OpenTween/Thumbnail/Services/MetaThumbnailService.cs

   1 // OpenTween - Client of Twitter
   2 // Copyright (c) 2012 kim_upsilon (@kim_upsilon) <https://upsilo.net/~upsilon/>
   3 // All rights reserved.
   4 //
   5 // This file is part of OpenTween.
   6 //
   7 // This program is free software; you can redistribute it and/or modify it
   8 // under the terms of the GNU General Public License as published by the Free
   9 // Software Foundation; either version 3 of the License, or (at your option)
  10 // any later version.
  11 //
  12 // This program is distributed in the hope that it will be useful, but
  13 // WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
  14 // or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
  15 // for more details.
  16 //
  17 // You should have received a copy of the GNU General Public License along
  18 // with this program. If not, see <http://www.gnu.org/licenses/>, or write to
  19 // the Free Software Foundation, Inc., 51 Franklin Street - Fifth Floor,
  20 // Boston, MA 02110-1301, USA.
  21
  22 using System;
  23 using System.Collections.Generic;
  24 using System.Linq;
  25 using System.Net;
  26 using System.Net.Http;
  27 using System.Text;
  28 using System.Text.RegularExpressions;
  29 using System.Threading;
  30 using System.Threading.Tasks;
  31
  32 namespace OpenTween.Thumbnail.Services
  33 {
  34     /// <summary>
  35     /// og:image や twitter:image をスクレイピングしてサムネイルURLを抽出する
  36     /// </summary>
  37     class MetaThumbnailService : IThumbnailService
  38     {
  39         protected static Regex metaPattern = new Regex("<meta (name|property)=[\"'](?<name>.+?)[\"'] (content|value)=[\"'](?<content>.+?)[\"']");
  40         protected static string[] propertyNames = { "twitter:image", "og:image" };
  41
  42         protected HttpClient http
  43         {
  44             get { return this.localHttpClient ?? HttpConnection.GlobalHttpClient; }
  45         }
  46         private readonly HttpClient localHttpClient;
  47
  48         protected readonly Regex regex;
  49
  50         public MetaThumbnailService(string urlPattern)
  51             : this(null, urlPattern)
  52         {
  53         }
  54
  55         public MetaThumbnailService(HttpClient http, string urlPattern)
  56         {
  57             this.localHttpClient = http;
  58             this.regex = new Regex(urlPattern);
  59         }
  60
  61         public override async Task<ThumbnailInfo> GetThumbnailInfoAsync(string url, PostClass post, CancellationToken token)
  62         {
  63             if (!this.regex.IsMatch(url))
  64                 return null;
  65
  66             try
  67             {
  68                 var content = await this.FetchImageUrlAsync(url, token)
  69                     .ConfigureAwait(false);
  70
  71                 var thumbnailUrl = this.GetThumbnailUrl(content);
  72                 if (string.IsNullOrEmpty(thumbnailUrl)) return null;
  73
  74                 return new ThumbnailInfo
  75                 {
  76                     ImageUrl = url,
  77                     ThumbnailUrl = thumbnailUrl,
  78                     TooltipText = null,
  79                 };
  80             }
  81             catch (HttpRequestException) { }
  82
  83             return null;
  84         }
  85
  86         protected virtual string GetThumbnailUrl(string html)
  87         {
  88             var matches = MetaThumbnailService.metaPattern.Matches(html);
  89
  90             foreach (Match match in matches)
  91             {
  92                 var propertyName = match.Groups["name"].Value;
  93                 if (MetaThumbnailService.propertyNames.Contains(propertyName))
  94                 {
  95                     return match.Groups["content"].Value;
  96                 }
  97             }
  98
  99             return null;
 100         }
 101
 102         protected virtual async Task<string> FetchImageUrlAsync(string url, CancellationToken token)
 103         {
 104             using (var response = await this.http.GetAsync(url, token).ConfigureAwait(false))
 105             {
 106                 response.EnsureSuccessStatusCode();
 107
 108                 return await response.Content.ReadAsStringAsync()
 109                     .ConfigureAwait(false);
 110             }
 111         }
 112     }
 113 }