Parsing and extracting information from (possibly malformed) HTML/XML documents


LTS Haskell 22.29:0.14.8
Stackage Nightly 2024-07-13:0.14.8
Latest on Hackage:0.14.8

See all snapshots tagsoup appears in

BSD-3-Clause licensed and maintained by Neil Mitchell
This version can be pinned in stack with:tagsoup-0.14.8@sha256:56b2023d2e9fdbff093719ce9af5285d2436b234c6c684d6a69f14595a8348ae,2117

TagSoup Hackage version Stackage version Linux build status Windows build status

TagSoup is a library for parsing HTML/XML. It supports the HTML 5 specification, and can be used to parse either well-formed XML, or unstructured and malformed HTML from the web. The library also provides useful functions to extract information from an HTML document, making it ideal for screen-scraping.

The library provides a basic data type for a list of unstructured tags, a parser to convert HTML into this tag type, and useful functions and combinators for finding and extracting information. This document gives two particular examples of scraping information from the web, while a few more may be found in the Sample file from the source repository. The examples we give are:

  • Obtaining the last modified date of the Haskell wiki
  • Obtaining a list of Simon Peyton Jones’ latest papers
  • A brief overview of some other examples

The intial version of this library was written in Javascript and has been used for various commercial projects involving screen scraping. In the examples general hints on screen scraping are included, learnt from bitter experience. It should be noted that if you depend on data which someone else may change at any given time, you may be in for a shock!

This library was written without knowledge of the Java version of TagSoup. They have made a very different design decision: to ensure default attributes are present and to properly nest parsed tags. We do not do this - tags are merely a list devoid of nesting information.


Thanks to Mike Dodds for persuading me to write this up as a library. Thanks to many people for debugging and code contributions, including: Gleb Alexeev, Ketil Malde, Conrad Parker, Henning Thielemann, Dino Morelli, Emily Mitchell, Gwern Branwen.

Potential Bugs

There are two things that may go wrong with these examples:

  • The Websites being scraped may change. There is nothing I can do about this, but if you suspect this is the case let me know, and I’ll update the examples and tutorials. I have already done so several times, it’s only a few minutes work.
  • The openURL method may not work. This happens quite regularly, and depending on your server, proxies and direction of the wind, they may not work. The solution is to use wget to download the page locally, then use readFile instead. Hopefully a decent Haskell HTTP library will emerge, and that can be used instead.

Last modified date of Haskell wiki

Our goal is to develop a program that displays the date that the wiki at wiki.haskell.org was last modified. This example covers all the basics in designing a basic web-scraping application.

Finding the Page

We first need to find where the information is displayed and in what format. Taking a look at the front web page, when not logged in, we see:

<ul id="f-list">
  <li id="lastmod"> This page was last modified on 9 September 2013, at 22:38.</li>
  <li id="copyright">Recent content is available under <a href="/HaskellWiki:Copyrights" title="HaskellWiki:Copyrights">a simple permissive license</a>.</li>
  <li id="privacy"><a href="/HaskellWiki:Privacy_policy" title="HaskellWiki:Privacy policy">Privacy policy</a></li>
  <li id="about"><a href="/HaskellWiki:About" title="HaskellWiki:About">About HaskellWiki</a></li>
  <li id="disclaimer"><a href="/HaskellWiki:General_disclaimer" title="HaskellWiki:General disclaimer">Disclaimers</a></li>

So, we see that the last modified date is available. This leads us to rule 1:

Rule 1: Scrape from what the page returns, not what a browser renders, or what view-source gives.

Some web servers will serve different content depending on the user agent, some browsers will have scripting modify their displayed HTML, some pages will display differently depending on your cookies. Before you can start to figure out how to start scraping, first decide what the input to your program will be. There are two ways to get the page as it will appear to your program.

Using the HTTP package

We can write a simple HTTP downloader with using the HTTP package:

module Main where

import Network.HTTP

openURL :: String -> IO String
openURL x = getResponseBody =<< simpleHTTP (getRequest x)

main :: IO ()
main = do
    src <- openURL "http://wiki.haskell.org/Haskell"
    writeFile "temp.htm" src

Now open temp.htm, find the fragment of HTML containing the hit count, and examine it.

Finding the Information

Now we examine both the fragment that contains our snippet of information, and the wider page. What does the fragment have that nothing else has? What algorithm would we use to obtain that particular element? How can we still return the element as the content changes? What if the design changes? But wait, before going any further:

Rule 2: Do not be robust to design changes, do not even consider the possibility when writing the code.

If the user changes their website, they will do so in unpredictable ways. They may move the page, they may put the information somewhere else, they may remove the information entirely. If you want something robust talk to the site owner, or buy the data from someone. If you try and think about design changes, you will complicate your design, and it still won’t work. It is better to write an extraction method quickly, and happily rewrite it when things change.

So now, let’s consider the fragment from above. It is useful to find a tag which is unique just above your snippet - something with a nice id or class attribute - something which is unlikely to occur multiple times. In the above example, an id with value lastmod seems perfect.

module Main where

import Data.Char
import Network.HTTP
import Text.HTML.TagSoup

openURL :: String -> IO String
openURL x = getResponseBody =<< simpleHTTP (getRequest x)

haskellLastModifiedDateTime :: IO ()
haskellLastModifiedDateTime = do
    src <- openURL "http://wiki.haskell.org/Haskell"
    let lastModifiedDateTime = fromFooter $ parseTags src
    putStrLn $ "wiki.haskell.org was last modified on " ++ lastModifiedDateTime
    where fromFooter = unwords . drop 6 . words . innerText . take 2 . dropWhile (~/= "<li id=lastmod>")

main :: IO ()
main = haskellLastModifiedDateTime

Now we start writing the code! The first thing to do is open the required URL, then we parse the code into a list of Tags with parseTags. The fromFooter function does the interesting thing, and can be read right to left:

  • First we throw away everything (dropWhile) until we get to an li tag containing id=lastmod. The (~==) and (~/=) operators are different from standard equality and inequality since they allow additional attributes to be present. We write "<li id=lastmod>" as syntactic sugar for TagOpen "li" [("id","lastmod")]. If we just wanted any open tag with the given id attribute we could have written (~== TagOpen "" [("id","lastmod")]) and this would have matched. Any empty strings in the second element of the match are considered as wildcards.
  • Next we take two elements: the <li> tag and the text node immediately following.
  • We call the innerText function to get all the text values from inside, which will just be the text node following the lastmod.
  • We split the string into a series of words and drop the first six, i.e. the words This, page, was, last, modified and on
  • We reassemble the remaining words into the resulting string 9 September 2013, at 22:38.

This code may seem slightly messy, and indeed it is - often that is the nature of extracting information from a tag soup.

Rule 3: TagSoup is for extracting information where structure has been lost, use more structured information if it is available.

Simon’s Papers

Our next very important task is to extract a list of all Simon Peyton Jones’ recent research papers off his home page. The largest change to the previous example is that now we desire a list of papers, rather than just a single result.

As before we first start by writing a simple program that downloads the appropriate page, and look for common patterns. This time we want to look for all patterns which occur every time a paper is mentioned, but no where else. The other difference from last time is that previous we grabbed an automatically generated piece of information - this time the information is entered in a more freeform way by a human.

First we spot that the page helpfully has named anchors, there is a current work anchor, and after that is one for Haskell. We can extract all the information between them with a simple take/drop pair:

takeWhile (~/= "<a name=haskell>") $
drop 5 $ dropWhile (~/= "<a name=current>") tags

This code drops until you get to the “current” section, then takes until you get to the “haskell” section, ensuring we only look at the important bit of the page. Next we want to find all hyperlinks within this section:

map f $ sections (~== "<A>") $ ...

Remember that the function to select all tags with name “A” could have been written as (~== TagOpen "A" []), or alternatively isTagOpenName "A". Afterwards we map each item with an f function. This function needs to take the tags starting just after the link, and find the text inside the link.

f = dequote . unwords . words . fromTagText . head . filter isTagText

Here the complexity of interfacing to human written markup comes through. Some of the links are in italic, some are not - the filter drops all those that are not, until we find a pure text node. The unwords . words deletes all multiple spaces, replaces tabs and newlines with spaces and trims the front and back - a neat trick when dealing with text which has spacing at the source code but not when displayed. The final thing to take account of is that some papers are given with quotes around the name, some are not - dequote will remove the quotes if they exist.

For completeness, we now present the entire example:

module Main where

import Network.HTTP
import Text.HTML.TagSoup

openURL :: String -> IO String
openURL x = getResponseBody =<< simpleHTTP (getRequest x)

spjPapers :: IO ()
spjPapers = do
        tags <- parseTags <$> openURL "http://research.microsoft.com/en-us/people/simonpj/"
        let links = map f $ sections (~== "<A>") $
                    takeWhile (~/= "<a name=haskell>") $
                    drop 5 $ dropWhile (~/= "<a name=current>") tags
        putStr $ unlines links
        f :: [Tag String] -> String
        f = dequote . unwords . words . fromTagText . head . filter isTagText

        dequote ('\"':xs) | last xs == '\"' = init xs
        dequote x = x

main :: IO ()
main = spjPapers

Other Examples

Several more examples are given in the Sample.hs file, including obtaining the (short) list of papers from my site, getting the current time and a basic XML validator. All use very much the same style as presented here - writing screen scrapers follow a standard pattern. We present the code from two for enjoyment only.

My Papers

module Main where

import Network.HTTP
import Text.HTML.TagSoup

openURL :: String -> IO String
openURL x = getResponseBody =<< simpleHTTP (getRequest x)

ndmPapers :: IO ()
ndmPapers = do
        tags <- parseTags <$> openURL "http://community.haskell.org/~ndm/downloads/"
        let papers = map f $ sections (~== "<li class=paper>") tags
        putStr $ unlines papers
        f :: [Tag String] -> String
        f xs = fromTagText (xs !! 2)

main :: IO ()
main = ndmPapers

UK Time

module Main where

import Network.HTTP
import Text.HTML.TagSoup

openURL :: String -> IO String
openURL x = getResponseBody =<< simpleHTTP (getRequest x)

currentTime :: IO ()
currentTime = do
    tags <- parseTags <$> openURL "http://www.timeanddate.com/worldclock/uk/london"
    let time = fromTagText (dropWhile (~/= "<span id=ct>") tags !! 1)
    putStrLn time

main :: IO ()
main = currentTime

Other Examples

In Sample.hs the following additional examples are listed:

  • Google Tech News
  • Package list form Hackage
  • Print names of story contributors on sequence.complete.org
  • Parse rows of a table

Related Projects


Changelog for TagSoup

0.14.8, released 2019-05-01
#78, add strMap
0.14.7, released 2018-09-18
#75, escape single quote (') characters as &#39;
#72, update some dead URLs
0.14.6, released 2018-02-27
Fix up source positions on bogus <! markers
0.14.5, released 2018-02-27
Fix up source positions on bogus <? markers
0.14.4, released 2018-02-26
#71, fix up source positions on bogus comments
0.14.3, released 2018-01-21
#70, fix up source positions on comments
0.14.2, released 2017-11-30
#66, make sure positions are correct for lone & characters
0.14.1, released 2017-02-25
#63, add maybeAttrib
0.14, released 2016-06-11
#14, eliminate Text.HTML.Download
0.13.10, released 2016-05-06
#51, improve the Haddock documentation
#52, fix some > 16bit HTML entities
0.13.9, released 2016-03-15
#50, fix a space leak
#36, fix the demo examples
#35, make IsString a superclass of StringLike
#33, make flattenTree O(n) instead of O(n^2)
0.13.8, released 2016-01-10
#30, add parse/render functions directly to the Tree module
0.13.7, released 2016-01-09
#32, make sure upper case &#X works in lookupEntity
0.13.6, released 2015-12-23
#28, some named entities require a trailing semicolon (e.g. mid)
0.13.5, released 2015-10-30
#26, rename the test program to test-tagsoup
0.13.4, released 2015-10-26
#24, add isTagComment function
Update the copyright year
0.13.3, released 2014-10-01
Work on GHC 7.9
0.13.2, released 2014-08-08
Remove all package upper bounds
Allow QuickCheck-2.6
0.13.1, released 2014-01-20
#562, treat <script> specially as per HTML5
0.13, released 2013-08-06
#616, extend to all HTML5 entities
Optimise lookupNamedEntity
Replace escapeXMLChar with escapeXML
Change all Entity functions to return String, not Int or Char
0.12.7, released 2012-08-19
Allow deepseq-1.3
Allow QuickCheck-2.5
Support bytestring-0.10 (NFData instances added)
0.12.6, released 2011-12-11
#515, don't crash on malformed characters (use ? instead)
0.12.5, released 2011-11-17
Add optRawTag to Render, to ensure script's don't get escaped
0.12.4, released 2011-11-05
#487, fix the behaviour of ~== for TagComment and others
0.12.3, released 2011-08-12
GHC 7.2 compatibility
0.12.2, released 2011-06-12
Add StringLike instances for Text
0.12.1, released 2011-06-01
Add parseOptionsEntities and improve documentation
0.12, released 2010-12-12
Upgrade to QuickCheck 2.4.*
Export toTagRep
Make the -download flag off by default
Eliminate HTTP dependency
Eliminate mtl dependency
0.11.1, released 2010-10-04
Support --flags=-download to eliminate the network dependency
0.11, released 2010-09-12
#326, <a "foo"> is no longer treated as an attribute
Add Eq/Ord instances to Tree
Don't mark Text.HTML.TagSoup.Tree as preliminary
#325, \r should be treated as whitespace
0.10.1, released 2010-07-22
#322, don't change ' to &apos; in render (do what the docs say)
0.10, released 2010-05-26
Improve the cabal file, make the test program off by default
Expose Text.HTML.TagSoup.Match again (was hidden accidentally)
0.9, released 2010-05-09
#294, let optEntityData see if there was a ';' (CHANGES API)
Numeric/hex entities in attributes were misparsed
#149, add escapeHTML function
0.8, released 2010-01-07
Parser now based on HTML5 specification
Tag is now parameterised by the string type
0.6, released 2008-04-23
Addition of Text.HTML.TagSoup.Tree and Text.HTML.TagSoup.Render
Text.HTML.TagSoup.Parser.Options renamed to ParseOptions
Text.HTML.TagSoup.Parser.options renamed to parseOptions
0.4, released 2008-01-14
Changelog started