private void AddBooksToPublisher(Publisher publisher, string url, List<string> visitedLinks, List<string> linksToVisit)
{
visitedLinks.Add(url);
linksToVisit.Remove(url);
HtmlDocument doc = new HtmlDocument();
int httpCode = 0;
doc.LoadHtml(GetContent(url, ref httpCode));
if (httpCode == 200)
{
HtmlNodeCollection books = doc.DocumentNode.SelectNodes("//div[@id='conteudo']//a[not(@class='l11')]");
if (books != null)
{
string href = string.Empty;
foreach (HtmlNode link in books)
{
href = link.Attributes["href"].Value;
if (href.StartsWith("/editora/livros/"))
{
href = "http://www.skoob.com.br" + href;
if (!visitedLinks.Contains(href) && !linksToVisit.Contains(href))
linksToVisit.Add(href);
}
else
{
href = href.Replace("/livro/", "").Split('-')[0];
publisher.Books.Add(href);
}
}
}
}
}
public void ParsePublishers(object info)
{
JavaScriptSerializer oSerializer = new JavaScriptSerializer();
Parameters parameters = (Parameters)info;
string url = string.Empty;
int httpCode = 0;
foreach (int id in parameters.Ids)
{
url = string.Format(baseUrl, "editora", id);
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(GetContent(url, ref httpCode));
if (httpCode == 200)
{
Publisher publisher = new Publisher();
publisher.Id = id;
HtmlNode name = doc.DocumentNode.SelectSingleNode("//h1[@class='titulo_editora']");
publisher.Name = string.Empty;
if (name != null)
publisher.Name = name.InnerText;
HtmlNode description = doc.DocumentNode.SelectSingleNode("//div[@id='historico']");
publisher.Description = string.Empty;
if (description != null)
publisher.Description = description.InnerHtml;
HtmlNode image = doc.DocumentNode.SelectSingleNode("//div[@id='menu']//img");
publisher.Image = string.Empty;
if (image != null)
publisher.Image = "http://www.skoob.com.br" + image.Attributes["src"].Value;
HtmlNodeCollection authors = doc.DocumentNode.SelectNodes("//div[@class='viewport']//ul[@class='overview']//li//img");
if (authors != null)
{
string[] parts;
foreach (HtmlNode link in authors)
{
parts = link.Attributes["src"].Value.Split('/');
publisher.Authors.Add(parts[parts.Length-2]);
}
}
HtmlNode blog = doc.DocumentNode.SelectSingleNode("//a[text()='Blog']");
publisher.Blog = string.Empty;
if (blog != null)
publisher.Blog = blog.Attributes["href"].Value;
HtmlNode site = doc.DocumentNode.SelectSingleNode("//a[text()='Site oficial']");
publisher.Site = string.Empty;
if (site != null)
publisher.Site = site.Attributes["href"].Value;
string network = string.Empty;
HtmlNodeCollection networks = doc.DocumentNode.SelectNodes("//a[child::img[@class='logoredes']]");
if (networks != null)
{
foreach (HtmlNode link in networks)
{
publisher.AddNetworkLink(link.Attributes["href"].Value);
}
}
List<string> visitedLinks = new List<string>();
List<string> linksToVisit = new List<string>();
url = string.Concat(string.Format(baseUrl, "editora/livros", id), "/mpage:1");
AddBooksToPublisher(publisher, url, visitedLinks, linksToVisit);
while (linksToVisit.Count > 0)
{
AddBooksToPublisher(publisher, linksToVisit[0], visitedLinks, linksToVisit);
}
CustomSave("Publishers", oSerializer.Serialize(publisher), 0, 100, id);
SaveImage(publisher.Image, 0, 100, id, "Publishers");
}
}
}
