Another Method of Expansion
But what if, on a rare occasion, you do need to get extra search results, but don’t want to modify the configuration file so as to affect all searches? This post demonstrates some simple IDOC code that will expand your search results.Sample HCSP
For this example, we’ll put the code in a simple HCSP page. The same strategies would work just as well inside fragments, components, or other code sources.Create a basic SearchResults.hcsp page. I always like to put in some basic HTML code first to aid with display. Once you have that set up, let’s start with the parameters for the search:
<!--$ QueryText = “” -->
<!--$ desiredResults = 500 -->
<!--$ maxResultsPerSearch = 200 -->
The QueryText variable is required to execute the search service. By setting it to blank, we are getting all documents in the system. You can put in your own properly formatted query here if you like.
The desiredResults variable is how many results we’d really like to see. The maxResultsPerSearch variable is set to 200, the default maximum that we’ll get from a single search. We can expand this code a little to accommodate a custom setting here:
<!--$ if MaxResults and MaxResults gt 0 -->
<!--$ maxResultsPerSearch = MaxResults -->
<!--$ endif -->
This tests for the presence of the MaxResults configuration setting and, if it finds one, will swap that value in for maxResultsPerSearch.
We’re going to need one more variable to keep track of how many total search results we’ve found to date.
<!--$ runningTotal = 0 -->
Now that we’ve got these variables established, we can get our search results. Remember, each time we run a search, we’ll be capped at 200 results, but we want to get 500. This means we’ll need to run two 200-result searches and one 100-result search. We’ll accomplish this by executing the search inside a loop.
<!--$ loopwhile runningTotal lt desiredResults -->
What we’re setting up here is a method to keep track of the total number of results we get. As long as that number is fewer that the desiredResults, we have more searching to do.
We need to add some additional search parameters that are specific to each iteration of the search execution:
<!--$ ResultCount = desiredResults - runningTotal -->
<!--$ StartRow = runningTotal + 1 -->
I’m cheating a little bit on the ResultCount here. Theoretically, this number could be well in excess of the search result cap. I’m relying on Oracle’s internal processes to enforce the limit of 200 documents. The StartRow will need to be different each time I run the search so that I don’t get duplicates.
Now that this is all set up, I can run the service and print the results:
<!--$ executeService("GET_SEARCH_RESULTS") -->
<!--$ loop SearchResults -->
<li><!--$ dDocTitle --></li>
<!--$ endloop -->
I have a touch of housekeeping to do to make sure the loop continues (and more importantly, ends) properly:
<!--$ runningTotal = runningTotal + SearchResults.#numRows -->
<!--$ if SearchResults.#numRows eq 0 -->
<!--$ break -->
<!--$ endif -->
I’m incrementing the runningTotal so I can keep track of my progress. I’m also checking for a zero result set. If that happens, then I’ve run out of results and need to break out of the loop.
All that’s left to do is end the loop!
<!--$ endloop -->
When you run this code, you should see 500 documents in your results.
Here is the code in its entirety (plus some minor formatting):
<html>
<head></head>
<body><!--$ QueryText="" -->
<!--$ desiredResults = 300 -->
<!--$ maxResultsPerSearch = 200 -->
<!--$ if MaxResults and MaxResults gt 0 -->
<!--$ maxResultsPerSearch = MaxResults -->
<!--$ endif -->
<!--$ runningTotal = 0 -->
<ol>
<!--$ loopwhile runningTotal lt desiredResults -->
<!--$ ResultCount = desiredResults - runningTotal -->
<!--$ StartRow = runningTotal + 1 -->
<!--$ executeService("GET_SEARCH_RESULTS") -->
<!--$ loop SearchResults -->
<li><!--$ dDocTitle --></li>
<!--$ endloop -->
<!--$ runningTotal = runningTotal + SearchResults.#numRows -->
<!--$ endloop -->
<ol>
<body>
<html>
One word of warning: this code will be executing a search service several times, so you may get some slower results.
This doesnt work, it works with a blank search, but as soon as you give it
ReplyDeleteor any type of query it doesnt work.
my comment got cropped. it is supposed to say:
ReplyDeleteThis doesnt work, it works with a blank search, but as soon as you give it
QueryText="dDoctype `Image`"
or any type of query it doesnt work.
The correct QueryText string in your above example should be:
ReplyDeleteQueryText="dDocType `Image`"
The sample code works fine for me.