Question

Let’s say i have 2 documents:

<doc1>
  <a>the dog</a>
  <a>the cat</a>
  <a>the human</a>
</doc1>

and

<doc2>
  <a>the dog</a>
  <a>foo</a>
  <a>bar</a>
</doc2>

Now for Marklogic, i want to search them for ‘the’ so I might run:

search:search(
  'a:the',
  <options xmlns="http://marklogic.com/appservices/search">
    <constraint name="a">
        <value>
            <term-option>case-insensitive</term-option>
            <element ns="" name="a"/>
        </value>
    </constraint>
  </options>
)

this works, but returns both documents (which is what i asked for)
But what if I wanted only results that had more than 1 match on the constraint…
ie. ‘give me the documents that have 2+ nodes with “the”‘

I have no idea where to start. Thanks!

Answer 1

The answers to your question at http://markmail.org/message/gaehhxnr7qb2un5p#query:+page:1+mid:vxjmjuh2wflofa67+state:results seemed good. Summarizing the thread, Rob and John both pointed out the min-occurs option for cts:search and Colleen added:

To enable that in the Search API you’d configure a word constraint on that element, adding <term-option>min-occurs=2</term-option>

Mary noted that the index lookups would be

…accurate for simple cases like this as long as you have the right positions enabled.

If you run xdmp:plan over it and see min-occurs in the final plan, you have
the right positions enabled.

Just to clarify: while Colleen mentioned a word constraint, http://docs.marklogic.com/cts:element-value-query shows the same option available for cts:element-value-query. So you should be able to use it with a value constraint too.