#include "Report_DomainSummary.h"

Report_DomainSummary::Report_DomainSummary(const SpiderManager *_sm, const int _domainId, const vector<unsigned int> *_pageIds): 
	Report(_sm), 
	m_domainId(_domainId), 
	m_pageIds(_pageIds) 
{
	//find the relevant domain
	InternetResource *ir = 0;
	vector<Domain*>::const_iterator i = m_sm->m_domains.begin();
	StringMap<size_t>::const_iterator iStat;
	const Parser *parser;
	const Filter *f;
	const FilterGroup *fg;

	while (i != m_sm->m_domains.end() && (*i)->m_domainID != m_domainId) i++;
	if (i != m_sm->m_domains.end()) {
		Domain *domain = *i;
		opentag("report");
			//parameters
			addAttribute("domainId", domain->m_domainID);
			write(" pageIds=\"");
			char str[32];
			for (vector<unsigned int>::const_iterator j = m_pageIds->begin(); j != m_pageIds->end(); j++) {
				_SNPRINTF(str, 32, "%u", *j);
				write(str);
				write(",");
			}
			write("\"");

			//domain
			opentag("domain");
				addAttribute("id",         domain->m_domainID);
				addAttribute("name",       domain->m_domain);
				addAttribute("repeaters",  domain->m_repeaters);
				addAttribute("memory",     domain->m_memoryUsage);
				addAttribute("startURL",   domain->m_startURL);
				addAttribute("persistent", domain->m_persistent);
				opentag("parsers");
					for (vector<pair<FilterGroup*, Parser*> >::const_iterator iParser = domain->m_parsers.begin(); iParser != domain->m_parsers.end(); iParser++) {
						opentag("parser");
						parser = iParser->second;
						if (parser) {
							addAttribute("pagegroupid", parser->m_pagegroupid);
							addAttribute("parsertype",  parser->parsertype());
							addAttribute("type",        parser->m_type);
							for (FilterGroup::const_iterator iFilter = iParser->first->begin(); iFilter != iParser->first->end(); iFilter++) {
								f = iFilter->second;
								opentag("area");
									if (f) CDATA(f->regEx());
								closetag("area");
							}
						}
						closetag("parser");
					}
				closetag("parsers");
				opentag("browseAreas");
					fg = domain->m_browseAreas;
					if (fg) {
						for (FilterGroup::const_iterator iFilter = domain->m_browseAreas->begin(); iFilter != domain->m_browseAreas->end(); iFilter++) {
							f = iFilter->second;
							opentag("area");
								if (f) CDATA(f->regEx());
							closetag("area");
						}
					}
				closetag("browseAreas");
				opentag("pages");
					page(domain->m_rootPage);
				closetag("pages");
				opentag("repeaters");
					for (StringMap<InternetURIRequest*>::const_iterator iPage=domain->m_pages.begin(); iPage!=domain->m_pages.end(); iPage++)
						if (iPage->second->isRepeating()) page(iPage->second);
				closetag("repeaters");
				opentag("codes");
					for (map<int, size_t>::const_iterator iCode=domain->m_HTTPReturnCodes.begin(); iCode!=domain->m_HTTPReturnCodes.end(); iCode++) {
						if (iCode->first!=200) {
							opentag("code");
							addAttribute("count", iCode->second);
							addAttribute("id", iCode->first);
							for (StringMap<InternetURIRequest*>::const_iterator iPage = domain->m_pages.begin(); iPage != domain->m_pages.end(); iPage++) {
								ir = iPage->second->m_ir;
								if (ir && iCode->first == ir->m_responseCode) page(iPage->second);
							}
							closetag("code");
						}
					}
				closetag("codes");
				opentag("resourceSizes");
					for (map<size_t, size_t>::const_iterator i = domain->m_resourceSizes.begin(); i != domain->m_resourceSizes.end(); i++) {
						opentag("size");
						addAttribute("kb",    i->first);
						addAttribute("count", i->second);
						closetag("size");
					}
				closetag("resourceSizes");
				opentag("unknownProtocols");
					for (iStat = domain->m_unknownProtocol.begin(); iStat != domain->m_unknownProtocol.end(); iStat++ ) {
						opentag("unknownProtocol");CDATA(iStat->first);closetag("unknownProtocol");
					}
				closetag("unknownProtocols");
				opentag("unknownFileExtensions");
					for (iStat = domain->m_unknownFileExtension.begin(); iStat != domain->m_unknownFileExtension.end(); iStat++ ) {
						opentag("unknownFileExtension");CDATA(iStat->first);closetag("unknownFileExtension");
					}
				closetag("unknownFileExtensions");
				opentag("unknownCharSets");
					for (iStat = domain->m_unknownCharSet.begin(); iStat != domain->m_unknownCharSet.end(); iStat++ ) {
						opentag("unknownCharSet");CDATA(iStat->first);closetag("unknownCharSet");
					}
				closetag("unknownCharSets");
				opentag("unknownContentTypes");
					for (iStat = domain->m_unknownContentType.begin(); iStat != domain->m_unknownContentType.end(); iStat++ ) {
						opentag("unknownContentType");CDATA(iStat->first);closetag("unknownContentType");
					}
				closetag("unknownContentTypes");
				opentag("unknownResponseProtocols");
					for (iStat = domain->m_unknownResponseProtocol.begin(); iStat != domain->m_unknownResponseProtocol.end(); iStat++ ) {
						opentag("unknownResponseProtocol");CDATA(iStat->first);closetag("unknownResponseProtocol");
					}
				closetag("unknownResponseProtocols");
				opentag("unknownLanguages");
					for (iStat = domain->m_unknownLanguage.begin(); iStat != domain->m_unknownLanguage.end(); iStat++ ) {
						opentag("unknownLanguage");CDATA(iStat->first);closetag("unknownLanguage");
					}
				closetag("unknownLanguages");
				opentag("unknownContentEncodings");
					for (iStat = domain->m_unknownContentEncoding.begin(); iStat != domain->m_unknownContentEncoding.end(); iStat++ ) {
						opentag("unknownContentEncoding");CDATA(iStat->first);closetag("unknownContentEncoding");
					}
				closetag("unknownContentEncodings");
				opentag("unknownCompressions");
					for (iStat = domain->m_unknownCompression.begin(); iStat != domain->m_unknownCompression.end(); iStat++ ) {
						opentag("unknownCompression");CDATA(iStat->first);closetag("unknownCompression");
					}
				closetag("unknownCompressions");
				opentag("unknownConversationEndSchemes");
					for (iStat = domain->m_unknownConversationEndScheme.begin(); iStat != domain->m_unknownConversationEndScheme.end(); iStat++ ) {
						opentag("unknownConversationEndScheme");CDATA(iStat->first);closetag("unknownConversationEndScheme");
					}
				closetag("unknownConversationEndSchemes");
			closetag("domain");
		closetag("report");
	}
}

void Report_DomainSummary::page(const InternetURIRequest *p) {
	if (p->m_ir) {
		//make sure that this page has not been displayed already
		vector<unsigned int>::const_iterator iExpanded=m_expanded.begin();
		const unsigned int id = p->m_ir->m_id;
		while (iExpanded!=m_expanded.end() && *iExpanded!=id) iExpanded++;
		if (iExpanded == m_expanded.end()) {
			//ok show page
			m_expanded.push_back(id);
			opentag("page");
				addAttribute("id", id);
				addAttribute("url", p->absoluteURL(), true);
				addAttribute("httpResponseCode", p->m_ir->m_responseCode);
				addAttribute("checkDigit", p->m_ir->m_checkDigit);
				addAttribute("childcount", p->size());
				//addAttribute("memory", p->m_memoryUsage);
				//show children only if this id is in the expanded list
				vector<unsigned int>::const_iterator iIDs = m_pageIds->begin();
				while (iIDs != m_pageIds->end() && *iIDs != id) iIDs++;
				if (iIDs != m_pageIds->end())
					//the page ID is in the expanded list so iterate through the children
					for (InternetURIRequest::const_iterator iPage=p->begin(); iPage!=p->end(); iPage++) 
						page(*iPage);
			closetag("page");
		}
	} else {
		//just a link, not retrieved yet
		opentag("page");
			addAttribute("url", p->absoluteURL(), true);
		closetag("page");
	}
}