Examiness hints and tips from the trenches

Ismail Mayat
Senior Web Developer
@ The Cogworks

Examiness
Hints and tips from the trenches

What this talk is not
• How to install
• How to configure

What we will cover
• Tools to help you
• Hints and tips regarding indexing
• GatheringNodeData event is your friend!
• Indexing media (pdf,word etc)
• Deep in the bowels with DocumentWriting event
• Search highlighting
• Deployment to staging / production environments
• Faceting (Not exactly examine but still useful)
• Food for thought
• Questions and answers

Tools to help you
“Use the source Luke!”
http://code.google.com/p/luke/

Tools to help you
• http://luke.codeplex.com/ (.net port)
• Subset of common features present
• Scripting with Rhino missing etc

Using Luke
• Writing out generated queries to test in luke
var criteria = searcher.CreateSearchCriteria(IndexTypes.Content);

IBooleanOperation query = criteria.NodeTypeAlias("NewsItem");

query = query.Not().Field("umbracoNaviHide", 1.ToString());

var results = searcher.Search(query.Compile());
criteria.ToString();

Generates the following query
SearchIndexType: content, LuceneQuery: +(+__NodeTypeAlias:newsitem -umbracoNaviHide:1)
+__IndexType:content

Tools to help you
http://our.umbraco.org/projects/developer-tools/examine-dashboard

GatheringNode Data
• Examine has rich event system
• In all my implementations I have used
GatheringNode
– Merge into one contents field
– Searching on path
– Adding nodeTypeAlias field into pdf index

GatheringNode Data
Merge into contents field
• Example query
var query =
searchCriteria.Field("nodeName","hello").Or().Field("metaTit
le","hello").Field("metaDescription","hello").Compile();

GatheringNode Data
Merge to contents field
public class ExamineEvents:ApplicationBase {

public ExamineEvents() {

ExamineManager.Instance.IndexProviderCollection[Constants.ATGMainIndexerName].GatheringNodeDa
ta += ATGMainExamineEvents_GatheringNodeData;

}

void ATGMainExamineEvents_GatheringNodeData(object sender, IndexingNodeDataEventArgs e) {
AddToContentsField(e);
}

private void AddToContentsField(IndexingNodeDataEventArgs e) {

var fields = e.Fields;
var combinedFields = new StringBuilder();

foreach (var keyValuePair in fields) {
combinedFields.AppendLine(keyValuePair.Value);
}
e.Fields.Add("contents", combinedFields.ToString());
}
}

GatheringNode Data
Merge to contents field
• Query now looks like
query.Field(“contents”,”hello”)
• Adding new fields is just case of rebuild index

GatheringNode Data
Creating a searchable path
• Path is in index as 1,1056,1078 not tokenised
• Add new field with , replaced with space

GatheringNode Data
• How to query when no value e.g sql query like
select where value=‘’
• Select all
• Cannot do query like this in Examine / Lucene
• However can use GatheringNode data event
to inject in some arbitrary value then query on
that.
• E.g. field noData_Title value 1

GatheringNode Data
• Re Indexing errors
• MNTP field referencing a node that no longer
exists
• Use try catch and log the offending node

Document writing event
• You need lower level Lucene access
• E.g. boosting a field
• What is boosting? Not all documents are equal you need to artificially give
higher ranking to certain documents . When sort by is just not enough e.g.

– Person doc type. If they have important title they need to appear at
top of person search list
– Boost documents by age. Penalize older documents useful for news
and business documents.
– Boost based on unique views (would need to know up front also base
on time slots e.g last month, last week)
– Documents with more likes (custom like functionality)
– Tagging using XFS Term selector with weighting
http://our.umbraco.org/projects/website-utilities/xfs-term-selector

Document writing event
var indexer =
(UmbracoContentIndexer)ExamineManager.Instance.IndexProviderColle
ction[Constants.ATGMDirectoryIndexerName];

indexer.DocumentWriting += indexer_DocumentWriting;

void indexer_DocumentWriting(object sender,
Examine.LuceneEngine.DocumentWritingEventArgs e) {

var title= e.Document.GetField("title");

if(title==“Partner”){
e.Document.SetBoost(1.5f);
}
}

Indexing media
• Pdf indexer. Only indexes pdf content.
• CogUmbracoExamineMediaIndexer (Available as package on our)
– Uses apache tika. Indexes content and any associated meta data
– XML and derived formats
– Microsoft Office document formats
– OpenDocument Format
– Portable Document Format
– Electronic Publication Format
– Rich Text Format
– Compression and packaging formats
– Text formats
– Audio formats (MP3 etc)
– Image formats
– Video formats
– Java class files and archives
– The mbox format

Search highlighting
• Lucene contrib package Highlighter.net
• Highlights occurrences of your search term in
search results summary fragment.
• Wiki on our http://our.umbraco.org/wiki/how-
tos/how-to-highlight-text-in-examine-search-
results

Deployment to staging / production
environments
• Cannot copy index
• Can check in but could corrupt
• Selenium with ashx to rebuild index

environments
public class RebuildIndexes : IHttpHandler
{
readonly List<string> indexes = new List<string> { "ATGIndexer", "InternalIndexer", "directoryIndexer" };
public void ProcessRequest(HttpContext context)
{
context.Response.ContentType = "text/plain";
try
{
if(string.IsNullOrEmpty(context.Request.QueryString["index"]))
{
foreach (var index in indexes)
{
ExamineManager.Instance.IndexProviderCollection[index].RebuildIndex();
}

}
else
{
ExamineManager.Instance.IndexProviderCollection[context.Request.QueryString["index"]].RebuildIndex();
}
context.Response.Write("done");
}
catch(Exception ex)
{
context.Response.Write(ex.ToString());
}
}

public bool IsReusable
{
get
{
return false;
}
}
}

environments
[SetUp]
public void SetupTest()
{
selenium = new DefaultSelenium("localhost", 4444, "*chrome", "http://mydevsite");
selenium.Start();
_verificationErrors = new StringBuilder();
}

[Test]
public void RebuildIndex()
{
//not proper test but a hack to get indexes rebuilt after a deployment
try
{
selenium.Open("/umbraco/webservices/RebuildIndexes.ashx");

}
catch (SeleniumException se)
{
if (!se.Message.StartsWith("Timed out"))
{
throw;
}
}
catch (AssertionException e)
{
_verificationErrors.Append(e.Message);
}
}

Faceting
• Faceted search, also called faceted navigation or faceted
browsing, is a technique for accessing information organized
according to a faceted classification system, allowing users to
explore a collection of information by applying multiple filters
• Amazon, LinkedIn
http://www.linkedin.com/search/fpsearch?type=people&key
words=umbraco&pplSearchOrigin=GLHD&pageKey=member-
home&search=Search
• LinkedIn uses Bobo browser. Written in java it has been
ported to .net http://bobo.codeplex.com/
• Demo is SimpleFacetHandler others are available e.g
RangeFacet,PathFacet, GetFacet

Food for thought
• Using the index as object db ala RavenDb
• Scenario: You have nodes with large number of multi tree node pickers used as look ups

Food for thought
• In index node ids are stored as CSV list if MNTP
set to csv.
• Use GatheringNodeData event to do lookups
create a POCO with lookup data, serialise POCO
to JSON and store that in index.
• Advantage: Instant lookup all data ready to use
• Disadvantage: Need to keep up with lookup
changes. E.g. If Country code changes then you
would need to lookup code already in use and
update.
• Nice approach if lookup data is fairly static

Food for thought
• POCO hydration using activelucenenet ala
USiteBuilder
• Create pocos and decorate with attributes
public class Product
{
[LuceneField(“sku")]
public string Sku { get; set; }

[LuceneField(“productName")]
public string ProductName { get; set; }
}

Food for thought
var luceneProductDoc = GetItFromLucene(1234);
var product = LuceneMediator<Product>.ToRecord(luceneProductDoc );

Would need to use Lucene directly as there is a no way of getting the lucene
document from examine search result wrapper?

Take home today
• Use the index!!!

Questions
• ????
• http://twitter.com/ismailmayat

Examiness hints and tips from the trenches

More Related Content

What's hot (20)

Similar to Examiness hints and tips from the trenches (20)

Examiness hints and tips from the trenches

Editor's Notes