使用 nokogiri 解析 xml 文件

Question

I need to parse a xml file on ruby on rails, i'm using nokogiri gem to parse it.

I can parse like this to appear the parent and his children, but its appearing like this:

PARENT: Example Parent 1

CHILD: Example Children 1Example Children 2Example Children 3

PARENT: Example Parent 2

CHILD:

Why it's missing the children of the second parent node? If I call the array with a for each, it appears all the children. I did this like this:

In the controller:

  @codes = []
    doc.xpath('//Node').each do |parent| 
       @parentN =parent.xpath('///ancestor::*/@name')


      @codes << parent.xpath('Node/@name').text

    end

And the view:

<% for x in 0...@parentN.count %>

    <p> PARENT: <%= @parentN[x]  %>  </p>

  <p> CHILD:  <%= @codes[x] %>  </p>

    <%   end %>

How can I "connect" the parent with the childs? Presenting the parent and his children, and then other parent and childrens...

This is my xml file:

   <Report>
       <Node name="Example Parent 1" color="red">
          <Node name="Example Children 1" color="red" rank="very high" />
          <Node name="Example Children 2" color="red" rank="high" />
          <Node name="Example Children 3" color="yellow" rank="moderate" />
       </Node>
       <Node name="Example Parent 2" color="yellow">
          <Node name="Example Children 1" color="yellow" rank="moderate" />
       </Node>
    </Report>

Answer 1

Problem #1

In this line:

       @parentN =parent.xpath('///ancestor::*/@name')

you override the previous value of @parentN .

Problem #2

By running

<% for x in 0...@parentN.count %>

You will be getting 2 values for a single valued array. .count is equivalent to the last index +1 (for an array with only [0] .count is 1. Your @parentN is assigned to an object

Recommendation (simple)

Use a single array to hold the nested values (as a hash) rather than two variables.

#xmlController.rb
@codes = []
doc.xpath('Report/Node').each do |parent| 
  @codes << { parent.xpath('@name') => parent.xpath('Node').map { |child| child.text }
end



#show.html.erb

<% @codes.each do |parent, children| %>
  <p> PARENT: <%= @parent  %>  </p>
  <p> CHILDREN:  <%= @children.each { |child| p child } %>  </p>

Recommendation based on comments below

The above was shown to demonstrate the simpilest way to think about the problem. Now that we are ready to parse all the data in the node, we need to change our xpath and our map. The doc.xpath('Report/Node') is used to select the parent node, and that can stay the same. We will want to set the @codes key to the actual value of the string embedded in the Node which is not parent.xpath('@name') but actually parent.xpath('@name')[0].value . There could be multiple xml representations of nodes with the attribute 'name' and we want the first ( [0] ) one. The value of the name attribute is returned using the .value method.

Make a class so the nodes become objects

Your Parent node has a name and a color and your children have name, color, and rank. It looks like you have a model for Node that looks like:

class Node
  include ActiveModel::Model
  attr_accessor :name, :color, :rank, :children
end

I'm simplifying things by not using persistence here, but you may want to save your records to disk, and if you do look into the slew of things ActiveRecord does on RailsGuides

Now when we go through the xml document, we will create an array of objects rather than the hash of strings (which both happen to be objects, but I'll leave that quandry for you to check out).

Parse the Xpath to get attributes of Node Objects

A quick way to set the name and color attributes of the parent looks like this:

@node = Node.new(doc.xpath('Report/Node').first.attributes.inject({}) { |attrs, value| attrs[value[0].to_sym] = value[1].value; attrs })

OK, so maybe that wasn't all that easy. What we do is take the Enumerable result of the XPath, navigate to the first attributes and make a hash of string attribute names (name, color, rank) and their corresponding values. Once we have the hash we pass it to our Node class' new method to instanciate (create) a node. This will pass us an object that we can use:

@node.name
#=> "Example Parent 1"

Extend the Class for children

Once we have the parent node, we can give it children, creating new nodes in an array. To facilitate this, we extend the definition of the model to include an overridden initializer (new()).

class Node
  include ActiveModel::Model
  attr_accessor :name, :color, :rank, :children

  def initialize(*args)
    self.children = []
    super(*args)
  end
end

Adding children

doc.xpath('Report/Node').each do |parent|
  node = Node.new(parent.attributes.inject({}) { |attrs, value| attrs[value[0].to_sym] = value[1].value; attrs }))
  node.children = parent.xpath('Node').map do |child|
    Node.new(child.attributes.inject({}) { |attrs, value| attrs[value[0].to_sym] = value[1].value; attrs }))
  end
end

We can automate this process now that we know how to create a Node object using .first and a child of it using .first with the previous enumeration.

 doc.xpath('Report/Node').each do |parent| node = Node.new(parent.attributes.inject({}) { |attrs, value| attrs[value[0].to_sym] = value[1].value; attrs })) node.children = parent.xpath('Node').map do |child| Node.new(child.attributes.inject({}) { |attrs, value| attrs[value[0].to_sym] = value[1].value; attrs })) end end

Ugly controller code

Move it to the model

But Wait! That isn't very DRY! Let's move the logic that hurts our eyes to look at into the model to make it easier to work with.

 class Node include ActiveModel::Model attr_accessor :name, :color, :rank, :children def initialize(*args) self.children = [] super(*args) end def self.new_from_xpath(xml_node) self.new(xml_node.attributes.inject({}) { |attrs, value| attrs[value[0].to_sym] = value[1].value; attrs }) end end

Final controller

Now the controller looks like this:

 @nodes = [] doc.xpath('Report/Node').each do |parent| node = Node.new_from_xpath(parent) node.children = parent.xpath('Node').map do |child| Node.new_from_xpath(child) end @nodes << node end

Using this in the view

In the view you can use the @nodes like this:

 <% for @node in @nodes %> Parent: <%= @node.name %> Children: <% for @child in @node.children %> <%= @child.name %> is <%= @child.color %> <% end %> <% end %>