#61 (#70)

* Typo in LoadHtmlDocumentAsync * Improved logging on webscraper * Further improved error reporting * Added exceptions for webscraper * Refactored webscraper * Updated project version
izqalan · Jan 23, 2022 · 4fa923b · 4fa923b
1 parent 8f0b983
commit 4fa923b
Show file tree

Hide file tree

Showing 6 changed files with 166 additions and 66 deletions.
diff --git a/src/CyberdropDownloader.Avalonia/CyberdropDownloader.Avalonia.csproj b/src/CyberdropDownloader.Avalonia/CyberdropDownloader.Avalonia.csproj
@@ -4,17 +4,18 @@
     <TargetFramework>net5.0</TargetFramework>
     <Nullable>enable</Nullable>
     <AssemblyName>Cyberdrop Downloader</AssemblyName>
-    <Copyright>Copyright ©  2021</Copyright>
+    <Copyright>Copyright ©  2022</Copyright>
     <Product>Cyberdrop Downloader</Product>
-    <AssemblyVersion>2.2.3.0</AssemblyVersion>
-    <FileVersion>2.2.2.0</FileVersion>
+    <AssemblyVersion>2.2.4.0</AssemblyVersion>
+    <FileVersion>2.2.4.0</FileVersion>
     <NeutralLanguage>en</NeutralLanguage>
     <Company />
     <Authors />
     <PackageIconUrl />
     <ApplicationIcon>Assets\duck.ico</ApplicationIcon>
     <Win32Resource />
-    <Version>2.2.3</Version>
+    <Version>2.2.4</Version>
+    <PackageProjectUrl>https://github.com/izqalan/cy-client</PackageProjectUrl>
   </PropertyGroup>
   <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|AnyCPU'">
     <DebugType>none</DebugType>

diff --git a/src/CyberdropDownloader.Avalonia/ViewModels/MainWindowViewModel.cs b/src/CyberdropDownloader.Avalonia/ViewModels/MainWindowViewModel.cs
@@ -10,6 +10,7 @@
 using System.Reflection;
 using System.Threading;
 using System.Threading.Tasks;
+using CyberdropDownloader.Core.Exceptions;
 
 namespace CyberdropDownloader.Avalonia.ViewModels
 {
@@ -115,8 +116,29 @@ await Task.Run(async () =>
                             _cancellationTokenSource = new CancellationTokenSource();
                         }
 
-                        // Load the album
-                        await _webScraper.LoadAlbumAsync(url);
+                        try
+                        {
+                            await _webScraper.LoadAlbumAsync(url);
+                        }
+                        catch(Exception exception)
+                        {
+                            switch(exception)
+                            {
+                                case NullAlbumTitleException:
+                                    Log("Failed to fetch album title.");
+                                    break;
+
+                                case NullAlbumSizeException:
+                                    Log("Failed to fetch album size.");
+                                    break;
+
+                                case NullAlbumFilesException:
+                                    Log("Failed to fetch album files.");
+                                    break;
+                            }
+
+                            continue;
+                        }
 
                         // If the album url is invalid, then log and skip over it
                         if(!_webScraper.Successful)

diff --git a/src/CyberdropDownloader.Core/Exceptions/NullAlbumFilesException.cs b/src/CyberdropDownloader.Core/Exceptions/NullAlbumFilesException.cs
@@ -0,0 +1,24 @@
+using System;
+using System.Runtime.Serialization;
+
+namespace CyberdropDownloader.Core.Exceptions
+{
+    public class NullAlbumFilesException : Exception
+    {
+        public NullAlbumFilesException()
+        {
+        }
+
+        public NullAlbumFilesException(string message) : base(message)
+        {
+        }
+
+        public NullAlbumFilesException(string message, Exception innerException) : base(message, innerException)
+        {
+        }
+
+        protected NullAlbumFilesException(SerializationInfo info, StreamingContext context) : base(info, context)
+        {
+        }
+    }
+}
diff --git a/src/CyberdropDownloader.Core/Exceptions/NullAlbumSizeException.cs b/src/CyberdropDownloader.Core/Exceptions/NullAlbumSizeException.cs
@@ -0,0 +1,24 @@
+using System;
+using System.Runtime.Serialization;
+
+namespace CyberdropDownloader.Core.Exceptions
+{
+    public class NullAlbumSizeException : Exception
+    {
+        public NullAlbumSizeException()
+        {
+        }
+
+        public NullAlbumSizeException(string message) : base(message)
+        {
+        }
+
+        public NullAlbumSizeException(string message, Exception innerException) : base(message, innerException)
+        {
+        }
+
+        protected NullAlbumSizeException(SerializationInfo info, StreamingContext context) : base(info, context)
+        {
+        }
+    }
+}
diff --git a/src/CyberdropDownloader.Core/Exceptions/NullAlbumTitleException.cs b/src/CyberdropDownloader.Core/Exceptions/NullAlbumTitleException.cs
@@ -0,0 +1,24 @@
+using System;
+using System.Runtime.Serialization;
+
+namespace CyberdropDownloader.Core.Exceptions
+{
+    public class NullAlbumTitleException : Exception
+    {
+        public NullAlbumTitleException()
+        {
+        }
+
+        public NullAlbumTitleException(string message) : base(message)
+        {
+        }
+
+        public NullAlbumTitleException(string message, Exception innerException) : base(message, innerException)
+        {
+        }
+
+        protected NullAlbumTitleException(SerializationInfo info, StreamingContext context) : base(info, context)
+        {
+        }
+    }
+}
diff --git a/src/CyberdropDownloader.Core/WebScraper.cs b/src/CyberdropDownloader.Core/WebScraper.cs
@@ -1,4 +1,5 @@
 using CyberdropDownloader.Core.DataModels;
+using CyberdropDownloader.Core.Exceptions;
 using HtmlAgilityPack;
 using System;
 using System.Collections.Generic;
@@ -7,76 +8,80 @@
 
 namespace CyberdropDownloader.Core
 {
-	public class WebScraper
-	{
-		private string _url;
-		private HtmlDocument _htmlDocument;
+    public class WebScraper
+    {
+        private Album _album;
+        private bool _successful;
 
-		private Album _album;
-		private bool _successful;
+        public Album Album => _album;
+        public bool Successful => _successful;
 
-		public Album Album => _album;
-		public bool Successful => _successful;
+        public async Task LoadAlbumAsync(string url)
+        {
+            await Task.Run(async () =>
+            {
+                // Load webpage
+                HtmlDocument htmlDocument = await new HtmlWeb().LoadFromWebAsync(url);
 
-		public async Task LoadAlbumAsync(string url)
-		{
-			_url = url;
-			_successful = false;
+                if(htmlDocument != null)
+                {
+                    try
+                    {
+                        (string title, string size, Queue<AlbumFile> files) albumData = FetchAlbumData(htmlDocument);
 
-			await Task.Run(async () =>
-			{
-				await LoadHtmlDocumenteAsync();
+                        _album = new Album(albumData.title, albumData.size, albumData.files);
+                        _successful = true;
+                    }
+                    catch(Exception)
+                    {
+                        _successful = false;
+                    }
+                }
+            });
+        }
 
-				if (_htmlDocument != null)
-				{
-					try
-					{
-						// Insatiate new album with title, size, and files.
-						_album = new Album(FetchAlbumTitle(), FetchAlbumSize(), FetchAlbumFiles());
-						_successful = true;
-					}
-					catch
-					{
-						_successful = false;
-					}
-				}
-			});
-		}
+        #region Load Album
+        private (string title, string size, Queue<AlbumFile> files) FetchAlbumData(HtmlDocument htmlDocument)
+        {
+            return (FetchAlbumTitle(htmlDocument), FetchAlbumSize(htmlDocument), FetchAlbumFiles(htmlDocument));
+        }
 
-		#region Load Album
-		private string FetchAlbumTitle() => _htmlDocument.DocumentNode.SelectNodes("//div/h1[@id='title']").First().Attributes["title"].Value;
+        private string FetchAlbumTitle(HtmlDocument htmlDocument)
+        {
+            string title = htmlDocument.DocumentNode.SelectNodes("//div/h1[@id='title']").First().Attributes["title"].Value;
 
-		private string FetchAlbumSize() => _htmlDocument.DocumentNode.SelectNodes("//div/p[@class='title']")[1].InnerHtml;
+            return title ?? throw new NullAlbumTitleException();
+        }
 
-		private Queue<AlbumFile> FetchAlbumFiles()
-		{
-			Queue<AlbumFile> urls = new Queue<AlbumFile>();
+        private string FetchAlbumSize(HtmlDocument htmlDocument)
+        {
+            string size = htmlDocument.DocumentNode.SelectNodes("//div/p[@class='title']")[1].InnerHtml;
 
-			HtmlNodeCollection nodes = _htmlDocument.DocumentNode.SelectNodes("//a[@class='image'][@href]");
+            return size ?? throw new NullAlbumFilesException();
+        }
 
-			foreach (HtmlNode link in nodes)
-			{
-				urls.Enqueue(new AlbumFile()
-				{
-					Name = link.Attributes["title"].Value,
-					Url = link.Attributes["href"].Value
-				});
-			}
+        private Queue<AlbumFile> FetchAlbumFiles(HtmlDocument htmlDocument)
+        {
+            Queue<AlbumFile> urls = new Queue<AlbumFile>();
 
-			return urls;
-		}
+            HtmlNodeCollection files = htmlDocument.DocumentNode.SelectNodes("//a[@class='image'][@href]");
 
-		private async Task LoadHtmlDocumenteAsync()
-		{
-			try
-			{
-				_htmlDocument = await new HtmlWeb().LoadFromWebAsync(_url);
-			}
-			catch (Exception)
-			{
-				_htmlDocument = null!;
-			}
-		}
-		#endregion
-	}
+            if(files == null)
+            {
+                throw new NullAlbumFilesException();
+            }
+
+            foreach(HtmlNode link in files)
+            {
+                urls.Enqueue(new AlbumFile()
+                {
+                    Name = link.Attributes["title"].Value,
+                    Url = link.Attributes["href"].Value
+                });
+            }
+
+            return urls;
+        }
+        #endregion
+    }
 }