/descendant::span[1] should not be equivalent to //span[1]

[yayuyokitano]

Note: I have experienced this issue using htmlquery, but it seems to not be fixed here either, so I am raising issue here.

By xpath spec, /descendant::span[1] and //span[1] are not the same.

/descendant::span[1] should get the first span descendant matched within the current ancestor, while //span[1] should get the first span descendant matched within the parent of each respective span element. This package seems to currently be treating both like //span[1]

Example:

<div id="wrapper">
  <span>span one</span>
  <div>
    <span>span two</span>
  </div>
</div>

In this case, //div[@id='wrapper']//span[1] should be matching both span one and span two, because both are the first child of their respective parents. However, //div[@id='wrapper']/descendant::span[1] should only match span one, as it goes by the position within div[@id='wrapper'], not position within direct parent.

See xpath spec 2.5

NOTE: The location path //para[1] does not mean the same as the location path /descendant::para[1]. The latter selects the first descendant para element; the former selects all descendant para elements that are the first para children of their parents.

[yayuyokitano]

Very likely tightly related observation that may or may not be an issue: following-sibling does not seem to take into account context.

Example:

<div id="wrapper">
  <a>anchor 1</a>
  <div>div 1</div>
  <a>anchor 2</a>
  <div>div 2</div>
  <div>div 3</div>
</div>

If I here run a queryall on //div[@id='wrapper']/a/following-sibling::div[1], I expected that both div 1 and div 2 are returned, as there are two matching anchors, both with their own first following sibling. However, only one match (div 1) seems to be returned.

However, I am not entirely sure if this is even an issue, as in initial checking I did not see this as clearly spelled out in the xpath spec as the main issue (though I did not read through it very thoroughly either). All I know is that the way I expected it to work is how the chromium devtools xpath implementation works.