Interfacing with RSS (v 0.9x, 2.x, 1.0) + Atom feeds.


Version on this page:
LTS Haskell 22.9:
Stackage Nightly 2024-07-21:
Latest on Hackage:

See all snapshots feed appears in

BSD-3-Clause licensed by Sigbjorn Finne
Maintained by Adam Bergmark
This version can be pinned in stack with:feed-,4036


feed Build Status


Interfacing with RSS (v 0.9x, 2.x, 1.0) + Atom feeds.

  • Parsers
  • Pretty Printers
  • Querying

To help working with the multiple feed formats we’ve ended up with this set of modules providing parsers, pretty printers and some utility code for querying and just generally working with a concrete representation of feeds in Haskell.

For basic reading and editing of feeds, consult the documentation of the Text.Feed.* hierarchy.


Building an Atom feed is similar to building an RSS feed, but we’ll arbitrarily pick Atom here:

We’d like to generate the XML for a minimal working example. Constructing our base Feed can use the smart constructor called nullFeed:

This is a pattern the library maintains for smart constructors. If you want the minimum viable ‘X’, use the ‘nullX’ constructor.

{-# LANGUAGE OverloadedStrings #-}

module Main where

import Prelude.Compat hiding (take)

import Data.Text
import Data.XML.Types as XML
import qualified Data.Text.Lazy as Lazy
import qualified Text.Atom.Feed as Atom
import qualified Text.Atom.Feed.Export as Export (textFeed)

myFeed :: Atom.Feed
myFeed = Atom.nullFeed
    "http://example.com/atom.xml"       -- ^ id
    (Atom.TextString "Example Website") -- ^ title
    "2017-08-01"                        -- ^ last updated

Now we can export the feed to Text.

renderFeed :: Atom.Feed -> Maybe Lazy.Text
renderFeed = Export.textFeed
> renderFeed myFeed
<?xml version="1.0" encoding="UTF-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
  <title type="text">Example Website</title>

The TextContent sum type allows us to specify which type of text we’re providing.

data TextContent
  = TextString Text
  | HTMLString Text
  | XHTMLString XML.Element
  deriving (Show)

A feed isn’t very useful without some content though, so we’ll need to build up an Entry.

data Post
  = Post
  { _postedOn :: Text
  , _url :: Text
  , _content :: Text

examplePosts :: [Post]
examplePosts =
  [ Post "2000-02-02T18:30:00Z" "http://example.com/2" "Bar."
  , Post "2000-01-01T18:30:00Z" "http://example.com/1" "Foo."

Our Post data type will need to be converted into an Entry in order to use it in the top level Feed. The required fields for an entry are an url “id” from which an entry’s presence can be validated, a title for the entry, and a posting date. In this example we’ll also add authors, link, and the entries actual content, since we have all of this available in the Post provided.

toEntry :: Post -> Atom.Entry
toEntry (Post date url content) =
     url -- The ID field. Must be a link to validate.
     (Atom.TextString (take 20 content)) -- Title
  { Atom.entryAuthors = [Atom.nullPerson {Atom.personName = "J. Smith"}]
  , Atom.entryLinks = [Atom.nullLink url]
  , Atom.entryContent = Just (Atom.HTMLContent content)

From the base feed we created earlier, we can add further details (Link and Entry content) as well as map our toEntry function over the posts we’d like to include in the feed.

feed :: Atom.Feed
feed =
  myFeed { Atom.feedEntries = fmap toEntry examplePosts
         , Atom.feedLinks = [Atom.nullLink "http://example.com/"]
> renderFeed feed
<?xml version="1.0" encoding="UTF-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
  <title type="text">Example Website</title>
  <link href="http://example.com/"/>
    <title type="text">Bar.</title>
      <name>J. Smith</name>
    <content type="text">Bar.</content>
    <link href="http://example.com/2"/>
    <title type="text">Foo.</title>
      <name>J. Smith</name>
    <content type="text">Foo.</content>
    <link href="http://example.com/1"/>

See here for this content as an uninterrupted running example.

-- Dummy main needed to compile this file with markdown-unlit
main :: IO ()
main = return ()


  • Add a test to check that validation works on a simple entry.
  • Change attribute handling when validating so that type attribute is recognised properly on content.

  • Reverted changes to EntryContent that came in Thanks to Tomas Janousek.

  • Get rid of xmlns:ns and ns:-prefixed attributes that confused some applications, thanks to Tomas Janousek.
  • GHC 8.8 support

Updated EntryContent’s HTMLContent to wrap an XML.Element instead of Text. Thanks to Jake Keuhlen.

  • parseFeedFromFile now returns IO (Maybe Feed) instead of IO Feed to distinguish IO exceptions from parse failures. Thanks to Jake Keuhlen.

  • Support for GHC 8.6.x libraries

  • Add textFeed and textRSS helpers (thanks to Francesco Ariis)

  • Thanks to Dmitry Dzhus feed has been modernized to use the text, xml-types and xml-conduit libraries.

  • Adds support for some fallback parsing of atom feeds to XMLFeed (thanks to Joey Hess)

  • Add missing file to dist of test-suite (thanks to Sergei Trofimovich)

  • Add toFeedDateStringUTC which uses UTCTime rather than ClockTime (thanks to Emanuel Borsboom)
  • Now with explicit export lists!

  • Fix toFeedDateString time format, It used %s (seconds since epoch) instead of %S (seconds of minute), and %m (month) instead of %M (minute) (thanks to Emanuel Borsboom)

  • RSS Export: avoid attribute (thanks to Roman Cheplyaka)

  • The Atom RFC says that when a link element doesn’t specify the “rel” attribute, i.e. link relation, it should be interpreted as an “alternate” relation. This makes the feed and item query functions treat a missing relation as “alternate”.

  • Allow HUnit 1.3.*

  • Add Text.Feed.Import.parseFeedSource :: XmlSource s => s -> Maybe Feed (thanks to Dmitry Dzhus)

  • Add missing modules in test-suite

  • Fix typo “skipDayss” -> “skipDays” in Text.RSS.Export (thanks to Daniele Francesconi)

  • Update maintainer information
  • Add test suite

  • Add support for utf8-string >= 1 && < 1.1 and time 1.5.*

  • add ref to github repo + .cabal tidying.


  • tidy up compilation with ghc-7.6(.3), bumped version (but no functional changes.)


  • cabal build fixes.


  • parsing: made be optional.
  • parsing: try if is missing.