Building XML by enumerating through array & hashes

Question

data =
[ { 'name' => 'category1',
    'subCategory' => [ {'name' => 'subCategory1',
                        'product' => [ {'name' => 'prodcutName1',
                                        'desc' => 'desc1'},
                                       {'name' => 'prodcutName2',
                                        'desc' => 'desc2'}]
                        } ]

  },
  { 
    #category2 ...and so on 
  }
]

Just recently finished a small project with Ruby. I used the above array of hashes to produce a XML.

I think my solution is quite messy nesting loops and builder tags. Can anyone help me out with a more elegant approach?

builder = Nokogiri::XML::Builder.new { |xml|
data.each { |category| # go through every category in the data array
    xml.category {
        xml.name category['name']
        category['subCategory'].each { |subCategory| # go through each subCategory in the category
            xml.subCategory {
                xml.name subCategory['name']
                subCategory['product'].each { |product| # go though each products and print their data
                    xml.name product['name']
                    xml.desc product['desc']
                }
            }
        }
    }
}


puts builder.to_xml

Answer 1

Here's a nice recursive solution, that creates <key>value</key> from 'key'=>'value' entries in your hash. If the value is an array, it instead recurses, using the key name as a wrapper element.

require 'nokogiri'

def process_array(label,array,xml)
  array.each do |hash|
    xml.send(label) do                 # Create an element named for the label
      hash.each do |key,value|
        if value.is_a?(Array)
          process_array(key,value,xml) # Recurse
        else
          xml.send(key,value)          # Create <key>value</key> (using variables)
        end
      end
    end
  end
end

builder = Nokogiri::XML::Builder.new do |xml|
  xml.root do                           # Wrap everything in one element.
    process_array('category',data,xml)  # Start the recursion with a custom name.
  end
end

puts builder.to_xml

When used with this data…

data = [
  { 'name' => 'category1',
    'subCategory' => [
      { 'name' => 'subCategory1',
        'product' => [
          { 'name' => 'productName1',
            'desc' => 'desc1' },
          { 'name' => 'productName2',
            'desc' => 'desc2' } ]
      } ]
  },
  { 'name' => 'category2',
    'subCategory' => [
      { 'name' => 'subCategory2.1',
        'product' => [
          { 'name' => 'productName2.1.1',
            'desc' => 'desc1' },
          { 'name' => 'productName2.1.2',
            'desc' => 'desc2' } ]
      } ]
  },
]

…you get this result:

<?xml version="1.0"?>
<root>
  <category>
    <name>category1</name>
    <subCategory>
      <name>subCategory1</name>
      <product>
        <name>productName1</name>
        <desc>desc1</desc>
      </product>
      <product>
        <name>productName2</name>
        <desc>desc2</desc>
      </product>
    </subCategory>
  </category>
  <category>
    <name>category2</name>
    <subCategory>
      <name>subCategory2.1</name>
      <product>
        <name>productName2.1.1</name>
        <desc>desc1</desc>
      </product>
      <product>
        <name>productName2.1.2</name>
        <desc>desc2</desc>
      </product>
    </subCategory>
  </category>
</root>

---

However, if I had control over the XML schema, I'd do this instead:

require 'nokogiri'

def process_array(label,array,xml)
  array.each do |hash|
    kids,attrs = hash.partition{ |k,v| v.is_a?(Array) }
    xml.send(label,Hash[attrs]) do
      kids.each{ |k,v| process_array(k,v,xml) }
    end
  end
end

builder = Nokogiri::XML::Builder.new do |xml|
  xml.root{ process_array('category',data,xml) }
end

puts builder.to_xml

<?xml version="1.0"?>
<root>
  <category name="category1">
    <subCategory name="subCategory1">
      <product name="productName1" desc="desc1"/>
      <product name="productName2" desc="desc2"/>
    </subCategory>
  </category>
  <category name="category2">
    <subCategory name="subCategory2.1">
      <product name="productName2.1.1" desc="desc1"/>
      <product name="productName2.1.2" desc="desc2"/>
    </subCategory>
  </category>
</root>

…but perhaps you're dealing with some terrible XML schema like PList.

Answer 2

A very clean approach is to use the xml-simple gem. Simply use xml_out with two options:

• RootName to specify the XML root element
• AnonymousTag to provide a tag name for your top-level hash

Other options are documented here.

Code:

require 'xmlsimple'
XmlSimple.xml_out(data, {"RootName" => "categories", "AnonymousTag => "category"})

Output:

<categories>
  <category name="category1">
    <subCategory name="subCategory1">
      <product name="prodcutName1" desc="desc1" />
      <product name="prodcutName2" desc="desc2" />
    </subCategory>
  </category>
  <category></category>
</categories>

Answer 3

Here is a single function to construct XML from a Hash with special handling for attributes and allows a mix of Array and Hash to represent the elements.

You can also begin the Builder outside and call this function after setting up all necessary namespaces.

require 'nokogiri'

def generate_xml(data, parent = false, opt = {})
    return if data.to_s.empty?
    return unless data.is_a?(Hash)

    unless parent
        # assume that if the hash has a single key that it should be the root
        root, data = (data.length == 1) ? data.shift : ["root", data]
        builder = Nokogiri::XML::Builder.new(opt) do |xml|
            xml.send(root) {
                generate_xml(data, xml)
            }
        end

        return builder.to_xml
    end

    data.each { |label, value|
        if value.is_a?(Hash)
            attrs = value.fetch('@attributes', {})
            # also passing 'text' as a key makes nokogiri do the same thing
            text = value.fetch('@text', '') 
            parent.send(label, attrs, text) { 
                value.delete('@attributes')
                value.delete('@text')
                generate_xml(value, parent)
            }

        elsif value.is_a?(Array)
            value.each { |el|
                # lets trick the above into firing so we do not need to rewrite the checks
                el = {label => el}
                generate_xml(el, parent) 
            }

        else
            parent.send(label, value)
        end
    }
end

puts generate_xml(
    {'myroot' => 
        {
            'num' => 99, 
            'title' => 'something witty', 
            'nested' => { 'total' => [99, 98], '@attributes' => {'foo' => 'bar', 'hello' => 'world'}}, 
            'anothernest' => {
                '@attributes' => {'foo' => 'bar', 'hello' => 'world'}, 
                'date' => [
                    'today', 
                    {'day' => 23, 'month' => 'Dec', 'year' => {'y' => 1999, 'c' => 21}, '@attributes' => {'foo' => 'blhjkldsaf'}}
                ]
            }
    }})
puts puts
puts generate_xml({
            'num' => 99, 
            'title' => 'something witty', 
            'nested' => { 'total' => [99, 98], '@attributes' => {'foo' => 'bar', 'hello' => 'world'}}, 
            'anothernest' => {
                '@attributes' => {'foo' => 'bar', 'hello' => 'world'}, 
                'date' => [
                    'today', 
                    {'day' => [23,24], 'month' => 'Dec', 'year' => {'y' => 1999, 'c' => 21}, '@attributes' => {'foo' => 'blhjkldsaf'}}
                ]
            }
    })

And the resulting XML output:

<?xml version="1.0"?>
<myroot>
  <num>99</num>
  <title>something witty</title>
  <nested foo="bar" hello="world">
    <total>99</total>
    <total>98</total>
  </nested>
  <anothernest foo="bar" hello="world">
    <date>today</date>
    <date foo="blhjkldsaf">
      <day>23</day>
      <month>Dec</month>
      <year>
        <y>1999</y>
        <c>21</c>
      </year>
    </date>
  </anothernest>
</myroot>


<?xml version="1.0"?>
<root>
  <num>99</num>
  <title>something witty</title>
  <nested foo="bar" hello="world">
    <total>99</total>
    <total>98</total>
  </nested>
  <anothernest foo="bar" hello="world">
    <date>today</date>
    <date foo="blhjkldsaf">
      <day>23</day>
      <day>24</day>
      <month>Dec</month>
      <year>
        <y>1999</y>
        <c>21</c>
      </year>
    </date>
  </anothernest>
</root