When you are building your Endeca implementation, you need to determine how you can import your source data, and whether a content adapter is necessary.
In general, regardless of the format of your source data, you need to first load it into the Data Foundry by using a record adapter. After the data is loaded through the record adapter, it can be processed further by Forge and indexed by Dgidx.
The Endeca software is packaged with predefined record adapters. The record adapters provide the build-in facility for processing data from many formats, including delimited, XML, binary, and others.
To decide whether you need a content adapter:
Determine if any of the standard record adapters can handle your source data. For information on record adapters, see the Endeca Forge Guide and the Oracle Endeca Developer Studio Help. If you can use a record adapter, import your source data with it.
If the format of your source data cannot be handled by any of the predefined record adapters shipped with Endeca, create your own content adapter in Java to process your data, using the CADK.
Note
Instead of a content adapter that will transform your source data, you can create a Java manipulator for the same purpose, since Java manipulators are a general version of the content adapters.
Here is how content adapters differ from Java manipulators:
Content adapters cannot take any Forge components as an input. They can only read data from multiple external sources. (You need a separate content adapter for each data source).
Java manipulators can read in source data as well, but the main purpose of Java manipulators is to process records that already exist within the pipeline in Forge.
To summarize, Java manipulators can take either source data, or records that have already been passed through record adapters or manipulators, and transform the records further. Content adapters can only transform source data.
