public SiteReader(string url)
{
if (String.IsNullOrWhiteSpace(url))
{
throw new Exception("Invalid url.");
}
this.SiteInfo = new SiteInfo();
this.SiteInfo.Content = GetContent(url);
TryParseMeta(SiteInfo.Content, SiteInfo);
this.SiteInfo.ImageUrls = ParseImages(SiteInfo.Content);
}
private bool TryParseMeta(string strIn, SiteInfo siteInfo)
{
try
{
// Parse the title
Match titleMatch = Regex.Match(strIn, @"<title>([^<]*)</title>", RegexOptions.IgnoreCase | RegexOptions.Multiline);
siteInfo.MetaTitle = titleMatch.Groups[1].Value;
// Parse the meta keywords
Match keywordMatch = Regex.Match(strIn, @"<meta\s*name=""keywords""\s*content=""([^<]*)\""\s*/?>", RegexOptions.IgnoreCase | RegexOptions.Multiline);
siteInfo.MetaKeywords = keywordMatch.Groups[1].Value;
// Parse the meta description
Match descriptionMatch = Regex.Match(strIn, @"<meta\s*name=""description""\s*content=""([^<]*)""\s*/?>", RegexOptions.IgnoreCase | RegexOptions.Multiline);
siteInfo.MetaDescription = descriptionMatch.Groups[1].Value;
return true;
}
catch
{
return false;
}
}
