コード例 #1
0
ファイル: Program.cs プロジェクト: Amichai/TwitterScrape
        private static void addTweet(VisualizationEntities db, DateTime date, string handle, string tweet, string url)
        {
            if (counter++ < Properties.Settings.Default.TweetIndex - 1) {
                return;
            }
            Debug.Print("Tweet counter: " + counter);
            Debug.Print("Tweet: " + tweet);
            TweetData tweetData = new TweetData(tweet);
            string tweetId = url.Split('/').Last();
            Tweet newTweet = new DB2.Tweet() { TweetText = tweet, Date = date, Username = handle.Split(' ').First(), TweetID = tweetId };
            if (!db.Tweets.Select(i => i.TweetID).Contains(tweetId)) {
                db.Tweets.AddObject(newTweet);
                db.SaveChanges();
                Properties.Settings.Default.TweetIndex = counter;
                Properties.Settings.Default.Save();
            } else {
                return;
            }

            foreach (var hashtag in tweetData.Hashtags()) {
                Hashtag extantTag;
                if (!db.Hashtags.Select(i => i.Tag).Contains(hashtag)) {
                    Debug.Print("new tag: " + hashtag);
                    extantTag = new Hashtag() { Tag = hashtag };
                    db.Hashtags.AddObject(extantTag);
                    db.SaveChanges();
                } else {
                    extantTag = db.Hashtags.Where(i => i.Tag == hashtag).Single();
                }

                db.TweetHashtags.AddObject(new TweetHashtag() { TweetID = newTweet.ID, HashtagID = extantTag.ID });
                db.SaveChanges();
            }

            foreach (string matchedUrl in tweetData.Urls()) {
                string uniqueUrl = "";
                try {
                    uniqueUrl = matchedUrl.ExpandUrl().UnuiqeUrl();
                } catch (Exception ex) {
                    continue;
                }

                ///Find matched websites:
                var knownSite = db.Websites.Where(i => i.Url == uniqueUrl);
                if (knownSite.Count() > 0) {
                    ///
                    var tweetWebsite = new TweetWebsite() { TweetID = newTweet.ID, WebsiteID = knownSite.First().ID };
                    db.TweetWebsites.AddObject(tweetWebsite);
                    db.SaveChanges();
                    continue;
                }
                ///This is a brand new website
                if (uniqueUrl.Count() > 600 || db.Websites.Any(i => i.Url == uniqueUrl)) {
                    continue;
                }
                Debug.Print("Website: " + uniqueUrl);
                string websiteUrl;
                try {
                    var page = new PageScraper(uniqueUrl);
                    string longUrlTitle = "";
                    string description = "";

                    try {
                        longUrlTitle = page.Url.GetTitle();
                    } catch { Debug.Print("Failed to get title"); }
                    try {
                        description = page.Url.GetDescription();
                    } catch { Debug.Print("Failed to get description"); }
                    var website = new DB2.Website() {
                        Url = page.Url,
                        Title = string.Concat((page.Title() ?? "").Take(300)),
                        TweetID = newTweet.ID,
                        LongUrlTitle = string.Concat(longUrlTitle.Take(100)),
                        Description = string.Concat(description.Take(500))
                    };
                    db.Websites.AddObject(website);
                    db.SaveChanges();

                    var tweetWebsite = new TweetWebsite() { TweetID = newTweet.ID, WebsiteID = website.ID };
                    db.TweetWebsites.AddObject(tweetWebsite);
                    db.SaveChanges();
                    foreach (var m2 in page.Media()) {
                        if (m2.Link.Count() > 600 || db.Media.Any(i => i.Url == m2.Link)) {
                            continue;
                        }
                        try {
                            DB2.Medium media = new DB2.Medium() { Type = m2.Type, Url = m2.Link, SourceSiteID = website.ID };
                            if (m2.Type == "image") {
                                var request = WebRequest.Create(m2.Link);

                                using (var response = request.GetResponse())
                                using (var stream = response.GetResponseStream())
                                using (var b = System.Drawing.Bitmap.FromStream(stream)) {
                                    media.Width = b.Width;
                                    media.Height = b.Height;
                                }
                            }
                            db.Media.AddObject(media);
                            db.SaveChanges();
                            Debug.Print("Media element: " + m2.Link);
                        } catch {
                            break;
                        }
                    }
                    websiteUrl = website.Url;
                } catch {
                    Debug.Print("Failed to scrape page: " + uniqueUrl);
                }
            }
            db.SaveChanges();
        }
コード例 #2
0
 /// <summary>
 /// Deprecated Method for adding a new object to the TweetWebsites EntitySet. Consider using the .Add method of the associated ObjectSet&lt;T&gt; property instead.
 /// </summary>
 public void AddToTweetWebsites(TweetWebsite tweetWebsite)
 {
     base.AddObject("TweetWebsites", tweetWebsite);
 }
コード例 #3
0
ファイル: Program.cs プロジェクト: Amichai/TwitterScrape
 private static void associateTweetsAndSites()
 {
     var db = GetDataContext();
     foreach (var website in db.Websites.ToList()) {
         TweetWebsite tweetWebsite = new TweetWebsite() { TweetID = website.TweetID,
         WebsiteID = website.ID };
         db.TweetWebsites.AddObject(tweetWebsite);
         db.SaveChanges();
     }
 }
コード例 #4
0
 /// <summary>
 /// Create a new TweetWebsite object.
 /// </summary>
 /// <param name="id">Initial value of the ID property.</param>
 /// <param name="tweetID">Initial value of the TweetID property.</param>
 /// <param name="websiteID">Initial value of the WebsiteID property.</param>
 public static TweetWebsite CreateTweetWebsite(global::System.Int32 id, global::System.Int32 tweetID, global::System.Int32 websiteID)
 {
     TweetWebsite tweetWebsite = new TweetWebsite();
     tweetWebsite.ID = id;
     tweetWebsite.TweetID = tweetID;
     tweetWebsite.WebsiteID = websiteID;
     return tweetWebsite;
 }