private async Task SaveArticleToFile(Article article
, string content, IProgress<DownloadStringTaskAsyncExProgress> progress)
{
if (progress != null)
{
progress.Report(new DownloadStringTaskAsyncExProgress()
{
Text = string.Format("{0} [Save] {1} ({2}) {0}"
, Environment.NewLine
, article.Title
, article.URL)
});
}
var filePath = Path.Combine(BaseFolder, article.Title.ToValidFileName() + ".htm");
await SaveToFileAsync(content, filePath);
}
public abstract Task<bool> ExtractArticleContent(Article article
, IProgress<DownloadStringTaskAsyncExProgress> progress);
private async Task<List<Catalog>> ExtractCatalogsFromWebPage(string url)
{
List<Catalog> reusltCatalogs = new List<Catalog>();
var uri = new Uri(url);
var browser1 = new ScrapingBrowser();
browser1.Encoding = Encoding.UTF8;
var html1 = browser1.DownloadString(uri);
var doc = new HtmlDocument();
doc.LoadHtml(html1);
var html = doc.DocumentNode;
foreach (var script in doc.DocumentNode.Descendants("script").ToArray())
{
script.Remove();
}
foreach (var style in doc.DocumentNode.Descendants("style").ToArray())
{
style.Remove();
}
foreach (var comment in doc.DocumentNode.SelectNodes("//comment()").ToArray())
{
comment.Remove();
}
var days = html.CssSelect("div.day");
if (!days.Any())
{
days = html.CssSelect("div#container >div#wrapper >div#content");
}
foreach (var day in days)
{
var catalog = new Catalog();
catalog.IsChecked = true;
var title = day.CssSelect("div.dayTitle").FirstOrDefault();
if(title!=null)
{
catalog.Title = title.InnerText.ClearNotWords();
var atricles = day.CssSelect("div.postTitle");
foreach (var atricle in atricles)
{
var article = new Article();
article.Title = atricle.InnerText.ClearNotWords();
var articleTitleEl = atricle.CssSelect("a.postTitle2");
article.URL = articleTitleEl.First().Attributes["href"].Value;
catalog.Articles.Add(article);
}
}
else
{
catalog.Title = "CataLog"+DateTime.Now.ToShortTimeString();
var atricles = day.CssSelect("div.post");
foreach (var atricle in atricles)
{
var article = new Article();
var articleTitleEl = atricle.CssSelect("a.PostTitle");
article.Title = articleTitleEl.First().InnerText.ClearNotWords();
article.URL = articleTitleEl.First().Attributes["href"].Value;
catalog.Articles.Add(article);
}
}
reusltCatalogs.Add(catalog);
}
return reusltCatalogs;
}
/// <summary>
/// Initializes a new instance of the ArchiveViewModel class.
/// </summary>
public ArticleViewModel(Article article)
{
CurrentEntity = article;
}
public override async Task<bool> ExtractArticleContent(Article article
,IProgress<DownloadStringTaskAsyncExProgress> progress )
{
if (article.IsLoaded)
{
return true;
}
web.URL = article.URL;
if (progress != null)
{
progress.Report(new DownloadStringTaskAsyncExProgress()
{
Text = string.Format("{0} [Load] {1} ({2}) {0}"
, Environment.NewLine, article.Title
, article.URL)
});
}
string content = await Task.Run(() => web.Get()).ConfigureAwait(false);
Match match = reg_con.Match(content);
if (match.Success)
{
content = match.Groups[1].Value.Trim();
article.Content= htmlString.Replace("{0}", article.Title).Replace("\n{1}", content);
article.IsLoaded=true;
}
return match.Success;
}
