How to combine field and attribute searches in XML

Last Reviewed: August 30, 2004

Article: DTS0195

Applies to: dtSearch Engine 6

dtSearch indexes attributes in XML files as if they were fields. For example:

This is indexed as it if were:

<name>

<last>Smith</last>

</name>

In some cases, though, the intent of an attribute is to modify the field name, not to provide a value. Example:

<name type="last">Smith</name>

The default treatment of XML attributes in dtSearch makes it difficult to search for a first name that was stored with this use of attributes. (A search for "(/name/type contains first) and (/name contains John)" will match any document that contains a "first" value in the "/name/type" field and also "John" in a /name field. However, these may be different /name fields in the same document.) A simple transformation can be performed on the XML that will cause the fields to be indexed in a way that permits searching on a first name. To do this, replace the attribute with a generated field that combines the attribute and named value, like this:

<name>

<TypeLast>Smith</type_last>

</name>

With this field structure, a search can be limited to the first name field as follows:

/name/TypeFirst contains John