使用 xpath 检索空节点和非空节点

Question

I am trying to get a good representation of an XML... To keep it simple, let's say we have the following XML

<div>
    <em>5</em>
    <em></em>
    <em></em>
    <em>A</em>
</div>

Ideally I would like to convert this to some table having one column:

| em |
------
| "5"| 
| "" |
| "" |
| "A"|

(I used quotes here to clearly show that I want the empty nodes as well)

I tried several xpath queries.. the easiest one is something I tested with R, here I would get

z = read_xml("<div>
        <em>5</em>
        <em></em>
        <em></em>
        <em>A</em>
</div>")
z

xml_find_all(z,"//*[name() = 'em']/text()")

{xml_nodeset (2)}
[1] 5
[2] A

Most other questions are about only detecting empty/non-empty cells.. or selecting the first non-empty one.. but I don't see how I can use that here.

One idea I had was trying to use concat... to add some string to all nodes (including the empty ones). However, this is an Xpath 2.0 solution (AFAIK) and this will not be a viable solution.

The final solution (extracting information from this XML) will be implemented in Hive. I use some Serde functionality to get the information.. which is then stored as arrays.. then I want to convert it to a normal table.. but this is not possible if the missing values are not retrieved because of length differences

Answer 1

in R you can do:

library(xml2)
library(magrittr)
z = read_xml("<div>
             <em>5</em>
             <em></em>
             <em></em>
             <em>A</em>
        </div>")
z %>% 
    xml_find_all('em') %>% 
    xml_text()

#> [1] "5" ""  ""  "A"

Or, without the piping:

library(xml2) 
xml_text(xml_find_all(z, 'em'))
#> [1] "5" ""  ""  "A"

Answer 2

It is possible to do it in Hive with xpath(). Unfortunately Hive implements xpath 1.0. Therefore functions that would help deal with missing values in a more elegant way are not accessible.

The only way I could deal with it is by using an 'or' statement in the xpath expression that would output an default value when xpath value is empty. In your case there is no default element, so I create one with regexp_replace():

select xplode.*
     from (select 0) t
     lateral view explode(xpath(regexp_replace('<div><em>5</em><em></em><em></em><em>A</em></div>', '<em>','<em dflt = "">'),'div/em/text()| div/em[not(./text())]/@dflt')) xplode  as em;