Added unit tests to cookie parser logic, added regex for comma separa…

…ted parsing, PR #23

ScrapySharp.Tests/Network/Cookie.txt

@@ -0,0 +1 @@
+key1=204=eBL9WYALz-YPXedwGc-utNT3YXlx7moCeN-dvDDl-7xYtHIZtPEr0OZ2bKWtEJWInrKuEnlR_-JFXZ4mMYIVQrPcTxGCZCCHoeUpdz1kU3cMa38TGrn-uaB6gH7D7A_XKa5bJvjbkatI3mvnPNgjJfd4QrHeDu8hms-c9b6to04; key2=this is a test value; expires=Tue, 08-Dec-2020 01:44:28 GMT; path=/; domain=.localhost.fakedomain; HttpOnly

ScrapySharp.Tests/ScrapySharp.Tests.csproj

@@ -2,6 +2,12 @@
   <PropertyGroup>
     <TargetFrameworks>netcoreapp2.0</TargetFrameworks>
   </PropertyGroup>
+  <ItemGroup>
+    <Compile Remove="When_build_HtmlDom.cs" />
+  </ItemGroup>
+  <ItemGroup>
+    <None Remove="Network\Cookie.txt" />
+  </ItemGroup>
   <ItemGroup>
     <PackageReference Include="HtmlAgilityPack" Version="1.7.*" />
     <PackageReference Include="Microsoft.NET.Test.Sdk" Version="15.7.0" />
@@ -41,5 +47,13 @@
     <Content Include="Html\Page1.htm">
       <CopyToOutputDirectory>Always</CopyToOutputDirectory>
     </Content>
+    <Content Include="When_build_HtmlDom.cs">
+      <CopyToOutputDirectory>Always</CopyToOutputDirectory>
+    </Content>
+  </ItemGroup>
+  <ItemGroup>
+    <Content Include="Network\Cookie.txt">
+      <CopyToOutputDirectory>Always</CopyToOutputDirectory>
+    </Content>
   </ItemGroup>
 </Project>

ScrapySharp.Tests/When_parse_cookies.cs

@@ -0,0 +1,56 @@
+// ReSharper disable InconsistentNaming
+
+using System.IO;
+using NUnit.Framework;
+using ScrapySharp.Html.Dom;
+using ScrapySharp.Html.Parsing;
+using System.Linq;
+using ScrapySharp.Network;
+using System.Net;
+
+namespace ScrapySharp.Tests
+{
+    [TestFixture]
+    public class When_parse_cookies
+    {
+        [Test]
+        public void When_parse_standard_cookie()
+        {
+            var cookie = GetCookie();
+
+            CookiesParser parser = new CookiesParser(".localhost.fakedomain");
+            var cookieList = parser.ParseCookies(cookie);
+
+            Assert.AreEqual(2, cookieList.Count);
+        }
+
+        [Test]
+        public void When_parse_csv_cookie()
+        {
+            var csvCookie = GetCookie().Replace(";",",");            
+
+            CookiesParser parser = new CookiesParser(".localhost.fakedomain");
+            var cookieList = parser.ParseCookies(csvCookie);
+
+            Assert.AreEqual(2, cookieList.Count);
+        }
+
+        [Test]
+        public void When_parse_csv_invalid_cookie()
+        {
+            string invalidCookie = GetCookie().Replace(";",";,");
+            CookiesParser parser = new CookiesParser(".localhost.fakedomain");
+
+            Assert.Throws<CookieException>(()=> { parser.ParseCookies(invalidCookie); });
+        }
+
+        private static string GetCookie()
+        {
+            var cookie = File.ReadAllText("Network/Cookie.txt");
+
+            return cookie;
+        }
+    }
+}
+
+// ReSharper restore InconsistentNaming

ScrapySharp.Tests/When_parses_using_CssSelector_with_fsharp_tokenizer.cs

@@ -1,11 +1,10 @@
 // ReSharper disable InconsistentNaming
 
-using System.Linq;
 using HtmlAgilityPack;
 using NUnit.Framework;
 using ScrapySharp.Core;
 using ScrapySharp.Extensions;
-using ScrapySharp.Core;
+using System.Linq;
 
 namespace ScrapySharp.Tests
 {

ScrapySharp/Network/CookiesParser.cs

@@ -8,8 +8,9 @@ namespace ScrapySharp.Network
     public class CookiesParser
     {
         private readonly string defaultDomain;
-        private static readonly Regex splitCookiesRegex = new Regex(@"\s*(?<name>[^=]+)=(?<val>[^;]+)?[,;]+", RegexOptions.Compiled);
-
+        private static readonly Regex splitCookiesRegex = new Regex(@"\s*(?<name>[^=]+)=(?<val>[^;]+)?[;]+", RegexOptions.Compiled);
+        private static readonly Regex splitCookiesCsvRegex = new Regex(@"\s*(?<name>[^=]+)=(?<val>.*?),(?=[^,]+?(?:=|$))+", RegexOptions.Compiled);
+
         public CookiesParser(string defaultDomain)
         {
             this.defaultDomain = defaultDomain;
@@ -19,24 +20,23 @@ public List<KeyValuePair<string, string>> ParseValuePairs(string cookiesExpressi
         {
             var list = new List<KeyValuePair<string, string>>();
 
-            var cookiesArr = cookiesExpression.Split(",".ToCharArray());
+            Match match;
+            if (cookiesExpression.Contains(";"))
+                match = splitCookiesRegex.Match(cookiesExpression);
+            else
+                match = splitCookiesCsvRegex.Match(cookiesExpression);
 
-            foreach (var cookieString in cookiesArr)
+            while (match.Success)
             {
-                var match = splitCookiesRegex.Match(cookieString);
-
-                while (match.Success)
+                if (match.Groups["name"].Success && match.Groups["val"].Success)
                 {
-                    if (match.Groups["name"].Success && match.Groups["val"].Success)
+                    try
                     {
-                        try
-                        {
-                            list.Add(new KeyValuePair<string, string>(match.Groups["name"].Value, match.Groups["val"].Value));
-                        }
-                        catch (CookieException) { }
+                        list.Add(new KeyValuePair<string, string>(match.Groups["name"].Value, match.Groups["val"].Value));
                     }
-                    match = match.NextMatch();
+                    catch (CookieException) { }
                 }
+                match = match.NextMatch();
             }
 
             return list;
@@ -50,7 +50,7 @@ public List<Cookie> ParseCookies(string cookiesExpression)
             for (int i = 0; i < keyValuePairs.Count; i++)
             {
                 var pair = keyValuePairs[i];
-                if (pair.Key.Equals("path", StringComparison.InvariantCultureIgnoreCase) 
+                if (pair.Key.Equals("path", StringComparison.InvariantCultureIgnoreCase)
                     || pair.Key.Equals("domain", StringComparison.InvariantCultureIgnoreCase)
                     || pair.Key.Equals("expires", StringComparison.InvariantCultureIgnoreCase))
                     continue;