/*
	CXml2Txt implementation

	Last modified: 14.12.03
*/

#include	"xml2txt.h"

CXml2Txt::CXml2Txt()
 :
 m_fin(NULL),
 m_fout(NULL),
 m_sSrc(""),
 m_sDst(""),
 m_tags_per_part(0),
 m_pause(false),
 m_verbose(false),
 m_ch(0),
 m_last_ch(0),
 m_line_cnt(1),
 m_state(eText)
 {
 m_szError[0] = '\0';
 m_buf[0] = '\0';
 m_idx = m_max = -1;
 }

CXml2Txt::~CXml2Txt()
 {
 if(m_fin)
	 fclose(m_fin);
 if(m_fout)
	 fclose(m_fout);
 }

void CXml2Txt::SetPause(bool flag /* = true */)
 {
 m_pause= flag;
 }

void CXml2Txt::BeVerbose(bool flag /* = true */)
 {
 m_verbose = flag;
 }

void CXml2Txt::SplitOutput(unsigned long n_tags /* = 0 */)
 {
 m_tags_per_part = n_tags;
 }

void CXml2Txt::Open(string sSrc, string sDst)
 {
 CXml2Txt::Close();

 if( (m_fin = fopen(sSrc.c_str(), "rb")) == NULL)
	{
	throw Error("Can't open source file.");
	}
 if( (m_fout = fopen(sDst.c_str(), "wb")) == NULL)
	{
	throw Error("Can't create destination file.");
	}
 
 // Save filenames:
 m_sSrc = sSrc;
 m_sDst = sDst;
 }

unsigned long CXml2Txt::Convert(const char sep /* = '\t' */)
 {
 m_sep = sep;
 
 while( true )
  {
  CXml2Txt::GetNext();

  switch(m_state)
	{
	case eText:		CXml2Txt::ParseText();
	break;
	
	case eTag:  	CXml2Txt::ParseTag();
    break;
    
	case eError:	throw(Error(m_szError));
					
	
	case eEOF:      CXml2Txt::On_EndOfFile();
					return m_line_cnt - 1;
	}
  }
 }

void CXml2Txt::Close(f_close f /* =  f_close::ALL */)
{
 if( m_fin && (f == ALL || f == SRC) )
	{
	//m_sSrc = "";
	fclose(m_fin);
	m_fin = NULL;
	}
 if( m_fout && (f == ALL || f == DST) )
	{
	//m_sDst = "";
	fclose(m_fout);
	m_fout = NULL;
	}
}

void CXml2Txt::On_EndOfFile()
	{
	if( ferror(m_fin) )
		printf("\nError while reading data from source file.");
	else
		printf("\nAll data was successfully read.");
	if( ferror(m_fout) )
		printf("\nError while writing data to destination file.");
	else
		printf("\nAll data was successfully written.");
	}

inline char CXml2Txt::GetNext()
 {
 m_last_ch = m_ch;

 if( m_idx == m_max )
	 {
	 if( (m_max = fread(m_buf, 1, STREAM_BUF_SIZE, m_fin)) < 1 || ferror(m_fout) )
		 ChangeState(eEOF);
	 else
		 {
		 m_idx = 0;

		 // Progress:
		 if( m_verbose )
			{
			fpos_t			pos;
		 
			fgetpos(m_fin, &pos);
			printf("\rConverting: %-I64u bytes complete.", pos);
			}
		 }
	 }
 
 m_ch = m_buf[m_idx++];

 return m_ch;

 /* // No buffer:
 m_last_ch = m_ch;
 
 if ( (m_ch = fgetc(m_fin)) == EOF)
	{
	ChangeState(eEOF);
	}
 return m_ch;
 */
 }

void CXml2Txt::ChangeState(state eNewState)
 {
	 m_prev_state	= m_state;
	 m_state		= eNewState;
 }

void CXml2Txt::ParseText()
 {
 switch(m_ch)
	 {
	 case  '<':		ChangeState(eTag);
	 break;
	 
	 case '\n':		m_line_cnt++;
	 break;
	 
	 default:		break;
	 }
 
 // Skip other chars.
 }

void CXml2Txt::ParseTag()
 {
 tag					tTag;
 attrib					pAttrib;
 
 ParseTagName(tTag.sName);
 
 while( m_state == eTag )
	 {
		pAttrib.sName = ""; 
		ParseTagAttribName(pAttrib.sName);
	
		if(m_state != eTag)
			break;
		
		pAttrib.sValue = "";
		ParseTagAttribVal(pAttrib.sValue);
		
		if(m_state != eTag)
			break;

		tTag.lAttribList.push_back(pAttrib);

		ParseTagEnd();
	 }
 
 AnalyzeTag(tTag);

 if( !tTag.lAttribList.empty() )
	tTag.lAttribList.erase(tTag.lAttribList.begin(), tTag.lAttribList.end());
 }

void CXml2Txt::ParseTagName(string &sTagName)
{
	unsigned int chars = 0;

	// Skip spaces:
	while( m_ch == ' ' && m_state == eTag ) CXml2Txt::GetNext();

	while( m_state == eTag )
		{
		switch(m_ch)
			{
			case  '>':	ChangeState(eText);
			case  ' ':  return;
			
			case '\n':  m_line_cnt++;
			break;
			}

		if( m_ch != '\r' && m_ch != '\n' )
			{
			if( chars < MAX_TAG_NAME )
				{
				if( m_ch == m_sep)
					{
					//printf("\nWarning: separator in tag name. Skipped.");
					}
				else
					{
					sTagName += m_ch;
					chars++;
					}
				}
			else
				{
				ChangeState(eError);
				
				_snprintf(m_szError, ERROR_BUF_SIZE, "Increase maximum tag name buffer. [%u]", chars);
				m_szError[ERROR_BUF_SIZE - 1] = '\0';
				
				return;
				}
			}

		CXml2Txt::GetNext();
		} 
}

void CXml2Txt::ParseTagAttribName(string &sTagAttribName)
{
	unsigned int chars = 0;

	// Skip spaces:
	while( m_ch == ' ' && m_state == eTag ) CXml2Txt::GetNext();

	while( m_state == eTag )
		{
		switch(m_ch)
			{
			case  '>':	ChangeState(eText);
						return;
			
			case  '=':	CXml2Txt::GetNext();
						return;

			case '\n':	m_line_cnt++;
			break;

			case ' ':	CXml2Txt::GetNext();
						continue;
			}
		
		if( m_ch != '\r' && m_ch != '\n' )
			{
			if( chars < MAX_TAG_ATTRIB_NAME )
				{
				if( m_ch == m_sep )
					{
					//printf("\nWarning: separator in attribute name. Skipped.");
					}
				else
					{
					sTagAttribName += m_ch;
					chars++;
					}
				}
			else
				{
				ChangeState(eError);
				
				_snprintf(m_szError, ERROR_BUF_SIZE, "Increase maximum tag attribute name buffer. [%u]", chars);
				m_szError[ERROR_BUF_SIZE - 1] = '\0';
				
				return;
				}
			}

		CXml2Txt::GetNext();
		} 
}

void CXml2Txt::ParseTagAttribVal(string &sTagAttribVal)
{
	char			used_quotes;
	unsigned int	chars = 0;

	// Skip spaces:
	while( m_ch == ' ' && m_state == eTag ) CXml2Txt::GetNext();

	while( m_state == eTag )
		{     
		switch(m_ch)
			{
			case  '>':	ChangeState(eText);
						return;

			case '\n':	m_line_cnt++;
			break;

			case ' ':	CXml2Txt::GetNext();
						continue;
			}

		if( m_ch == '\'' || m_ch == '"' )
				{
				char	wnd[3], i;

				wnd[0] = used_quotes = m_ch;
				
				// Fill the window:
				for(i = 1; i < 3; i++)
					wnd[i] = CXml2Txt::GetNext();

				// Check for empty value case:
				if( wnd[0] == used_quotes && wnd[1] == used_quotes && wnd[2] !=	used_quotes)
					{
					return;
					}
				else
					{
					// Assume wnd[0] = first char of value:
					for(i = 1; i < 3; i++)
						wnd[i-1] = wnd[i];
					wnd[2] = CXml2Txt::GetNext();
					}
				
				// Read the value:
				while( m_state == eTag && chars < MAX_TAG_ATTRIB_VAL) 
					{
					// Skip one quote:
					if( wnd[0] == used_quotes && wnd[1] == used_quotes)
						{
						// Move window:
						for(i = 1; i < 3; i++)
							wnd[i-1] = wnd[i];
						wnd[2] = CXml2Txt::GetNext();
						}

					// Store the char:
					if( wnd[0] == m_sep )
						{
						//printf("\nWarning: separator in attribute value. Skipped.");
						}
					else
					if( wnd[0] == '\n' )
						{
						m_line_cnt++;
						}
					else
						{
						if( wnd[0] != '\r' )
							{
							sTagAttribVal += wnd[0];
							chars++;
							}
						}

					// Check for ending quote:
					if( (wnd[1] == used_quotes && wnd[2] != used_quotes) || m_state != eTag )
						break;

					// Move window:
					for(i = 1; i < 3; i++)
						wnd[i-1] = wnd[i];
					wnd[2] = CXml2Txt::GetNext();
					}
				
				if ( chars >= MAX_TAG_ATTRIB_VAL )
					{
					ChangeState(eError);

					_snprintf(m_szError, ERROR_BUF_SIZE, "Increase maximum tag attribute value buffer. [%u]", chars);
					m_szError[ERROR_BUF_SIZE - 1] = '\0';

					return;
					}
				
				return;
				}
		else
				{
				ChangeState(eError);

				_snprintf(m_szError, ERROR_BUF_SIZE, "Line %lu: value is not quoted.", m_line_cnt);
				m_szError[ERROR_BUF_SIZE - 1] = '\0';

				return;
				}    
		}
}   

void CXml2Txt::ParseTagEnd()
{
	// Skip spaces:
	while( m_ch == ' ' && m_state == eTag ) CXml2Txt::GetNext();

	switch(m_ch)
		{
		case '>':	ChangeState(eText);
		break;

		case '/':	CXml2Txt::GetNext();
		break;
		}
}

void CXml2Txt::AnalyzeTag(tag &tTag)
 {
 fprintf(m_fout, "%s", tTag.sName.c_str());
 
 while( tTag.lAttribList.size() > 0 )
	 {
	 fprintf(m_fout, "%c%s=%s", m_sep,
								tTag.lAttribList.front().sName.c_str(),
								tTag.lAttribList.front().sValue.c_str());
	 
	 tTag.lAttribList.pop_front();
	 }
 
 fprintf(m_fout, "\r\n");
 }

//