Question

I found that the following code worked on a small set of my data, but I didn’t realize that I hadn’t taken any samples with multiple comments. When I tried to apply the code to the actual database, which has multiple comments per entry, I received the error mentioned above.

Current code:

for $doc in doc('test')
let $results :=
(
  let $pKeywords := ('best clients', 'Very', '20')
  return
    for $kw in $pKeywords
    return
    (
      $doc/set/entry[contains(comment, concat('!', $kw))],
      $doc/set/entry[contains(comment, $kw)]
    )
  [not(position() gt 2)]
)
for $i in (1 to count($results))
return
(
  subsequence($results/comment, $i, 1),
  subsequence($results/buyer, $i, 1)
)

Document:

<set>
  <entry>
    <comment>The client is only 20 years old.  Do not be surprised by his youth.</comment>
    <buyer></buyer>
    <id>1282</id>
    <industry>International Trade; Fish and Game</industry>
  </entry>
  <entry>
    <comment>!On leave in October.</comment>
    <comment>!Planning to make a large purchase before Christmas.</comment>
    <buyer></buyer>
    <id>709</id>
    <industry>Real Estate</industry>
  </entry>
    <entry>
    <comment>Is often !out between 1 and 3 p.m.</comment>
    <buyer></buyer>
    <id>127</id>
    <industry>Virus Software Marketting</industry>
  </entry>
  <entry>
    <comment>Very personable.  One of our best clients.</comment>
    <buyer></buyer>
    <id>14851</id>
    <industry>Administrative support.</industry>
  </entry>
  <entry>
    <comment>!Very difficult to reach, but one of our top buyers.</comment>
    <comment>His wife often answers the phone.  That means he is out of the office.</comment>
    <buyer></buyer>
    <id>1458</id>
    <industry>Construction</industry>
  </entry>
  <entry>
    <comment></comment>
    <buyer></buyer>
    <id>276470</id>
    <industry>Bulk Furniture Sales</industry>
  </entry>
  <entry>
    <comment>A bit of an eccentric.  One of our best clients.</comment>
    <buyer></buyer>
    <id>1506</id>
    <industry>Sports Analysis</industry>
  </entry>
  <entry>
    <comment>Very gullible, so please !be sure she needs what you sell her.  She's one of our best clients.</comment>
    <buyer></buyer>
    <id>1523</id>
    <industry>International Trade</industry>
  </entry>
  <entry>
    <comment>He wants to buy everything, but !he has a tight budget.</comment>
    <comment>!His company may be closing soon.</comment>
    <buyer></buyer>
    <id>1524</id>
    <industry>Public Relations</industry>
  </entry>
</set>

The result:

Stopped at line 9, column 22: [XPTY0004] document-node()(...): function(item()*) as item()* expected, document-node() found.

I ran into a similar error and was able to fix it, but when I try to apply the fixes, that did not work. Example:

  $doc('test')/set/entry[contains(., concat('!', $kw))],
  $doc('test')/set/entry[contains(., $kw)]

returns the same result.

Walking through the desired result:

The first return should return every entry and its children if the entry’s comment child contains any of the three keywords in $pKeywords.

concat('!', $kw) is supposed to make !-containing comments the priority.

The second return slices the comment and buyer nodes from the results of the first return.

As long as there is exactly 1 comment-named node in every entry, the code executes fine. When there are 2 or more comment-named nodes, the code fails, and the compiler returns the error mentioned above:

Stopped at line 9, column 22: [XPTY0004] document-node()(...): function(item()*) as item()* expected, document-node() found.

-Edit-

Desired result:

<comment>The client is only 20 years old.  Do not be surprised by his youth.</comment>
<buyer/>
<comment>Very personable.  One of our best clients.</comment>
<buyer/>
<comment>!Very difficult to reach, but one of our top buyers.</comment>
<buyer/>
<comment>A bit of an eccentric.  One of our best clients.</comment>
<buyer/>

Clarifying the desired result:

//contains ! and the first keyword, "best clients"; so, the first result should come from this entry.
  <entry>
    <comment>Very gullible, so please !be sure she needs what you sell her.  She's one of our best clients.</comment>
    <buyer></buyer>
    <id>1523</id>
    <industry>International Trade</industry>
  </entry>

//Only one entry contains ! and "best clients".  So, the first result containing "best clients" contains nodes for the second result.
  <entry>
    <comment>Very personable.  One of our best clients.</comment>
    <buyer></buyer>
    <id>14851</id>
    <industry>Administrative support.</industry>
  </entry>

//This contains ! and the second keyword, "Very", but it is a duplicate.  So, ideally its children should not be returned.
  <entry>
    <comment>!Very difficult to reach, but one of our top buyers.</comment>
    <comment>His wife often answers the phone.  That means he is out of the office.</comment>
    <buyer></buyer>
    <id>1458</id>
    <industry>Construction</industry>
  </entry>

//This contains ! and a string, "very" (part of everything).  Nodes from this entry should be returned as the third result.
  <entry>
    <comment>He wants to buy everything, but !he has a tight budget.</comment>
    <comment>!His company may be closing soon.</comment>
    <buyer></buyer>
    <id>1524</id>
    <industry>Public Relations</industry>
  </entry>

//The only entry whose comment child contains the keyword '20'.  There is no '!'-containing comment with 20, so this nodes is the top and only node whose children should be returned.
  <entry>
    <comment>The client is only 20 years old.  Do not be surprised by his youth.</comment>
    <buyer></buyer>
    <id>1282</id>
    <industry>International Trade; Fish and Game</industry>
  </entry>

-Edit 2-

Next pass gives a better idea of what I’m trying to accomplish, but there are some obvious syntax errors (for example, I’m still discovering how to work with arrays, as seen on line 8). I will update this as I resolve the syntax errors:

<set>
{
    let $kw := ('best clients', 'Very', '20')
    let $entry := doc('test')/set/entry
    let $priority := '!'

    for $i in (1, count($kw))
    let $priority_result[$i] :=
    (
        for $entries in $entry
        where $entry contains(., $priority) and where $entry contains $kw[$i]
        return subsequence($priority_result[$i], 1, 2)
    )

    if $priority_result[$i] < 2
    for $i in (1, count($kw))
    let $secondary_result[$i] :=
    (
        for $entries in $entry
        where $entry contains $kw[$i] and where $entry not($priority_result) and where $entry not($secondary_result[1..($i-1)])
        return $secondary_result[$i]
    )
    else let $secondary_result[$i] := ''

    for $i in (1, count($kw))
    return
    (
        $primary_result[$i],
        $secondary_result[$i]
    )
}
</set>

And the suggested change, which returns a null result:

for $doc in doc('test')
let $results :=
(
  let $pKeywords := ('best clients', 'Very', '20')
  return
    for $kw in $pKeywords
    return
    (
      $doc/set/entry/comment[contains(., concat('!', $kw))],
      $doc/set/entry/comment[contains(., $kw)]
    )
  [not(position() gt 2)]
)
for $i in (1 to count($results))
return
(
  subsequence($results/comment, $i, 1),
  subsequence($results/buyer, $i, 1)
)

Answer 1

The error message seems to be complaining about trying to call a document-node() as a function.

$doc('test') vs $doc

---

Either that, or comments(...) only works for a single node, not a node-set.

contains(comment, $kw) vs comment/contains(.,$kw)
or comment[contains(.,$kw)]
or comment[contains(text(),$kw)]

---

This worked for me:

<set>{
    for $entry in doc('test')/set/entry
    let $kw := (
        for $prefix in ('!','')
        for $kw in ('best clients', 'Very', '20')
        where exists($entry/comment[contains(., concat($prefix,$kw))])
        return concat($prefix,$kw)
    )[1]
    where exists($kw)
    order by not(starts-with($kw,'!'))
    return <entry keyword="{$kw}">{
      ( $entry/comment,
        $entry/buyer )
    }</entry>
}</set>

Result (multiple comments per <entry>):

<set>
   <entry keyword="!Very">
      <comment>!Very difficult to reach, but one of our top buyers.</comment>
      <comment>His wife often answers the phone.  That means he is out of the office.</comment>
      <buyer/>
   </entry>
   <entry keyword="20">
      <comment>The client is only 20 years old.  Do not be surprised by his youth.</comment>
      <buyer/>
   </entry>
   <entry keyword="best clients">
      <comment>Very personable.  One of our best clients.</comment>
      <buyer/>
   </entry>
   <entry keyword="best clients">
      <comment>A bit of an eccentric.  One of our best clients.</comment>
      <buyer/>
   </entry>
   <entry keyword="best clients">
      <comment>Very gullible, so please !be sure she needs what you sell her.  She's one of our best clients.</comment>
      <buyer/>
   </entry>
</set>

---

This will give you separate entries for each comment:

<set>{
    for $entry in doc('test')/set/entry
    for $comment in $entry/comment
    let $kw := (
        for $prefix in ('!','')
        for $kw in ('best clients', 'Very', '20')
        where exists($comment[contains(., concat($prefix,$kw))])
        return concat($prefix,$kw)
    )[1]
    where exists($kw)
    order by not(starts-with($kw,'!'))
    return <entry keyword="{$kw}">{
      ( $comment,
        $entry/buyer )
    }</entry>
}</set>

Output:

<set>
   <entry keyword="!Very">
      <comment>!Very difficult to reach, but one of our top buyers.</comment>
      <buyer/>
   </entry>
   <entry keyword="20">
      <comment>The client is only 20 years old.  Do not be surprised by his youth.</comment>
      <buyer/>
   </entry>
   <entry keyword="best clients">
      <comment>Very personable.  One of our best clients.</comment>
      <buyer/>
   </entry>
   <entry keyword="best clients">
      <comment>A bit of an eccentric.  One of our best clients.</comment>
      <buyer/>
   </entry>
   <entry keyword="best clients">
      <comment>Very gullible, so please !be sure she needs what you sell her.  She's one of our best clients.</comment>
      <buyer/>
   </entry>
</set>