/// <summary>
///
/// </summary>
/// <param name="mediaWiki"></param>
/// <returns></returns>
public String[] Convert(mediawiki mediaWiki)
{
List <String> arParts = new List <string>();
if (mediaWiki == null)
{
return(null);
}
arParts.Add(mediaWiki.page.title);
arParts.Add(String.Format("http://en.wikipedia.org/wiki/{0}", mediaWiki.page.title));
arParts.Add(mediaWiki.page.revision.text.Value);
String text = mediaWiki.page.revision.text.Value;
//Remove all the comments
Regex regex = new Regex(@"<{1}.*?>{1}");
text = regex.Replace(text, this.MatchEvaluatorSimple);
//Process only the Introduction to save time and space
int firstHeaderIndex = text.IndexOf("==");
if (firstHeaderIndex != -1)
{
text = text.Substring(0, firstHeaderIndex);
}
//Remove all the special text
regex = new Regex(@"{{([^{{])+?}}");
while (regex.IsMatch(text))
{
text = regex.Replace(text, this.MatchEvaluatorSimple);
}
//Remove all the unneccesary text like "(, )"
regex = new Regex(@"\(([^a..zA..Z0..9])+?\)");
while (regex.IsMatch(text))
{
text = regex.Replace(text, this.MatchEvaluatorSimple);
}
//Simplify all the link text
regex = new Regex(@"\[\[([^\|\]]+)(\|{0,1})([^\|\]]+)\]\](<nowiki>(.+)?</nowiki>){0,1}");
text = regex.Replace(text, this.MatchEvaluator);
text = text.Replace("'''", "");
text = text.Replace("''", "");
text = text.Substring(0, text.Length > 500 ? 500 : text.Length - 1);
text += ".....";
arParts.Add(text);
return(arParts.ToArray());
}
/// <summary>
///
/// </summary>
/// <param name="mediaWiki"></param>
/// <returns></returns>
public String[] Convert(mediawiki mediaWiki)
{
List<String> arParts = new List<string>();
if (mediaWiki == null)
return null;
arParts.Add(mediaWiki.page.title);
arParts.Add(String.Format("http://en.wikipedia.org/wiki/{0}", mediaWiki.page.title));
arParts.Add(mediaWiki.page.revision.text.Value);
String text = mediaWiki.page.revision.text.Value;
//Remove all the comments
Regex regex = new Regex(@"<{1}.*?>{1}");
text = regex.Replace(text, this.MatchEvaluatorSimple);
//Process only the Introduction to save time and space
int firstHeaderIndex = text.IndexOf("==");
if (firstHeaderIndex != -1)
text = text.Substring(0, firstHeaderIndex);
//Remove all the special text
regex = new Regex(@"{{([^{{])+?}}");
while (regex.IsMatch(text))
text = regex.Replace(text, this.MatchEvaluatorSimple);
//Remove all the unneccesary text like "(, )"
regex = new Regex(@"\(([^a..zA..Z0..9])+?\)");
while (regex.IsMatch(text))
text = regex.Replace(text, this.MatchEvaluatorSimple);
//Simplify all the link text
regex = new Regex(@"\[\[([^\|\]]+)(\|{0,1})([^\|\]]+)\]\](<nowiki>(.+)?</nowiki>){0,1}");
text = regex.Replace(text, this.MatchEvaluator);
text = text.Replace("'''", "");
text = text.Replace("''", "");
text = text.Substring(0, text.Length > 500 ? 500 : text.Length - 1);
text += ".....";
arParts.Add(text);
return arParts.ToArray();
}
