The Solr web service allows you to return results in either XML format or JSON (the format is XML by default; to return results in JSON format, add the parameter “wt=json” to your Solr query string). Suppose we run a query on the Solr collection myindex. There are three ways we can search this collection: (1) we could use the <cfsearch> tag; (2) we could use the Solr web service and return XML; or (3) we could use the Solr web service and return JSON. We’ve talked about <cfsearch> and its limitations before so I’ll limit this discussion to the second and third options.

XML:

<cfhttp url="http://localhost:8983/solr/myindex?q=*:*&fl=*,score" result="myresult" />

JSON:

<cfhttp url="http://localhost:8983/solr/myindex?q=*:*&fl=*,score&wt=json" result="myresult" />

In each case (XML or JSON), ColdFusion has powerful functions that help you manage the data without needing to parse it manually. Typically I create a friendly query that I can return to the front-end developer that he can use in place of a SQL query or <cfsearch> result.

<cfset search_results = queryNew("id,title,description,pubdate,journal_name,author_name,num_reads,score"
                               , "CF_SQL_INTEGER,CF_SQL_VARCHAR,CF_SQL_CLOB,CF_SQL_DATE,CF_SQL_VARCHAR,CF_SQL_VARCHAR,CF_SQL_INTEGER,CF_SQL_DOUBLE")
/>
<!--- We're using our articles index mentioned in a previous post --->

XML:

<cfif isXML(myresult)>
    <cfset xml_result = XMLParse(myresult.fileContent) />
    <!--- continue processing --->
</cfif>

JSON:

<cfif isJSON(myresult)>
    <cfset json_result = deserializeJSON(myresult.fileContent) />
    <!--- continue processing --->
</cfif>

In either case we will iterate over our results structure and put the results in our friendly query.

JSON:
The JSON structure is particularly friendly for our purposes. First, let’s get the number of results and the maximum score:

<cfset result_cnt = json_result.response.numFound />
<cfset max_score = json_result.response.maxScore />

These fields aren’t absolutely necessary, of course. In particular, maxScore seems superfluous but it might be useful if you’re converting from Verity to Solr and need scores in a % format. The Verity search engine scores results between 0 and 1, which is easily converted to a percentage; while Solr has no maximum score. So to get a Verity-style score, simply divide each result’s score by maxScore.

Now we simply loop over the docs array of the response:

<cfloop array="#json_result.response.docs#" index="doc">
    <cfif structKeyExists(doc, "id") AND structKeyExists(doc, "title")>
        <!--- Both id and title are absolutely required --->
        <cfset queryAddRow(search_results) />
        <cfset querySetCell(search_results, "id", doc[id]) />
        <cfset querySetCell(search_results, "title", doc[title]) />
        <cfset querySetCell(search_results, "score", doc[score] / max_score) />
        <cfif structKeyExists(doc, "description")>
            <cfset querySetCell(search_results, "description", doc[description]) />
        </cfif>
        <cfif structKeyExists(doc, "pub_date")>
            <cfset querySetCell(search_results, "pub_date", doc[pub_date]) />
        </cfif>
        <cfif structKeyExists(doc, "journal_name")>
            <cfset querySetCell(search_results, "journal_name", doc[journal_name]) />
        </cfif>
        <cfif structKeyExists(doc, "author_name")>
            <cfset querySetCell(search_results, "author_name", doc[author_name]) />
        </cfif>
        <cfif structKeyExists(doc, "read_cnt")>
            <cfset querySetCell(search_results, "read_cnt", doc[read_cnt]) />
        </cfif>
    </cfif>
</cfloop>

Dealing with XML results is a bit more complicated so I will discuss that in a subsequent post or in an update to this one. Suffice it to say that the approach will be very similar – we will iterate over the results and store them in a friendly query.