Handle not supported content encoding

src/HtmlAgilityPack.Net35/HtmlAgilityPack.Net35.csproj

@@ -18,7 +18,7 @@
     <DebugType>full</DebugType>
     <Optimize>false</Optimize>
     <OutputPath>bin\Debug\</OutputPath>
-    <DefineConstants>DEBUG;TRACE</DefineConstants>
+    <DefineConstants>DEBUG;TRACE;FX35</DefineConstants>
     <ErrorReport>prompt</ErrorReport>
     <WarningLevel>4</WarningLevel>
   </PropertyGroup>

src/HtmlAgilityPack.Net40-client/HtmlAgilityPack.Net40-client.csproj

@@ -44,7 +44,7 @@
     <DebugType>full</DebugType>
     <Optimize>false</Optimize>
     <OutputPath>bin\Debug\</OutputPath>
-    <DefineConstants>DEBUG;TRACE</DefineConstants>
+    <DefineConstants>DEBUG;TRACE;FX40</DefineConstants>
     <ErrorReport>prompt</ErrorReport>
     <WarningLevel>4</WarningLevel>
     <DocumentationFile>bin\Debug\HtmlAgilityPack.XML</DocumentationFile>

src/HtmlAgilityPack.Net40/Properties/AssemblyInfo.cs

@@ -18,6 +18,7 @@
 #endif
 #endif
 [assembly: InternalsVisibleTo("HtmlAgilityPack.Tests, PublicKey=002400000480000094000000060200000024000052534131000400000100010027dc71d8e0b968c7324238e18a4cee4a367f1bf50c9d7a52d91ed46c6a1a584b9142c1d4234c4011d25437c909924079660c434eebe6d2c46412f30520a276e7ca8d8fa7075bb8b9e1c7502ef0e50423b32d469ba750012823fde16989ab42d8428ca5fdd0b06b801788a17239b78e0f75900012a50c5038ab93abbe2ac0d6ee")]
+[assembly: InternalsVisibleTo("DynamicProxyGenAssembly2, PublicKey=0024000004800000940000000602000000240000525341310004000001000100c547cac37abd99c8db225ef2f6c8a3602f3b3606cc9891605d02baa56104f4cfc0734aa39b93bf7852f7d9266654753cc297e7d2edfe0bac1cdcf9f717241550e0a7b191195b7667bb4f64bcb8e2121380fd1d9d46ad2d92d2d15605093924cceaf74c4861eff62abf69b9291ed0a340e113be11e6a7d3113e92484cf7045cc7")]
 [assembly: AssemblyConfiguration("")]
 [assembly: AssemblyCompany("ZZZ Projects Inc.")]
 [assembly: AssemblyProduct("Html Agility Pack")]

src/HtmlAgilityPack.Net45/Properties/AssemblyInfo.cs

@@ -18,6 +18,7 @@
 #endif
 #endif
 [assembly: InternalsVisibleTo("HtmlAgilityPack.Tests, PublicKey=002400000480000094000000060200000024000052534131000400000100010027dc71d8e0b968c7324238e18a4cee4a367f1bf50c9d7a52d91ed46c6a1a584b9142c1d4234c4011d25437c909924079660c434eebe6d2c46412f30520a276e7ca8d8fa7075bb8b9e1c7502ef0e50423b32d469ba750012823fde16989ab42d8428ca5fdd0b06b801788a17239b78e0f75900012a50c5038ab93abbe2ac0d6ee")]
+[assembly: InternalsVisibleTo("DynamicProxyGenAssembly2, PublicKey=0024000004800000940000000602000000240000525341310004000001000100c547cac37abd99c8db225ef2f6c8a3602f3b3606cc9891605d02baa56104f4cfc0734aa39b93bf7852f7d9266654753cc297e7d2edfe0bac1cdcf9f717241550e0a7b191195b7667bb4f64bcb8e2121380fd1d9d46ad2d92d2d15605093924cceaf74c4861eff62abf69b9291ed0a340e113be11e6a7d3113e92484cf7045cc7")]
 [assembly: AssemblyConfiguration("")]
 [assembly: AssemblyCompany("ZZZ Projects Inc.")]
 [assembly: AssemblyProduct("Html Agility Pack")]

src/HtmlAgilityPack.NetStandard2_0/AssemblyInfo.cs

@@ -1,3 +1,7 @@
 using System;
+using System.Runtime.CompilerServices;
 
-[assembly: CLSCompliant(true)]
+[assembly: CLSCompliant(true)]
+
+[assembly: InternalsVisibleTo("HtmlAgilityPack.Tests, PublicKey=002400000480000094000000060200000024000052534131000400000100010027dc71d8e0b968c7324238e18a4cee4a367f1bf50c9d7a52d91ed46c6a1a584b9142c1d4234c4011d25437c909924079660c434eebe6d2c46412f30520a276e7ca8d8fa7075bb8b9e1c7502ef0e50423b32d469ba750012823fde16989ab42d8428ca5fdd0b06b801788a17239b78e0f75900012a50c5038ab93abbe2ac0d6ee")]
+[assembly: InternalsVisibleTo("DynamicProxyGenAssembly2, PublicKey=0024000004800000940000000602000000240000525341310004000001000100c547cac37abd99c8db225ef2f6c8a3602f3b3606cc9891605d02baa56104f4cfc0734aa39b93bf7852f7d9266654753cc297e7d2edfe0bac1cdcf9f717241550e0a7b191195b7667bb4f64bcb8e2121380fd1d9d46ad2d92d2d15605093924cceaf74c4861eff62abf69b9291ed0a340e113be11e6a7d3113e92484cf7045cc7")]

src/HtmlAgilityPack.Shared/EncodingNotSupportedException.cs

@@ -0,0 +1,39 @@
+// Description: Html Agility Pack - HTML Parsers, selectors, traversors, manupulators.
+// Website & Documentation: http://html-agility-pack.net
+// Forum & Issues: https://github.com/zzzprojects/html-agility-pack
+// License: https://github.com/zzzprojects/html-agility-pack/blob/master/LICENSE
+// More projects: http://www.zzzprojects.com/
+// Copyright � ZZZ Projects Inc. 2014 - 2017. All rights reserved.
+
+using System;
+using System.Text;
+
+namespace HtmlAgilityPack
+{
+    public class EncodingNotSupportedException : Exception
+    {
+        #region Fields
+
+        private string _encoding;
+
+        #endregion
+
+        #region Constructors
+
+        internal EncodingNotSupportedException(string encoding)
+        {
+            _encoding = encoding;
+        }
+
+        #endregion
+
+        #region Properties
+
+        public string Encoding
+        {
+            get { return _encoding; }
+        }
+
+        #endregion
+    }
+}

src/HtmlAgilityPack.Shared/HtmlAgilityPack.Shared.projitems

@@ -17,6 +17,7 @@
   </PropertyGroup>
   <ItemGroup>
     <Compile Include="$(MSBuildThisFileDirectory)crc32.cs" />
+    <Compile Include="$(MSBuildThisFileDirectory)EncodingNotSupportedException.cs" />
     <Compile Include="$(MSBuildThisFileDirectory)EncodingFoundException.cs" />
     <Compile Include="$(MSBuildThisFileDirectory)HtmlAttribute.cs" />
     <Compile Include="$(MSBuildThisFileDirectory)HtmlAttributeCollection.cs" />
@@ -57,5 +58,11 @@
     <Compile Include="$(MSBuildThisFileDirectory)Trace.cs" />
     <Compile Include="$(MSBuildThisFileDirectory)Trace.FullFramework.cs" />
     <Compile Include="$(MSBuildThisFileDirectory)Utilities.cs" />
+    <Compile Include="$(MSBuildThisFileDirectory)IHttpWebRequest.cs" />
+    <Compile Include="$(MSBuildThisFileDirectory)HttpWebRequestWrapper.cs" />
+    <Compile Include="$(MSBuildThisFileDirectory)IHttpWebResponse.cs" />
+    <Compile Include="$(MSBuildThisFileDirectory)HttpWebResponseWrapper.cs" />
+    <Compile Include="$(MSBuildThisFileDirectory)IHttpWebRequestFactory.cs" />
+    <Compile Include="$(MSBuildThisFileDirectory)HttpWebRequestFactory.cs" />
   </ItemGroup>
 </Project>

src/HtmlAgilityPack.Shared/HtmlWeb.cs

@@ -83,6 +83,10 @@ public partial class HtmlWeb
 
         #region Fields
 
+#if !(NETSTANDARD1_3 || NETSTANDARD1_6)
+        private IHttpWebRequestFactory _requestFactory;
+#endif
+
         private bool _autoDetectEncoding = true;
         private bool _cacheOnly;
 
@@ -928,6 +932,21 @@ public bool UsingCache
 
         #endregion
 
+        #region Constructors
+
+#if !(NETSTANDARD1_3 || NETSTANDARD1_6)
+        public HtmlWeb() : this(new HttpWebRequestFactory())
+        {
+        }
+
+        internal HtmlWeb(IHttpWebRequestFactory requestFactory)
+        {
+            _requestFactory = requestFactory;
+        }
+#endif
+
+        #endregion
+
         #region Public Methods
 
 #if !(NETSTANDARD1_3 || NETSTANDARD1_6)
@@ -1547,10 +1566,10 @@ private static long SaveStream(Stream stream, string path, DateTime touchDate, i
             ICredentials creds)
         {
             string cachePath = null;
-            HttpWebRequest req;
+            IHttpWebRequest req;
             bool oldFile = false;
 
-            req = WebRequest.Create(uri) as HttpWebRequest;
+            req = _requestFactory.Create(uri);
             req.Method = method;
             req.UserAgent = UserAgent;
             if (CaptureRedirect)
@@ -1617,7 +1636,7 @@ private static long SaveStream(Stream stream, string path, DateTime touchDate, i
             if (PreRequest != null)
             {
                 // allow our user to change the request at will
-                if (!PreRequest(req))
+                if (!PreRequest(req.Request))
                 {
                     return HttpStatusCode.ResetContent;
                 }
@@ -1632,16 +1651,16 @@ private static long SaveStream(Stream stream, string path, DateTime touchDate, i
                 //                }
             }
 
-            HttpWebResponse resp;
+            IHttpWebResponse resp;
 
             try
             {
-                resp = req.GetResponse() as HttpWebResponse;
+                resp = req.GetResponse();
             }
             catch (WebException we)
             {
                 _requestDuration = Environment.TickCount - tc;
-                resp = (HttpWebResponse) we.Response;
+                resp = we.Response == null ? null : new HttpWebResponseWrapper((HttpWebResponse) we.Response);
                 if (resp == null)
                 {
                     if (oldFile)
@@ -1668,7 +1687,7 @@ private static long SaveStream(Stream stream, string path, DateTime touchDate, i
             // allow our user to get some info from the response
             if (PostResponse != null)
             {
-                PostResponse(req, resp);
+                PostResponse(req.Request, resp.Response);
             }
 
             _requestDuration = Environment.TickCount - tc;
@@ -1677,11 +1696,22 @@ private static long SaveStream(Stream stream, string path, DateTime touchDate, i
             bool html = IsHtmlContent(resp.ContentType);
             bool isUnknown = string.IsNullOrEmpty(resp.ContentType);
 
-            Encoding respenc = !string.IsNullOrEmpty(resp.ContentEncoding)
-                ? Encoding.GetEncoding(resp.ContentEncoding)
-                : null;
-            if (OverrideEncoding != null)
-                respenc = OverrideEncoding;
+            Encoding respenc = OverrideEncoding;
+            if (respenc == null && !string.IsNullOrEmpty(resp.ContentEncoding))
+            {
+                try
+                {
+                    Encoding.GetEncoding(resp.ContentEncoding);
+                }
+                catch (ArgumentException ex)
+                {
+                    if (ex.ParamName == "name")
+                    {
+                        throw new EncodingNotSupportedException(resp.ContentEncoding);
+                    }
+                    throw ex;
+                }
+            }
 
             if (CaptureRedirect)
             {
@@ -2092,7 +2122,7 @@ private HtmlDocument LoadUrl(Uri uri, string method, IWebProxy proxy, ICredentia
         }
 #endif
 #if !(NETSTANDARD1_3 || NETSTANDARD1_6)
-        private void SaveCacheHeaders(Uri requestUri, HttpWebResponse resp)
+        private void SaveCacheHeaders(Uri requestUri, IHttpWebResponse resp)
         {
             // we cache the original headers aside the cached document.
             string file = GetCacheHeadersPath(requestUri);

src/HtmlAgilityPack.Shared/HttpWebRequestFactory.cs

@@ -0,0 +1,31 @@
+// Description: Html Agility Pack - HTML Parsers, selectors, traversors, manupulators.
+// Website & Documentation: http://html-agility-pack.net
+// Forum & Issues: https://github.com/zzzprojects/html-agility-pack
+// License: https://github.com/zzzprojects/html-agility-pack/blob/master/LICENSE
+// More projects: http://www.zzzprojects.com/
+// Copyright © ZZZ Projects Inc. 2014 - 2017. All rights reserved.
+
+#if !(NETSTANDARD1_3 || NETSTANDARD1_6 || METRO)
+
+using System;
+using System.Net;
+
+namespace HtmlAgilityPack
+{
+    /// <summary>
+    /// Implement the initialization of HttpWebRequest.
+    /// </summary>
+    internal class HttpWebRequestFactory : IHttpWebRequestFactory
+    {
+        public HttpWebRequestFactory()
+        {
+        }
+
+        public IHttpWebRequest Create(Uri uri)
+        {
+            return new HttpWebRequestWrapper((HttpWebRequest)HttpWebRequest.Create(uri));
+        }
+    }
+}
+
+#endif

src/HtmlAgilityPack.Shared/HttpWebRequestWrapper.cs

@@ -0,0 +1,44 @@
+// Description: Html Agility Pack - HTML Parsers, selectors, traversors, manupulators.
+// Website & Documentation: http://html-agility-pack.net
+// Forum & Issues: https://github.com/zzzprojects/html-agility-pack
+// License: https://github.com/zzzprojects/html-agility-pack/blob/master/LICENSE
+// More projects: http://www.zzzprojects.com/
+// Copyright © ZZZ Projects Inc. 2014 - 2017. All rights reserved.
+
+#if !(NETSTANDARD1_3 || NETSTANDARD1_6 || METRO)
+
+using System;
+using System.Net;
+
+namespace HtmlAgilityPack
+{
+    /// <summary>
+    /// Wraps HttpWebRequest.
+    /// </summary>
+    internal class HttpWebRequestWrapper : IHttpWebRequest
+    {
+        HttpWebRequest _request;
+        public HttpWebRequest Request { get { return _request; } }
+
+        public HttpWebRequestWrapper(HttpWebRequest request)
+        {
+            _request = request;
+        }
+
+        public string Method { get { return _request.Method; } set { _request.Method = value; } }
+        public string UserAgent { get { return _request.UserAgent; } set { _request.UserAgent = value; } }
+        public bool AllowAutoRedirect { get { return _request.AllowAutoRedirect; } set { _request.AllowAutoRedirect = value; } }
+        public ICredentials Credentials { get { return _request.Credentials; } set { _request.Credentials = value; } }
+        public IWebProxy Proxy { get { return _request.Proxy; } set { _request.Proxy = value; } }
+        public Uri RequestUri { get { return _request.RequestUri; } }
+        public DateTime IfModifiedSince { get { return _request.IfModifiedSince; } set { _request.IfModifiedSince = value; } }
+        public CookieContainer CookieContainer { get { return _request.CookieContainer; } set { _request.CookieContainer = value; } }
+
+        public IHttpWebResponse GetResponse()
+        {
+            return new HttpWebResponseWrapper((HttpWebResponse)_request.GetResponse());
+        }
+    }
+}
+
+#endif

src/HtmlAgilityPack.Shared/HttpWebResponseWrapper.cs

@@ -0,0 +1,55 @@
+// Description: Html Agility Pack - HTML Parsers, selectors, traversors, manupulators.
+// Website & Documentation: http://html-agility-pack.net
+// Forum & Issues: https://github.com/zzzprojects/html-agility-pack
+// License: https://github.com/zzzprojects/html-agility-pack/blob/master/LICENSE
+// More projects: http://www.zzzprojects.com/
+// Copyright © ZZZ Projects Inc. 2014 - 2017. All rights reserved.
+
+#if !(NETSTANDARD1_3 || NETSTANDARD1_6 || METRO)
+
+using System;
+using System.IO;
+using System.Net;
+
+namespace HtmlAgilityPack
+{
+    /// <summary>
+    /// Wraps HttpWebResponse.
+    /// </summary>
+    internal class HttpWebResponseWrapper : IHttpWebResponse
+    {
+        HttpWebResponse _response;
+        public HttpWebResponse Response { get { return _response; } }
+
+        public HttpWebResponseWrapper(HttpWebResponse response)
+        {
+            _response = response;
+        }
+
+        public Uri ResponseUri { get { return _response.ResponseUri; } }
+        public HttpStatusCode StatusCode { get { return _response.StatusCode; } }
+        public string ContentType { get { return _response.ContentType; } }
+        public string ContentEncoding { get { return _response.ContentEncoding; } }
+        public WebHeaderCollection Headers { get { return _response.Headers; } }
+        public DateTime LastModified { get { return _response.LastModified; } }
+
+        public void Close()
+        {
+            _response.Close();
+        }
+
+        public void Dispose()
+        {
+#if !(FX20 || FX35 || FX40)
+			_response.Dispose();
+#endif
+        }
+
+        public Stream GetResponseStream()
+        {
+            return _response.GetResponseStream();
+        }
+    }
+}
+
+#endif

src/HtmlAgilityPack.Shared/IHttpWebRequest.cs

@@ -0,0 +1,34 @@
+// Description: Html Agility Pack - HTML Parsers, selectors, traversors, manupulators.
+// Website & Documentation: http://html-agility-pack.net
+// Forum & Issues: https://github.com/zzzprojects/html-agility-pack
+// License: https://github.com/zzzprojects/html-agility-pack/blob/master/LICENSE
+// More projects: http://www.zzzprojects.com/
+// Copyright © ZZZ Projects Inc. 2014 - 2017. All rights reserved.
+
+#if !(NETSTANDARD1_3 || NETSTANDARD1_6 || METRO)
+
+using System;
+using System.Net;
+
+namespace HtmlAgilityPack
+{
+    internal interface IHttpWebRequest
+    /// <summary>
+    /// Abstracts HttpWebRequest.
+    /// </summary>
+    {
+        HttpWebRequest Request { get; }
+        string Method { get; set; }
+        string UserAgent { get; set; }
+        bool AllowAutoRedirect { get; set; }
+        ICredentials Credentials { get; set; }
+        IWebProxy Proxy { get; set; }
+        Uri RequestUri { get; }
+        DateTime IfModifiedSince { get; set; }
+        CookieContainer CookieContainer { get; set; }
+
+        IHttpWebResponse GetResponse();
+    }
+}
+
+#endif