how to exclude a table inside in another table in xpath? - xpath

Question

I have the follow html file:
<table class="pd-table">
<caption> Tech </caption>
<tbody>
<tr data-group="1">
<td> Electrical </td>
<td> Design </td>
<tr data-group="1">
<td> Output </td>
<td> Function </td>
<tr data-group="7">
<td> EMC </td>
<table>
<tbody>
<tr>
<td> EN 6547 ESD </td>
<td> EN 8901 ESD </td>
<tr data-group="8">
<td> Weight [8] </td>
<td> 27.7 </td>
I can isolate EN 6547 ESD and EN 8901 ESD with the follow xpath:
//table[#class="pd-table"]//tbody//tr//td/table//tr//td/text()').getall()
Any other way is always welcome :)
Another data which I would like to get is to get all the rest of the data without the previous isolated.
Is there any way to do it? :)

Answer 1

Looks like table tag is not closed properly in data-group-7...
Anyway in such cases you can stick to text content of the cell using contains() or text()="some exact text"
response.xpath('//td[contains(text(), "EMC")]').css('td~table tbody td::text').extract()

Answer 2

Your used Xpath uses a lot of unwanted double slash.
See meaning of double slash in Xpath.
The less you use double slash, the better it will perform.
So just use single slash like this:
//table[#class="pd-table"]/tbody/tr/td/table/tr/td/text()
Another way of selecting td's that have two ancestor::table
//td[count(ancestor::table)=2]/text()
And that leads to the answer of your second question:
//td[count(ancestor::table)=1]/text()
An other possibility would just be:
//table[#class="pd-table"]/tbody/tr/td/text()
Or(assuming the second tabel does not have tr's with #data-group):
//tr[#data-group]/td/text()
So you see there are many Xpath's lead to Rome ;-).