Search code examples
marklogic

Marklogic Search options to restrict search for a JSON nested property

I have a json which has the following structure 

{"scientist":{
         "username": "XXX",
         "name":"XXXX"
          ...
    },
    "registrar":{
      "username": "YYY",
         "name":"aaaaa"
          ...
    }
   }

I want to restrict the search to search a nested property in the above json for eg: -- want to search for "registrar/username".. I was trying to use the search -constraint "container".. but unable to do the search.. following is my code, and it does not seem to work..

Is there any other mechanism for the nested json field search ?

```

let $search := "(username:WHITEMK) AND (concept.registrar: )"

let $options := 
  <options xmlns="http://marklogic.com/appservices/search">
    <constraint name="concept.registrar">
      <container>
        <json-property>registrar</json-property>
      </container>
    </constraint>
    <constraint name="username">
      <value>
        <json-property>username</json-property>
        <term-option>case-insensitive</term-option>
        <term-option>wildcarded</term-option>
      </value>
    </constraint>
    <return-facets>false</return-facets>
    <return-values>false</return-values>
    <return-constraints>false</return-constraints>
    <return-frequencies>false</return-frequencies>
    <return-qtext>false</return-qtext>
    <search-option>unfiltered</search-option>
    <search-option>unfaceted</search-option>
    <search-option>format-json</search-option>
    <search-option>score-simple</search-option>
  </options>

let $start := 1
let $page-length :=10000

return search:search($search, $options, $start, $page-length)

```

Thanks, Ravi

Solution

  • I was able to get a more generic version without adding nested fields is by building a custom search options, which parses the nested JSON property and builds the search query using cts:json-property-scope-query..
    for eg: Let us say you want to search for some 3 level nested property

    let $search := "(concept:orfs.aminoAcids.predictedMatureSeqs.domains.heavyChainIsoType:igg1)"

    Using custom parser I was able to transform this to the following

    <cts:json-property-scope-query xmlns:cts="http://marklogic.com/cts">
<cts:property>orfs</cts:property>
<cts:json-property-scope-query>
<cts:property>aminoAcids</cts:property>
<cts:json-property-scope-query>
<cts:property>predictedMatureSeqs</cts:property>
<cts:json-property-scope-query>
<cts:property>domains</cts:property>
<cts:json-property-scope-query>
<cts:property>heavyChainIsoType</cts:property>
<cts:word-query>
<cts:text xml:lang="en">igg1</cts:text>
<cts:option>case-insensitive</cts:option>
<cts:option>punctuation-insensitive</cts:option>
<cts:option>whitespace-insensitive</cts:option>
<cts:option>wildcarded</cts:option>
</cts:word-query>
</cts:json-property-scope-query>
</cts:json-property-scope-query>
</cts:json-property-scope-query>
</cts:json-property-scope-query>
</cts:json-property-scope-query>

    If anyone is interested following is the code for the custom parser

    xquery version "1.0-ml";
module namespace gbrsso="http://marklogic.com/gbrs/modules";
import module namespace search = "http://marklogic.com/appservices/search" at "/MarkLogic/appservices/search/search.xqy";
(:
   This module tries to build a custom constraint.. and takes into account the nested query as well
   for eg: Concept:registrar.username:mamidrx
:)

(: This is parse function which is invoked my marklogic when it sees the custom constraing :)
declare function gbrsso:parse($constraint-qtext as xs:string, $right as schema-element(cts:query)) as schema-element(cts:query){
 let $log := xdmp:log("Constraint-qtext : " || $constraint-qtext)
 let $log := xdmp:log("$right : " || fn:string($right//cts:text/text()))
 let $queryText := fn:string($right//cts:text/text())
 let $qparts := fn:tokenize($right//cts:text/text(), ":")
 let $queryText := $qparts[last()]
 let $log := xdmp:log("$queryText : " || $queryText)

  let $qpartsCnt := fn:count($qparts)

  let $query :=
      if(fn:count($qparts) = 1) then
        <root>{
          cts:word-query($queryText, ("case-insensitive", "wildcarded", "punctuation-insensitive", "whitespace-insensitive"))
          }</root>/*
      else 
        <root>{
          let $qparts := fn:tokenize($qparts[1],"\.") 
          let $endPart := $qparts[last()]
          (: remove the last part in sequence as we need to create word query with it :)
          let $qparts := fn:remove($qparts, fn:count($qparts))
          let $queryExp :=
            if(fn:count($qparts) = 0) then (: checks if it is nested... :)
               cts:json-property-scope-query($endPart, cts:word-query($queryText, ("case-insensitive", "wildcarded", "punctuation-insensitive", "whitespace-insensitive")))
            else
              let $xy := cts:json-property-scope-query($endPart, cts:word-query($queryText, ("case-insensitive", "wildcarded", "punctuation-insensitive", "whitespace-insensitive")))
              return gbrsso:buildQuery($xy, $qparts)

          return $queryExp
        }</root>/*

  return $query

};