private static void ConvertPostLink(SiteUris uris, XmlReader r, SequentialTurtleWriter w, UnknownValueStore <string> unknownLinkTypeIds)
{
if (r.MoveToAttribute("LinkTypeId"))
{
switch (r.Value)
{
case "1": // linked
if (r.MoveToAttribute("PostId"))
{
w.StartTriple(uris.CreatePostUri(r.Value));
if (r.MoveToAttribute("RelatedPostId"))
{
w.AddToTriple(uris.GeneralUris.LinkProperty, uris.CreatePostUri(r.Value));
}
}
break;
case "3": // duplicate
if (r.MoveToAttribute("RelatedPostId"))
{
w.StartTriple(uris.CreatePostUri(r.Value));
if (r.MoveToAttribute("PostId"))
{
w.AddToTriple(uris.GeneralUris.DuplicateProperty, uris.CreatePostUri(r.Value));
}
}
break;
default:
unknownLinkTypeIds.RegisterUnknownValue(r.Value);
break;
}
}
else
{
r.MoveToElement();
ConsoleHelper.WriteErrorLine("No LinkTypeId attribute found on element {0}. Skipping element.", r.ReadOuterXml());
}
}
private static void ConvertSiteList(GeneralUris generalUris, string tempDir, Uri baseUri, string destDir, VDS.RDF.INamespaceMapper nsMapper)
{
string srcFile = Path.Combine(tempDir, "Sites.xml");
ConsoleHelper.WriteMilestone("Downloading site list ...");
using (var client = new WebClient()) {
client.DownloadFile(new Uri(baseUri, "Sites.xml"), srcFile);
}
Console.WriteLine(" done.");
using (var destWriter = new SequentialTurtleWriter(File.CreateText(Path.Combine(destDir, "_sites.ttl")), nsMapper)) {
using (var fs = File.OpenRead(srcFile)) {
using (var reader = XmlReader.Create(fs)) {
while (reader.NodeType != XmlNodeType.Element)
{
if (!reader.Read())
{
ConsoleHelper.WriteErrorLine("No contents found in file {0}.", srcFile);
return;
}
}
if (reader.LocalName == "sitelist")
{
ConvertSites(generalUris, reader, destWriter);
}
else
{
ConsoleHelper.WriteWarningLine("Unknown root element \"{0}\". Skipping document.", reader.LocalName);
}
}
}
GlobalData.UpdateStats(destWriter);
}
Console.WriteLine("Conversion of site list completed.");
}
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 ConvertPostHistoryItem(SiteUris uris, XmlReader r, SequentialTurtleWriter w, UnknownValueStore <string> unknownPostHistoryTypeIds)
{
Uri subjectUri;
if (r.MoveToAttribute("Id"))
{
subjectUri = uris.CreatePostHistoryUri(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("PostHistoryTypeId"))
{
switch (r.Value)
{
case "1": // initial title
break;
case "2": // initial body
break;
case "3": // initial tags
break;
case "4": // edit title
break;
case "5": // edit body
break;
case "6": // edit tags
break;
case "7": // rollback title
break;
case "8": // rollback body
break;
case "9": // rollback tags
break;
case "10": // post closed
w.AddToTriple(uris.GeneralUris.TypeProperty, uris.GeneralUris.PostClosureType);
if (r.MoveToAttribute("Comment"))
{
switch (r.Value)
{
case "1": // Exact Duplicate
case "101": // duplicate
w.AddToTriple(uris.GeneralUris.CloseReasonProperty, uris.GeneralUris.DuplicateCloseReason);
break;
case "2": // Off-topic
case "102": // Off-topic
w.AddToTriple(uris.GeneralUris.CloseReasonProperty, uris.GeneralUris.OffTopicCloseReason);
break;
case "3": // Subjective and argumentative
case "105": // Primarily opinion-based
w.AddToTriple(uris.GeneralUris.CloseReasonProperty, uris.GeneralUris.SubjectiveCloseReason);
break;
case "4": // Not a real question
w.AddToTriple(uris.GeneralUris.CloseReasonProperty, uris.GeneralUris.NotAQuestionCloseReason);
break;
case "7": // Too localized
w.AddToTriple(uris.GeneralUris.CloseReasonProperty, uris.GeneralUris.TooLocalizedCloseReason);
break;
case "10": // General reference
w.AddToTriple(uris.GeneralUris.CloseReasonProperty, uris.GeneralUris.GeneralReferenceCloseReason);
break;
case "20": // Noise or pointless
w.AddToTriple(uris.GeneralUris.CloseReasonProperty, uris.GeneralUris.NoiseCloseReason);
break;
case "103": // Unclear what you're asking
w.AddToTriple(uris.GeneralUris.CloseReasonProperty, uris.GeneralUris.UnclearCloseReason);
break;
case "104": // Too broad
w.AddToTriple(uris.GeneralUris.CloseReasonProperty, uris.GeneralUris.TooBroadCloseReason);
break;
default:
ConsoleHelper.WriteWarningLine("Unknown post close reason: {0}", r.Value);
break;
}
}
if (r.MoveToAttribute("CreationDate"))
{
w.AddToTriple(uris.GeneralUris.DateProperty, DateTime.Parse(r.Value, System.Globalization.CultureInfo.InvariantCulture));
}
AddParticipants(uris, r, w);
LinkToPost(uris, subjectUri, r, w);
break;
case "11": // post reopened
w.AddToTriple(uris.GeneralUris.TypeProperty, uris.GeneralUris.PostReopeningType);
if (r.MoveToAttribute("CreationDate"))
{
w.AddToTriple(uris.GeneralUris.DateProperty, DateTime.Parse(r.Value, System.Globalization.CultureInfo.InvariantCulture));
}
AddParticipants(uris, r, w);
LinkToPost(uris, subjectUri, r, w);
break;
case "12": // post deleted
w.AddToTriple(uris.GeneralUris.TypeProperty, uris.GeneralUris.PostDeletionType);
if (r.MoveToAttribute("CreationDate"))
{
w.AddToTriple(uris.GeneralUris.DateProperty, DateTime.Parse(r.Value, System.Globalization.CultureInfo.InvariantCulture));
}
AddParticipants(uris, r, w);
LinkToPost(uris, subjectUri, r, w);
break;
case "13": // post undeleted
w.AddToTriple(uris.GeneralUris.TypeProperty, uris.GeneralUris.PostUndeletionType);
if (r.MoveToAttribute("CreationDate"))
{
w.AddToTriple(uris.GeneralUris.DateProperty, DateTime.Parse(r.Value, System.Globalization.CultureInfo.InvariantCulture));
}
AddParticipants(uris, r, w);
LinkToPost(uris, subjectUri, r, w);
break;
case "14": // post locked
w.AddToTriple(uris.GeneralUris.TypeProperty, uris.GeneralUris.PostLockingType);
if (r.MoveToAttribute("CreationDate"))
{
w.AddToTriple(uris.GeneralUris.DateProperty, DateTime.Parse(r.Value, System.Globalization.CultureInfo.InvariantCulture));
}
AddParticipants(uris, r, w);
LinkToPost(uris, subjectUri, r, w);
break;
case "15": // post unlocked
w.AddToTriple(uris.GeneralUris.TypeProperty, uris.GeneralUris.PostUnlockingType);
if (r.MoveToAttribute("CreationDate"))
{
w.AddToTriple(uris.GeneralUris.DateProperty, DateTime.Parse(r.Value, System.Globalization.CultureInfo.InvariantCulture));
}
AddParticipants(uris, r, w);
LinkToPost(uris, subjectUri, r, w);
break;
case "16": // community owned
break;
case "17": // post migrated superseded with 35/36
break;
case "18": // question merged
break;
case "19": // question protected
w.AddToTriple(uris.GeneralUris.TypeProperty, uris.GeneralUris.PostProtectionType);
if (r.MoveToAttribute("CreationDate"))
{
w.AddToTriple(uris.GeneralUris.DateProperty, DateTime.Parse(r.Value, System.Globalization.CultureInfo.InvariantCulture));
}
AddParticipants(uris, r, w);
LinkToPost(uris, subjectUri, r, w);
break;
case "20": // question unprotected
w.AddToTriple(uris.GeneralUris.TypeProperty, uris.GeneralUris.PostUnprotectionType);
if (r.MoveToAttribute("CreationDate"))
{
w.AddToTriple(uris.GeneralUris.DateProperty, DateTime.Parse(r.Value, System.Globalization.CultureInfo.InvariantCulture));
}
AddParticipants(uris, r, w);
LinkToPost(uris, subjectUri, r, w);
break;
case "21": // post disassociated
break;
case "22": // question unmerged
break;
case "24": // suggested edit applied
break;
case "25": // post tweeted
break;
case "31": // comment discussion moved to chat
break;
case "33": // post notice added
break;
case "34": // post notice removed
break;
case "35": // post migrated away replaces 17
break;
case "36": // post migrated here replaces 17
break;
case "37": // post merge source
break;
case "38": // post merge destination
break;
default:
unknownPostHistoryTypeIds.RegisterUnknownValue(r.Value);
break;
}
}
else
{
r.MoveToElement();
ConsoleHelper.WriteErrorLine("No PostHistoryTypeId attribute found on element {0}. Skipping element.", r.ReadOuterXml());
}
}
public static void Convert(GeneralUris generalUris, string srcFile, string destDir, SiteInfo website)
{
string fileNameOnly = Path.GetFileName(srcFile);
Console.WriteLine("Processing {0} ...", fileNameOnly);
var siteUris = new SiteUris(generalUris, website);
var nsMapper = siteUris.CreateNamespaceMapper();
using (var destWriter = new SequentialTurtleWriter(File.CreateText(Path.Combine(destDir, website + "-" + Path.GetFileNameWithoutExtension(srcFile) + ".ttl")), nsMapper)) {
using (var fs = File.OpenRead(srcFile)) {
using (var reader = XmlReader.Create(fs)) {
while (reader.NodeType != XmlNodeType.Element)
{
if (!reader.Read())
{
ConsoleHelper.WriteErrorLine("No contents found in file {0}.", srcFile);
return;
}
}
switch (reader.LocalName)
{
case "badges":
ConsoleHelper.WriteInfoLine("List of badges identified.");
ConvertBadges(siteUris, reader, destWriter);
break;
case "comments":
ConsoleHelper.WriteInfoLine("List of comments identified.");
ConvertComments(siteUris, reader, destWriter);
break;
case "posthistory":
ConsoleHelper.WriteInfoLine("List of posthistory identified.");
ConvertPostHistory(siteUris, reader, destWriter);
break;
case "postlinks":
ConsoleHelper.WriteInfoLine("List of postlinks identified.");
ConvertPostLinks(siteUris, reader, destWriter);
break;
case "posts":
ConsoleHelper.WriteInfoLine("List of posts identified.");
ConvertPosts(siteUris, reader, destWriter);
break;
case "tags":
ConsoleHelper.WriteInfoLine("List of tags identified.");
ConvertTags(siteUris, reader, destWriter);
break;
case "users":
ConsoleHelper.WriteInfoLine("List of users identified.");
ConvertUsers(siteUris, reader, destWriter);
break;
case "votes":
ConsoleHelper.WriteInfoLine("List of votes identified.");
ConvertVotes(siteUris, reader, destWriter);
break;
default:
ConsoleHelper.WriteWarningLine("Unknown root element \"{0}\". Skipping document.", reader.LocalName);
break;
}
}
}
GlobalData.UpdateStats(destWriter);
}
Console.WriteLine("Conversion of {0} completed.", fileNameOnly);
}
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());
}
}
/// <summary>
/// Downloads and converts the data.
/// </summary>
/// <param name="generalUris">An object that provides general URIs used in the exported dataset.</param>
/// <param name="filelist">The URL of a filelist Xml file.</param>
/// <exception cref="ArgumentNullException">Any of the arguments is <see langword="null"/>.</exception>
private static void RetrieveData(GeneralUris generalUris, Uri filelist)
{
if (generalUris == null)
{
throw new ArgumentNullException("generalUris");
}
if (filelist == null)
{
throw new ArgumentNullException("filelist");
}
string tempDir = Path.Combine(BaseDir, "tmp");
string destDir = Path.Combine(BaseDir, "rdf");
Directory.CreateDirectory(tempDir);
Directory.CreateDirectory(destDir);
DateTime startTime = DateTime.Now;
ConsoleHelper.WriteInfoLine("Current time: {0:yyyy-MM-dd HH:mm:ss}", startTime);
if (!GlobalData.Options.OntologyOnly)
{
ConsoleHelper.WriteMilestone("Downloading files list ...");
using (var client = new WebClient()) {
client.DownloadFile(filelist, Path.Combine(tempDir, "files.xml"));
}
Console.WriteLine(" done.");
var files = LoadFilesList(Path.Combine(tempDir, "files.xml")).OrderBy(f => f).ToArray();
ConsoleHelper.WriteInfoLine("{0} file(s) in list, totalling to a compressed size of {1:F1} GB.",
files.Length, (double)files.Sum(f => f.Size) / 1024 / 1024 / 1024);
int processedFilesCount = 0;
foreach (var f in files)
{
SiteInfo siteInfo;
try {
siteInfo = f.RetrieveSiteInfo();
}
catch (ArgumentException ex) {
ConsoleHelper.WriteErrorLine("Skipping file {0}, as it cannot be associated with a website.\n{1}", f, ex);
continue;
}
if (IncludeSite(siteInfo))
{
GlobalData.Sites[siteInfo.Id] = siteInfo;
if (!GlobalData.Options.SiteListOnly)
{
string fn = f.Download(filelist, tempDir);
string[] rawFiles = null;
switch (Path.GetExtension(fn))
{
case ".7z":
rawFiles = ExtractSevenZipArchive(fn);
break;
default:
ConsoleHelper.WriteWarningLine("File {0} has an unknown file extension.", fn);
break;
}
if (rawFiles != null)
{
ConsoleHelper.WriteInfoLine("{0} file(s) extracted.", rawFiles.Length);
foreach (var rawFile in rawFiles)
{
Converter.Convert(generalUris, rawFile, destDir, siteInfo);
}
}
}
processedFilesCount++;
if (processedFilesCount >= GlobalData.Options.MaxFileCount)
{
break;
}
}
}
}
GlobalInformationConverter.Convert(generalUris, tempDir, filelist, destDir);
if (!GlobalData.Options.KeepTemporaryFiles)
{
Console.Write("Removing temporary files ... ");
try {
Directory.Delete(tempDir, true);
Console.WriteLine(" done.");
}
catch {
ConsoleHelper.WriteErrorLine("Please remove the directory {0} manually.", tempDir);
}
}
Console.WriteLine();
GlobalData.PrintStats();
DateTime endTime = DateTime.Now;
ConsoleHelper.WriteInfoLine("Current time: {0:yyyy-MM-dd HH:mm:ss}", endTime);
ConsoleHelper.WriteInfoLine("Total duration: {0}", endTime - startTime);
}
