Performance update with HTMLAgilityPack

Added Branches
Added LinksToScrape
Added ScrapedLinks

UWLecturePlan/Controllers/LectureController.cs

@@ -25,6 +25,7 @@ namespace UWLecturePlan.Controllers
             model.LectureEvents = db.LectureEvents.Include(x=>x.Lecture)
                 .Where(x => x.From >= from)
                 .Where(x => x.Lecture.Semester == model.CurrentSemester)
+                .Where(x => x.Lecture.Branch == model.BranchFilter || model.BranchFilter == null)
                 .ToList();
 
             if (model.LocationFilter != null)
@@ -33,6 +34,8 @@ namespace UWLecturePlan.Controllers
                     .Where(x => x.Location.Contains(model.LocationFilter)).ToList();
             }
 
+            model.Branches = db.LectureEvents.Select(x => x.Lecture.Branch).Distinct().OrderBy(x=>x).ToList();
+          
 
             return View(model);
         }

UWLecturePlan/Models/LecturesViewModel.cs

@@ -7,8 +7,12 @@ namespace UWLecturePlan.Models
 
         public string CurrentSemester { get; set; }
 
-        public string LocationFilter { get; set; }
+        public string? LocationFilter { get; set; }
+
+        public string? BranchFilter { get; set; }
 
         public List<LectureEvent> LectureEvents { get; set; }
+
+        public List<string> Branches { get; set; }
     }
 }

UWLecturePlan/Views/Lecture/Index.cshtml

@@ -1,4 +1,21 @@
-@model LecturesViewModel
+@using System.Text.RegularExpressions
+@model LecturesViewModel
+@{
+    string GetBranchName(string branch)
+    {
+        Regex regex = new Regex(@".* - (.*)");
+
+        var match = regex.Match(branch);
+        if (match.Success)
+        {
+            return match.Groups[1].Value;
+        }
+        else
+        {
+            return branch;
+        }
+    }
+}
 
 
 <style>
@@ -30,9 +47,17 @@
 <form method="get" >
     Semester: <input type="text" asp-for="CurrentSemester" style="width:3.5em;text-align:center;" maxlength="5" />
     Ort: <input type="text" asp-for="LocationFilter" />
+    Studiengang: 
+    <select type="text" asp-for="BranchFilter" >
+        <option value="">Alle</option>
+        @foreach (var branch in Model.Branches.OrderBy(x => GetBranchName(x)))
+        {
+                <option value="@branch">
+                @GetBranchName(branch)
+                </option>
+        }
+    </select>
     <input type="submit" value="Filter" />
-
-    
 </form>
 
 

UWLib/Lecture.cs

@@ -12,11 +12,11 @@ namespace UWLib
     public class Lecture
     {
         public int Id { get; set; }
-
         public string Semester { get; set; }
         public string Title { get; set; }
         public string Url { get; set; }
         public string? Description { get; set; }
         public List<LectureEvent> Events { get; set; } = new List<LectureEvent>();
+        public string Branch { get; set; }
     }
 }

UWLib/LectureContext.cs

@@ -12,6 +12,10 @@ namespace UWLib
         public DbSet<Lecture> Lectures { get; set; }    
         public DbSet<LectureEvent> LectureEvents { get; set; }
 
+        public DbSet<ScrapedLink> ScrapedLinks { get; set; }
+
+        public DbSet<LinkToScrape> LinksToScrape { get; set; }
+
 
         public string DbPath { get; }
 
@@ -19,7 +23,7 @@ namespace UWLib
         public LectureContext(string path) 
         {
             DbPath = path;
-            this.Database.Migrate();
+            Database.Migrate();
         }
 
         public LectureContext()
@@ -29,7 +33,7 @@ namespace UWLib
             DbPath = System.IO.Path.Join(path, "lecture.db");
             DbPath = "lecture.db";
 
-            this.Database.Migrate();
+            Database.Migrate();
         }
 
         // The following configures EF to create a Sqlite database file in the

UWLib/LinkToScrape.cs

@@ -0,0 +1,16 @@
+using System;
+using System.Collections.Generic;
+using System.ComponentModel.DataAnnotations;
+using System.Linq;
+using System.Text;
+using System.Threading.Tasks;
+
+namespace UWLib
+{
+    public class LinkToScrape
+    {
+        [Key]
+        public string Url { get; set; }
+        public int Sort { get; set; }
+    }
+}

UWLib/Migrations/20241012131142_ScrapedLinks.Designer.cs

@@ -0,0 +1,108 @@
+// <auto-generated />
+using System;
+using Microsoft.EntityFrameworkCore;
+using Microsoft.EntityFrameworkCore.Infrastructure;
+using Microsoft.EntityFrameworkCore.Migrations;
+using Microsoft.EntityFrameworkCore.Storage.ValueConversion;
+using UWLib;
+
+#nullable disable
+
+namespace UWLib.Migrations
+{
+    [DbContext(typeof(LectureContext))]
+    [Migration("20241012131142_ScrapedLinks")]
+    partial class ScrapedLinks
+    {
+        /// <inheritdoc />
+        protected override void BuildTargetModel(ModelBuilder modelBuilder)
+        {
+#pragma warning disable 612, 618
+            modelBuilder.HasAnnotation("ProductVersion", "8.0.10");
+
+            modelBuilder.Entity("UWLib.Lecture", b =>
+                {
+                    b.Property<int>("Id")
+                        .HasColumnType("INTEGER");
+
+                    b.Property<string>("Semester")
+                        .HasColumnType("TEXT");
+
+                    b.Property<string>("Description")
+                        .HasColumnType("TEXT");
+
+                    b.Property<string>("Title")
+                        .IsRequired()
+                        .HasColumnType("TEXT");
+
+                    b.Property<string>("Url")
+                        .IsRequired()
+                        .HasColumnType("TEXT");
+
+                    b.HasKey("Id", "Semester");
+
+                    b.ToTable("Lectures");
+                });
+
+            modelBuilder.Entity("UWLib.LectureEvent", b =>
+                {
+                    b.Property<int>("Id")
+                        .ValueGeneratedOnAdd()
+                        .HasColumnType("INTEGER");
+
+                    b.Property<DateTime>("From")
+                        .HasColumnType("TEXT");
+
+                    b.Property<int>("LectureId")
+                        .HasColumnType("INTEGER");
+
+                    b.Property<string>("LectureSemester")
+                        .IsRequired()
+                        .HasColumnType("TEXT");
+
+                    b.Property<string>("Location")
+                        .IsRequired()
+                        .HasColumnType("TEXT");
+
+                    b.Property<DateTime>("To")
+                        .HasColumnType("TEXT");
+
+                    b.HasKey("Id");
+
+                    b.HasIndex("LectureId", "LectureSemester");
+
+                    b.ToTable("LectureEvents");
+                });
+
+            modelBuilder.Entity("UWLib.ScrapedLink", b =>
+                {
+                    b.Property<string>("Url")
+                        .HasColumnType("TEXT");
+
+                    b.Property<DateTime>("LastScrape")
+                        .HasColumnType("TEXT");
+
+                    b.HasKey("Url");
+
+                    b.ToTable("ScrapedLinks");
+                });
+
+            modelBuilder.Entity("UWLib.LectureEvent", b =>
+                {
+                    b.HasOne("UWLib.Lecture", "Lecture")
+                        .WithMany("Events")
+                        .HasForeignKey("LectureId", "LectureSemester")
+                        .OnDelete(DeleteBehavior.Cascade)
+                        .IsRequired();
+
+                    b.Navigation("Lecture");
+                });
+
+            modelBuilder.Entity("UWLib.Lecture", b =>
+                {
+                    b.Navigation("Events");
+                });
+#pragma warning restore 612, 618
+        }
+    }
+}

UWLib/Migrations/20241012131142_ScrapedLinks.cs

@@ -0,0 +1,34 @@
+using System;
+using Microsoft.EntityFrameworkCore.Migrations;
+
+#nullable disable
+
+namespace UWLib.Migrations
+{
+    /// <inheritdoc />
+    public partial class ScrapedLinks : Migration
+    {
+        /// <inheritdoc />
+        protected override void Up(MigrationBuilder migrationBuilder)
+        {
+            migrationBuilder.CreateTable(
+                name: "ScrapedLinks",
+                columns: table => new
+                {
+                    Url = table.Column<string>(type: "TEXT", nullable: false),
+                    LastScrape = table.Column<DateTime>(type: "TEXT", nullable: false)
+                },
+                constraints: table =>
+                {
+                    table.PrimaryKey("PK_ScrapedLinks", x => x.Url);
+                });
+        }
+
+        /// <inheritdoc />
+        protected override void Down(MigrationBuilder migrationBuilder)
+        {
+            migrationBuilder.DropTable(
+                name: "ScrapedLinks");
+        }
+    }
+}

UWLib/Migrations/20241012131329_Branch.Designer.cs

@@ -0,0 +1,112 @@
+// <auto-generated />
+using System;
+using Microsoft.EntityFrameworkCore;
+using Microsoft.EntityFrameworkCore.Infrastructure;
+using Microsoft.EntityFrameworkCore.Migrations;
+using Microsoft.EntityFrameworkCore.Storage.ValueConversion;
+using UWLib;
+
+#nullable disable
+
+namespace UWLib.Migrations
+{
+    [DbContext(typeof(LectureContext))]
+    [Migration("20241012131329_Branch")]
+    partial class Branch
+    {
+        /// <inheritdoc />
+        protected override void BuildTargetModel(ModelBuilder modelBuilder)
+        {
+#pragma warning disable 612, 618
+            modelBuilder.HasAnnotation("ProductVersion", "8.0.10");
+
+            modelBuilder.Entity("UWLib.Lecture", b =>
+                {
+                    b.Property<int>("Id")
+                        .HasColumnType("INTEGER");
+
+                    b.Property<string>("Semester")
+                        .HasColumnType("TEXT");
+
+                    b.Property<string>("Branch")
+                        .IsRequired()
+                        .HasColumnType("TEXT");
+
+                    b.Property<string>("Description")
+                        .HasColumnType("TEXT");
+
+                    b.Property<string>("Title")
+                        .IsRequired()
+                        .HasColumnType("TEXT");
+
+                    b.Property<string>("Url")
+                        .IsRequired()
+                        .HasColumnType("TEXT");
+
+                    b.HasKey("Id", "Semester");
+
+                    b.ToTable("Lectures");
+                });
+
+            modelBuilder.Entity("UWLib.LectureEvent", b =>
+                {
+                    b.Property<int>("Id")
+                        .ValueGeneratedOnAdd()
+                        .HasColumnType("INTEGER");
+
+                    b.Property<DateTime>("From")
+                        .HasColumnType("TEXT");
+
+                    b.Property<int>("LectureId")
+                        .HasColumnType("INTEGER");
+
+                    b.Property<string>("LectureSemester")
+                        .IsRequired()
+                        .HasColumnType("TEXT");
+
+                    b.Property<string>("Location")
+                        .IsRequired()
+                        .HasColumnType("TEXT");
+
+                    b.Property<DateTime>("To")
+                        .HasColumnType("TEXT");
+
+                    b.HasKey("Id");
+
+                    b.HasIndex("LectureId", "LectureSemester");
+
+                    b.ToTable("LectureEvents");
+                });
+
+            modelBuilder.Entity("UWLib.ScrapedLink", b =>
+                {
+                    b.Property<string>("Url")
+                        .HasColumnType("TEXT");
+
+                    b.Property<DateTime>("LastScrape")
+                        .HasColumnType("TEXT");
+
+                    b.HasKey("Url");
+
+                    b.ToTable("ScrapedLinks");
+                });
+
+            modelBuilder.Entity("UWLib.LectureEvent", b =>
+                {
+                    b.HasOne("UWLib.Lecture", "Lecture")
+                        .WithMany("Events")
+                        .HasForeignKey("LectureId", "LectureSemester")
+                        .OnDelete(DeleteBehavior.Cascade)
+                        .IsRequired();
+
+                    b.Navigation("Lecture");
+                });
+
+            modelBuilder.Entity("UWLib.Lecture", b =>
+                {
+                    b.Navigation("Events");
+                });
+#pragma warning restore 612, 618
+        }
+    }
+}

UWLib/Migrations/20241012131329_Branch.cs

@@ -0,0 +1,29 @@
+using Microsoft.EntityFrameworkCore.Migrations;
+
+#nullable disable
+
+namespace UWLib.Migrations
+{
+    /// <inheritdoc />
+    public partial class Branch : Migration
+    {
+        /// <inheritdoc />
+        protected override void Up(MigrationBuilder migrationBuilder)
+        {
+            migrationBuilder.AddColumn<string>(
+                name: "Branch",
+                table: "Lectures",
+                type: "TEXT",
+                nullable: false,
+                defaultValue: "");
+        }
+
+        /// <inheritdoc />
+        protected override void Down(MigrationBuilder migrationBuilder)
+        {
+            migrationBuilder.DropColumn(
+                name: "Branch",
+                table: "Lectures");
+        }
+    }
+}

UWLib/Migrations/20241012140426_LinksToScrape.Designer.cs

@@ -0,0 +1,125 @@
+// <auto-generated />
+using System;
+using Microsoft.EntityFrameworkCore;
+using Microsoft.EntityFrameworkCore.Infrastructure;
+using Microsoft.EntityFrameworkCore.Migrations;
+using Microsoft.EntityFrameworkCore.Storage.ValueConversion;
+using UWLib;
+
+#nullable disable
+
+namespace UWLib.Migrations
+{
+    [DbContext(typeof(LectureContext))]
+    [Migration("20241012140426_LinksToScrape")]
+    partial class LinksToScrape
+    {
+        /// <inheritdoc />
+        protected override void BuildTargetModel(ModelBuilder modelBuilder)
+        {
+#pragma warning disable 612, 618
+            modelBuilder.HasAnnotation("ProductVersion", "8.0.10");
+
+            modelBuilder.Entity("UWLib.Lecture", b =>
+                {
+                    b.Property<int>("Id")
+                        .HasColumnType("INTEGER");
+
+                    b.Property<string>("Semester")
+                        .HasColumnType("TEXT");
+
+                    b.Property<string>("Branch")
+                        .IsRequired()
+                        .HasColumnType("TEXT");
+
+                    b.Property<string>("Description")
+                        .HasColumnType("TEXT");
+
+                    b.Property<string>("Title")
+                        .IsRequired()
+                        .HasColumnType("TEXT");
+
+                    b.Property<string>("Url")
+                        .IsRequired()
+                        .HasColumnType("TEXT");
+
+                    b.HasKey("Id", "Semester");
+
+                    b.ToTable("Lectures");
+                });
+
+            modelBuilder.Entity("UWLib.LectureEvent", b =>
+                {
+                    b.Property<int>("Id")
+                        .ValueGeneratedOnAdd()
+                        .HasColumnType("INTEGER");
+
+                    b.Property<DateTime>("From")
+                        .HasColumnType("TEXT");
+
+                    b.Property<int>("LectureId")
+                        .HasColumnType("INTEGER");
+
+                    b.Property<string>("LectureSemester")
+                        .IsRequired()
+                        .HasColumnType("TEXT");
+
+                    b.Property<string>("Location")
+                        .IsRequired()
+                        .HasColumnType("TEXT");
+
+                    b.Property<DateTime>("To")
+                        .HasColumnType("TEXT");
+
+                    b.HasKey("Id");
+
+                    b.HasIndex("LectureId", "LectureSemester");
+
+                    b.ToTable("LectureEvents");
+                });
+
+            modelBuilder.Entity("UWLib.LinkToScrape", b =>
+                {
+                    b.Property<string>("Url")
+                        .HasColumnType("TEXT");
+
+                    b.Property<int>("Sort")
+                        .HasColumnType("INTEGER");
+
+                    b.HasKey("Url");
+
+                    b.ToTable("LinksToScrape");
+                });
+
+            modelBuilder.Entity("UWLib.ScrapedLink", b =>
+                {
+                    b.Property<string>("Url")
+                        .HasColumnType("TEXT");
+
+                    b.Property<DateTime>("LastScrape")
+                        .HasColumnType("TEXT");
+
+                    b.HasKey("Url");
+
+                    b.ToTable("ScrapedLinks");
+                });
+
+            modelBuilder.Entity("UWLib.LectureEvent", b =>
+                {
+                    b.HasOne("UWLib.Lecture", "Lecture")
+                        .WithMany("Events")
+                        .HasForeignKey("LectureId", "LectureSemester")
+                        .OnDelete(DeleteBehavior.Cascade)
+                        .IsRequired();
+
+                    b.Navigation("Lecture");
+                });
+
+            modelBuilder.Entity("UWLib.Lecture", b =>
+                {
+                    b.Navigation("Events");
+                });
+#pragma warning restore 612, 618
+        }
+    }
+}

UWLib/Migrations/20241012140426_LinksToScrape.cs

@@ -0,0 +1,33 @@
+using Microsoft.EntityFrameworkCore.Migrations;
+
+#nullable disable
+
+namespace UWLib.Migrations
+{
+    /// <inheritdoc />
+    public partial class LinksToScrape : Migration
+    {
+        /// <inheritdoc />
+        protected override void Up(MigrationBuilder migrationBuilder)
+        {
+            migrationBuilder.CreateTable(
+                name: "LinksToScrape",
+                columns: table => new
+                {
+                    Url = table.Column<string>(type: "TEXT", nullable: false),
+                    Sort = table.Column<int>(type: "INTEGER", nullable: false)
+                },
+                constraints: table =>
+                {
+                    table.PrimaryKey("PK_LinksToScrape", x => x.Url);
+                });
+        }
+
+        /// <inheritdoc />
+        protected override void Down(MigrationBuilder migrationBuilder)
+        {
+            migrationBuilder.DropTable(
+                name: "LinksToScrape");
+        }
+    }
+}

UWLib/Migrations/LectureContextModelSnapshot.cs

@@ -25,6 +25,10 @@ namespace UWLib.Migrations
                     b.Property<string>("Semester")
                         .HasColumnType("TEXT");
 
+                    b.Property<string>("Branch")
+                        .IsRequired()
+                        .HasColumnType("TEXT");
+
                     b.Property<string>("Description")
                         .HasColumnType("TEXT");
 
@@ -71,6 +75,32 @@ namespace UWLib.Migrations
                     b.ToTable("LectureEvents");
                 });
 
+            modelBuilder.Entity("UWLib.LinkToScrape", b =>
+                {
+                    b.Property<string>("Url")
+                        .HasColumnType("TEXT");
+
+                    b.Property<int>("Sort")
+                        .HasColumnType("INTEGER");
+
+                    b.HasKey("Url");
+
+                    b.ToTable("LinksToScrape");
+                });
+
+            modelBuilder.Entity("UWLib.ScrapedLink", b =>
+                {
+                    b.Property<string>("Url")
+                        .HasColumnType("TEXT");
+
+                    b.Property<DateTime>("LastScrape")
+                        .HasColumnType("TEXT");
+
+                    b.HasKey("Url");
+
+                    b.ToTable("ScrapedLinks");
+                });
+
             modelBuilder.Entity("UWLib.LectureEvent", b =>
                 {
                     b.HasOne("UWLib.Lecture", "Lecture")

UWLib/ScrapedLink.cs

@@ -0,0 +1,17 @@
+using System;
+using System.Collections.Generic;
+using System.ComponentModel.DataAnnotations;
+using System.Linq;
+using System.Text;
+using System.Threading.Tasks;
+
+namespace UWLib
+{
+    public class ScrapedLink
+    {
+        [Key]
+        public string Url{ get; set; }
+
+        public DateTime LastScrape { get; set; }
+    }
+}

UWScraper/Program.cs

@@ -1,11 +1,15 @@
 
+using System.Collections.Generic;
 using System.Globalization;
 using System.Net.NetworkInformation;
+using System.Text.Encodings.Web;
 using System.Web;
 using System.Xml.Linq;
+using HtmlAgilityPack;
 using OpenQA.Selenium;
 using OpenQA.Selenium.BiDi.Modules.Script;
 using OpenQA.Selenium.Chrome;
+using OpenQA.Selenium.DevTools.V127.Target;
 using OpenQA.Selenium.Support.UI;
 using UWLib;
 using static Microsoft.EntityFrameworkCore.DbLoggerCategory;
@@ -13,110 +17,287 @@ using static Microsoft.EntityFrameworkCore.DbLoggerCategory;
 
 internal class Program
 {
+    readonly static List<string> checkedUrls = [];
+    readonly static List<string> urlsToCheck = [];
 
-    static List<string> checkedUrls = new List<string>();
-    static List<string> urlsToCheck = new List<string>();
+    static bool refreshLectures = false;
 
-    static LectureContext db = new LectureContext();
-    static IWebDriver driver = null;
+    static readonly LectureContext db = new();
+    static ChromeDriver? driver = null;
     private static void Main(string[] args)
     {
+        AppDomain.CurrentDomain.ProcessExit += (s, e) =>
+        {
+            Done();
+        };
+
+        var service = ChromeDriverService.CreateDefaultService();
+        service.HideCommandPromptWindow = true;
+
         var options = new ChromeOptions();
         //options.AddArgument("--headless");
-        driver = new ChromeDriver(options);
+        driver = new ChromeDriver(service, options);
 
-        checkedUrls.AddRange(db.Lectures.Select(x => x.Url).ToList());
+        if (args.Any(x => x == "-r" || x == "--refresh-lectures"))
+        {
+            refreshLectures = true;
 
+            Console.WriteLine("Refreshing lectures");
+        }
+
+        if (args.Length == 0)
+        {
+            urlsToCheck.AddRange([.. db.LinksToScrape.Select(x => x.Url)]);
+
+            //add all urls that were checked in the last 24 hours
+            checkedUrls.AddRange([.. db.ScrapedLinks.Where(x => x.LastScrape > DateTime.Now.AddDays(-1)).Select(x => x.Url)]);
+
+            //remove all checked urls from the urls to check
+            foreach (var url in checkedUrls)
+            {
+                urlsToCheck.Remove(url);
+            }
+
+            if (urlsToCheck.Count == 0)
+            {
+                // add the first url to check / Vorlesungsverzeichnis
                 urlsToCheck.Add("https://ufind.univie.ac.at/de/vvz.html");
+            }
+        }
+
+        if (refreshLectures)
+        {
+            urlsToCheck.AddRange([.. db.Lectures.Select(x => x.Url)]);
+        }
+
+        Console.CursorVisible = false;
+
+        int top = Console.CursorTop;
 
         while (urlsToCheck.Count > 0)
         {
             var url = urlsToCheck.First();
 
+            try
+            {
                 FindUrls(url);
             }
+            catch (Exception e)
+            {
+                Console.Clear();
+                Console.WriteLine($"Url: {url}");
+                Console.WriteLine();
+                Console.WriteLine(e);
+                return;
+            }
+
+            Console.CursorLeft = 0;
+            Console.CursorTop = top;
+            Console.WriteLine($"Urls checked: {checkedUrls.Count}");
+            Console.WriteLine($"Urls to check: {urlsToCheck.Count}");
+        }
+
+        Console.WriteLine("Done");
+    }
+
+    private static void Done()
+    {
+        driver?.Quit();
+
+        Console.CursorVisible = true;
+
+        Console.ReadLine();
+    }
+
+    static string GetAbsoluteUrl(string relativeUrl, string baseUrl)
+    {
+        var uri = new Uri(baseUrl);
+        var baseUri = new Uri(uri, relativeUrl);
+        return HttpUtility.HtmlDecode(baseUri.AbsoluteUri);
     }
 
     private static void FindUrls(string source)
     {
-        if (checkedUrls.Contains(source))
-        {
-            urlsToCheck.Remove(source);
-            return;
-        }
+        driver?.Navigate().GoToUrl(source);
 
-        if (db.Lectures.Any(x=>x.Url == source))
-        {
-            checkedUrls.Add(source);
-            urlsToCheck.Remove(source);
-            return;
-        }
-
-
-        driver.Navigate().GoToUrl(source);
-        WebDriverWait wait = new WebDriverWait(driver, TimeSpan.FromSeconds(2));
+        // wait for the page to load
+        WebDriverWait wait = new(driver, TimeSpan.FromSeconds(2));
         wait.Until(d => d.FindElements(By.XPath("//a[starts-with(@href, 'vvz_sub.html')]")).Count > 0 ||
         d.FindElements(By.XPath("//a[starts-with(@href, 'course.html')]")).Count > 0 ||
         d.FindElements(By.XPath("//h1/*/*[@class='what']")).Count > 0
         );
 
+        HtmlDocument doc = new();
+        doc.LoadHtml(driver?.PageSource);
+        var root = doc.DocumentNode;
 
         if (source.Contains("course.html"))
         {
-            Lecture lecture = new Lecture();
+            CreateLecture(source, root);
+        }
 
-            lecture.Url = source;
+        if (!refreshLectures)
+        {
+            FindPathLinks(source, root);
+            FindCourseLinks(source, root);
+        }
+
+        RefreshScrapedLink(source);
+        RemoveLinkToScrape(source);
+        db.SaveChanges();
+
+        checkedUrls.Add(source);
+        urlsToCheck.Remove(source);
+    }
+
+    private static void RefreshScrapedLink(string source)
+    {
+        var scrapedLink = db.ScrapedLinks.Find(source);
+        if (scrapedLink == null)
+        {
+            scrapedLink = new ScrapedLink { Url = source };
+            db.ScrapedLinks.Add(scrapedLink);
+        }
+        scrapedLink.LastScrape = DateTime.Now;
+    }
+
+    private static void FindCourseLinks(string source, HtmlNode root)
+    {
+        var courseLinkParents = root.SelectNodes("//a[starts-with(@href, 'course.html')]/..");
+        if (courseLinkParents != null)
+        {
+            foreach (var parent in courseLinkParents)
+            {
+                if (parent.SelectSingleNode("abbr[contains(@title,'Vorlesung')]") != null)
+                {
+                    var link = parent.SelectSingleNode("a[contains(@href, 'course.html')]");
+
+                    var url = GetAbsoluteUrl(link.GetAttributeValue("href", ""), source);
+
+                    if (!checkedUrls.Contains(url))
+                    {
+                        db.LinksToScrape.Add(new LinkToScrape { Url = url, Sort = 0 });
+                        urlsToCheck.Insert(0, url);
+                    }
+                }
+            }
+        }
+    }
+
+    private static void FindPathLinks(string source, HtmlNode root)
+    {
+        var pathLinks = root.SelectNodes("//a[starts-with(@href, 'vvz_sub.html')]");
+
+        if (pathLinks != null)
+        {
+            foreach (var link in pathLinks)
+            {
+                var url = GetAbsoluteUrl(link.GetAttributeValue("href", ""), source);
+
+                if (!checkedUrls.Contains(url))
+                {
+                    int sort = 1;
+                    AddLinkToScrape(url, sort);
+                    urlsToCheck.Add(url);
+                }
+            }
+        }
+    }
+
+    private static void RemoveLinkToScrape(string url)
+    {
+        var link = db.LinksToScrape.Find(url);
+        if (link != null)
+        {
+            db.LinksToScrape.Remove(link);
+        }
+    }
+
+    private static void AddLinkToScrape(string url, int sort)
+    {
+        var link = db.LinksToScrape.Find(url);
+        if (link == null)
+        {
+            db.LinksToScrape.Add(new LinkToScrape { Url = url, Sort = sort });
+        }
+    }
+
+    private static void CreateLecture(string source, HtmlNode root)
+    {
+        var branch = root.SelectSingleNode("/html/body/main/div[1]/div[1]/a");
+        var what = root.SelectSingleNode("//h1/*/*[@class='what']");
+        var when = root.SelectSingleNode("//h1/*/*[@class='when']");
+        var info = root.SelectSingleNode("//*[@class='info list']");
+        var events = root.SelectNodes("//ul[@class='classes events list']/li");
 
         var uri = new Uri(source);
 
         var query = HttpUtility.ParseQueryString(uri.Query);
 
+        Lecture? lecture = db.Lectures.FirstOrDefault(db => db.Url == source);
+
+        if (lecture == null)
+        {
+            lecture = new Lecture();
+
             if (query.AllKeys.Contains("lv"))
             {
-                lecture.Id = int.Parse(query["lv"]);
+                lecture.Id = int.Parse(query["lv"] ?? "0");
             }
 
-            int year = DateTime.Now.Year;
-
-
-            var what = driver.FindElements(By.XPath("//h1/*/*[@class='what']"));
-            var when = driver.FindElements(By.XPath("//h1/*/*[@class='when']"));
-            var info = driver.FindElements(By.XPath("//*[@class='info list']"));
-            var events = driver.FindElements(By.XPath("//ul[@class='classes events list']/li"));
-
-            if (what.Count > 0)
+            if (when != null)
             {
-                lecture.Title = what.First().Text;
+                lecture.Semester = when.InnerText;
             }
 
+            lecture.Url = source;
 
-            if (when.Count > 0)
+            db.Lectures.Add(lecture);
+        }
+
+        if (branch != null)
         {
-                lecture.Semester = when.First().Text;
-                year = int.Parse(when.First().Text.Substring(0, 4));
+            lecture.Branch = branch.InnerText;
         }
 
-            if (info.Count > 0)
+        if (what != null)
         {
-                lecture.Description = info.First().GetAttribute("innerHTML");
+            lecture.Title = what.InnerText;
         }
 
+        if (info != null)
+        {
+            lecture.Description = info.InnerHtml;
+        }
+
+        if (events != null)
+        {
+            CreateLectureEvents(events, lecture);
+        }
+    }
+
+    private static void CreateLectureEvents(HtmlNodeCollection events, Lecture lecture)
+    {
+        int year = int.Parse(lecture.Semester[..4]);
+
+        db.RemoveRange(lecture.Events);
+        lecture.Events.Clear();
         foreach (var item in events)
         {
-                LectureEvent lectureEvent = new LectureEvent();
+            LectureEvent lectureEvent = new();
 
-                var day = item.FindElements(By.XPath("*[@class='date']"));
-                var time = item.FindElements(By.XPath("*[@class='time']"));
-                var room = item.FindElements(By.XPath("*[@class='room']"));
-                DateTime date = new DateTime();
-                if (day.Count > 0)
+            var day = item.SelectSingleNode("*[@class='date']");
+            var time = item.SelectSingleNode("*[@class='time']");
+            var room = item.SelectSingleNode("*[@class='room']");
+            DateTime date = new();
+            if (day != null)
             {
-                    date = DateTime.ParseExact(day.First().Text + year.ToString(), "dd.MM.yyyy", CultureInfo.InvariantCulture);
+                date = DateTime.ParseExact(day.InnerText + year.ToString(), "dd.MM.yyyy", CultureInfo.InvariantCulture);
             }
 
-                if (time.Count > 0)
+            if (time != null)
             {
-                    var text = time.First().Text;
+                var text = time.InnerText;
 
                 var times = text.Split(" - ");
 
@@ -127,56 +308,12 @@ internal class Program
                 lectureEvent.To = date.Add(to);
             }
 
-                if (room.Count > 0)
+            if (room != null)
             {
-                    lectureEvent.Location = room.First().Text;
+                lectureEvent.Location = room.InnerText;
             }
 
             lecture.Events.Add(lectureEvent);
         }
-
-            try
-            {
-                db.Lectures.Add(lecture);
-            }
-            catch (Exception e)
-            {
-            }
-
-            db.SaveChanges();
-        }
-
-        var pathLinks = driver.FindElements(By.XPath("//a[starts-with(@href, 'vvz_sub.html')]"));
-        foreach (var link in pathLinks)
-        {
-            var url = link.GetAttribute("href");
-
-            if (!checkedUrls.Contains(url))
-            {
-                urlsToCheck.Add(link.GetAttribute("href"));
-            }
-        }
-
-        var courseLinkParents = driver.FindElements(By.XPath("//a[starts-with(@href, 'course.html')]/.."));
-
-        foreach (var parent in courseLinkParents)
-        {
-            if (parent.FindElements(By.XPath("abbr[@title='Vorlesung']")).Count > 0)
-            {
-                var link = parent.FindElement(By.XPath("a[starts-with(@href, 'course.html')]"));
-
-                var url = link.GetAttribute("href");
-
-                if (!checkedUrls.Contains(url))
-                {
-                    urlsToCheck.Insert(0, link.GetAttribute("href"));
-                }
-            }
-        }
-
-        //        driver.Close();
-
-        checkedUrls.Add(source);
-        urlsToCheck.Remove(source);
     }
 }

UWScraper/Properties/launchSettings.json

@@ -0,0 +1,8 @@
+{
+  "profiles": {
+    "UWScraper": {
+      "commandName": "Project",
+      "commandLineArgs": "-r"
+    }
+  }
+}

UWScraper/UWScraper.csproj

@@ -9,6 +9,10 @@
 
   <ItemGroup>
     <PackageReference Include="HtmlAgilityPack" Version="1.11.67" />
+    <PackageReference Include="Microsoft.EntityFrameworkCore.Design" Version="8.0.10">
+      <PrivateAssets>all</PrivateAssets>
+      <IncludeAssets>runtime; build; native; contentfiles; analyzers; buildtransitive</IncludeAssets>
+    </PackageReference>
     <PackageReference Include="Selenium.WebDriver" Version="4.25.0" />
   </ItemGroup>