There is actually a scenario where you'll want to use a custom crawler over multiple crawlers:
If you use the indexing.getDependencies pipeline to force an item to be (re)indexed when another item is indexed, that pipeline is only executed within the context of a single crawler. That means that an item from one crawler cannot cause the item of another crawler to be re-indexed.
Here's a basic MultiRootCrawler that we use in production that gets around the above limitation:
/// <summary>
/// Handles multiple root locations, specified by a pipe delimited Root value
/// </summary>
public class MultiRootCrawler : SitecoreItemCrawler
{
protected override bool IsAncestorOf(Item item)
{
using (new SecurityDisabler())
using (new WriteCachesDisabler())
{
var rootItems = GetRootItems();
return rootItems.Any(x => item.Paths.LongID.StartsWith(x.Paths.LongID,
StringComparison.InvariantCulture));
}
}
public override void RebuildFromRoot(IProviderUpdateContext context,
IndexingOptions indexingOptions, CancellationToken cancellationToken)
{
Assert.ArgumentNotNull(context, "context");
if (!base.ShouldStartIndexing(indexingOptions))
{
return;
}
var indexableRoots = this.GetRootItems()
.Select(x => (SitecoreIndexableItem)x)
.ToList();
IDocumentBuilderOptions documentOptions = base.DocumentOptions;
Assert.IsNotNull(documentOptions, "DocumentOptions");
var eventEmitter = context.Index.Locator.GetInstance<IEvent>();
foreach (var indexableRoot in indexableRoots)
{
eventEmitter.RaiseEvent("indexing:addingrecursive",
context.Index.Name, indexableRoot.UniqueId, indexableRoot.AbsolutePath);
this.AddHierarchicalRecursive(indexableRoot, context,
this.index.Configuration, cancellationToken);
eventEmitter.RaiseEvent("indexing:addedrecursive",
context.Index.Name, indexableRoot.UniqueId, indexableRoot.AbsolutePath);
}
}
public override int GetContextIndexRanking(IIndexable indexable)
{
var sitecoreIndexableItem = indexable as SitecoreIndexableItem;
if (sitecoreIndexableItem == null)
{
return 2147483647;
}
Item item = sitecoreIndexableItem;
using (new SecurityDisabler())
using (new SitecoreCachesDisabler())
{
foreach (var rootItem in GetRootItems())
{
if (item.Paths.LongID.StartsWith(
rootItem.Paths.LongID, StringComparison.InvariantCulture))
{
return item.Axes.Level - rootItem.Axes.Level;
}
}
}
return 2147483647;
}
/*
* RefreshFromRoot does refer to GetIndexableRoot, but (in 8.1 at least)
only after a scenario where it would have thrown a NullReferenceException, so
we can safetly assume it doesn't actually occur.
*/
public new string Root
{
get { return base.Root; }
set
{
if (value == null)
{
this.Roots = new string[0];
base.Root = null;
return;
}
this.Roots = value.Split('|');
base.Root = this.Roots[0];
}
}
public string[] Roots
{
get; set;
}
private List<Item> rootItems;
private List<Item> GetRootItems()
{
if (this.rootItems == null)
{
var db = ContentSearchManager.Locator.GetInstance<IFactory>()
.GetDatabase(this.Database);
using (new SecurityDisabler())
{
this.rootItems = this.Roots
.Select(path => db.GetItem(path))
.Where(item => item != null)
.ToList();
}
}
return this.rootItems;
}
}
