ILM: Scheduled job and server to get all pages that is not updated from a past date [https://devblog.gosso.se/?p=953]

ILMPageCheckJob.cs

using EPiServer;
using EPiServer.Core;
using EPiServer.PlugIn;
using EPiServer.Security;
using EPiServer.ServiceLocation;
using EPiServer.Web;
using EPiServer.Web.Routing;
using Gosso.Mvc.Business.Initialization;
using Gosso.Mvc.Helpers;
using log4net;
using System;
using System.Configuration;
using System.Diagnostics;
using System.Linq;
using System.Reflection;
using System.Text;

namespace Gosso.Mvc.Business
{
    [ScheduledPlugIn(DisplayName = "ILM: Job to evaluate updated pages", SortIndex = 100)]
    public class ILMPageCheckJob : EPiServer.Scheduler.ScheduledJobBase
    {
        private readonly ILog _log = LogManager.GetLogger(MethodBase.GetCurrentMethod().DeclaringType);
        private static bool _stop;
        private string _returnMessage = string.Empty;

        private readonly IContentRepository _contentRepository;
        private readonly IUserImpersonation _userImpersonation;
        private readonly PageUpdateService _pageUpdateService;
        private readonly ISiteDefinitionRepository _siteDefinitionRepository;
        private readonly UrlResolver _urlResolver;

        public PageUpdateJob()
            : base()
        {
            // Make the job interruptable
            this.IsStoppable = true;

            _contentRepository = ServiceLocator.Current.GetInstance<IContentRepository>();
            _userImpersonation = ServiceLocator.Current.GetInstance<IUserImpersonation>();
            _pageUpdateService = ServiceLocator.Current.GetInstance<PageUpdateService>();
            _siteDefinitionRepository = ServiceLocator.Current.GetInstance<ISiteDefinitionRepository>();
            _urlResolver = ServiceLocator.Current.GetInstance<UrlResolver>();
        }

        public override string Execute()
        {
            Stopwatch sw = Stopwatch.StartNew();
            _log.Info("PageUpdate Job started at: " + DateTime.Now);
            _stop = false;
            try
            {
                string RunBy = "Job executed manually";
                if (PrincipalInfo.CurrentPrincipal.Identity.Name == string.Empty)
                {
                    string userName = ConfigurationManager.AppSettings["JobExecutorUserName"] as string ?? string.Empty;
                    PrincipalInfo.CurrentPrincipal = _userImpersonation.CreatePrincipal(userName);
                    RunBy = "Job executed by Scheduler";
                }

                _returnMessage += "<br/>" + RunBy + "<br/>";

                int count = _siteDefinitionRepository.List().Count();

                int cc = 1;
                foreach (SiteDefinition site in _siteDefinitionRepository.List())
                {
                    if (!ContentReference.IsNullOrEmpty(site.StartPage))
                    {
                        this.OnStatusChanged("Evaluating " + site.Name + " - site " + cc++ + " of total " + count);
                        HandleSite(site.Name, site.SiteUrl, site.StartPage);
                    }
                }
            }
            catch (Exception ex)
            {
                _log.Debug("Error in Page Update job", ex);
                throw new Exception(_returnMessage);
            }

            _log.Info("Page Update Job ended at: " + DateTime.Now);
            sw.Stop();

            _returnMessage += "Time for execution: " + sw.ElapsedMilliseconds + " ms<br/>";

            // Return message indicating finished status  
            return string.Format(cctotal + "pages not updated<br/>" + _returnMessage).Replace("<br/>", " | ");
        }

        private const string htmlTemplate = "<tr><td>{2}</td><td>{1}</td><td>{3}</td><td><a href=\"{0}\">{0}</a></td><td><a href=\"{5}\">{5}</a></td><td>{4}</td></tr>";

        private void HandleSite(string siteName, Uri siteUri, ContentReference startPage)
        {

            if (_stop)
                return;

            StringBuilder outputHtml = new StringBuilder();
            outputHtml.Append("<table width='100%' cellspacing='0' cellpadding='2'border='0'>");
            outputHtml.Append("<tr><th><b>changed</b></th><th><b>PageTypeName</b></th>" +
                " <th><b>Changed By</b></th><th><b>Link for content</b></th><th><b>Web Resource</b></th></tr>");
            var pageList = _pageUpdateService.GetPagesNotUpdated(DateTime.Now.AddMonths(-13), startPage);
            
            var siteUrl = SiteDefinition.Current.SiteUrl;

             EditUrlResolver editUrlResolver = ServiceLocator.Current.GetInstance<EditUrlResolver>();
            if (pageList.Count > 0)
            {
                pageList = pageList.OrderBy(x => x.Created).ToList();
                foreach (var page in pageList)
                {
                    if (_stop)
                        return;
                    string pageUrl = editUrlResolver.GetEditViewUrl(page.ContentLink).ToString();           
                    outputHtml.AppendLine(string.Format(htmlTemplate, siteUrl + pageUrl, page.PageTypeName, page.Changed.ToString("yyyy-MM-dd"), page.ChangedBy, page.PageName, _urlResolver.GetUrl(page.ContentLink)));
                }
                cctotal = pageList.Count;
                
                outputHtml.Append("</table>");
                //implement your own emailing
                MailHelper.SendMail("ILM: Page granting found "+ pageList.Count+ " pages",
                    outputHtml.ToString(), "[email protected]",
                    "[email protected]", 
                    IsBodyHtml:true);
            }
        }
        

    public int cctotal
        { get; set; }

   
        public override void Stop()
        {
            _stop = true;
        }
    }

}

PageUpdateService.cs

using EPiServer;
using EPiServer.Core;
using EPiServer.Security;
using EPiServer.ServiceLocation;
using Gosso.Mvc.Models.Pages;
using System;
using System.Collections.Generic;
using System.Linq;

namespace Gosso.Mvc.Business.Initialization
{
    [ServiceConfiguration(ServiceType = typeof(PageUpdateService),Lifecycle = ServiceInstanceScope.Singleton)]
    public class PageUpdateService
    {

        private readonly IContentRepository _contentRepository;

        public PageUpdateService(IContentRepository contentRepository)
        {
            _contentRepository = contentRepository;
        }
        /// <summary>
        /// Service to get all pages that is not updated from a past date
        /// </summary>
        /// <param name="date"></param>
        /// <param name="startPage"></param>
        /// <returns></returns>
        public List<PageData> GetPagesNotUpdated(DateTime date, ContentReference startPage)
        {

            var pageList = new List<PageData>();

            foreach (var pr in _contentRepository.GetDescendents(startPage))
            {
                BasePageData pd = null;
                try
                {
                    pd = _contentRepository.Get<BasePageData>(pr);
                }
                catch
                {
                    pd = null;
                }
                if (checkAccess("Everyone", pd))
                {
                    if (pd != null)
                    {
                        foreach (var language in pd.ExistingLanguages)
                        {
                            if (_contentRepository.Get<PageData>(pd.ContentGuid, language) is BasePageData pdLang && pdLang.CheckPublishedStatus(PagePublishedStatus.Published) && pdLang.Changed < date)
                            {
                                if (!pageList.Contains(pdLang))
                                {
                                    if (pdLang.LinkType != PageShortcutType.External && CheckPageType(pdLang))
                                    {
                                        pageList.Add(pdLang);
                                    }
                                }
                            }
                        } // foreach lang
                    }
                }
            }
            return pageList;
        }

        private bool CheckPageType(BasePageData page)
        {
            //if you rather dont want to evaluate som pagetypes, put them here
            if (page is NewsArticlePage
                || page is NewsArticleListPage
            )
                return false;

            return true;
        }
        private bool checkAccess(string role, PageData pd)
        {
            if (pd == null)
                return false;

            try
            {
                var list = pd.ACL.Entries.FirstOrDefault(x => x.Name == "Everyone");
                if (list != null && list.Access == AccessLevel.Read)
                {
                    return true;
                }
            }
            catch (Exception)
            {

                return false;
            }

            return false;
        }
    }
}