MIME Type Identification

One of the core tasks of Aperture is full-text and metadata extraction. To choose the right Extractor for a given document, one must first establish the document's MIME type. Therefore, we have designed a MimeTypeIdentifier API that fulfills this task and developed a general purpose implementation for it, MagicMimeTypeIdentifier.

We know from experience that simple heuristics based on for example file name extensions or HTTP result headers break rather easily. Our implementation is therefore primarily based on magic numbers (the first few bytes in a file that can often be used to recognize the file type with great accuracy) and uses file extensions only as a fall-back when a certain file type does not have a magic number (e.g. BinHex files) or when the magic number is shared by a number of different formats (e.g. MS Office files and some other file formats using the OLE structure all have the same magic number).

The MIME types and their magic numbers and file name extensions are described in an XML file delivered with the MagicMimeTypeIdentifier. The XML file to use can be specified, so the same code can also be used with a different set of MIME type descriptions. The format of the XML file and the algorithm for applying them is documented in the default XML file.


See the page on Extractors for a combined example of how to use a MimeTypeIdentifier and the Extractor framework.