/// <summary>
/// Gets the first 5 videos related to the specified video.
/// </summary>
public async Task<RelatedVideos> GetRelatedVideos(RelatedVideosQuery query)
{
// Set the base URL of the REST client to use the first node in the Cassandra cluster
string nodeIp = _session.Cluster.AllHosts().First().Address.Address.ToString();
_restClient.BaseUrl = new Uri(string.Format("http://{0}:8983/solr", nodeIp));
//WebRequest mltRequest = WebRequest.Create("http://127.0.2.15:8983/solr/killrvideo.videos/mlt?q=videoid%3Asome-uuid&wt=json&indent=true&qt=mlt&mlt.fl=name&mlt.mindf=1&mlt.mintf=1");
var request = new RestRequest("killrvideo.videos/mlt");
request.AddParameter("q", string.Format("videoid:\"{0}\"", query.VideoId));
request.AddParameter("wt", "json");
// Paging information
int start;
if (query.PagingState == null || int.TryParse(query.PagingState, out start) == false)
start = 0;
request.AddParameter("start", start);
request.AddParameter("rows", query.PageSize);
//MLT Fields to Consider
request.AddParameter("mlt.fl", "name,description,tags");
//MLT Minimum Document Frequency - the frequency at which words will be ignored which do not occur in at least this many docs.
request.AddParameter("mlt.mindf", 2);
//MLT Minimum Term Frequency - the frequency below which terms will be ignored in the source doc.
request.AddParameter("mlt.mintf", 2);
IRestResponse<MLTQueryResult> response = await _restClient.ExecuteTaskAsync<MLTQueryResult>(request).ConfigureAwait(false);
// Check for network/timeout errors
if (response.ResponseStatus != ResponseStatus.Completed)
{
Logger.Error(response.ErrorException, "Error while querying Solr video suggestions from {host} for {query}", nodeIp, query);
return new RelatedVideos { VideoId = query.VideoId, Videos = Enumerable.Empty<VideoPreview>(), PagingState = null };
}
// Check for HTTP error codes
if (response.StatusCode != HttpStatusCode.OK)
{
Logger.Error("HTTP status code {code} while querying Solr video suggestions from {host} for {query}", (int) response.StatusCode,
nodeIp, query);
return new RelatedVideos { VideoId = query.VideoId, Videos = Enumerable.Empty<VideoPreview>(), PagingState = null };
}
// Success
int nextPageStartIndex = response.Data.Response.Start + response.Data.Response.Docs.Count;
string pagingState = nextPageStartIndex == response.Data.Response.NumFound ? null : nextPageStartIndex.ToString();
return new RelatedVideos { VideoId = query.VideoId, Videos = response.Data.Response.Docs, PagingState = pagingState };
}
/// <summary>
/// Gets the first 4 videos related to the specified video. Does not support paging.
/// </summary>
public async Task<RelatedVideos> GetRelatedVideos(RelatedVideosQuery queryParams)
{
// Lookup the tags for the video
PreparedStatement tagsForVideoPrepared = await _statementCache.NoContext.GetOrAddAsync("SELECT tags FROM videos WHERE videoid = ?");
BoundStatement tagsForVideoBound = tagsForVideoPrepared.Bind(queryParams.VideoId);
RowSet tagRows = await _session.ExecuteAsync(tagsForVideoBound).ConfigureAwait(false);
Row tagRow = tagRows.SingleOrDefault();
if (tagRow == null)
return new RelatedVideos { VideoId = queryParams.VideoId, Videos = Enumerable.Empty<VideoPreview>(), PagingState = null };
var tagsValue = tagRow.GetValue<IEnumerable<string>>("tags");
var tags = tagsValue == null ? new List<string>() : tagsValue.ToList();
// If there are no tags, we can't find related videos
if (tags.Count == 0)
return new RelatedVideos { VideoId = queryParams.VideoId, Videos = Enumerable.Empty<VideoPreview>(), PagingState = null };
var relatedVideos = new Dictionary<Guid, VideoPreview>();
PreparedStatement videosForTagPrepared = await _statementCache.NoContext.GetOrAddAsync("SELECT * FROM videos_by_tag WHERE tag = ? LIMIT ?");
var inFlightQueries = new List<Task<RowSet>>();
for (var i = 0; i < tags.Count; i++)
{
// Use the number of results we ultimately want * 2 when querying so that we can account for potentially having to filter
// out the video Id we're using as the basis for the query as well as duplicates
const int pageSize = RelatedVideosToReturn * 2;
// Kick off a query for each tag and track them in the inflight requests list
string tag = tags[i];
IStatement query = videosForTagPrepared.Bind(tag, pageSize);
inFlightQueries.Add(_session.ExecuteAsync(query));
// Every third query, or if this is the last tag, wait on all the query results
if (inFlightQueries.Count == 3 || i == tags.Count - 1)
{
RowSet[] results = await Task.WhenAll(inFlightQueries).ConfigureAwait(false);
foreach (RowSet rowSet in results)
{
foreach (Row row in rowSet)
{
VideoPreview preview = MapRowToVideoPreview(row);
// Skip self
if (preview.VideoId == queryParams.VideoId)
continue;
// Skip videos we already have in the results
if (relatedVideos.ContainsKey(preview.VideoId))
continue;
// Add to results
relatedVideos.Add(preview.VideoId, preview);
// If we've got enough, no reason to continue
if (relatedVideos.Count >= RelatedVideosToReturn)
break;
}
// If we've got enough, no reason to continue
if (relatedVideos.Count >= RelatedVideosToReturn)
break;
}
// See if we've got enough results now to fulfill our requirement
if (relatedVideos.Count >= RelatedVideosToReturn)
break;
// We don't have enough yet, so reset the inflight requests to allow another batch of tags to be queried
inFlightQueries.Clear();
}
}
return new RelatedVideos
{
VideoId = queryParams.VideoId,
Videos = relatedVideos.Values,
PagingState = null
};
}
