Google search (~g) VASTLY improved! closes #402

src/NadekoBot/Modules/Searches/Searches.cs

@@ -18,6 +18,10 @@ using NadekoBot.Extensions;
 using System.IO;
 using NadekoBot.Modules.Searches.Commands.OMDB;
 using NadekoBot.Modules.Searches.Commands.Models;
+using AngleSharp.Parser.Html;
+using AngleSharp;
+using AngleSharp.Dom.Html;
+using AngleSharp.Dom;
 
 namespace NadekoBot.Modules.Searches
 {
@@ -214,6 +218,9 @@ namespace NadekoBot.Modules.Searches
                                                             .ConfigureAwait(false);
         }
 
+        //private readonly Regex googleSearchRegex = new Regex(@"<h3 class=""r""><a href=""(?:\/url?q=)?(?<link>.*?)"".*?>(?<title>.*?)<\/a>.*?class=""st"">(?<text>.*?)<\/span>", RegexOptions.Compiled);
+        //private readonly Regex htmlReplace = new Regex(@"(?:<b>(.*?)<\/b>|<em>(.*?)<\/em>)", RegexOptions.Compiled);
+
         [NadekoCommand, Usage, Description, Aliases]
         [RequireContext(ContextType.Guild)]
         public async Task Google(IUserMessage umsg, [Remainder] string terms = null)
@@ -224,8 +231,50 @@ namespace NadekoBot.Modules.Searches
             if (string.IsNullOrWhiteSpace(terms))
                 return;
 
-            await channel.SendMessageAsync($"https://google.com/search?q={ WebUtility.UrlEncode(terms).Replace(' ', '+') }")
+            terms = WebUtility.UrlEncode(terms).Replace(' ', '+');
-                           .ConfigureAwait(false);
+
+            var fullQueryLink = $"https://www.google.com/search?q={ terms }&gws_rd=cr,ssl";
+            var config = Configuration.Default.WithDefaultLoader();
+            var document = await BrowsingContext.New(config).OpenAsync(fullQueryLink);
+
+            var elems = document.QuerySelectorAll("div.g");
+
+            var resultsElem = document.QuerySelectorAll("#resultStats").FirstOrDefault();
+            var totalResults = resultsElem?.TextContent;
+            //var time = resultsElem.Children.FirstOrDefault()?.TextContent
+            //^ this doesn't work for some reason, <nobr> is completely missing in parsed collection
+            if (!elems.Any())
+                return;
+
+            var results = elems.Select<IElement, GoogleSearchResult?>(elem =>
+            {
+                var aTag = (elem.Children.FirstOrDefault().Children.FirstOrDefault() as IHtmlAnchorElement); // <h3> -> <a>
+                var href = aTag?.Href;
+                var name = aTag?.TextContent;
+                if (href == null || name == null)
+                    return null;
+
+                var txt = elem.QuerySelectorAll(".st").FirstOrDefault()?.TextContent;
+
+                if (txt == null)
+                    return null;
+
+                return new GoogleSearchResult(name, href, txt);
+            }).Where(x => x != null).Take(5);
+
+            var embed = new EmbedBuilder()
+                .WithOkColor()
+                .WithAuthor(eab => eab.WithName("Search For: " + terms)
+                    .WithUrl(fullQueryLink)
+                    .WithIconUrl("http://i.imgur.com/G46fm8J.png"))
+                .WithTitle(umsg.Author.Mention)
+                .WithFooter(efb => efb.WithText(totalResults));
+            string desc = "";
+            foreach (GoogleSearchResult res in results)
+            {
+                desc += $"[{Format.Bold(res.Title)}]({res.Link})\n{res.Text}\n\n";
+            }
+            await channel.EmbedAsync(embed.WithDescription(desc).Build()).ConfigureAwait(false);
         }
 
         [NadekoCommand, Usage, Description, Aliases]

src/NadekoBot/project.json

@@ -25,8 +25,10 @@
     "Discord.Net": {
       "target": "project"
     },
+    "AngleSharp": "0.9.9",
     "Google.Apis.Urlshortener.v1": "1.19.0.138",
     "Google.Apis.YouTube.v3": "1.20.0.701",
+    "Google.Apis.Customsearch.v1": "1.20.0.466",
     "ImageSharp": "1.0.0-alpha-000079",
     "Microsoft.EntityFrameworkCore": "1.1.0",
     "Microsoft.EntityFrameworkCore.Design": "1.1.0",