Knowledgebase: MarkLogic Server
XPath not() with path fields
22 November 2016 07:01 PM

Introduction

Fields are a great way of restricting what parts of your documents to search, based on XML element QNames or JSON propertyNames .  Fields are extremely useful when you have content in one or more elements or JSON properties that you want to query simply and efficiently as a single unit. But can you use field names you've created with XPath's fn:not ()? In other words, given a field name "test-field-name" can you do something like fn:not(//test-field-name)? Unfortunately, you can not, as the server will return an XDMP-UNINDEXABLEPATH error. There is, however, a workaround.

Workaround

The workaround is to create two fields, then to query across those two fields using cts:not-in-query ( http://docs.marklogic.com/cts:not-in-query), Consider two documents:

Document 1

xdmp:document-insert(
  "/test/fields-001.xml",
  <doc>
      <content>
          <courtcase>
              <metadata>
                  <docinfo>
                      <hier>
                          <hierlev>
                              <heading>
                                  <title>1900</title>
                              </heading>
                              <hierlev>
                                  <heading>
                                      <title>Volume 10</title>
                                  </heading>
                                  <hierlev>
                                      <heading>
                                          <title>test title - (1900) 10 Ch.D. 900</title>
                                      </heading>
                                  </hierlev>
                              </hierlev>
                          </hierlev>
                      </hier>
                  </docinfo>
              </metadata>
          </courtcase>
      </content>
</doc> ,
xdmp:default-permissions(),
("test", "fields")
)

Document 2

xdmp:document-insert(
  "/test/fields-002.xml",
<doc>
    <content>
        <courtcase>
              <metadata>
                     <docinfo>
                            <hier>
                                  <hierlev>
                                       <heading>
                                              <title>1879</title>
                                        </heading>
                                        <hierlev>
                                            <heading>
                                                  <title>John had a little lamb</title>
                                            </heading>
                                            <hierlev>
                                                  <heading>
                                                        <title>Mary had a little lamb</title>
                                                  </heading>
                                              </hierlev>
                                         </hierlev>
                                   </hierlev>
                               </hier>
                          </docinfo>
                     </metadata>
                </courtcase>
            </content>
</doc> ,
xdmp:default-permissions(),
("test", "fields")
)

Say you're interested in three different paths:

1) All titles, Which Should be defined as fn:collection()//heading/title

2) Titles with lower-level titles, Which Should be defined as fn:collection()//hierlev[.//hierlev/heading/title]/heading/title

3) Titles with NO lower-level titles, Which Should be defined as fn:collection()//hierlev[fn:not(.//hierlev/heading/title)]/heading/title

Unfortunately, while we can express #3 in full XPath, we can not express #3 in the subset of XPath used to describe path fields. However, you can emulate #3 by defining fields corresponding to #1 & #2, then combining them in a cts:not-in-query.

Create the path fields
  1. All titles

  Create a Path Field with name "titles-all" path "//heading/title"

  1. Titles with lower-level titles

  Create a Path Field with name "titles-with-lower-level titles," path "//hierlev[.//hierlev/heading/title]/heading/title"

Emulate the XPath you want by combining these two newly created path fields in a cts: not-in-query ()

for $doc in cts:search(

  fn:collection("fields"),

  cts:not-in-query(

    cts:field-word-query(

      "titles-all",

      $term

      ) ,

    cts:field-word-query(

      "titles-with-lower-level-titles",

      $term

      )

    )

  )

return

  xdmp:node-uri($doc)

(0 vote(s))
Helpful
Not helpful
Comments (0)