This solution was built on 10.2.1.1, but should work on all versions after 10.2. Versions prior to 10.2 will need some rewriting as this is based on the Prompt API.

One of the biggest complaints I get about search prompts is the amount of real estate they take from reports. The multiselect prompts are huge, and users hate them for it. So I was asked to find a way to shrink it down.

Obviously the solution was to everything but the search input and button. The keyword and instruction text can be controlled by the properties, simply setting it to “Specified Text” is sufficient to hide them:

That leaves the Options link and the search results. Personally I don’t believe the results list need to appear unless there are actually results. The options should be at the mercy of the report author, so my solution includes a way to choose a default value and to hide the options.

First a look at the JS

/* function paulScripts.advancedSnS - Paul Mendelson - 2013-09-16
 * Will convert a standard search and select. It will add a button to expand/collapse the 
 * are selection area. The button will display how many matches have been found, and how 
 * many selected. 
 * 
 * The Option param will control the default search method. 
 * 1 = Starts with any of these keywords **DEFAULT**
 * 2 = Starts with the first keyword and contains all of the remaining keywords
 * 3 = Contains any of these keywords
 * 4 = Contains all of these keywords
 *
 * Hide the options link by setting 1 in the hideOption param.
 */ 
paulScripts.advancedSnS = function(promptName,option,hideOption){
  var prompt = paulScripts.getControl(promptName)
    , elm = document.getElementsByName(prompt.getParameterName())[0].parentNode
    , selectTable = elm.getElementsByTagName('table')[4]
    , selectDiv = document.createElement('div')
    , selectOptions = cognos.Report.getReport("_THIS_").prompt.getControlByName(promptName).getSelectOptions()
    , selectChoices = cognos.Report.getReport("_THIS_").prompt.getControlByName(promptName).getSelectChoices()
    , tr = elm.getElementsByTagName('button')[0].parentNode.parentNode.parentNode.appendChild(document.createElement('tr'))
    , td= tr.insertCell(0)
    , slideDiv = document.createElement('div')
    , btn = document.createElement('input');


    if(option) elm.getElementsByTagName('input')[option+2].checked=true;
    if(hideOption) tr.parentNode.parentNode.parentNode.parentNode.nextSibling.style.display='none';
    td.setAttribute('colSpan',2);

//Hacky – Building a new dispatcher request to save the state of the SnS. Do any IBMers have any comments? 
  if(typeof oReq ==='undefined') oReq = new ViewerDispatcherEntry(window['oCV'+'_THIS_'].getRV().getCV());
  if(!oReq.getFormFields().m_aValues[promptName]) oReq.addFormField(promptName,'hidden')

  btn.type='button';
  btn.value='';

  if(!selectOptions&&!selectChoices) btn.value='click to open';
  if(selectOptions)  btn.value = selectOptions.length + ' found. ';
  if(selectChoices)  btn.value += selectChoices.c.length + ' selected.';

  selectTable.parentNode.insertBefore(selectDiv,selectTable);
  selectDiv.appendChild(selectTable);
  
  //the right side of the table is a bit ugly; not enough padding.
  selectDiv.style.paddingRight='5px';

  // Sets the table of the select options to position absolute. This will prevent the page from expanding vertically
  // when the table is visible;
  selectDiv.style.position='absolute';
  selectDiv.style.overflow='hidden';
  selectDiv.style.backgroundColor='white';

  btn.onclick = function() {
    if(selectDiv.style.visibility=='hidden')
      {slidedown(selectDiv);
oReq.addFormField(promptName,'visible');}
    else {slideup(selectDiv);oReq.addFormField(promptName,'hidden');}
      return true;

  }

  selectDiv.style.visibility=oReq.getFormFields().m_aValues[promptName];
  td.appendChild(btn);
  return true;
}

The JS itself is fairly straight forward. The variables define the DOM element, a JSON array of the found options, another JSON array of the selected objects, and a couple of generated elements.

If a default search option is selected, it will set that.

if(option) elm.getElementsByTagName('input')[option+2].checked=true;

If hideOption is set to true, it will hide the options link.

 if(hideOption) tr.parentNode.parentNode.parentNode.parentNode.nextSibling.style.display='none';

Then it will insert a button, and finally the “results” lists are moved into a generated div which can then be slid open or shut.

  td.appendChild(btn);

Whenever I create a solution that changes the appearance of the page, showing or hiding an element, the users demand that the change persist through a page reset. The easiest way would be to create a new parameter to store the display state. Possibly a hidden text box with box type set to none. But if we always did things the easy way, we’d never learn anything.

 //Hacky – Building a new dispatcher request to save the state of the SnS. Do any IBMers have any comments? 
  if(typeof oReq ==='undefined') oReq = new ViewerDispatcherEntry(window['oCV'+'_THIS_'].getRV().getCV());
  if(!oReq.getFormFields().m_aValues[promptName]) oReq.addFormField(promptName,'hidden')

I haven’t tested this on a wide scale, so I don’t know of any performance impact. To be honest, while I love figuring out little hacks like these, without a more in-depth understanding of what it’s doing, it may not be wise to push this to all reports. The attached report contains two functions, the first using the hacky dispatcher request, and the second using a secondary text box to retain the state. The disadvantage of using the text box is that you’ll need to remember to add that in.

Once the JS is in place, we can call the functions with a simple:

paulScripts.advancedSnS ('Retailers', 3,1);

It will look for the prompt named Retailers, set the option to “Contains any of these keywords”, and hide the options link.

And the final product:

Report XML:
Advanced SnS (4749 downloads)

I’ve been somewhat negligent in responding to the comments lately. One of the most common questions appears to be how to change the default option in a select and search prompt.

The following should work for all versions of Cognos.

Put an HTML item before the prompt, with the expression:

<div id="mySnS">

Now put another HTML item after the prompt with the expression:

</div>
<script>
/* Select and search options
 * 2 - Starts with any of these keywords * DEFAULT
 * 3 - Starts with the first keyword and contains all of the remaining keywords
 * 4 - Contains any of these keywords
 * 5 - Contains all of these keywords
 */
document.getElementById('mySnS').getElementsByTagName('input')[5].click();
</script>

The script will locate the mySnS element that surrounds the prompt, generate an array of the input tags, and click() on the one specified. In the example, 5 is selected so it will click on the sixth element (0 based array).

Going through the Cognos JavaScript files, it looks like there is supposed to be a function in the new Prompt API that would let you do it in an easier manner, unfortunately it appears to be an incomplete function, maybe it will be released in the new fixpack.

In theory, you should be able to do something like the following:

var acme = {};
acme.getControl = function(promptName)
{
  var ocr = cognos.Report.getReport("_THIS_");
  return ocr.prompt.getControlByName(promptName);
};

acme.getControl('mySnS2').setProperty( "caseInsensitive", false);
acme.getControl('mySnS2').setProperty( "searchType", "containsAny");

But, as I said, the function doesn’t appear to be complete in this version. Use the other method for now.