown ( );
+
+	void testLZ77Decompress1 ( );
+	void testLZ77Decompress2 ( );
+	void testLZ77Decompress3 ( );
+};
+
+#endif // LZ77DECOMPRESS_TEST
\ No newline at end of file

alib2algo/test-src/string/compress/LZ77Test.cpp

+#include "LZ77Test.h"
+
+#include <string/String.h>
+#include <string/compress/LZ77.h>
+
+#define CPPUNIT_IMPLY( x, y )    CPPUNIT_ASSERT ( !( x ) || ( y ) )
+
+#include <iostream>
+#include <tuple>
+
+
+CPPUNIT_TEST_SUITE_NAMED_REGISTRATION ( LZ77Test, "stringology" );
+CPPUNIT_TEST_SUITE_REGISTRATION ( LZ77Test );
+
+void LZ77Test::setUp ( ) {
+}
+
+void LZ77Test::tearDown ( ) {
+}
+
+void LZ77Test::testLZ77_1 ( ) {
+	std::string testStr = "aabaacaaaaaababab";
+
+	string::String string = string::stringFrom ( testStr );
+	std::vector < std::tuple < unsigned int, unsigned int, DefaultSymbolType > > res = string::compress::LZ77::compress ( string, 6, 4 );
+
+	std::vector < std::tuple < unsigned int, unsigned int, DefaultSymbolType > > expectedOutput;
+
+	expectedOutput.push_back(std::make_tuple(0, 0, DefaultSymbolType ( 'a' )));
+	expectedOutput.push_back(std::make_tuple(1, 1, DefaultSymbolType ( 'b' )));
+	expectedOutput.push_back(std::make_tuple(3, 2, DefaultSymbolType ( 'c' )));
+	expectedOutput.push_back(std::make_tuple(3, 2, DefaultSymbolType ( 'a' )));
+	expectedOutput.push_back(std::make_tuple(1, 3, DefaultSymbolType ( 'b' )));
+	expectedOutput.push_back(std::make_tuple(2, 3, DefaultSymbolType ( 'b' )));
+
+	CPPUNIT_ASSERT ( res == expectedOutput );
+
+}
+
+void LZ77Test::testLZ77_2 ( ) {
+
+	std::string testStr = "aacaacabcabaaac";
+
+	string::String string = string::stringFrom ( testStr );
+	std::vector < std::tuple < unsigned int, unsigned int, DefaultSymbolType > > res = string::compress::LZ77::compress ( string, 6, 4 );
+
+	std::vector < std::tuple < unsigned int, unsigned int, DefaultSymbolType > > expectedOutput;
+
+	expectedOutput.push_back(std::make_tuple(0, 0, DefaultSymbolType ( 'a' )));
+	expectedOutput.push_back(std::make_tuple(1, 1, DefaultSymbolType ( 'c' )));
+	expectedOutput.push_back(std::make_tuple(3, 4, DefaultSymbolType ( 'b' )));
+	expectedOutput.push_back(std::make_tuple(3, 3, DefaultSymbolType ( 'a' )));
+	expectedOutput.push_back(std::make_tuple(1, 2, DefaultSymbolType ( 'c' )));
+
+	CPPUNIT_ASSERT ( res == expectedOutput );
+}

alib2algo/test-src/string/compress/LZ77Test.h

+#ifndef LZ77_TEST
+#define LZ77_TEST
+
+#include <cppunit/extensions/HelperMacros.h>
+
+class LZ77Test : public CppUnit::TestFixture {
+	CPPUNIT_TEST_SUITE ( LZ77Test );
+	CPPUNIT_TEST ( testLZ77_1 );
+	CPPUNIT_TEST ( testLZ77_2 );
+	CPPUNIT_TEST_SUITE_END ( );
+
+public:
+	void setUp ( );
+	void tearDown ( );
+
+	void testLZ77_1 ( );
+	void testLZ77_2 ( );
+};
+
+#endif // LZ77_TEST
\ No newline at end of file