Scrapy / XPATH : how to extract ONLY text from descendants and self - xpath

Question

I have the following simple, nested structure:
<main>
<em>bla-bla</em>
<div class="1">1.1</div>
<div class="2">2.1</div>
<div class="2">2.2</div>
<div class="1">1.2</div>
<div class="2">
<span>
<em>2.3</em>
</span>
</div>
<div class="2">2.4</div>
</main>
I would like to extract now all text from all nodes, but struggle with the nested node ( etc.).
The expected output should be:
2.1
2.2
2.3
2.4
Trying something like:
//div[contains(#class,"2")]/text()
gives
2.1
2.2
<div class="2"><span><em>2.3</em></span></div>
<div class="2"><span><em>2.3</em></span></div>
2.4
Instead of using straight XPATH, I also tried using several steps in Scrapy, like:
divs = response.xpath("//div[contains(#class,"2")]")
for div in divs:
# now check somehow that the div contains an "em" node
Using
div.xpath("//em")
does not work, since it gives all nodes. Using div.extract() here and looking at the returned string, I could of course find using string search, but this is rather a hack and does not look like the proper Scrapy solution.
Any suggestions how to solve this either directly with Xpath or with Scrapy in general would be grealy appreciated.

Answer 1

What do you think about [i.strip() for i in response.xpath('//div[contains(#class, "2")]//text()').extract() if i.strip()]?
Without stripping it gives some empty cases also:
>>> response.xpath('//div[contains(#class, "2")]//text()').extract()
[u'2.1', u'2.2', u'\n ', u'\n ', u'2.3', u'\n ', u'\n ', u'2.4']
So I filter them with strip:
>>> [i.strip() for i in response.xpath('//div[contains(#class, "2")]//text()').extract() if i.strip()]
[u'2.1', u'2.2', u'2.3', u'2.4']