private Comment ScrapeComment(HtmlNode commentNode)
{
Comment comment = new Comment()
{
AccessedDate = DateTime.UtcNow.ToUniversalTime(),
};
commentNode.NullCheck();
HtmlNode headerNode = commentNode.SelectSingleNode(CommentsPageXPath.HeaderInfo);
IList<HtmlNode> innerHeaderNodes = headerNode.ChildNodes.Where(x => x.Name == "a").ToList();
/// userUrl, url, username, id
if (innerHeaderNodes != null && innerHeaderNodes.Count == 2)
{
if (innerHeaderNodes[0].Attributes["href"] != null)
{
comment.UserUrl = innerHeaderNodes[0].Attributes["href"].Value.SafeTrim().ToFullRtvSloUrl();
comment.UserId = this.GetIdStringFromUrl(comment.UserUrl);
}
else
{
this._logger.ErrorFormat("ScrapingService, ScrapeComment - User url is null - NODE: {0}", commentNode.SerializeHtmlNode());
}
comment.UserName = innerHeaderNodes[0].InnerText.SafeTrim();
if (innerHeaderNodes[1].Attributes["href"] != null)
{
comment.Url = innerHeaderNodes[1].Attributes["href"].Value.SafeTrim().ToFullRtvSloUrl();
comment.Id = this.GetIdFromUrl(comment.Url);
}
else
{
this._logger.ErrorFormat("ScrapingService, ScrapeComment - Comment url is null - NODE: {0}", commentNode.SerializeHtmlNode());
}
}
/// created date time
string dateCreatedString = headerNode.LastChild.InnerText.SafeTrim();
DateTime created;
if (dateCreatedString.TryParseExactLogging(ParsingHelper.ShortDateTimeParseExactPattern, this.cultureInfo, DateTimeStyles.None, out created))
{
comment.DateCreated = created.ToUniversalTime();
}
HtmlNode contentNode = commentNode.SelectSingleNode(CommentsPageXPath.Content);
if (contentNode != null)
{
string content = contentNode.InnerText.SafeTrimAndEscapeHtml();
comment.Content = content;
}
else
{
this._logger.ErrorFormat("ScrapingService, ScrapeComment - Comment content is null - URL: {0}", comment.Url);
}
/// rating
HtmlNode ratingNode = commentNode.SelectSingleNode(CommentsPageXPath.Rating);
string plusRatingString = ratingNode.SelectSingleNode(CommentsPageXPath.PlusRating).InnerText.SafeTrim();
string minusRatingString = ratingNode.SelectSingleNode(CommentsPageXPath.MinusRating).InnerText.SafeTrim();
int plusRating = this.ScrapeCommentRating(plusRatingString, comment.Url);
int minusRating = this.ScrapeCommentRating(minusRatingString, comment.Url);
comment.Rating = plusRating + minusRating;
return comment;
}
/// <summary>
/// Save new comment in RDF format
/// </summary>
/// <param name="comment"></param>
/// <param name="update"></param>
/// <returns>Guid url</returns>
public string SaveComment(Comment comment, bool update = false)
{
using (SesameHttpProtocolConnector connector = new SesameHttpProtocolConnector(RtvSloConfig.RepositoryUrl, RtvSloConfig.RepositoryName))
{
if (connector.IsReady)
{
/// SELECT ?guidUrl
/// WHERE {
/// ?guidUrl rdf:type news:Comment
/// ; news:ID "id"
/// }
string query = string.Format("SELECT ?guidUrl WHERE {{ ?guidUrl {0} {1} ; {2} \"{3}\" }}",
Predicate.RdfType, Predicate.NewsComment, Predicate.NewsId, comment.Id.ToSafeString());
if (update)
{
/// SELECT ?guidUrl ?predicate ?object
/// WHERE {
/// ?guidUrl rdf:type news:Comment
/// ; news:ID "id"
/// ; ?predicate ?object
/// }
query = string.Format("SELECT ?guidUrl ?predicate ?object WHERE {{ ?guidUrl {0} {1} ; {2} \"{3}\" ; ?predicate ?object }}",
Predicate.RdfType, Predicate.NewsComment, Predicate.NewsId, comment.Id.ToSafeString());
}
SparqlResultSet queryResult = connector.QueryFormat(query);
string guidUrl = null;
INode guidUrlNode = null;
if (queryResult != null && !queryResult.Results.IsEmpty())
{
/// comment already exist
guidUrlNode = queryResult.Results.First().Value("guidUrl");
guidUrl = ((UriNode)guidUrlNode).Uri.AbsoluteUri;
if (!update)
{
return guidUrl;
}
}
/// save new or update
using (IGraph g = new Graph())
{
g.BaseUri = RepositoryHelper.BaseUrl.ToUri();
IList<Triple> newTriples = new List<Triple>();
IList<Triple> removeTriples = new List<Triple>();
if (string.IsNullOrEmpty(guidUrl))
{
update = false;
guidUrl = string.Format(RepositoryHelper.CommentUrlPattern, Guid.NewGuid().ToString());
}
INode subject = guidUrlNode != null ? guidUrlNode.CopyNode(g) : guidUrl.ToUriNode(g);
comment.RepositoryGuidUrl = guidUrl;
/// save new
if (!update)
{
/// initialize
newTriples.Add(new Triple(subject, Predicate.RdfType.ToUriNode(g), Predicate.NewsComment.ToUriNode(g)));
/// ID
newTriples.Add(new Triple(subject, Predicate.NewsId.ToUriNode(g),
comment.Id.ToString().ToLiteralNode(g, dataType: RepositoryHelper.IntegerDataType)));
/// see also
newTriples.Add(new Triple(subject, Predicate.RdfsSeeAlso.ToUriNode(g), comment.Url.ToUriNode(g)));
/// link with post
if (!string.IsNullOrEmpty(comment.PostGuidUrl))
{
newTriples.Add(new Triple(comment.PostGuidUrl.ToUriNode(g), Predicate.SiocHasReply.ToUriNode(g), subject));
}
else
{
this._logger.FatalFormat("RepositoryService, SaveComment, PostGuidUrl IS NULL - COMMENT: {0}", comment.Url);
}
/// has creator
if (!string.IsNullOrEmpty(comment.UserGuidUrl))
{
newTriples.Add(new Triple(subject, Predicate.SiocHasCreator.ToUriNode(g), comment.UserGuidUrl.ToUriNode(g)));
}
else
{
this._logger.FatalFormat("RepositoryService, SaveComment, UserGuidUrl IS NULL - COMMENT: {0}", comment.Url);
}
}
/// accessed date
newTriples.Add(new Triple(subject, Predicate.NewsAccessed.ToUriNode(g),
comment.AccessedDate.ToString(RepositoryHelper.DateTimeFormat).ToLiteralNode(g, dataType: RepositoryHelper.DateTimeDataType)));
/// content
newTriples.Add(new Triple(subject, Predicate.SiocContent.ToUriNode(g), comment.Content.ToLiteralNode(g)));
/// date created
newTriples.Add(new Triple(subject, Predicate.DctCreated.ToUriNode(g),
comment.DateCreated.Value.ToString(RepositoryHelper.DateTimeFormat).ToLiteralNode(g, dataType: RepositoryHelper.DateTimeDataType)));
/// rating
newTriples.Add(new Triple(subject, Predicate.MmcRating.ToUriNode(g),
comment.Rating.ToString().ToLiteralNode(g, dataType: RepositoryHelper.IntegerDataType)));
/// remove old triples
if (update)
{
this.RemoveTriples(removeTriples, queryResult, g, subject,
new string[] { Predicate.NewsAccessed, Predicate.SiocContent, Predicate.DctCreated, Predicate.MmcRating });
}
/// save
connector.UpdateGraph(g.BaseUri, newTriples, removeTriples);
return guidUrl;
}
}
else
{
this._logger.FatalFormat("RepositoryService, SaveComment, SesameHttpProtocolConnector is not ready");
}
}
return null;
}
