Beispiel #1
0
        async Task <River> WriteRiver(Uri uri, River river)
        {
            river = await MaybeArchiveRiver(uri.AbsoluteUri, river);

            await feedStore.WriteRiver(uri, river);

            return(river);
        }
Beispiel #2
0
        public async Task <River> FetchRiver(Uri uri)
        {
            River river = await feedStore.LoadRiverForFeed(uri);

            while (river.Metadata.LastStatus == HttpStatusCode.MovedPermanently)
            {
                river = await feedStore.LoadRiverForFeed(river.Metadata.OriginUrl);
            }
            return(river);
        }
Beispiel #3
0
        public async Task <River> UpdateAsync(River river)
        {
            FetchResult fetchResult = await FetchAsync(
                river.Metadata.OriginUrl,
                river.Metadata.Etag,
                river.Metadata.LastModified
                );

            var updatedFeeds = river.UpdatedFeeds;

            if (fetchResult.Feed != null)
            {
                var    feed     = fetchResult.Feed;
                Item[] newItems = await this.feedItemStore.StoreItems(
                    river.Metadata.OriginUrl, feed.Items.ToArray());

                // TODO: Filter this out once we've loaded a bit.
                var existingItems = new HashSet <string>(
                    from existingFeed in river.UpdatedFeeds.Feeds
                    from item in existingFeed.Items
                    where item.Id != null
                    select item.Id
                    );
                newItems = newItems.Where(item => !existingItems.Contains(item.Id)).ToArray();
                if (newItems.Length > 0)
                {
                    Uri baseUri = SyndicationUtil.TryParseAbsoluteUrl(feed.WebsiteUrl) ?? feed.FeedUrl;
                    for (int i = 0; i < newItems.Length; i++)
                    {
                        newItems[i] = Rebase(newItems[i], baseUri);
                    }

                    newItems = await this.thumbnailExtractor.LoadItemThumbnailsAsync(baseUri, newItems);

                    await this.feedItemStore.UpdateItemThumbs(river.Metadata.OriginUrl, newItems);

                    feed         = feed.With(items: newItems);
                    updatedFeeds = river.UpdatedFeeds.With(feeds: river.UpdatedFeeds.Feeds.Insert(0, feed));
                }
            }

            var metadata = river.Metadata.With(
                etag: fetchResult.Etag,
                lastModified: fetchResult.LastModified,
                originUrl: fetchResult.FeedUrl,
                lastStatus: fetchResult.Status);

            return(river.With(updatedFeeds: updatedFeeds, metadata: metadata));
        }
Beispiel #4
0
        public async Task <River> RefreshAggregateRiverWithFeeds(string id, IList <Uri> feedUrls)
        {
            Stopwatch aggregateTimer = Stopwatch.StartNew();

            River river = await this.aggregateStore.LoadAggregate(id);

            Log.AggregateRefreshStart(id, feedUrls.Count);
            DateTimeOffset lastUpdated = river.UpdatedFeeds.Feeds.Count > 0
                ? river.UpdatedFeeds.Feeds.Max(f => f.WhenLastUpdate)
                : DateTimeOffset.MinValue;

            var parser = new RiverLoader();

            River[] rivers = await Task.WhenAll(from url in feedUrls select parser.FetchRiver(url));

            Log.AggregateRefreshPulledRivers(id, rivers.Length);

            var newFeeds = new List <FeedSegment>();

            foreach (River feedRiver in rivers)
            {
                FeedSegment[] newUpdates =
                    feedRiver.UpdatedFeeds.Feeds.Where(rf => rf.WhenLastUpdate > lastUpdated).ToArray();
                Log.AggregateNewUpdates(id, feedRiver.Metadata.OriginUrl, newUpdates.Length);

                if (newUpdates.Length > 0)
                {
                    Item[]         newItems      = newUpdates.SelectMany(rf => rf.Items).ToArray();
                    DateTimeOffset biggestUpdate = newUpdates.Max(rf => rf.WhenLastUpdate);
                    Log.AggregateFeedState(id, feedRiver.Metadata.OriginUrl, newUpdates.Length, biggestUpdate);
                    newFeeds.Add(newUpdates[0].With(whenLastUpdate: biggestUpdate, items: newItems));
                }
            }

            // Sort all the new feeds by time they were updated (latest time first).
            Log.AggregateHasNewFeeds(id, newFeeds.Count);
            newFeeds = newFeeds.OrderByDescending(f => f.WhenLastUpdate).ToList();
            River newRiver = river.With(
                updatedFeeds: river.UpdatedFeeds.With(feeds: newFeeds.Concat(river.UpdatedFeeds.Feeds)));

            newRiver = await MaybeArchiveRiver(id, newRiver);

            Log.UpdatingAggregate(id);
            await this.aggregateStore.WriteAggregate(id, newRiver);

            Log.AggregateRefreshed(id, aggregateTimer);
            return(newRiver);
        }
Beispiel #5
0
        async Task <River> MaybeArchiveRiver(string id, River river)
        {
            if (river.UpdatedFeeds.Feeds.Count > UpdateLimit)
            {
                Log.SplittingFeed(id, river);
                ImmutableList <FeedSegment> feeds    = river.UpdatedFeeds.Feeds;
                IEnumerable <FeedSegment>   oldFeeds = feeds.Skip(UpdateSize);
                IEnumerable <FeedSegment>   newFeeds = feeds.Take(UpdateSize);

                River  oldRiver   = river.With(updatedFeeds: river.UpdatedFeeds.With(feeds: oldFeeds));
                string archiveKey = await this.archiveStore.WriteRiverArchive(oldRiver);

                Log.WroteArchive(id, river, archiveKey);

                river = river.With(
                    updatedFeeds: river.UpdatedFeeds.With(feeds: newFeeds),
                    metadata: river.Metadata.With(next: archiveKey)
                    );
            }

            return(river);
        }
Beispiel #6
0
        public async Task <River> FetchAndUpdateRiver(Uri uri)
        {
            River river = await feedStore.LoadRiverForFeed(uri);

            if ((river.Metadata.LastStatus != HttpStatusCode.MovedPermanently) &&
                (river.Metadata.LastStatus != HttpStatusCode.Gone))
            {
                river = await UpdateAsync(river);

                if (river.Metadata.LastStatus == HttpStatusCode.MovedPermanently)
                {
                    // The first time this happens we need to move all the items across.
                    await feedItemStore.UpdateOriginUrl(uri, river.Metadata.OriginUrl);
                }
                await WriteRiver(uri, river);
            }

            if (river.Metadata.LastStatus == HttpStatusCode.MovedPermanently)
            {
                return(await FetchAndUpdateRiver(river.Metadata.OriginUrl));
            }

            return(river);
        }