travis Flexparser

Provides an easy to use DSL for flexible, robust xml parsers. The goal of flexparser is to be able to write One Parser to parse them all.


Add this line to your application's Gemfile:

gem 'flexparser'

Or simply

gem install flexparser



Including the Flexparser module in any Class turns it into a parser. Lets start simple:

class WebParser
  include Flexparser

  property 'url'

Now this class is able to parse xml similar to this:


Now your parser can do this:

# Assuming the xml variable holds the xml code mentioned above
website = WebParser.parse xml
webite.url #=> 'www.my-page.com'


A node command will only return the first value it finds. When you have multiple nodes that interest you, you can get a collection of them.

books = '
        <work author="H.P. Lovecraft">
                 <story>The Call of Cthulhu</story>
                 <story>The Nameless City</story>

class LovecraftParser
  include Flexparser

  property 'story', collection: true

work = LovecraftParser.parse books
work.story #=> ['The Call of Cthulhu', 'Dagon', 'The Nameless City']

Nested Parsers

Sometimes you want more than to just extract a String. This way you can make your parser return complex Objects and nest parsers as deep as you like.

library = "
  <author>J. R. R. Tolkien</author>
  <title>The Hobbit</title>
  <author>Suzanne Collins</author>
  <title>The Hunger Games</title>

class LibraryParser
  include Flexparser

  property 'book', collection: true do
    attr_accessor :isbn
    property 'author'
    property 'title'

lib = LibraryParser.parse library
lib.book.second.authro #=> 'Suzanne Collins'
lib.book.first.title #=> 'The Hobbit'
lib.book.first.isbn = '9780582186552'
lib.book.first.isbn #=> '9780582186552'

With nested parsers, anonymous classes are defined inside an existing parser. Therefore you can define methods all you like (should the need arise).

Tag Definitions

You might not always know (or it might not always be the same), what the information you are looking for is called. If that is the case, you can define multiple tags for the same property. Here are a few examples:

class UniParser
  include Flexparser

  # Creates accessors called 'url' and 'url=' but will look for nodes with the name url, link and website. Will return the first thing it finds.
  property %w[url link website]

  # Creates a property called main_header and will look for message and title
  property %w[message title], name: 'main_header'

  # This will define a property called width and will look for an attribute of the same name
  property '@width'

  # This will define a property called `image_url` that will look for a node called 'image' and extract its 'url' attribute
  property 'image/@url'

  # This will look for a tag called encoded with the namespace content
  property 'content:encoded'

  # Here we define a transformation to make the parser return an integer
  property 'height', transform: :to_i

  # An alternative to the transformation is a type. The type must have a #parse method that receives a string
  property 'url', type: URI

  # A little bit of everything
  property %w[image picture @img media:image], name: 'visual', type: URI, collection: true

Defining a parser with a block

When defining nested parsers, you would use a block. Like this:

class ParserClass
  include Flexparser

  property 'story', collection: true do
    property 'author'
    property 'title'

When passing a block to a parser definition, a new class is created that basically looks like this:

Class.new { include Flexparser }

The block is then class_evaled on this anonymous class. Thats gives you a lot of flexibility in definen your parsers.


You can configure Flexparser by using a block (for example in an initializer) like so:

Flexparser.configure do |config|
  config.option = value

At time of writing there are two Options:


Default: true If this is true you need to specify a :name for your property everytime there is more than one tag in your tag-list. Example:

# Bad!
property %w[url link website]

# Good!
property %w[url link website], name: 'website'

# Don't care! Unambiguous!
property 'url'
property ['width']


Default: true If true, Flexparser will add a second set of xpaths to the list of tags you specified, that will ignore namespaces completely. Example:

Flexparser.configure { |c| c.retry_without_namespaces = false }
class SomeParser
  property 'inventory'

xml = '<inventory xmlns="www.my-inventory.com">james</inventory>'

# The inventory can't be found because it is namespaced.
SomeParser.parse(xml).inventory #=> nil :(

Flexparser.configure { |c| c.retry_without_namespaces = true }
class SomeBetterParser
  property 'inventory'

xml = '<inventory xmlns="www.my-inventory.com">james</inventory>'

# The inventory can be found because we don't care.
SomeParser.parse(xml).inventory #=> 'james'

The Xpath used here adheres to xpath version 1.X.X and uses the name property .//[name()='inventory']


After checking out the repo, run bin/setup to install dependencies. Then, run rake test to run the tests. You can also run bin/console for an interactive prompt that will allow you to experiment.

How it works (in a nutshell)

The Flexparser module defines certain class methods. Most importantly property works in similar ways. property takes a String or an array of strings as well as some options. The property method instantiates a TagParser and adds it to the @tags property of the class that is including Flexparser (we'll call it MainClass from here on out), which holds an array of all the TagParsers and CollectionParsers . It also defines accessors for the name of the property the property parser should extract.

The Parsers use an instance of Flexparser::XPaths to handle the array of tags that they are passed. When everything is setup (i.e. the class is loaded), you can call ::parse on your MainClass and pass it an XML string. At this point the MainClass instantiates itself and the TagParsers and CollectionParsers extract a value from the xml, that is then assigned to the newly created MainClass instance.