#include <iostream>

#include "HTMLPage.h"

pthread_mutex_t HTMLPage::m_hFirst_mutex = PTHREAD_MUTEX_INITIALIZER;
FilterGroup     HTMLPage::m_fgURLs;

HTMLPage::HTMLPage(TIPsDatabase *_db, InternetURIRequest *_ir, const char *_body, const int _responseCode, const char *_ifModifiedSince, const bool _manageBuffer):
	InternetResource(_db, _ir, _body, _responseCode, _ifModifiedSince, 0, _manageBuffer)
{
	//these will be cleared on destruction of the program (static)
	pthread_mutex_lock(&m_hFirst_mutex);
	if (!m_fgURLs.size()) {
		DEBUGPRINT("[%s]: HTMLPage static load of [fltsURLs]", DEBUG_LINE, m_ir->domain()->m_domain);
		m_fgURLs.addFilter(new Filter(58,10,"<[^>]+ href *= *['\"]?([^\"' <>]{1,1024})","\\1",0,"any href (a, image map, etc)",regex::NOCASE,false));
		m_fgURLs.addFilter(new Filter(59,20,"document.location *= *['\"]([^\"' <>,)]{1,1024})","\\1",0,"javascript",regex::NOCASE,false));
		m_fgURLs.addFilter(new Filter(60,30,"<[^>]+ src *= *['\"]?([^\"' <>]{1,1024})","\\1",0,"src (iframe, frame, link, etc)",regex::NOCASE,false));
		m_fgURLs.addFilter(new Filter(68,40,"window.open\\( *['\"]?([^\"'<>, )]{1,1024})","\\1",0,"popups",regex::NOCASE,false));
		m_fgURLs.addFilter(new Filter(70,50,"http://[^ \"'<>]{1,1024}","\\0",0,"absolute links",regex::NOCASE,false));
		//<meta http-equiv="refresh" content="0 ; url=http://www.pan-international.org/panint/?q=node/33"> 
	}
	pthread_mutex_unlock(&m_hFirst_mutex);
	memoryDelta((int)sizeof(HTMLPage), this);  //inform the domain which will, in turn, inform the Spider
}

HTMLPage::~HTMLPage() {
	memoryDelta(-(int)sizeof(HTMLPage), this); //inform the domain which will, in turn, inform the Spider
}

vector<const char*> *HTMLPage::links() {
	return InternetResource::links(&m_fgURLs);
}