html.cpp

一套linux下的C++开发库

/***************************************************************************
                          html.cpp  -  description
                             -------------------
    begin                : Fri Jul 20 2001
    copyright            : (C) 2001 by Mark
    email                : [e-mail]
 ***************************************************************************/

/***************************************************************************
 *                                                                         *
 *   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 2 of the License, or     *
 *   (at your option) any later version.                                   *
 *                                                                         *
 ***************************************************************************/


#include <fstream.h>
#include <stdio.h>

#include "html.h"


// ex: <$test$><$/test$>
static const char s_sPartStartTag[] = "<$";
static const char s_sPartEndTag[] = "$>";
static const char s_chPartTailTag = '/';

// ex: <%test%>
static const char s_sReplaceStartTag[] = "<%";
static const char s_sReplaceEndTag[] = "%>";


const string CHtmlDocument::top = "";


void CReplacePart::ParsePart(const string& strHtml)
{
	int iPos = 0;
	while (true)
	{
		if (!FindTextElement(strHtml, iPos))	//end of html
			return;
			
		FindTagElement(strHtml, iPos);
	}
}


bool CReplacePart::FindTextElement(const string& strHtml, int& iPos)
{
	int iTagPos;
	if ((iTagPos = strHtml.find(s_sReplaceStartTag, iPos)) == string::npos)
	{
		m_stPart.push_back(CReplaceElement(CReplaceElement::TEXT, strHtml.substr(iPos)));
		return false;
	}

	m_stPart.push_back(CReplaceElement(CReplaceElement::TEXT, strHtml.substr(iPos, iTagPos-iPos)));
	iPos = iTagPos;
	return true;
}


bool CReplacePart::FindTagElement(const string& strHtml, int& iPos)
{
	string strTagName;
	GetReplaceTagName(strTagName, strHtml, iPos);
	m_stPart.push_back(CReplaceElement(CReplaceElement::TAG, strTagName));
	return true;
}


void CReplacePart::GetReplaceTagName(string& strTagName, const string& strHtml, int& iPos)
{
	iPos += strlen(s_sReplaceStartTag);
	int iTagPos = strHtml.find(s_sReplaceEndTag, iPos);
	if (iTagPos == string::npos)
	{
		string strException("can not find ");
		strException += s_sReplaceEndTag;
		strException += ".";
		throw CHtmlException(strException.c_str(), __FILE__, __LINE__);
	}
	
	if (iTagPos == iPos)
		throw CHtmlException("replace tag name is empty.", __FILE__, __LINE__);
	
	strTagName = strHtml.substr(iPos, iTagPos-iPos);
	iPos = iTagPos + strlen(s_sReplaceEndTag);
}


void CReplacePart::Replace(ostream& os, CReplaceMap& stReplaceMap)
{
	for (int i = 0; i < m_stPart.size(); i++)
	{
		if (m_stPart[i].m_iType == CReplaceElement::TEXT)
		{
			os << m_stPart[i].m_strContent;
			continue;
		}
		
		if (m_stPart[i].m_iType == CReplaceElement::TAG)
			os << stReplaceMap[m_stPart[i].m_strContent];
	}
}


CHtmlDocument::CHtmlDocument(const char* sHtmlFilePath)
{
	Open(sHtmlFilePath);
}


void CHtmlDocument::Open(const char* sHtmlFilePath)
{
	ifstream file(sHtmlFilePath, ios::in | ios::nocreate);
	if (file.fail())
		throw CFileException("open html file fail.", __FILE__, __LINE__);
		
	file.seekg(0, ios::end);
	streampos sp = file.tellg();
	if (file.fail())
		throw CFileException("get html file length fail.", __FILE__, __LINE__);
	
	char* achHtml = new char[sp];
	if (achHtml == NULL)
		throw CMemException("out of memory.", __FILE__, __LINE__);

	file.seekg(0, ios::beg);
	file.read(achHtml, sp);
	if (file.fail())
	{
		delete [] achHtml;
		throw CFileException("read html file fail.", __FILE__, __LINE__);
	}
		
	string strHtml;
	strHtml.append(achHtml, sp);
	delete achHtml;
	ParseHtml(strHtml);
}


CHtmlDocument::~CHtmlDocument()
{
}

	
void CHtmlDocument::ParseHtml(const string& strHtml)
{
	int iPos = 0;
	string strTagName = CHtmlDocument::top;
	while (true)
	{
		if (!FindTextPart(strTagName, strHtml, iPos))	//end of html
			return;
			
		FindReplacePart(strTagName, strHtml, iPos);
	}
}


bool CHtmlDocument::FindTextPart(const string& strTagName, const string& strHtml, int& iPos)
{
	int iTagPos;
	if ((iTagPos = strHtml.find(s_sPartStartTag, iPos)) == string::npos)
	{
		m_stTextParts[strTagName] = strHtml.substr(iPos);
		return false;
	}

	m_stTextParts[strTagName] = strHtml.substr(iPos, iTagPos-iPos);
	iPos = iTagPos;
	return true;
}


bool CHtmlDocument::FindReplacePart(string& strTagName, const string& strHtml, int& iPos)
{
	string strReplacePart;
	GetPartTagName(strTagName, strHtml, iPos);
	GetPartString(strReplacePart, strTagName, strHtml, iPos);
	m_stReplaceParts[strTagName] = CReplacePart(strReplacePart);
	return true;
}


void CHtmlDocument::GetPartTagName(string& strTagName, const string& strHtml, int& iPos)
{
	iPos += strlen(s_sPartStartTag);
	int iTagPos = strHtml.find(s_sPartEndTag, iPos);
	if (iTagPos == string::npos)
	{
		string strException("can not find ");
		strException += s_sPartEndTag;
		strException += ".";
		throw CHtmlException(strException.c_str(), __FILE__, __LINE__);
	}
	
	if (iTagPos == iPos)
		throw CHtmlException("part tag name is empty.", __FILE__, __LINE__);
	
	strTagName = strHtml.substr(iPos, iTagPos-iPos);
	iPos = iTagPos + strlen(s_sPartEndTag);
}


void CHtmlDocument::GetPartString(string& strReplacePart, const string& strTagName, const string& strHtml, int& iPos)
{
	string strTag = s_sPartStartTag;
	strTag += s_chPartTailTag;
	strTag += strTagName;
	strTag += s_sPartEndTag;
	int iTagPos = strHtml.find(strTag.c_str(), iPos);
	if (iTagPos == string::npos)
	{
		string strException("can not find ");
		strException += strTag;
		strException += ".";
		throw CHtmlException(strException.c_str(), __FILE__, __LINE__);	
	}
	
	strReplacePart = strHtml.substr(iPos, iTagPos-iPos);
	iPos = iTagPos + strTag.length();
}


void CHtmlDocument::Replace(ostream& os, const char* strTagName, ...)
{
	va_list ap;
	va_start(ap, strTagName);

	char *pchName, *pchValue;
	CReplaceMap stReplaceMap;
	while ((pchName = va_arg(ap, char*)))
	{
		pchValue = va_arg(ap, char*);
		stReplaceMap[pchName] = pchValue;
	}
	
	va_end(ap);
	Replace(os, strTagName, stReplaceMap);
}


void CHtmlDocument::ReplaceF(ostream& os, const char* strTagName, ...)
{
	va_list ap;
	va_start(ap, strTagName);

	char *pchName, *pchValue;
	int iValue;
	long lValue;
	double dValue;
	int iFormat;
	CReplaceMap stReplaceMap;
	char sTmp[64];
	while ((pchName = va_arg(ap, char*)))
	{
		iFormat = va_arg(ap, int);
		switch (iFormat)
		{
			case STRING:
				pchValue = va_arg(ap, char*);
				stReplaceMap[pchName] = pchValue;
				break;
				
			case INT:
				iValue = va_arg(ap, int);
				sprintf(sTmp, "%d", iValue);
				stReplaceMap[pchName] = sTmp;
				break;
				
			case LONG:
				lValue = va_arg(ap, long);
				sprintf(sTmp, "%ld", lValue);
				stReplaceMap[pchName] = sTmp;
				break;
				
			case DOUBLE:
				dValue = va_arg(ap, double);
				sprintf(sTmp, "%f", dValue);
				stReplaceMap[pchName] = sTmp;
				break;
		}		
	}
	
	va_end(ap);
	Replace(os, strTagName, stReplaceMap);
}


string& UrlEncode(const string& strSrc, string& strDest)
{
	for (int i = 0; i < strSrc.length(); i++)
	{
		switch (strSrc[i])
		{
			case '&':
				strDest += "%26";
				break;
				
			case '=':
				strDest += "%3D";
				break;
				
			case '%':
				strDest += "%25";
				break;
				
			case '+':
				strDest += "%2B";
				break;
				
			case '?':
				strDest += "%3F";
				break;

			case ' ':
				strDest += "+";
				break;								
				
			default:
				strDest += strSrc[i];
				break;
		}
	}
	
	return strDest;
}


string& UrlDecode(const string& strSrc, string& strDest)
{
	char sTmp[3];
	sTmp[2] = 0;
	unsigned int uiTmp;
	int i = 0;
	while (i < strSrc.length())
	{
		switch (strSrc[i])
		{
			case '+':
				strDest += ' ';
				i++;
				break;
			
			case '%':
				if (i+2 >= strSrc.length())
					throw CHtmlException("source url error.", __FILE__, __LINE__);
				sTmp[0] = strSrc[i+1];
				sTmp[1] = strSrc[i+2];
				sscanf(sTmp, "%x", &uiTmp);
				strDest += (char)uiTmp;
				i += 3;
				break;
				
			default:
				strDest += strSrc[i];
				i++;
				break;
		}
	}
	
	return strDest;
}