MARC record specs for export to libraries

[Karen Coyle]

This says how to export MARC records for scanned books. This specifies the most common format that library catalogs can accept.

Any full MARC records that have been received with the scanned books can be used as the basis for a MARC export. The MARC export must:
  -- be a MARC "binary" file
  -- use the MARC-8 character set

The link to the scanned book needs to be in the 856 field, with these subfields:

 -- $u [the URL]
 -- $z [note for users, e.g. "PDF" or "DejaVu" if we are sending multiple formats]
[those aren't really dollar signs, they are MARC subfield codes, hex 1F]

If there will be more than one URL, each one is in a separate 856 field.

In the 856 field, the first indicator must be set to "4", meaning "http".

e.g.
8564 $uhttp://www.archive.org/download/truebenjaminfran00fishiala/truebenjaminfran00fishiala.pdf$zPDF
8564 $uhttp://www.archive.org/download/truebenjaminfran00fishiala/truebenjaminfran00fishiala.djvu$zDjVu
8564 $uhttp://www.archive.org/stream/truebenjaminfran00fishiala$zRead online

*Ideally* we would also replace the following fields in the record being exported:

001 -- should contain an identifier for the MARC record (if we don't have one, it should be left as it is. 001 is a required field)
003 -- put in "archive.org" to identify the source of the record
005 -- the date and time that the record is created, in format yyyymmddhhmmss.x (just date is fine, any other positions can be set to zero)

Note that after adding fields the directory and leader must be recalculated. It may be easiest to use one of the existing MARC packages, which do this automatically on export. (Like Marc4J or MARCPM)