如何使用Nokogiri在标签下获取所有文本？

Question

In this example I am trying to get the text from within the <td> tag of a table. First, the html code.

<table>
  <tbody>
  <tr>
    <td>Single line of text</td>
  </tr>
  <tr>
    <td>Text here<p>First line</p><p>Second line</p></td>
  </tr>
  </tbody>
</table>

Then the ruby code here.

require 'nokogiri'
require 'pp'

html = File.open('test.html').read
doc = Nokogiri::HTML(html)
rows = doc.xpath('//table[1]/tbody/tr')

data = rows.collect do |row|
  row.at_xpath('td[1]/text()').to_s
end

pp data

And the result that I get is.

["Single line of text", "Text here"]

How can I get all of the text in the second <td> tag?

Answer 1

There are two changes you will need to make to get all the text nodes. First at_xpath will only ever return a single node, so to get multiple nodes you'll need to use xpath .

Second, to get all descendant nodes, not just child nodes, use // instead of / .

Combining these, the line of code would be:

row.xpath('td[1]//text()').to_s

This will concatenate all the text nodes together, giving the result:

["Single line of text", "Text hereFirst lineSecond line"]

which may not be what you want. Rather than just call to_s on the resulting nodeset you will need to process to fit your needs.

Answer 2

How about this?

pp doc.search("//tr[2]//td//text()").map { |item| item.text }

As matt says, you can get all descendants using // .

You can also index the second tr if you want that one specifically. Just leave out the indexing to get all the tr s.

And you can filter the resulting text objects to get only those that have a td upstream.

Finally, map over each Nokogiri object, plucking out the text into the final array, which looks like this:

["Text here", "First line", "Second line"]

Answer 3

You want the text method of Nokogiri::XML::Node if you want to get all the text for any element:

p doc.xpath('//table[1]/tbody/tr').map{ |tr| tr.text.strip }
#=> ["Single line of text", "Text hereFirst lineSecond line"]

(The strip method just gets rid of leading and trailing whitespace.)