Umbraco-CMS/src/Umbraco.Web/Routing/PublishedContentRequestEngine.cs

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading;
using System.Globalization;
using System.IO;

using Umbraco.Core;
using Umbraco.Core.IO;
using Umbraco.Core.Logging;

using umbraco;
using umbraco.cms.businesslogic.web;
using umbraco.cms.businesslogic.language;
using umbraco.cms.businesslogic.template;
using umbraco.cms.businesslogic.member;

namespace Umbraco.Web.Routing
{
	internal class PublishedContentRequestEngine
	{
		private PublishedContentRequest _pcr;
		private RoutingContext _routingContext;

		/// <summary>
		/// Initializes a new instance of the <see cref="PublishedContentRequestEngine"/> class with a content request.
		/// </summary>
		/// <param name="pcr">The content request.</param>
		public PublishedContentRequestEngine(PublishedContentRequest pcr)
		{
			_pcr = pcr;
			_routingContext = pcr.RoutingContext;

			var umbracoContext = _routingContext.UmbracoContext;
			if (_routingContext == null) throw new ArgumentException("pcr.RoutingContext is null.");
			if (umbracoContext == null) throw new ArgumentException("pcr.RoutingContext.UmbracoContext is null.");
			if (umbracoContext.RoutingContext != _routingContext) throw new ArgumentException("RoutingContext confusion.");
			// no! not set yet.
			//if (umbracoContext.PublishedContentRequest != _pcr) throw new ArgumentException("PublishedContentRequest confusion.");
		}

		#region Public

		/// <summary>
		/// Prepares the request.
		/// </summary>
		public void PrepareRequest()
		{
			// note - at that point the original legacy module did something do handle IIS custom 404 errors
			//   ie pages looking like /anything.aspx?404;/path/to/document - I guess the reason was to support
			//   "directory urls" without having to do wildcard mapping to ASP.NET on old IIS. This is a pain
			//   to maintain and probably not used anymore - removed as of 06/2012. @zpqrtbnk.
			//
			//   to trigger Umbraco's not-found, one should configure IIS and/or ASP.NET custom 404 errors
			//   so that they point to a non-existing page eg /redirect-404.aspx
			//   TODO: SD: We need more information on this for when we release 4.10.0 as I'm not sure what this means.

			//find domain
			FindDomain();

			// if request has been flagged to redirect then return
			// whoever called us is in charge of actually redirecting
			if (_pcr.IsRedirect)
				return;

			// set the culture on the thread - once, so it's set when running document lookups
			Thread.CurrentThread.CurrentUICulture = Thread.CurrentThread.CurrentCulture = _pcr.Culture;

			// find the document & template
			FindPublishedContentAndTemplate();

			// set the culture on the thread -- again, 'cos it might have changed due to a wildcard domain
			Thread.CurrentThread.CurrentUICulture = Thread.CurrentThread.CurrentCulture = _pcr.Culture;

			// find the rendering engine
			FindRenderingEngine();

			// trigger the Prepared event - at that point it is still possible to change about anything
			_pcr.OnPrepared();

			// set the culture on the thread -- again, 'cos it might have changed in the event handler
			Thread.CurrentThread.CurrentUICulture = Thread.CurrentThread.CurrentCulture = _pcr.Culture;

			// if request has been flagged to redirect then return
			// whoever called us is in charge of actually redirecting
			if (_pcr.IsRedirect)
				return;

			// safety
			if (!_pcr.HasPublishedContent)
				_pcr.Is404 = true;

			// handle 404 : return
			// whoever called us is in charge of doing what's appropriate
			if (_pcr.Is404)
				return;

			// can't go beyond that point without a PublishedContent to render
			// it's ok not to have a template, in order to give MVC a chance to hijack routes

			// assign the legacy page back to the docrequest
			// handlers like default.aspx will want it and most macros currently need it
			_pcr.UmbracoPage = new page(_pcr);

			// these two are used by many legacy objects
			_routingContext.UmbracoContext.HttpContext.Items["pageID"] = _pcr.PublishedContentId;
			_routingContext.UmbracoContext.HttpContext.Items["pageElements"] = _pcr.UmbracoPage.Elements;
		}

		/// <summary>
		/// Updates the request when there is no template to render the content.
		/// </summary>
		/// <remarks>This is called from Mvc when there's a document to render but no template.</remarks>
		public void UpdateRequestOnMissingTemplate()
		{
			// clear content
			var content = _pcr.PublishedContent;
			_pcr.PublishedContent = null;

			HandlePublishedContent(); // will go 404
			FindTemplate();
			FindRenderingEngine();

			// if request has been flagged to redirect then return
			// whoever called us is in charge of redirecting
			if (_pcr.IsRedirect)
				return;

			if (!_pcr.HasPublishedContent)
			{
				// means the engine could not find a proper document to handle 404
				// restore the saved content so we know it exists
				_pcr.PublishedContent = content;
				return;
			}

			if (!_pcr.HasTemplate)
			{
				// means we may have a document, but we have no template
				// at that point there isn't much we can do and there is no point returning
				// to Mvc since Mvc can't do much either
				return;
			}

			// assign the legacy page back to the docrequest
			// handlers like default.aspx will want it and most macros currently need it
			_pcr.UmbracoPage = new page(_pcr);

			// these two are used by many legacy objects
			_routingContext.UmbracoContext.HttpContext.Items["pageID"] = _pcr.PublishedContentId;
			_routingContext.UmbracoContext.HttpContext.Items["pageElements"] = _pcr.UmbracoPage.Elements;
		}

		#endregion

		#region Domain

		/// <summary>
		/// Finds the site root (if any) matching the http request, and updates the PublishedContentRequest accordingly.
		/// </summary>        
		/// <returns>A value indicating whether a domain was found.</returns>
		internal bool FindDomain()
		{
			const string tracePrefix = "FindDomain: ";

			// note - we are not handling schemes nor ports here.

			LogHelper.Debug<PublishedContentRequestEngine>("{0}Uri=\"{1}\"", () => tracePrefix, () => _pcr.Uri);

			// try to find a domain matching the current request
			var domainAndUri = DomainHelper.DomainMatch(Domain.GetDomains(), _pcr.Uri, false);

			// handle domain
			if (domainAndUri != null)
			{
				// matching an existing domain
				LogHelper.Debug<PublishedContentRequestEngine>("{0}Matches domain=\"{1}\", rootId={2}, culture=\"{3}\"",
												 () => tracePrefix,
												 () => domainAndUri.Domain.Name,
												 () => domainAndUri.Domain.RootNodeId,
												 () => domainAndUri.Domain.Language.CultureAlias);

				_pcr.Domain = domainAndUri.Domain;
				_pcr.DomainUri = domainAndUri.Uri;
				_pcr.Culture = new CultureInfo(domainAndUri.Domain.Language.CultureAlias);

				// canonical? not implemented at the moment
				// if (...)
				// {
				//  _pcr.RedirectUrl = "...";
				//  return true;
				// }
			}
			else
			{
				// not matching any existing domain
				LogHelper.Debug<PublishedContentRequestEngine>("{0}Matches no domain", () => tracePrefix);

				var defaultLanguage = Language.GetAllAsList().FirstOrDefault();
				_pcr.Culture = defaultLanguage == null ? CultureInfo.CurrentUICulture : new CultureInfo(defaultLanguage.CultureAlias);
			}

			LogHelper.Debug<PublishedContentRequestEngine>("{0}Culture=\"{1}\"", () => tracePrefix, () => _pcr.Culture.Name);

			return _pcr.Domain != null;
		}

		/// <summary>
		/// Looks for wildcard domains in the path and updates <c>Culture</c> accordingly.
		/// </summary>
		private void HandleWildcardDomains()
		{
			const string tracePrefix = "HandleWildcardDomains: ";

			if (!_pcr.HasPublishedContent)
				return;

			var nodePath = _pcr.PublishedContent.Path;
			LogHelper.Debug<PublishedContentRequestEngine>("{0}Path=\"{1}\"", () => tracePrefix, () => nodePath);
			var rootNodeId = _pcr.HasDomain ? _pcr.Domain.RootNodeId : (int?)null;
			var domain = DomainHelper.LookForWildcardDomain(Domain.GetDomains(), nodePath, rootNodeId);

			if (domain != null)
			{
				_pcr.Culture = new CultureInfo(domain.Language.CultureAlias);
				LogHelper.Debug<PublishedContentRequestEngine>("{0}Got domain on node {1}, set culture to \"{2}\".", () => tracePrefix,
					() => domain.RootNodeId, () => _pcr.Culture.Name);
			}
			else
			{
				LogHelper.Debug<PublishedContentRequestEngine>("{0}No match.", () => tracePrefix);
			}
		}

		#endregion

		#region Rendering engine

		/// <summary>
		/// Finds the rendering engine to use, and updates the PublishedContentRequest accordingly.
		/// </summary>
		internal void FindRenderingEngine()
		{
			//First, if there is no template, we will default to use MVC because MVC supports Hijacking routes which
			//sometimes don't require a template since the developer may want full control over the rendering. 
			//Webforms doesn't support this so MVC it is. MVC will also handle what to do if no template or hijacked route
			//is there (i.e. blank page)
			if (!_pcr.HasTemplate)
			{
				_pcr.RenderingEngine = RenderingEngine.Mvc;
				return;
			}

			//NOTE: Not sure how the alias is actually saved with a space as this shouldn't ever be the case? 
			// but apparently this happens. I think what should actually be done always is the template alias 
			// should be saved using the ToUmbracoAlias method and then we can use this here too, that way it
			// it 100% consistent. I'll leave this here for now until further invenstigation.
			var templateAlias = _pcr.Template.Alias.Replace(" ", string.Empty);
			//var templateAlias = _pcr.Template.Alias.ToUmbracoAlias(StringAliasCaseType.PascalCase);

			Func<DirectoryInfo, string, string[], RenderingEngine, bool> determineEngine =
				(directory, alias, extensions, renderingEngine) =>
				{
					//so we have a template, now we need to figure out where the template is, this is done just by the Alias field					
					//ensure it exists
					if (!directory.Exists) Directory.CreateDirectory(directory.FullName);
					var file = directory.GetFiles()
						.FirstOrDefault(x => extensions.Any(e => x.Name.InvariantEquals(alias + e)));

					if (file != null)
					{
						//it is mvc since we have a template there that exists with this alias
						_pcr.RenderingEngine = renderingEngine;
						return true;
					}
					return false;
				};

			//first determine if it is MVC, we will favor mvc if there is a template with the same name in both 
			// folders, if it is then MVC will be selected
			if (!determineEngine(
				new DirectoryInfo(IOHelper.MapPath(SystemDirectories.MvcViews)),
				templateAlias,
				new[] { ".cshtml", ".vbhtml" },
				RenderingEngine.Mvc))
			{
				//if not, then determine if it is webforms (this should def match if a template is assigned and its not in the MVC folder)
				// if it doesn't match, then MVC will be used by default anyways.
				determineEngine(
					new DirectoryInfo(IOHelper.MapPath(SystemDirectories.Masterpages)),
					templateAlias,
					new[] { ".master" },
					RenderingEngine.WebForms);
			}

		}

		#endregion

		#region Document and template

		/// <summary>
		/// Finds the Umbraco document (if any) matching the request, and updates the PublishedContentRequest accordingly.
		/// </summary>
		/// <returns>A value indicating whether a document and template were found.</returns>
		private bool FindPublishedContentAndTemplate()
		{
			const string tracePrefix = "FindPublishedContentAndTemplate: ";
			LogHelper.Debug<PublishedContentRequestEngine>("{0}Path=\"{1}\"", () => tracePrefix, () => _pcr.Uri.AbsolutePath);

			// run the document finders
			FindPublishedContent();

			// not handling umbracoRedirect here but after LookupDocument2
			// so internal redirect, 404, etc has precedence over redirect

			// handle not-found, redirects, access...
			HandlePublishedContent();

			// find a template
			FindTemplate();

			// handle umbracoRedirect
			FollowExternalRedirect();

			// handle wildcard domains
			HandleWildcardDomains();

			return _pcr.HasPublishedContent && _pcr.HasTemplate;
		}

		/// <summary>
		/// Tries to find the document matching the request, by running the IPublishedContentFinder instances.
		/// </summary>
		internal void FindPublishedContent()
		{
			const string tracePrefix = "FindPublishedContent: ";

			// look for the document
			// the first successful finder, if any, will set this.PublishedContent, and may also set this.Template
			// some finders may implement caching

			using (DisposableTimer.DebugDuration<PluginManager>(
				() => string.Format("{0}Begin finders", tracePrefix),
				() => string.Format("{0}End finders, {1}", tracePrefix, (_pcr.HasPublishedContent ? "a document was found" : "no document was found"))))
			{
				_routingContext.PublishedContentFinders.Any(lookup => lookup.TryFindDocument(_pcr));
			}

			// indicate that the published content (if any) we have at the moment is the
			// one that was found by the standard finders before anything else took place.
			_pcr.IsInitialPublishedContent = true;
		}

		/// <summary>
		/// Handles the published content (if any).
		/// </summary>
		/// <remarks>
		/// Handles "not found", internal redirects, access validation...
		/// things that must be handled in one place because they can create loops
		/// </remarks>
		private void HandlePublishedContent()
		{
			const string tracePrefix = "HandlePublishedContent: ";

			// because these might loop, we have to have some sort of infinite loop detection 
			int i = 0, j = 0;
			const int maxLoop = 8;
			do
			{
				LogHelper.Debug<PublishedContentRequestEngine>("{0}{1}", () => tracePrefix, () => (i == 0 ? "Begin" : "Loop"));

				// handle not found
				if (!_pcr.HasPublishedContent)
				{
					_pcr.Is404 = true;
					LogHelper.Debug<PublishedContentRequestEngine>("{0}No document, try last chance lookup", () => tracePrefix);

					// if it fails then give up, there isn't much more that we can do
					var lastChance = _routingContext.PublishedContentLastChanceFinder;
					if (lastChance == null || !lastChance.TryFindDocument(_pcr))
					{
						LogHelper.Debug<PublishedContentRequestEngine>("{0}Failed to find a document, give up", () => tracePrefix);
						break;
					}

					LogHelper.Debug<PublishedContentRequestEngine>("{0}Found a document", () => tracePrefix);
				}

				// follow internal redirects as long as it's not running out of control ie infinite loop of some sort
				j = 0;
				while (FollowInternalRedirects() && j++ < maxLoop) ;
				if (j == maxLoop) // we're running out of control
					break;

				// ensure access
				if (_pcr.HasPublishedContent)
					EnsurePublishedContentAccess();

				// loop while we don't have page, ie the redirect or access
				// got us to nowhere and now we need to run the notFoundLookup again
				// as long as it's not running out of control ie infinite loop of some sort

			} while (!_pcr.HasPublishedContent && i++ < maxLoop);

			if (i == maxLoop || j == maxLoop)
			{
				LogHelper.Debug<PublishedContentRequestEngine>("{0}Looks like we're running into an infinite loop, abort", () => tracePrefix);
				_pcr.PublishedContent = null;
			}

			LogHelper.Debug<PublishedContentRequestEngine>("{0}End", () => tracePrefix);
		}

		/// <summary>
		/// Follows internal redirections through the <c>umbracoInternalRedirectId</c> document property.
		/// </summary>
		/// <returns>A value indicating whether redirection took place and led to a new published document.</returns>
		/// <remarks>
		/// <para>Redirecting to a different site root and/or culture will not pick the new site root nor the new culture.</para>
		/// <para>As per legacy, if the redirect does not work, we just ignore it.</para>
		/// </remarks>
		private bool FollowInternalRedirects()
		{
			const string tracePrefix = "FollowInternalRedirects: ";

			if (_pcr.PublishedContent == null)
				throw new InvalidOperationException("There is no PublishedContent.");

			bool redirect = false;
			var internalRedirect = _pcr.PublishedContent.GetPropertyValue<string>("umbracoInternalRedirectId");

			if (!string.IsNullOrWhiteSpace(internalRedirect))
			{
				LogHelper.Debug<PublishedContentRequestEngine>("{0}Found umbracoInternalRedirectId={1}", () => tracePrefix, () => internalRedirect);

				int internalRedirectId;
				if (!int.TryParse(internalRedirect, out internalRedirectId))
					internalRedirectId = -1;

				if (internalRedirectId <= 0)
				{
					// bad redirect - log and display the current page (legacy behavior)
					//_pcr.Document = null; // no! that would be to force a 404
					LogHelper.Debug<PublishedContentRequestEngine>("{0}Failed to redirect to id={1}: invalid value", () => tracePrefix, () => internalRedirect);
				}
				else if (internalRedirectId == _pcr.PublishedContentId)
				{
					// redirect to self
					LogHelper.Debug<PublishedContentRequestEngine>("{0}Redirecting to self, ignore", () => tracePrefix);
				}
				else
				{
					// redirect to another page
					var node = _routingContext.PublishedContentStore.GetDocumentById(_routingContext.UmbracoContext, internalRedirectId);

					_pcr.PublishedContent = node;
					if (node != null)
					{
						redirect = true;
						LogHelper.Debug<PublishedContentRequestEngine>("{0}Redirecting to id={1}", () => tracePrefix, () => internalRedirectId);
					}
					else
					{
						LogHelper.Debug<PublishedContentRequestEngine>("{0}Failed to redirect to id={1}: no such published document", () => tracePrefix, () => internalRedirectId);
					}
				}
			}

			return redirect;
		}

		/// <summary>
		/// Ensures that access to current node is permitted.
		/// </summary>
		/// <remarks>Redirecting to a different site root and/or culture will not pick the new site root nor the new culture.</remarks>
		private void EnsurePublishedContentAccess()
		{
			const string tracePrefix = "EnsurePublishedContentAccess: ";

			if (_pcr.PublishedContent == null)
				throw new InvalidOperationException("There is no PublishedContent.");

			var path = _pcr.PublishedContent.Path;

			if (Access.IsProtected(_pcr.PublishedContentId, path))
			{
				LogHelper.Debug<PublishedContentRequestEngine>("{0}Page is protected, check for access", () => tracePrefix);

				System.Web.Security.MembershipUser user = null;
				try
				{
					user = System.Web.Security.Membership.GetUser();
				}
				catch (ArgumentException)
				{
					LogHelper.Debug<PublishedContentRequestEngine>("{0}Membership.GetUser returned ArgumentException", () => tracePrefix);
				}

				if (user == null || !Member.IsLoggedOn())
				{
					LogHelper.Debug<PublishedContentRequestEngine>("{0}Not logged in, redirect to login page", () => tracePrefix);
					var loginPageId = Access.GetLoginPage(path);
					if (loginPageId != _pcr.PublishedContentId)
						_pcr.PublishedContent = _routingContext.PublishedContentStore.GetDocumentById(_routingContext.UmbracoContext, loginPageId);
				}
				else if (!Access.HasAccces(_pcr.PublishedContentId, user.ProviderUserKey))
				{
					LogHelper.Debug<PublishedContentRequestEngine>("{0}Current member has not access, redirect to error page", () => tracePrefix);
					var errorPageId = Access.GetErrorPage(path);
					if (errorPageId != _pcr.PublishedContentId)
						_pcr.PublishedContent = _routingContext.PublishedContentStore.GetDocumentById(_routingContext.UmbracoContext, errorPageId);
				}
				else
				{
					LogHelper.Debug<PublishedContentRequestEngine>("{0}Current member has access", () => tracePrefix);
				}
			}
			else
			{
				LogHelper.Debug<PublishedContentRequestEngine>("{0}Page is not protected", () => tracePrefix);
			}
		}

		/// <summary>
		/// Finds a template for the current node.
		/// </summary>
		private void FindTemplate()
		{
			// NOTE: at the moment there is only 1 way to find a template, and then ppl must
			// use the Prepared event to change the template if they wish. Should we also
			// implement an ITemplateFinder logic?

			const string tracePrefix = "FindTemplate: ";

			if (_pcr.PublishedContent == null)
				throw new InvalidOperationException("There is no PublishedContent.");

			// read the alternate template alias, from querystring, form, cookie or server vars,
			// only if the published content is the initial once, else the alternate template
			// does not apply
			string altTemplate = _pcr.IsInitialPublishedContent 
				? _routingContext.UmbracoContext.HttpContext.Request["altTemplate"] 
				: null;

			if (string.IsNullOrWhiteSpace(altTemplate))
			{
				// we don't have an alternate template specified. use the current one if there's one already,
				// which can happen if a content lookup also set the template (LookupByNiceUrlAndTemplate...),
				// else lookup the template id on the document then lookup the template with that id.

				if (_pcr.HasTemplate)
				{
					LogHelper.Debug<PublishedContentRequest>("{0}Has a template already, and no alternate template.", () => tracePrefix);
					return;
				}

				// TODO: When we remove the need for a database for templates, then this id should be irrelavent,
				// not sure how were going to do this nicely.

				var templateId = _pcr.PublishedContent.TemplateId;

				if (templateId > 0)
				{
					LogHelper.Debug<PublishedContentRequestEngine>("{0}Look for template id={1}", () => tracePrefix, () => templateId);
					// don't use the Template ctor as the result is not cached... instead use this static method
					var template = Template.GetTemplate(templateId);
					if (template == null)
						throw new InvalidOperationException("The template with Id " + templateId + " does not exist, the page cannot render");
					_pcr.Template = template;
					LogHelper.Debug<PublishedContentRequestEngine>("{0}Got template id={1} alias=\"{2}\"", () => tracePrefix, () => template.Id, () => template.Alias);
				}
				else
				{
					LogHelper.Debug<PublishedContentRequestEngine>("{0}No specified template.", () => tracePrefix);
				}
			}
			else
			{
				// we have an alternate template specified. lookup the template with that alias
				// this means the we override any template that a content lookup might have set
				// so /path/to/page/template1?altTemplate=template2 will use template2

				// ignore if the alias does not match - just trace

				if (_pcr.HasTemplate)
					LogHelper.Debug<PublishedContentRequestEngine>("{0}Has a template already, but also an alternate template.", () => tracePrefix);
				LogHelper.Debug<PublishedContentRequestEngine>("{0}Look for alternate template alias=\"{1}\"", () => tracePrefix, () => altTemplate);

				var template = Template.GetByAlias(altTemplate, true);
				if (template != null)
				{
					_pcr.Template = template;
					LogHelper.Debug<PublishedContentRequestEngine>("{0}Got template id={1} alias=\"{2}\"", () => tracePrefix, () => template.Id, () => template.Alias);
				}
				else
				{
					LogHelper.Debug<PublishedContentRequestEngine>("{0}The template with alias=\"{1}\" does not exist, ignoring.", () => tracePrefix, () => altTemplate);
				}
			}

			if (!_pcr.HasTemplate)
			{
				LogHelper.Debug<PublishedContentRequestEngine>("{0}No template was found.", () => tracePrefix);

				// initial idea was: if we're not already 404 and UmbracoSettings.HandleMissingTemplateAs404 is true
				// then reset _pcr.Document to null to force a 404.
				//
				// but: because we want to let MVC hijack routes even though no template is defined, we decide that
				// a missing template is OK but the request will then be forwarded to MVC, which will need to take
				// care of everything.
				//
				// so, don't set _pcr.Document to null here
			}
			else
			{
				LogHelper.Debug<PublishedContentRequestEngine>("{0}Running with template id={1} alias=\"{2}\"", () => tracePrefix, () => _pcr.Template.Id, () => _pcr.Template.Alias);
			}
		}

		/// <summary>
		/// Follows external redirection through <c>umbracoRedirect</c> document property.
		/// </summary>
		/// <remarks>As per legacy, if the redirect does not work, we just ignore it.</remarks>
		private void FollowExternalRedirect()
		{
			if (_pcr.HasPublishedContent)
			{
				var redirectId = _pcr.PublishedContent.GetPropertyValue<int>("umbracoRedirect", -1);
				
				string redirectUrl = "#";
				if (redirectId > 0)
					redirectUrl = _routingContext.NiceUrlProvider.GetNiceUrl(redirectId);
				if (redirectUrl != "#")
					_pcr.RedirectUrl = redirectUrl;
			}
		}
	
		#endregion
	}
}