private static void WriteBadgeInfo(SiteUris uris, string badgeName, SequentialTurtleWriter w)
{
w.StartTriple(uris.CreateBadgeUri(badgeName));
w.AddToTriple(uris.GeneralUris.TypeProperty, uris.GeneralUris.BadgeType);
w.AddToTriple(uris.GeneralUris.LabelProperty, badgeName);
uris.LinkToSite(w);
}
private static void ConvertTag(SiteUris uris, XmlReader r, SequentialTurtleWriter w)
{
Uri subjectUri;
if (r.MoveToAttribute("TagName"))
{
subjectUri = uris.CreateTagUri(r.Value);
w.StartTriple(subjectUri);
}
else
{
r.MoveToElement();
ConsoleHelper.WriteErrorLine("No TagName attribute found on element {0}. Skipping element.", r.ReadOuterXml());
return;
}
w.AddToTriple(uris.GeneralUris.LabelProperty, r.Value);
w.AddToTriple(uris.GeneralUris.TypeProperty, uris.GeneralUris.TagType);
uris.LinkToSite(w);
if (r.MoveToAttribute("ExcerptPostId"))
{
w.AddToTriple(uris.GeneralUris.TagExcerptProperty, uris.CreatePostUri(r.Value));
}
if (r.MoveToAttribute("WikiPostId"))
{
w.AddToTriple(uris.GeneralUris.TagDescriptionProperty, uris.CreatePostUri(r.Value));
}
// TODO: Count
}
private static void ConvertComment(SiteUris uris, XmlReader r, SequentialTurtleWriter w)
{
Uri subjectUri;
if (r.MoveToAttribute("Id"))
{
subjectUri = uris.CreateCommentUri(r.Value);
w.StartTriple(subjectUri);
}
else
{
r.MoveToElement();
ConsoleHelper.WriteErrorLine("No Id attribute found on element {0}. Skipping element.", r.ReadOuterXml());
return;
}
w.AddToTriple(uris.GeneralUris.TypeProperty, uris.GeneralUris.CommentType);
uris.LinkToSite(w);
if (r.MoveToAttribute("Score"))
{
w.AddToTriple(uris.GeneralUris.ScoreProperty, long.Parse(r.Value));
}
if (r.MoveToAttribute("Text"))
{
w.AddToTriple(uris.GeneralUris.DescriptionProperty, r.Value);
}
if (r.MoveToAttribute("CreationDate"))
{
w.AddToTriple(uris.GeneralUris.DateProperty, DateTime.Parse(r.Value, System.Globalization.CultureInfo.InvariantCulture));
}
if (r.MoveToAttribute("UserId"))
{
w.AddToTriple(uris.GeneralUris.OwnerProperty, uris.CreateUserUri(r.Value));
}
if (r.MoveToAttribute("PostId"))
{
w.StartTriple(uris.CreatePostUri(r.Value));
w.AddToTriple(uris.GeneralUris.CommentProperty, subjectUri);
w.StartTriple(subjectUri);
}
else
{
ConsoleHelper.WriteWarningLine("Orphaned comment: {0}", subjectUri);
}
}
private static void ConvertUser(SiteUris uris, XmlReader r, SequentialTurtleWriter w, ICollection <string> malformedIris, ref long totalMalformedIriCount)
{
Uri subjectUri;
if (r.MoveToAttribute("Id"))
{
subjectUri = uris.CreateUserUri(r.Value);
w.StartTriple(subjectUri);
}
else
{
r.MoveToElement();
ConsoleHelper.WriteErrorLine("No Id attribute found on element {0}. Skipping element.", r.ReadOuterXml());
return;
}
w.AddToTriple(uris.GeneralUris.TypeProperty,
uris.GeneralUris.UserType);
uris.LinkToSite(w);
if (r.MoveToAttribute("DisplayName"))
{
w.AddToTriple(uris.GeneralUris.LabelProperty, r.Value);
w.AddToTriple(uris.GeneralUris.UserNameProperty, r.Value);
}
if (r.MoveToAttribute("CreationDate"))
{
w.AddToTriple(uris.GeneralUris.DateProperty, DateTime.Parse(r.Value, System.Globalization.CultureInfo.InvariantCulture));
}
if (r.MoveToAttribute("Reputation"))
{
w.AddToTriple(uris.GeneralUris.ReputationProperty, long.Parse(r.Value));
}
if (r.MoveToAttribute("Location"))
{
w.AddToTriple(uris.GeneralUris.LocationProperty, r.Value);
}
if (r.MoveToAttribute("Age"))
{
w.AddToTriple(uris.GeneralUris.AgeProperty, long.Parse(r.Value));
}
if (r.MoveToAttribute("Views"))
{
w.AddToTriple(uris.GeneralUris.ViewCountProperty, long.Parse(r.Value));
}
if (r.MoveToAttribute("UpVotes"))
{
w.AddToTriple(uris.GeneralUris.UpVotesProperty, long.Parse(r.Value));
}
if (r.MoveToAttribute("DownVotes"))
{
w.AddToTriple(uris.GeneralUris.DownVotesProperty, long.Parse(r.Value));
}
if (r.MoveToAttribute("WebsiteUrl"))
{
string websiteUrl = r.Value;
if (!string.IsNullOrWhiteSpace(websiteUrl))
{
if (websiteUrl.ToLowerInvariant().StartsWith("http"))
{
websiteUrl = "http" + websiteUrl.Substring(4);
}
if (!emptyWebsiteRegex.IsMatch(websiteUrl.ToLowerInvariant()))
{
if (websiteRegex.IsMatch(websiteUrl))
{
try {
Uri homepageUrl = new Uri(websiteUrl);
w.AddToTriple(uris.GeneralUris.WebsiteProperty, homepageUrl);
}
catch (UriFormatException) {
totalMalformedIriCount++;
if (malformedIris.Count < GlobalData.Options.MaxDisplayedMalformedIris)
{
malformedIris.Add(websiteUrl);
}
w.AddToTriple(uris.GeneralUris.WebsiteProperty, websiteUrl);
}
}
else
{
totalMalformedIriCount++;
if (malformedIris.Count < GlobalData.Options.MaxDisplayedMalformedIris)
{
malformedIris.Add(websiteUrl);
}
w.AddToTriple(uris.GeneralUris.WebsiteProperty, websiteUrl);
}
}
}
}
if (r.MoveToAttribute("AccountId"))
{
w.AddToTriple(uris.GeneralUris.AccountProperty, uris.GeneralUris.CreateAccountUri(r.Value));
GlobalData.AccountIds.Add(r.Value);
}
if (r.MoveToAttribute("AboutMe"))
{
string desc = r.Value;
if (!string.IsNullOrWhiteSpace(desc))
{
w.AddToTriple(uris.GeneralUris.DescriptionProperty, desc);
}
}
if (r.MoveToAttribute("LastAccessDate"))
{
w.AddToTriple(uris.GeneralUris.LastSeenProperty, DateTime.Parse(r.Value, System.Globalization.CultureInfo.InvariantCulture));
}
}
private static void ConvertVote(SiteUris uris, XmlReader r, SequentialTurtleWriter w, UnknownValueStore <string> unknownVoteTypeIds)
{
Uri subjectUri;
if (r.MoveToAttribute("Id"))
{
subjectUri = uris.CreateVoteUri(r.Value);
w.StartTriple(subjectUri);
}
else
{
r.MoveToElement();
ConsoleHelper.WriteErrorLine("No Id attribute found on element {0}. Skipping element.", r.ReadOuterXml());
return;
}
if (r.MoveToAttribute("VoteTypeId"))
{
switch (r.Value)
{
case "1": // acceptance
if (GlobalData.Options.FullTimeInfo)
{
w.AddToTriple(uris.GeneralUris.TypeProperty, uris.GeneralUris.AcceptanceType);
uris.LinkToSite(w);
if (r.MoveToAttribute("PostId"))
{
w.StartTriple(uris.CreatePostUri(r.Value));
w.AddToTriple(uris.GeneralUris.AcceptanceProperty, subjectUri);
w.StartTriple(subjectUri);
}
if (r.MoveToAttribute("CreationDate"))
{
w.AddToTriple(uris.GeneralUris.DateProperty, DateTime.Parse(r.Value, System.Globalization.CultureInfo.InvariantCulture));
}
}
break;
case "2": // upvote
if (GlobalData.Options.FullTimeInfo)
{
w.AddToTriple(uris.GeneralUris.TypeProperty, uris.GeneralUris.UpVoteType);
uris.LinkToSite(w);
if (r.MoveToAttribute("CreationDate"))
{
w.AddToTriple(uris.GeneralUris.DateProperty, DateTime.Parse(r.Value, System.Globalization.CultureInfo.InvariantCulture));
}
if (r.MoveToAttribute("PostId"))
{
w.StartTriple(uris.CreatePostUri(r.Value));
w.AddToTriple(uris.GeneralUris.VoteProperty, subjectUri);
w.StartTriple(subjectUri);
}
}
break;
case "3": // downvote
if (GlobalData.Options.FullTimeInfo)
{
w.AddToTriple(uris.GeneralUris.TypeProperty, uris.GeneralUris.DownVoteType);
uris.LinkToSite(w);
if (r.MoveToAttribute("CreationDate"))
{
w.AddToTriple(uris.GeneralUris.DateProperty, DateTime.Parse(r.Value, System.Globalization.CultureInfo.InvariantCulture));
}
if (r.MoveToAttribute("PostId"))
{
w.StartTriple(uris.CreatePostUri(r.Value));
w.AddToTriple(uris.GeneralUris.VoteProperty, subjectUri);
w.StartTriple(subjectUri);
}
}
break;
case "4": // offensive
break;
case "5": // favorite
if (GlobalData.Options.FullTimeInfo)
{
w.AddToTriple(uris.GeneralUris.TypeProperty, uris.GeneralUris.FavoriteType);
uris.LinkToSite(w);
if (r.MoveToAttribute("CreationDate"))
{
w.AddToTriple(uris.GeneralUris.DateProperty, DateTime.Parse(r.Value, System.Globalization.CultureInfo.InvariantCulture));
}
if (r.MoveToAttribute("PostId"))
{
w.AddToTriple(uris.GeneralUris.PostProperty, uris.CreatePostUri(r.Value));
}
if (r.MoveToAttribute("UserId"))
{
w.StartTriple(uris.CreateUserUri(r.Value));
w.AddToTriple(uris.GeneralUris.FavoriteProperty, subjectUri);
w.StartTriple(subjectUri);
}
}
else
{
if (r.MoveToAttribute("UserId"))
{
w.StartTriple(uris.CreateUserUri(r.Value));
if (r.MoveToAttribute("PostId"))
{
w.AddToTriple(uris.GeneralUris.FavoriteProperty, uris.CreatePostUri(r.Value));
}
w.StartTriple(subjectUri);
}
}
break;
case "6": // closed
break;
case "7": // reopened
break;
case "8": // bounty started
w.AddToTriple(uris.GeneralUris.TypeProperty, uris.GeneralUris.StartOfBountyType);
if (r.MoveToAttribute("PostId"))
{
w.AddToTriple(uris.GeneralUris.PostProperty, uris.CreatePostUri(r.Value));
}
if (r.MoveToAttribute("UserId"))
{
w.AddToTriple(uris.GeneralUris.DonorProperty, uris.CreateUserUri(r.Value));
}
if (r.MoveToAttribute("CreationDate"))
{
w.AddToTriple(uris.GeneralUris.DateProperty, DateTime.Parse(r.Value, System.Globalization.CultureInfo.InvariantCulture));
}
if (r.MoveToAttribute("BountyAmount"))
{
w.AddToTriple(uris.GeneralUris.OfferedAmountProperty, long.Parse(r.Value));
}
break;
case "9": // bounty closed
w.AddToTriple(uris.GeneralUris.TypeProperty, uris.GeneralUris.EndOfBountyType);
if (r.MoveToAttribute("PostId"))
{
w.AddToTriple(uris.GeneralUris.AnswerProperty, uris.CreatePostUri(r.Value));
}
if (r.MoveToAttribute("CreationDate"))
{
w.AddToTriple(uris.GeneralUris.DateProperty, DateTime.Parse(r.Value, System.Globalization.CultureInfo.InvariantCulture));
}
if (r.MoveToAttribute("BountyAmount"))
{
w.AddToTriple(uris.GeneralUris.TransferredAmountProperty, long.Parse(r.Value));
}
break;
case "10": // deletion
break;
case "11": // undeletion
break;
case "12": // spam
break;
case "13": // moderator informed
break;
case "15": // under moderator review
break;
case "16": // approved edit suggestion
break;
default:
unknownVoteTypeIds.RegisterUnknownValue(r.Value);
break;
}
}
else
{
r.MoveToElement();
ConsoleHelper.WriteErrorLine("No VoteTypeId attribute found on element {0}. Skipping element.", r.ReadOuterXml());
}
}
private static void ConvertPost(SiteUris uris, XmlReader r, SequentialTurtleWriter w, UnknownValueStore <string> unknownPostTypeIds)
{
Uri subjectUri;
if (r.MoveToAttribute("Id"))
{
subjectUri = uris.CreatePostUri(r.Value);
w.StartTriple(subjectUri);
}
else
{
r.MoveToElement();
ConsoleHelper.WriteErrorLine("No Id attribute found on element {0}. Skipping element.", r.ReadOuterXml());
return;
}
if (r.MoveToAttribute("PostTypeId"))
{
switch (r.Value)
{
case "1": // question
w.AddToTriple(uris.GeneralUris.TypeProperty, uris.GeneralUris.QuestionType);
uris.LinkToSite(w);
if (r.MoveToAttribute("AcceptedAnswerId"))
{
w.AddToTriple(uris.GeneralUris.AcceptedAnswerProperty, uris.CreatePostUri(r.Value));
}
if (r.MoveToAttribute("ViewCount"))
{
w.AddToTriple(uris.GeneralUris.ViewCountProperty, long.Parse(r.Value));
}
if (r.MoveToAttribute("Title"))
{
w.AddToTriple(uris.GeneralUris.TitleProperty, r.Value);
w.AddToTriple(uris.GeneralUris.LabelProperty, r.Value);
}
if (r.MoveToAttribute("Score"))
{
w.AddToTriple(uris.GeneralUris.ScoreProperty, long.Parse(r.Value));
}
break;
case "2": // answer
w.AddToTriple(uris.GeneralUris.TypeProperty, uris.GeneralUris.AnswerType);
uris.LinkToSite(w);
if (r.MoveToAttribute("ParentId"))
{
w.StartTriple(uris.CreatePostUri(r.Value));
w.AddToTriple(uris.GeneralUris.AnswerProperty, subjectUri);
w.StartTriple(subjectUri);
}
else
{
ConsoleHelper.WriteWarningLine("Orphaned answer: {0}", subjectUri);
}
if (r.MoveToAttribute("Score"))
{
w.AddToTriple(uris.GeneralUris.ScoreProperty, long.Parse(r.Value));
}
break;
case "3": // orphaned tag wiki
break;
case "4": // tag info excerpt
w.AddToTriple(uris.GeneralUris.TypeProperty, uris.GeneralUris.TagExcerptType);
break;
case "5": // tag description
w.AddToTriple(uris.GeneralUris.TypeProperty, uris.GeneralUris.TagDescriptionType);
break;
case "6": // moderator nomination
break;
case "7": // "Wiki placeholder" (seems to only be the election description)
//w.AddToTriple(uris.GeneralUris.TypeProperty, uris.GeneralUris.SiteInfoType);
break;
default:
unknownPostTypeIds.RegisterUnknownValue(r.Value);
break;
}
if (r.MoveToAttribute("CreationDate"))
{
w.AddToTriple(uris.GeneralUris.DateProperty, DateTime.Parse(r.Value, System.Globalization.CultureInfo.InvariantCulture));
}
if (r.MoveToAttribute("LastEditDate"))
{
w.AddToTriple(uris.GeneralUris.LastEditDateProperty, DateTime.Parse(r.Value, System.Globalization.CultureInfo.InvariantCulture));
}
if (r.MoveToAttribute("LastActivity"))
{
w.AddToTriple(uris.GeneralUris.LastActivityDateProperty, DateTime.Parse(r.Value, System.Globalization.CultureInfo.InvariantCulture));
}
if (r.MoveToAttribute("OwnerUserId"))
{
w.AddToTriple(uris.GeneralUris.OwnerProperty, uris.CreateUserUri(r.Value));
}
// TODO: LastEditorUserId (given in post history)
// TODO: FavoriteCount (linked to users?)
if (r.MoveToAttribute("Body"))
{
w.AddToTriple(uris.GeneralUris.DescriptionProperty, r.Value);
}
if (r.MoveToAttribute("Tags"))
{
w.AddToTriple(uris.GeneralUris.TagProperty,
tagRegex.Matches(r.Value).Cast <Match>().Select(m => uris.CreateTagUri(m.Groups[1].Value)));
}
}
else
{
r.MoveToElement();
ConsoleHelper.WriteErrorLine("No PostTypeId attribute found on element {0}. Skipping element.", r.ReadOuterXml());
}
}