/// <summary>
/// Parses the HTML.
/// </summary>
/// <param name="result">The result.</param>
/// <param name="content">The content.</param>
/// <param name="validateSpecification">if set to <c>true</c> [validate specification].</param>
/// <returns><see cref="OpenGraph"/></returns>
/// <exception cref="InvalidSpecificationException">The parsed HTML does not meet the open graph specification</exception>
private static OpenGraph ParseHtml(OpenGraph result, string content, bool validateSpecification = false)
{
int indexOfClosingHead = Regex.Match(content, "</head>").Index;
string toParse = content.Substring(0, indexOfClosingHead + 7);
toParse = toParse + "<body></body></html>\r\n";
HtmlDocument document = new HtmlDocument();
document.LoadHtml(toParse);
HtmlNodeCollection allMeta = document.DocumentNode.SelectNodes("//meta");
var urlPropertyPatterns = new[] { "image", "url^" };
var openGraphMetaTags = from meta in allMeta ?? new HtmlNodeCollection(null)
where (meta.Attributes.Contains("property") && meta.Attributes["property"].Value.StartsWith("og:")) ||
(meta.Attributes.Contains("name") && meta.Attributes["name"].Value.StartsWith("og:"))
select meta;
foreach (HtmlNode metaTag in openGraphMetaTags)
{
string value = GetOpenGraphValue(metaTag);
string property = GetOpenGraphKey(metaTag);
if (string.IsNullOrWhiteSpace(value))
{
continue;
}
if (result._openGraphData.ContainsKey(property))
{
continue;
}
foreach (var urlPropertyPattern in urlPropertyPatterns)
{
if (Regex.IsMatch(property, urlPropertyPattern))
{
value = HtmlDecodeUrl(value);
break;
}
}
result._openGraphData.Add(property, value);
}
string type;
result._openGraphData.TryGetValue("type", out type);
result.Type = type ?? string.Empty;
string title;
result._openGraphData.TryGetValue("title", out title);
result.Title = title ?? string.Empty;
try
{
string image;
result._openGraphData.TryGetValue("image", out image);
result.Image = new Uri(image ?? string.Empty);
}
catch (UriFormatException)
{
// do nothing
}
catch (ArgumentException)
{
// do nothing
}
try
{
string url;
result._openGraphData.TryGetValue("url", out url);
result.Url = new Uri(url ?? string.Empty);
}
catch (UriFormatException)
{
// do nothing
}
catch (ArgumentException)
{
// do nothing
}
if (validateSpecification)
{
foreach (string required in RequiredMeta)
{
if (!result.ContainsKey(required))
{
throw new InvalidSpecificationException("The parsed HTML does not meet the open graph specification");
}
}
}
return(result);
}
