static Item HandleAtomLink(Item item, XElement link)
{
string rel = link.Attribute(XNames.Atom.Rel)?.Value ?? "alternate";
string type = link.Attribute(XNames.Atom.Type)?.Value ?? "text/html";
string href = link.Attribute(XNames.Atom.Href)?.Value;
if (String.Equals(rel, "alternate", StringComparison.OrdinalIgnoreCase) &&
type.StartsWith("text/html", StringComparison.OrdinalIgnoreCase))
{
item = item.With(link: SyndicationUtil.ParseLink(href, link));
}
if (String.Equals(rel, "self", StringComparison.OrdinalIgnoreCase) &&
type.StartsWith("text/html", StringComparison.OrdinalIgnoreCase))
{
item = item.With(permaLink: SyndicationUtil.ParseLink(href, link));
}
if (link.Attribute(XNames.Atom.Rel)?.Value == "enclosure")
{
item = item.With(enclosures: item.Enclosures.Add(new Enclosure(
length: link.Attribute(XNames.Atom.Length)?.Value,
type: type,
url: SyndicationUtil.ParseLink(href, link)
)));
}
return(item);
}
static Item HandleThumbnail(Item item, XElement element)
{
if (
element.Name == XNames.Media.Content &&
element.Attribute(XNames.Media.Medium)?.Value == "image"
)
{
Uri url =
SyndicationUtil.TryParseUrl(
element.Attribute(XNames.Media.Url)?.Value,
null,
element
);
int width, height;
if (
url != null &&
Int32.TryParse(
element.Attribute(XNames.Media.Width)?.Value,
out width
) &&
Int32.TryParse(
element.Attribute(XNames.Media.Height)?.Value,
out height
)
)
{
item =
item.With(thumbnail: new Thumbnail(url, width, height));
}
}
return(item);
}
static Item LoadItem(XElement item)
{
var ri = new Item();
foreach (XElement xe in item.Elements())
{
Func <Item, XElement, Item> func;
if (ItemElements.TryGetValue(xe.Name, out func))
{
ri = func(ri, xe);
}
}
// Load the body; prefer explicit summaries to "description", which is ambiguous, to "content", which is
// explicitly intended to be the full entry content.
if (ri.Summary != null)
{
ri = ri.With(body: SyndicationUtil.ParseBody(ri.Summary));
}
else if (ri.Description != null)
{
ri = ri.With(body: SyndicationUtil.ParseBody(ri.Description));
}
else if (ri.Content != null)
{
ri = ri.With(body: SyndicationUtil.ParseBody(ri.Content));
}
if (ri.PermaLink == null)
{
ri = ri.With(permaLink: ri.Link);
}
if (ri.Id == null)
{
ri = ri.With(id: CreateItemId(ri));
}
if (String.IsNullOrWhiteSpace(ri.Title))
{
string title = null;
if (ri.PubDate != null)
{
title = ri.PubDate.ToString();
}
else if (ri.PermaLink != null)
{
title = ri.PermaLink.AbsoluteUri;
}
else if (ri.Id != null)
{
title = ri.Id;
}
if (title != null)
{
ri = ri.With(title: title);
}
}
return(ri);
}
static Item Rebase(Item item, Uri baseUri)
{
return(item.With(
link: SyndicationUtil.Rebase(item.Link, baseUri),
permaLink: SyndicationUtil.Rebase(item.PermaLink, baseUri),
enclosures: item.Enclosures.Select(e => e.With(url: SyndicationUtil.Rebase(e.Url, baseUri)))
));
}
static Item HandleEnclosure(Item item, XElement element)
{
return(item.With(enclosures: item.Enclosures.Add(new Enclosure(
length: element.Attribute(XNames.RSS.Length)?.Value,
type: element.Attribute(XNames.RSS.Type)?.Value,
url: SyndicationUtil.ParseLink(element.Attribute(XNames.RSS.Url)?.Value, element)
))));
}
static Item HandlePubDate(Item item, XElement element)
{
DateTime?date = SyndicationUtil.ParseDate(element);
if (date != null && (item.PubDate == null || date > item.PubDate))
{
return(item.With(pubDate: date));
}
return(item);
}
static Item HandleGuid(Item item, XElement element)
{
item = item.With(id: element.Value);
if (item.Id.StartsWith("http://", StringComparison.OrdinalIgnoreCase) ||
item.Id.StartsWith("https://", StringComparison.OrdinalIgnoreCase))
{
item = item.With(permaLink: SyndicationUtil.ParseLink(item.Id, element));
}
return(item);
}
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));
}
public static async Task <IList <Uri> > GetFeedUrls(
string originUrl,
bool findAll = false)
{
var allUrls = new List <Uri>();
Uri baseUri = FixupUrl(originUrl);
// Maybe... maybe this one is a feed?
Log.FindFeedCheckingBase(baseUri);
string data = await GetFeedData(baseUri);
if (LooksLikeFeed(data))
{
Log.FindFeedBaseWasFeed(baseUri);
return(new[] { baseUri });
}
// Nope, let's dive into the soup!
var parser = new HtmlParser();
IHtmlDocument document = parser.ParseDocument(data);
// Link elements.
Log.FindFeedCheckingLinkElements(baseUri);
List <Uri> linkUrls = new List <Uri>();
foreach (IElement element in document.GetElementsByTagName("link"))
{
string linkType = element.GetAttribute("type");
if (linkType != null && FeedMimeTypes.Contains(linkType))
{
Uri hrefUrl =
SyndicationUtil.TryParseAbsoluteUrl(
element.GetAttribute("href"),
baseUri
);
if (hrefUrl != null)
{
linkUrls.Add(hrefUrl);
}
}
}
await FilterUrlsByFeed(linkUrls);
if (linkUrls.Count > 0)
{
Log.FindFeedFoundLinkElements(baseUri, linkUrls);
linkUrls.Sort(UrlFeedComparison);
allUrls.AddRange(linkUrls);
if (!findAll)
{
return(allUrls);
}
}
// <a> tags
Log.FindFeedCheckingAnchorElements(baseUri);
List <Uri> localGuesses = new List <Uri>();
List <Uri> remoteGuesses = new List <Uri>();
foreach (IElement element in document.GetElementsByTagName("a"))
{
Uri hrefUrl =
SyndicationUtil.TryParseAbsoluteUrl(
element.GetAttribute("href"),
baseUri
);
if (hrefUrl != null)
{
if ((hrefUrl.Host == baseUri.Host) && IsFeedUrl(hrefUrl))
{
localGuesses.Add(hrefUrl);
}
else if (IsFeedishUrl(hrefUrl))
{
remoteGuesses.Add(hrefUrl);
}
}
}
Log.FindFeedFoundSomeAnchors(baseUri, localGuesses, remoteGuesses);
// (Consider ones on the same domain first.)
await FilterUrlsByFeed(localGuesses);
if (localGuesses.Count > 0)
{
Log.FindFeedsFoundLocalGuesses(baseUri, localGuesses);
localGuesses.Sort(UrlFeedComparison);
allUrls.AddRange(localGuesses);
if (!findAll)
{
return(localGuesses);
}
}
await FilterUrlsByFeed(remoteGuesses);
if (remoteGuesses.Count > 0)
{
Log.FindFeedsFoundRemoteGuesses(baseUri, remoteGuesses);
remoteGuesses.Sort(UrlFeedComparison);
allUrls.AddRange(remoteGuesses);
if (!findAll)
{
return(remoteGuesses);
}
}
List <Uri> randomGuesses =
FeedNames.Select(s => new Uri(baseUri, s)).ToList();
await FilterUrlsByFeed(randomGuesses);
if (randomGuesses.Count > 0)
{
Log.FindFeedsFoundRandomGuesses(baseUri, randomGuesses);
randomGuesses.Sort(UrlFeedComparison);
allUrls.AddRange(randomGuesses);
if (!findAll)
{
return(randomGuesses);
}
}
// All done, nothing. (Or... everything!)
Log.FindFeedFoundTotal(baseUri, allUrls);
return(allUrls);
}
