As Don pointed out, you can parse the result XML which contains information about each extracted file attachment, including the unique name created for the document, the filename that was originally associated with the document, and the content type, if it was originally provided.
The problem with using the filename for the extracted name of the document in the AssemblerResult documents map is that the filename is not guaranteed to be unique.
You can ask for the result XML without actually extracting the file attachments (extract="false"), and then use the information from the result XML to get exactly the file attachments you want.
The unique name created (the attachmentKey) actually provides information as to whether the file attachment was attached to the document in general or to a specific page, such as for a comment annotation.
The attachments.xsd schema is installed in the LC ES SDK to assist with parsing.
Here's an example result XML with one document-level and one page-level file attachment:
<?xml version="1.0" encoding="UTF-8"?>
<Attachment attachmentKey="doc.source_attach.0000.0001" name="Untitled Object">
<File creationDate="2004-08-05T01:34:18Z" mimeType="image/jpeg" modificationDate="2000-03-30T01:46:42Z" size="14359">
<Filename fromEncoding="ISO-8859-1" success="true">Origami.jpg</Filename>
<Description>This is Origami.jpg, a document-level attachment.</Description>
<File creationDate="2004-08-05T01:34:16Z" mimeType="image/gif" modificationDate="2002-09-10T01:51:22Z" size="3939">
<Filename fromEncoding="ISO-8859-1" success="true">dog.gif</Filename>
<Description>This is a dog.gif from page 3</Description>
<Location x="532.5" y="720.75"/>