// Site Mapper
//
// Douglas Thrift
//
// $Id$

#include <cxx/standard.hh>

#include <api/pcre/regex.hpp>
#include <api/exename.hpp>
#include <app/simple.hpp>

#include "SiteMapper.hpp"

int Main(const app::Options &options)
{
	SiteMapper::program = api::GetExecutablePath().GetName();

	cse::String siteIndex, siteMap;
	api::Pcre::RegEx index(_B("^-index=(.+)$")), map(_B("^-map=(.+)$"));

	_foreach (const app::ArgumentList, arg, app::GetArguments())
	{
		api::Pcre::RegEx::Match match;

		if (match = index(*arg))
		{
			siteIndex = match[1];
		}
		else if (match = map(*arg))
		{
			siteMap = match[1];
		}
		else if (*arg == _B("-D"))
		{
			if (!SiteMapper::debug) SiteMapper::debug = true;
		}
	}

	if (!siteIndex.IsEmpty() && !siteMap.IsEmpty())
		SiteMapper mapper(siteIndex, siteMap);
	else
		api::Cout << _B("Usage: ") << SiteMapper::program << _B(" -index=index -map=map [-D]") << ios::NewLine;

	return 0;
}

SiteMapper::SiteMapper(const cse::String &siteIndex, const cse::String &siteMap)
{
	oldMap(siteMap);
	newIndex(siteIndex);
	newMap(siteMap);
}

cse::String SiteMapper::program;
bool SiteMapper::debug(false);

void SiteMapper::oldMap(const cse::String &siteMap)
{
	ext::Handle<xml::Document> document(xml::Parse(siteMap));
	ext::Handle<xml::Node> list(*document/_B("page")/_B("section")/_B("list"));

	comment = *document/_B("comment()");

	if (debug) api::Cerr << _B("comment = ") << comment << ios::NewLine;

	oldMap(pages, list);
}

void SiteMapper::oldMap(ext::Vector<Page> &pages, xml::Node* list)
{
	xml::NodeSet nodes(*list/_B("item"));

	_foreach (xml::NodeSet, node, nodes)
	{
		cse::String url(**node/_B("link")/_B("@address")), title(**node/_B("link"));
		Page page(url, title);
		ext::Handle<xml::Node> list(**node/_B("list"));

		if (!list.IsEmpty()) oldMap(page.GetChildren(), list);

		pages.InsertLast(page);
	}
}

void SiteMapper::newIndex(const cse::String &siteIndex)
{
	ext::Handle<xml::Document> document(xml::Parse(siteIndex));
	xml::NodeSet nodes(*document/_B("index")/_B("page"));

	_foreach (xml::NodeSet, node, nodes)
	{
		_S<ios::String> address(**node/_B("address"));
	 	cse::String port(**node/_B("port"));

		if (!port.IsEmpty()) address << _B(":") << port;

		cse::String path(**node/_B("path")), title(**node/_B("title"));
		Page page(address, path, title);
		static api::Pcre::RegEx blog(_B("^Douglas\\sThrift's\\sWebsite\\s\\|\\sDouglas\\sThrift's\\sBlog:\\s(.+)$")), page_(_B("^Douglas\\sThrift's.+Website\\s\\|\\s(.+)$"));

		if (api::Pcre::RegEx::Match match = blog(page.GetTitle()))
		{
			static api::Pcre::RegEx archives(_B("^\\w+\\s\\d{4}\\sArchives$"));

			if (archives(match[1]))
			{
				page.SetTitle(match[1]);

				if (newIndex(pages, page))
					continue;
			}
			else
				continue;
		}
		else if (api::Pcre::RegEx::Match match = page_(page.GetTitle()))
		{
			page.SetTitle(match[1]);

			if (newIndex(pages, page))
				continue;
		}
		else
			continue;

		newPages[page.GetAddress()][page.GetChildOf()].InsertLast(page);
	}
}

bool SiteMapper::newIndex(ext::Vector<Page> &pages, Page &page)
{
	_foreach (ext::Vector<Page>, page_, pages)
	{
		if (*page_ == page.GetAddress())
		{
			if (*page_ == page)
			{
				page.SetChildren(page_->GetChildren());

				*page_ = page;

				api::Cout << _B("Updated: ") << page.GetUrl() << ios::NewLine;

				return true;
			}
			else if (page.GetPath().StartsWithAll(page_->GetPath()))
			{
				page.SetChildOf(page_->GetPath());

				api::Pcre::RegEx title(_S<ios::String>() << _B("^") << page_->GetTitle() << "\\s\\|\\s(.+)$");

				if (api::Pcre::RegEx::Match match = title(page.GetTitle()))
					page.SetTitle(match[1]);

				return newIndex(page_->GetChildren(), page);
			}
		}
	}

	return false;
}

void SiteMapper::newMap(const cse::String &siteMap)
{
	_S<api::FileWriter> file(siteMap);
	_S<ios::FormatWriter> fout(file);
	_S<xml::TextWriter> xml(file);

	// XXX: xml::TextWriter should have this kind of stuff, no?
	fout << ios::NewLine << _B("<?xml-stylesheet type=\"text/xsl\" href=\"stylesheets/sitemap.xsl\"?>") << ios::NewLine << _B("<!DOCTYPE page SYSTEM \"stylesheets/page.dtd\">");

	xml.OutputComment(comment);

	xml::ScopeElement page(xml, _B("page"));

	xml.OpenElement(_B("title"));
	xml.OutputText(_B("Sitemap"));
	xml.CloseElement();

	xml::ScopeElement section(xml, _B("section")), list(xml, _B("list"));

	_foreach (ext::Vector<Page>, page, pages)
	{
		if (newPages.Contains(page->GetAddress()))
			newMap(page->GetChildren(), page->GetPath(), newPages.Find(page->GetAddress())->Second());

		xml << *page;
	}
}

void SiteMapper::newMap(ext::Vector<Page> &pages, const cse::String &childOf, ext::RedBlackMap<cse::String, ext::Vector<Page>, LessThan> &newPages)
{
	_foreach (ext::Vector<Page>, page, pages)
		newMap(page->GetChildren(), page->GetPath(), newPages);

	_foreach (ext::Vector<Page>, page, newPages[childOf])
	{
		api::Cout << _B("Added: ") << page->GetUrl() << ios::NewLine;

		pages.InsertLast(*page);
	}

	newPages.Remove(childOf);
}