rthtmlregex.h

来自「vc++编写的界面源代码,对C++爱好者是很发的帮助。」· C头文件 代码 · 共 155 行

#pragma once

class CRTHtmlRegex
{
public:

	CRTHtmlRegex(void)
	{
	}

	virtual ~CRTHtmlRegex(void)
	{
	}

	BOOL Match(const char * input,char ** start,char **end,char **next)
	{
		if(input == NULL)return NULL;
		if(*input == NULL)return NULL;
		char *ptr = (char *)input;
		while(*ptr != '\0')
		{
			if(*ptr == '<')break;
			ptr ++;
		}
		if(*ptr == '\0')return FALSE;
		ptr ++;

		while(*ptr != '\0' && *ptr != '>')
		{
			if(*ptr == 'h' || *ptr == 'H')
			{
				if(MatchHref(ptr,start,end,next))
					return TRUE;
			}
			else if(*ptr == 's' || *ptr == 'S')
			{
				if(MatchSrc(ptr,start,end,next))
					return TRUE;
			}
			ptr ++;
		}

		if(*ptr == '\0')return FALSE;
		ptr ++;
		return Match(ptr,start,end,next);
	}

private:
	BOOL MatchHref(const char *input,char **start,char **end,char **next)
	{
		char *ptr = (char *)input;
		ptr++;
		if(*ptr != 'r' && *ptr != 'R')return FALSE;
		ptr++;
		if(*ptr != 'e' && *ptr != 'E')return FALSE;
		ptr ++;
		if(*ptr != 'f' && *ptr != 'F')return FALSE;

		//begin with href
		ptr ++;
		while(*ptr > 0 && *ptr < 33)
			ptr ++;
		if(*ptr != '=')return FALSE;
		ptr++;
		while(*ptr > 0 && *ptr < 33)
			ptr ++;

		return MatchLink(ptr,start,end,next);
		
	}

	BOOL MatchSrc(const char *input,char** start,char** end,char **next)
	{
		char *ptr = (char *)input;
		ptr++;
		if(*ptr != 'r' && *ptr != 'R')return FALSE;
		ptr++;
		if(*ptr != 'c' && *ptr != 'C')return FALSE;
		ptr ++;

		//begin with src
		while(*ptr > 0 && *ptr < 33)
			ptr ++;
		if(*ptr != '=')return FALSE;
		ptr++;
		while(*ptr > 0 && *ptr < 33)
			ptr ++;
		return MatchLink(ptr,start,end,next);
	}

	BOOL MatchLink(const char *input,char **start,char**end,char **next)
	{
		if(*input == NULL)return FALSE;
		char *ptr = (char *)input;
		char *ptrStart = NULL;
		char *ptrEnd = NULL;
		if(*ptr == '\"')
		{
			ptr++;
			ptrStart = ptr;
			while(*ptr != '\0')
			{
				if(*ptr == '>')break;
				if(*ptr == '\"' || *ptr == '#')
				{
                    if(ptrEnd == NULL)ptrEnd = ptr;
				}
				ptr ++;
			}
			if(ptrEnd == NULL)return FALSE;
		}
		else if(*ptr == '\'')
		{
			ptr++;
			ptrStart = ptr;
			while(*ptr != '\0')
			{
				if(*ptr == '>')break;
				if(*ptr == '\'' || *ptr == '#')
				{
                    if(ptrEnd == NULL)ptrEnd = ptr;
				}
				ptr ++;
			}
			if(ptrEnd == NULL)return FALSE;
		}
		else
		{
			ptr++;
			ptrStart = ptr;
			while(*ptr != '\0')
			{
				if(*ptr == '>')
				{
					if(ptrEnd == NULL)ptrEnd = ptr;
					break;
				}
				if(*ptr == '#' || (*ptr > 0 && *ptr < 33))
				{
                    if(ptrEnd == NULL)ptrEnd = ptr;
				}
				ptr ++;
			}
			if(ptrEnd == NULL)return FALSE;		
		}

		if(*ptr != NULL)ptr ++;
		
		*start = ptrStart;
		*end = ptrEnd;
		*next = ptr;
		return TRUE;
	}
};