Merge pull request #1275 from gsomix/feature/linereader

Bug fixes and re-factoring for LineReader and CircularBuffer

examples/undocumented/libshogun/io_linereader.cpp

@@ -1,6 +1,6 @@
 #include <shogun/base/init.h>
-
 #include <shogun/io/LineReader.h>
+#include <shogun/lib/DelimiterTokenizer.h>
 #include <shogun/lib/SGVector.h>
 
 #include <cstdio>
@@ -12,13 +12,17 @@ int main(int argc, char** argv)
 	init_shogun_with_defaults();
 
 	FILE* fin=fopen("io_linereader.cpp", "r");
-	CLineReader* reader=new CLineReader(fin);
+
+	CDelimiterTokenizer* tokenizer=new CDelimiterTokenizer();
+	tokenizer->delimiters['\n']=1;
+
+	CLineReader* reader=new CLineReader(fin, tokenizer);
 
 	int lines_count=0;
-	SGVector<char> tmp_string;
-	while (reader->has_next_line())
+	SGVector<char> tmp_string(true);
+	while (reader->has_next())
 	{
-		tmp_string=reader->get_next_line();
+		tmp_string=reader->read_line();
 		SG_SPRINT("%d %d ", lines_count, tmp_string.vlen);
 		for (int i=0; i<tmp_string.vlen; i++)
 			SG_SPRINT("%c", tmp_string.vector[i]);
@@ -27,7 +31,10 @@ int main(int argc, char** argv)
 	}
 	SG_SPRINT("total lines: %d\n", lines_count);
 
-	delete reader;
+	tmp_string=SGVector<char>();
+	SG_UNREF(tokenizer);
+	SG_UNREF(reader);
+
 	fclose(fin);
 
 	exit_shogun();

examples/undocumented/libshogun/library_circularbuffer.cpp

@@ -1,5 +1,6 @@
 #include <shogun/base/init.h>
 #include <shogun/lib/CircularBuffer.h>
+#include <shogun/lib/DelimiterTokenizer.h>
 #include <shogun/lib/SGVector.h>
 
 #include <cstdio>
@@ -13,7 +14,7 @@ int main(int argc, char** argv)
 {
 	init_shogun_with_defaults();
 
-	SGVector<char> test_string("all your bayes are belong to us! ", 33, false);
+	SGVector<char> test_string(const_cast<char* >("all your bayes are belong to us! "), 33, false);
 
 	CCircularBuffer* buffer=new CCircularBuffer(max_line_length);
 	CDelimiterTokenizer* tokenizer=new CDelimiterTokenizer();

src/shogun/io/LineReader.cpp

@@ -7,94 +7,125 @@
  * Written (W) 2013 Evgeniy Andreev (gsomix)
  */
 
-#include <cstdio>
 #include <shogun/io/LineReader.h>
 
 using namespace shogun;
 
 CLineReader::CLineReader()
-	: m_stream(NULL), m_max_line_length(0), m_next_line_length(-1)
 {
-	m_buffer=new CCircularBuffer(0);
-	m_tokenizer=NULL;
+	init();
+
+	m_buffer=new CCircularBuffer();
 }
 
-CLineReader::CLineReader(FILE* stream, char delimiter)
-	: m_stream(stream), m_max_line_length(10*1024*1024), m_next_line_length(-1)
+CLineReader::CLineReader(FILE* stream, CTokenizer* tokenizer)
 {
-	m_buffer=new CCircularBuffer(m_max_line_length);
-	m_tokenizer=new CDelimiterTokenizer();
-	m_tokenizer->delimiters[delimiter]=1;
-	m_buffer->set_tokenizer(m_tokenizer);
+	init();
+
+	m_stream=stream;
+	m_max_token_length=10*1024*1024;
+
+	m_buffer=new CCircularBuffer(m_max_token_length);
+	m_buffer->set_tokenizer(tokenizer);
+
+	m_tokenizer=tokenizer;
 }
 
-CLineReader::CLineReader(int32_t max_line_length, FILE* stream, char delimiter)
-	: m_stream(stream), m_max_line_length(max_line_length), m_next_line_length(-1)
+CLineReader::CLineReader(int32_t max_token_length, FILE* stream, CTokenizer* tokenizer)
 {
-	m_buffer=new CCircularBuffer(m_max_line_length);
-	m_tokenizer=new CDelimiterTokenizer();
-	m_tokenizer->delimiters[delimiter]=1;
-	m_buffer->set_tokenizer(m_tokenizer);
+	init();
+
+	m_stream=stream;
+	m_max_token_length=max_token_length;
+
+	m_buffer=new CCircularBuffer(m_max_token_length);
+	m_buffer->set_tokenizer(tokenizer);
+
+	m_tokenizer=tokenizer;
 }
 
 CLineReader::~CLineReader()
 {
-	SG_UNREF(m_tokenizer);
 	SG_UNREF(m_buffer);
 }
 
-bool CLineReader::has_next_line()
+bool CLineReader::has_next()
 {
-	if (m_stream==NULL || m_max_line_length==0)
+	if (m_stream==NULL || m_max_token_length==0 || m_tokenizer==NULL)
 	{
-		SG_ERROR("Class is not initialized");
+		SG_ERROR("CLineReader::has_next():: Class is not initialized\n");
 		return false;
 	}
 
 	if (ferror(m_stream))
 	{
-		SG_ERROR("Error reading file");
+		SG_ERROR("CLineReader::has_next():: Error reading file\n");
 		return false;
 	}
 
-	if (feof(m_stream) && m_buffer->num_bytes_contained()<=0)
+	if (feof(m_stream) && (m_buffer->num_bytes_contained()<=0 || !m_buffer->has_next()))
+	{
 		return false; // nothing to read
+	}
 
 	return true;	
 }
 
-SGVector<char> CLineReader::get_next_line()
+void CLineReader::skip_line()
+{	
+	int32_t bytes_to_skip=0;
+	m_next_token_length=read(bytes_to_skip);
+	if (m_next_token_length==-1)
+		return;
+	else
+		m_buffer->skip_characters(bytes_to_skip);
+}
+
+SGVector<char> CLineReader::read_line()
 {
 	SGVector<char> line;	
 
-	m_next_line_length=read_line();
-	if (m_next_line_length==-1)
+	int32_t bytes_to_skip=0;
+	m_next_token_length=read(bytes_to_skip);
+	if (m_next_token_length==-1)
 		line=SGVector<char>();
 	else
-		line=copy_line(m_next_line_length);
+	{
+		m_buffer->skip_characters(bytes_to_skip);
+		line=read_token(m_next_token_length-bytes_to_skip);
+	}
 
 	return line;
 }
 
-void CLineReader::set_delimiter(char delimiter)
+void CLineReader::set_tokenizer(CTokenizer* tokenizer)
 {
-	m_tokenizer->delimiters[delimiter]=1;
+	m_buffer->set_tokenizer(tokenizer);
+	m_tokenizer=tokenizer;
 }
 
-void CLineReader::clear_delimiters()
+void CLineReader::init()
 {
-	m_tokenizer->clear_delimiters();
+	m_buffer=NULL;
+	m_tokenizer=NULL;
+	m_stream=NULL;
+
+	m_max_token_length=0;
+	m_next_token_length=-1;
 }
 
-int32_t CLineReader::read_line()
+int32_t CLineReader::read(int32_t& bytes_to_skip)
 {
 	int32_t line_end=0;
-	int32_t bytes_to_skip=0;
 	int32_t bytes_to_read=0;
+	int32_t temp_bytes_to_skip=0;
 
 	while (1)
 	{
-		line_end+=m_buffer->next_token_idx(bytes_to_skip)-bytes_to_skip;
+		if (bytes_to_skip==line_end)
+			line_end=m_buffer->next_token_idx(bytes_to_skip);
+		else
+			line_end=m_buffer->next_token_idx(temp_bytes_to_skip);
 
 		if (m_buffer->num_bytes_contained()!=0 && line_end<m_buffer->num_bytes_contained())
 			return line_end;
@@ -104,10 +135,10 @@ int32_t CLineReader::read_line()
 		// if there is no delimiter in buffer
 		// try get more data from stream
 		// and write it into buffer
-		if (m_buffer->available() < m_max_line_length)
+		if (m_buffer->available() < m_max_token_length)
 			bytes_to_read=m_buffer->available();
 		else
-			bytes_to_read=m_max_line_length;
+			bytes_to_read=m_max_token_length;
 
 		if (feof(m_stream))
 			return line_end;
@@ -116,13 +147,13 @@ int32_t CLineReader::read_line()
 
 		if (ferror(m_stream))
 		{
-			SG_ERROR("Error reading file");
+			SG_ERROR("CLineReader::read(int32_t&):: Error reading file\n");
 			return -1;
 		}
 	}	
 }
 
-SGVector<char> CLineReader::copy_line(int32_t line_len)
+SGVector<char> CLineReader::read_token(int32_t line_len)
 {
 	SGVector<char> line;
 
@@ -131,7 +162,5 @@ SGVector<char> CLineReader::copy_line(int32_t line_len)
 	else
 		line=m_buffer->pop(line_len);
 
-	m_buffer->skip_characters(1);
-
 	return line;
 }

src/shogun/io/LineReader.h

@@ -10,14 +10,13 @@
 #ifndef __LINE_READER_H__
 #define __LINE_READER_H__
 
-#include <shogun/base/SGObject.h>
 #include <shogun/lib/SGVector.h>
+#include <shogun/lib/Tokenizer.h>
 #include <shogun/lib/CircularBuffer.h>
 
 namespace shogun
 {
-/** @brief Class for buffered reading lines from a ascii file
- */
+/** @brief Class for buffered reading from a ascii file */
 class CLineReader : public CSGObject
 {
 public:
@@ -28,77 +27,67 @@ class CLineReader : public CSGObject
 	 *
 	 * @param stream readable stream
 	 */
-	CLineReader(FILE* stream, char delimiter='\n');
+	CLineReader(FILE* stream, CTokenizer* tokenizer);
 
 	/** create object associated with the stream to read
 	 * and specify maximum length of a string that can be read
 	 *
 	 * @param stream readable stream
 	 * @param buffer_size size of internal buffer
 	 */
-	CLineReader(int32_t max_string_length, FILE* stream, char delimiter='\n');
+	CLineReader(int32_t max_string_length, FILE* stream, CTokenizer* tokenizer);
 
 	/** deconstructor */
-	~CLineReader();
+	virtual ~CLineReader();
 
 	/** check for next line in the stream
 	 *
 	 * @return true if there is next line, false - otherwise
 	 */
-	bool has_next_line();
+	virtual bool has_next();
 
-	/** get line from the buffer into SGVector
-	 * there is no warranty that after reading the caret will 
-	 * set at the beginning of a new line
-	 *
-	 * @return SGVector that contains line
-	 */
-	SGVector<char> get_next_line();
+	/** skip next line */
+	virtual void skip_line();
 
-	/** set delimiter active for tokenizing 
+	/** read string	*/
+	virtual SGVector<char> read_line();
+
+	/** set tokenizer
 	 *
-	 * @param delimiter delimiter
+	 * @param tokenizer tokenizer	
 	 */
-	void set_delimiter(char delimiter);
-
-	/** clear all delimiters */
-	void clear_delimiters();
+	void set_tokenizer(CTokenizer* tokenizer);
 
 	/** @return object name */
 	virtual const char* get_name() const { return "LineReader"; }
 
 private:
-	/** read one line into buffer
-	 *
-	 * @return length of line
- 	 */
-	int32_t read_line();
+	/** class initialization */
+	void init();
 
-	/** copy chars into SGVector from source
-	 *
-	 * @param line destination string
-	 * @param line_len length of line in source
-	 * @param source source array of chars
-	 */
-	SGVector<char> copy_line(int32_t line_len);
+	/** read file into memory */
+	int32_t read(int32_t& bytes_to_skip);
+
+	/** read token from internal buffer */
+	SGVector<char> read_token(int32_t line_len);
 
 private:
 	/** internal buffer for searching */
 	CCircularBuffer* m_buffer;
 
-	/** tokenizer */
-	CDelimiterTokenizer* m_tokenizer;
+	/** */
+	CTokenizer* m_tokenizer;
 
 	/** readable stream */
-	FILE* m_stream;	
+	FILE* m_stream;
 
 	/** maximum length of a line that can be read */
-	int32_t m_max_line_length;
+	int32_t m_max_token_length;
 
 	/** length of next line in the buffer */
-	int32_t m_next_line_length;	
+	int32_t m_next_token_length;	
 };
 
 }
 
-#endif /* __LINE_READER_H__ */
+#endif /* __FILE_READER_H__ */