Thanks Slace,

For the larger query mentioned in 1, would the .Skip & .Take alleviate any performance issue?

I'm just looking up the GatheringNodeData now - are there any online examples of its implementation with Umbraco?

Skip and Take have built-in performance considerations, hydration of items from the Lucene index doesn't happen until they are required, so if you skip 100 results you wont have them accessed from Lucene.

For GatheringNodeData you should check out shan's code demo, that's all the doco we've got so far (we're a bit behind in the writing of it)

Hi, can anyone tell me how i make the indexer store text in its orginal case for example a property called name containing "John Doe" currently is stored as "john doe" how can i retain the original casing in the lucene index.

You have to store both a case sensitive and case insensitive data as Lucene isn't really designed for data retrieval.

To do this with Examine you have to attach to the UmbracoExamine.LuceneExamineIndexer.DocumentWriting event (which may have moved into the LuceneEngine with the latest check ins, I'm not 100% sure).

This event is fired in a Lucene-scope as provides you with access to the Lucene Document object as it's being written to, and in which you'll need to add your un-analyzed version of the content.

Here's an example of how we did it in a recent project for showing in search results:

void indexer_DocumentWriting(object sender, DocumentWritingEventArgs e)
{
    var doc = e.Document;

    // Find the title
    string title = !e.Fields.ContainsKey("PageTitle") || string.IsNullOrEmpty(e.Fields["PageTitle"]) ? e.Fields["nodeName"] : e.Fields["PageTitle"];

    // Default content is nothing:
    string content = string.Empty;
    // Unless a description is found:
    if (e.Fields.ContainsKey("Description") && !string.IsNullOrEmpty(e.Fields["Description"]))
    {
        content = e.Fields["Description"];
    }
    // Or BodyContent is found:
    else if (e.Fields.ContainsKey("BodyContent") && !string.IsNullOrEmpty(e.Fields["BodyContent"]))
    {
        content = e.Fields["BodyContent"];
    }

    // Store the title and content with text casing unchanged
    doc.Add(new Field("__PageTitle", title, Field.Store.YES, Field.Index.NOT_ANALYZED));
    doc.Add(new Field("__Content", content, Field.Store.YES, Field.Index.NOT_ANALYZED));
}

Hi cant get any of this to work as attaching to events seems to be broken any ideas ?

When/ How are you attaching the events?

I blogged this last night: http://farmcode.org/post/2010/08/23/Text-casing-and-Examine.aspx

Hi,

I am (still) very new to Umbraco.

As some of the comments say on the umbraco.tv session called "Adding a search to your website", it would be very nice to have some examples on how to "further tweak the results to include page excepts (with possible keyowrd highlight), paging and any other common scenarios that would be required in a search for a site."

I have this exact requirement: showing a text snippet from the "hit" document(s) that contains the word/phrase - even highlightning would be a bonus! :-)

From a rather reliable source a have got a hint regarding IDictionary which should hold the indexed user fields and by which you should be able to access items containing the search term.

I will try this out, but maybe there's a bunch of other/better ways to accomplish this..(?)

Please knock me over with all your great ideas and - even better - examples :-)

Cheers

Folke

Folke,

There are contributions to the lucene project that do some of the above, Highlight which highlights your search term match in result set.  Paging you need to write something yourself I have a .net control to do it.  I highly recommend the lucene in action edition 2 book that covers alot of the contrib stuff.

Regards

Ismail

Hi Ismail,

I just might buy that book to really understand Lucene and it's possibilities.

An other thing that buzzles me is the fact that this site specifies the use of operators like "AND", "OR", "+", "-" and a lot more in the search query. But that doesn't seem to work via Examine or have I just totally missed it? If I use an "OR"-operator between two words that I know exist I get a zero hit - like:

byggeri OR tag

gives me 0 hits while the words byggeri and tag (in two seperate searches) both give me hits.

Doesn't Exmamine support these operators or do I just use them ... erh.. a bad/stupid way?

Kind regards

Folke

You would need to use the Lucene.Net contrib library for highlighting, and you may have to implement your own searcher to make it completely work (just overriding the built-in one should be sufficient). I've never had the need on a project so it's not something I've really looked into and since adding the contrib library adds another dependency we wont be doing it out of the box.

As for using the Lucene query syntax you have to pass that into the Fluent API's "RawQuery" method. This is because adding search using proper query-parser syntax adds overhead and it's a fairly advanced task so it's left to a more advanced API call.

Lastly check out this article for how to do paging: http://farmcode.org/post/2010/08/18/Paging-with-Examine.aspx

Hi Slace,

Thanks a lot for your comments. I actually had the feeling that I needed to "transfer" or parse the query into the Fluent API to make it work. Very nice to have that settled :-)

My "own searcher".. well that would be an interesting (challenging) task for me to accomplish, but I'll give it a go as well with the pagination and have some og my more experienced fellow devs. help me along the way :-)

Thanks a lot! :D

Cheers

Folke

From what I recall of the highlighter it's just a custom analyzer, so you may be able to get away without having to do anything but specify it in the Examine config (I'm not that far into the book yet, Ismail is a lot further than I am, I just skimmed that chapter and then started from page 1 of the book :P).

The only reason I can think you'd need your own searcher is because of how the results are retrieved and returned.

is working on a reply...