/// <summary>
/// Downloads the HTML of the specified URL and parses it for open graph content.
/// </summary>
/// <param name="url">The URL to download the HTML from.</param>
/// <param name="userAgent">The user agent to use when downloading content. The default is <c>"facebookexternalhit"</c> which is required for some site (like amazon) to include open graph data.</param>
/// <param name="validateSpecification">if set to <c>true</c> verify that the document meets the required attributes of the open graph specification.</param>
/// <returns><see cref="OpenGraph"/></returns>
public static OpenGraph ParseUrl(Uri url, string userAgent = "facebookexternalhit", bool validateSpecification = false)
{
OpenGraph result = new OpenGraph {
OriginalUrl = url
};
HttpDownloader downloader = new HttpDownloader(url, null, userAgent);
string html = "";//downloader.GetPage();
return(ParseHtml(result, html, validateSpecification));
}
/// <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="OpenGraph_Net.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 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;
}
result.openGraphData.Add(property, value);
}
string type = string.Empty;
result.openGraphData.TryGetValue("type", out type);
result.Type = type ?? string.Empty;
string title = string.Empty;
result.openGraphData.TryGetValue("title", out title);
result.Title = title ?? string.Empty;
try
{
string image = string.Empty;
result.openGraphData.TryGetValue("image", out image);
result.Image = new Uri(image);
}
catch (UriFormatException)
{
// do nothing
}
catch (ArgumentException)
{
// do nothing
}
try
{
string url = string.Empty;
result.openGraphData.TryGetValue("url", out url);
result.Url = new Uri(url);
}
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);
}
/// <summary>
/// Parses the HTML for open graph content.
/// </summary>
/// <param name="content">The HTML to parse.</param>
/// <param name="validateSpecification">if set to <c>true</c> verify that the document meets the required attributes of the open graph specification.</param>
/// <returns><see cref="OpenGraph"/></returns>
public static OpenGraph ParseHtml(string content, bool validateSpecification = false)
{
OpenGraph result = new OpenGraph();
return(ParseHtml(result, content, validateSpecification));
}
/// <summary>
/// Makes the graph.
/// </summary>
/// <param name="title">The title.</param>
/// <param name="type">The type.</param>
/// <param name="image">The image.</param>
/// <param name="url">The URL.</param>
/// <param name="description">The description.</param>
/// <param name="siteName">Name of the site.</param>
/// <param name="audio">The audio.</param>
/// <param name="video">The video.</param>
/// <param name="locale">The locale.</param>
/// <param name="localeAlternate">The locale alternate.</param>
/// <param name="determiner">The determiner.</param>
/// <returns><see cref="OpenGraph"/></returns>
public static OpenGraph MakeGraph(
string title,
string type,
string image,
string url,
string description = "",
string siteName = "",
string audio = "",
string video = "",
string locale = "",
IList <string> localeAlternate = null,
string determiner = "")
{
var graph = new OpenGraph
{
Title = title,
Type = type,
Image = new Uri(image, UriKind.Absolute),
Url = new Uri(url, UriKind.Absolute)
};
graph.openGraphData.Add("title", title);
graph.openGraphData.Add("type", type);
graph.openGraphData.Add("image", image);
graph.openGraphData.Add("url", url);
if (!string.IsNullOrWhiteSpace(description))
{
graph.openGraphData.Add("description", description);
}
if (!string.IsNullOrWhiteSpace(siteName))
{
graph.openGraphData.Add("site_name", siteName);
}
if (!string.IsNullOrWhiteSpace(audio))
{
graph.openGraphData.Add("audio", audio);
}
if (!string.IsNullOrWhiteSpace(video))
{
graph.openGraphData.Add("video", video);
}
if (!string.IsNullOrWhiteSpace(locale))
{
graph.openGraphData.Add("locale", locale);
}
if (!string.IsNullOrWhiteSpace(determiner))
{
graph.openGraphData.Add("determiner", determiner);
}
if (localeAlternate != null)
{
graph.localAlternatives = localeAlternate;
}
return(graph);
}