BDC Entity with TypeDescriptor System.String[] gives errors when crawling

Question

Is it possible to crawl a BDC content source with an entity that has a property of type string array (System.String[])?

At one point I had this working successfully in my dev environment but it longer works and I can't figure out why.

Environment is SharePoint 2010 and Search (not using FAST).

I've successfully created a BDC model that connects to a WCF Service and contains an Entity. The entity looks like this:

    <Entity Namespace="MyCompany.KM" Version="1.0.5.0" EstimatedInstanceCount="10000" Name="MyCompanyArchives" DefaultDisplayName="My Company Archives">
      <Properties>
        <Property Name="Title" Type="System.String">DisplayTitle</Property>
      </Properties>
      <Identifiers>
        <Identifier TypeName="System.Int32" Name="IdentifierId" />
      </Identifiers>
      <Methods>
        <Method IsStatic="false" Name="ReadItem">
          <Parameters>
            <Parameter Direction="In" Name="id">
              <TypeDescriptor TypeName="System.Int32" IdentifierName="IdentifierId" Name="id" />
            </Parameter>
            <Parameter Direction="Return" Name="ReadItem">
              <TypeDescriptor TypeName="BCSServiceProxy.MyCompanyArchives, MyCompanyArchives" Name="MyCompanyArchiveItem">
                <TypeDescriptors>
                  <TypeDescriptor TypeName="System.Int32" ReadOnly="true" IdentifierName="IdentifierId" Name="Record_No" />
                  <TypeDescriptor TypeName="System.DateTime" Name="Date_modified">
                    <Interpretation>
                      <NormalizeDateTime LobDateTimeMode="UTC" />
                    </Interpretation>
                  </TypeDescriptor>
                  <TypeDescriptor TypeName="System.String" Name="DisplayTitle" />
                  <TypeDescriptor TypeName="System.String[], mscorlib, Version=2.0.0.0, Culture=neutral, PublicKeyToken=b77a5c561934e089" IsCollection="true" Name="Personal_name">
                    <TypeDescriptors>
                      <TypeDescriptor TypeName="System.String" Name="Personal_nameElement" />
                    </TypeDescriptors>
                  </TypeDescriptor>
                </TypeDescriptors>
              </TypeDescriptor>
            </Parameter>
          </Parameters>
          <MethodInstances>
            <MethodInstance Type="SpecificFinder" ReturnParameterName="ReadItem" Default="true" Name="ReadItem" DefaultDisplayName="Read Item MyCompanyArchives">
              <Properties>
                <Property Name="LastDesignedOfficeItemType" Type="System.String">None</Property>
              </Properties>
            </MethodInstance>
          </MethodInstances>
        </Method>
        <Method IsStatic="false" Name="ReadIds">
          <FilterDescriptors>
            <FilterDescriptor Type="LastId" Name="LastIdSeen"/>
          </FilterDescriptors>
          <Parameters>
            <Parameter Direction="In" Name="LastIdSeen">
              <TypeDescriptor TypeName="System.Int32" IdentifierName="IdentifierId" AssociatedFilter="LastIdSeen" Name="LastIdSeen" />
            </Parameter>
            <Parameter Direction="Return" Name="ReadIds">
              <TypeDescriptor TypeName="BCSServiceProxy.MyCompanyArchives[], MyCompanyArchives" IsCollection="true" Name="ReadIds">
                <TypeDescriptors>
                  <TypeDescriptor TypeName="BCSServiceProxy.MyCompanyArchives, MyCompanyArchives" Name="MyCompanyArchivesItem">
                    <TypeDescriptors>
                      <TypeDescriptor TypeName="System.DateTime" Name="Date_modified">
                        <Interpretation>
                          <NormalizeDateTime LobDateTimeMode="UTC" />
                        </Interpretation>
                      </TypeDescriptor>
                      <TypeDescriptor TypeName="System.String" Name="DisplayTitle">
                        <Properties>
                          <Property Name="ShowInPicker" Type="System.Boolean">true</Property>
                        </Properties>
                      </TypeDescriptor>
                      <TypeDescriptor TypeName="System.String[], mscorlib, Version=2.0.0.0, Culture=neutral, PublicKeyToken=b77a5c561934e089" IsCollection="true" Name="Personal_name">
                        <TypeDescriptors>
                          <TypeDescriptor TypeName="System.String" Name="Personal_nameElement" />
                        </TypeDescriptors>
                      </TypeDescriptor>
                      <TypeDescriptor TypeName="System.Int32" ReadOnly="true" IdentifierName="IdentifierId" Name="Record_No" />
                    </TypeDescriptors>
                  </TypeDescriptor>
                </TypeDescriptors>
              </TypeDescriptor>
            </Parameter>
          </Parameters>
          <MethodInstances>
            <MethodInstance Type="Finder" ReturnParameterName="ReadIds" Default="true" Name="ReadIds" DefaultDisplayName="MyCompanyArchives Read Ids">
              <Properties>
                <Property Name="RootFinder" Type="System.String"></Property>
                <Property Name="UseClientCachingForSearch" Type="System.String"></Property>
                <Property Name="LastModifiedTimeStampField" Type="System.String">Date_modified</Property>
              </Properties>
            </MethodInstance>
          </MethodInstances>
        </Method>
      </Methods>
    </Entity>

I've created a search content source to crawl this BDC but I'm getting errors when I run a crawl. The problem seems to be due to the Personal_name property which is a string array (System.String[]).

In a crawl I get for example 169 successes and 551 errors. The 169 successes that show up in search results are all records that don't have a value for Personal_name and the WCF service returns a null object. The 551 errors that don't show up in search results are all records that do have a value for Personal_name and the WCF service returns a string array.

The error logs contain what seems to be a random assortment of the following errors:

FLTRDMN: Errorinfo is "Unexpected end of file has occurred. The following elements are not closed: SerializedEntityInstance. Line 54, position 4." [fltrsink.cxx:553] d:\office\source\search\native\mssdmn\fltrsink.cxx

FLTRDMN: Errorinfo is "There is an unclosed literal string. Line 54, position 28." [fltrsink.cxx:553] d:\office\source\search\native\mssdmn\fltrsink.cxx

FLTRDMN: Errorinfo is "Unexpected end of file while parsing Name has occurred. Line 52, position 38." [fltrsink.cxx:553] d:\office\source\search\native\mssdmn\fltrsink.cxx

Answer 1

I ran into the same problem and solved it by replacing the String[] by a IEnumerable<StringWrapper>, or in your case IEnumerable<Personal_nameElement>.

This could look like:

public class Personal_nameElement
{
  public String Value { get; set; }
}

And in the BDC:

<TypeDescriptor Name="Personal_name" TypeName="System.Collections.Generic.IEnumerable`1[[Personal_nameElement]]" IsCollection="true">
    <TypeDescriptors>
      <TypeDescriptor Name="Personal_nameElement" TypeName="Personal_nameElement" IsCollection="false">
        <TypeDescriptors>
          <TypeDescriptor Name="Value" TypeName="System.String" />
        </TypeDescriptors>
      </TypeDescriptor>
    </TypeDescriptors>
</TypeDescriptor>

The property of your entity class would be:

public IEnumerable<Personal_nameElement> Personal_name { get; set; }

In my case the errors went away and the list content was indexed properly.

Side note: Replacing String[] by IEnumarable<String> (without the additional wrapper) did not work and lead to the same errors as before.

Answer 2

Probably two years too late, but I've got the following multi-value property in my Entity that works fine:

<TypeDescriptor Name="Subjects" TypeName="System.String[]" />

Note that I didn't specify 'sub' TypeDescriptors, could that be the difference? Also, I'm using FAST, but I'm not sure if that is of any influence...

~Mark