/// <summary>
/// Checks this links URL.
///
/// All links start off in an Unchecked state. By filtering out the rest of the URL data and extracting the weapon name, we can determine which state to move the link into.Note that we can never discard the link(and thus move into a Discard state) without checking it first, since we do not have enough information from just the URL.
/// </summary>
public void Check()
{
if (DiscardedLinks.Contains(URL))
{
goto DISCARD;
}
{
string urllow = URL.ToLower();
if (urllow.StartsWith("/wiki/talk:") ||
urllow.StartsWith("/wiki/category:") ||
urllow.StartsWith("/wiki/category_talk:") ||
urllow.StartsWith("/wiki/template_talk:"))
{
goto DISCARD;
}
// If the link contains the word "weapon" in it, then we change the state of the link to Certain and download the needed information to the database.
if (urllow.Contains("weapon"))
{
if (STOPWORDS.Any(sw => URL.Contains(sw)))
{
goto DISCARD;
}
goto CERTAIN;
}
}
// If however, the link does not contain the word "weapon", then we must check if the word weapon is contained in the Categories section at the bottom of the article.
List <string> categories;
{
categories = WikipediaUtility.GetCategories(website);
if (categories == null)
{
goto DISCARD;
}
}
if (!categories.Any(c => c.ToLower().Contains("weapon")))
{
goto DISCARD;
}
if (STOPWORDS.Any(sw => categories.Any(c => c.Contains(sw))))
{
goto DISCARD;
}
CERTAIN:
State = LinkState.Certain;
return;
DISCARD:
State = LinkState.Discard;
}
/// <summary>
/// The details that are necessary in order to perform the information extraction are downloaded.
/// </summary>
public weapon DownloadDetails()
{
weapon w = new weapon();
w.weap_name = URL
.Substring(URL.LastIndexOf('/') + 1)
.Replace('_', ' ');
w.weap_url = URL;
List <string> articleText = WikipediaUtility.GetArticleText(website);
if (articleText == null || string.IsNullOrWhiteSpace(string.Join("", articleText)))
{
return(null);
}
w.weap_text = string.Join(" ", articleText);
List <string> categories = WikipediaUtility.GetCategories(website);
if (categories == null)
{
return(null);
}
w.weap_cat = string.Join(",", categories);
return(w);
}