XPath: get closest heading element (h1, h2, h3, etc.)

Question

I want to select the heading element (whether h1, h2, h3, h4, h5 or h6) that is closest above the form using XPath in PHP.

<h2>Foo</h2>
<h3>Bar</h3>
<form>
    <input />
</form>

The example above should return the h3 (Bar) because it is closest to the form.

<h4>Kee</h4>
<form>
    <input />
</form>

This example on the other hand should return the h4 (Kee) because it is closest.

This query (from https://stackoverflow.com/a/2216795/4391251) works fine for just the h2 tags. I could modify it for h1, h3, h4, h5, etc. but I want a catch-all query.

$headings = $xpath->query('((//form)[2]/ancestor::*/h2[.!=""])[last()]');

Basically I want something like this

$headings = $xpath->query('((//form)['.$i.']/ancestor::*/[h2 or h3][.!=""])[last()]');

Except for that doesn't return any results, and neither does (based on https://stackoverflow.com/a/7995095/4391251)

$headings = $xpath->query('((//form)['.$i.']/ancestor::*/[self::h2 or self::h3][.!=""])[last()]');

What query will give the desired results?

har07 · Accepted Answer · 2015-06-11T08:55:28.097

2

You can try something like this :

$xpath->query('//form['.$i.']/preceding-sibling::*[self::h2 or self::h3][1]')

basically, the xpath get the first preceding sibling of form[i] that is of type <h2> or <h3> (or whatever, just list all other elements as needed in the xpath predicate).

edited Jun 11 '15 at 08:55

answered Jun 11 '15 at 08:47

har07

88,338
12
84
137

1

Excellent! I ended up using it like this: $headings = $xpath->query('((//form)['.$i.']/preceding-sibling::*[self::h2 or self::h3])[1]'); – Theo van der Zee Jun 11 '15 at 09:04
could you post sample HTML that reproduce the problem? – har07 Jun 11 '15 at 09:12
*AFAICS*, as explained in this answer, the xpath should return *only the first* preceding sibling of `form`, there is no chance it can return more than one element. – har07 Jun 11 '15 at 09:16
1

I had already erased the comment I had posted earlier, it was a mistake on my end. The query returns multiple headings (as expected) because there were multiple forms on the page (each with their own heading). The query works as supposed! – Theo van der Zee Jun 11 '15 at 09:34

score 1 · Answer 2 · answered Jun 11 '15 at 08:48

1

Take the fist h before form

//form/preceding::*[starts-with(name(),'h')][1]

answered Jun 11 '15 at 08:48

splash58

26,043
3
22
34

This one unfortunately also returned the Head element (as it also starts with h) – Theo van der Zee Jun 11 '15 at 09:03
@Theo if there is no h elements between head and form :) – splash58 Jun 11 '15 at 09:07

score -1 · Answer 3 · edited May 03 '17 at 08:59

-1

/html/body/*[starts-with(name(),'h')]

edited May 03 '17 at 08:59

goto

7,908
10
48
58

answered May 03 '17 at 03:08

awuu

251
2
6

XPath: get closest heading element (h1, h2, h3, etc.)

3 Answers3