MARC21toMADS XSLT needs bugfixes and updates

[Mackenzie Johnson]

EG 3.8.0
PostgreSQL 10.something?

Okay, these issues stretch all the way back to the official .xsl file hosted on the MADS website, so it's not something that popped up as part of any direct EG developement. I suspect the XSLT for XML 1.0 is not getting much or any focus from LoC anymore.

First, a question: is there a reason why the contents of MARC21slimUtils.xsl is copied into the MARCtoMADS XSLT instead of being called in the schema? Does it have something to do with MARC21slimUtils.xsl not being part of the .xsl files included in the main branch of the repository (as far as I could see)?

Second, the XML 1.0 version of the XSLT that we use technically still calls to the MADS 2.0 schema in the mads:madsCollection node rather than MADS 2.1. The mads:RecordOrigin node also still says the XSLT is on revision 2.13 when the last official revision was 2.15.

Third, past fixes in the Geographic templates have actually broken other components of the Geographic subject crosswalk -- namely, the XSLT as it currently stands does not transform any 451$a or 751$a values, as an xsl:otherwise node that would cover both those fields (and acts similarly in comparable templates) has been commented out in an older revision, and the template call for the 751 field is also matching on the $z subfield instead of $a (the 781 field for linking geographic subdivisions does still match on $z).

Fourth, related to updating the schema to 2.1, a template could be made to parse out an X00$q value (the fuller form of a personal name) to become a distinct namePart node with a fullerForm attribute. Currently the $q is translated into MADS, but I believe it stays with the $a subfield

A mix of small but significant changes, small and less significant changes, and if there is a desire to fully make use of MADS 2.1 instead of MADS 2.0, the possibility to add a new small template. I've made a number of those fixes in a local copy, but again I figured I should create the bug before I share that (though I'll gladly attach it upon request).

Update (2023-11-21): Found more issues! Templates are also not properly applied across subfields in 450 and 550 fields.

[Mackenzie Johnson (mtjohnsonupei)]

Making this bit a comment rather than updating my original post, as it pertains to further changes needed to the schema in order to adhere to changes made in MADS 2.1 rather than bugs affecting the capacity to transform to MADS 2.0.

To match the changes made to MADS 2.1, the transformations for the 372 (field of activity) and 374 (occupation) MARC fields need to be changed so that they become subelements of a <mads:fieldOfEndeavor> element.

Currently, 372 transforms in at <mads:fieldOfActivity> element (with any start and end dates include as part of the one text node), while 374 currently is captured in the <mads:extension> element as <mads:profession>, with <mads:professionTerm>, <mads:startDate>, and <mads:endDate> as children. The start and end dates are also subelements under the new <mads:fieldOfEndeavor> element, but <mads:profession> does not have a further professionTerm subelement in the 2.1 schema.

[Mackenzie Johnson (mtjohnsonupei)]

I emailed the MODS/MADS listserv to ask about another error I found in there, and in the reply back I did hear that the editorial committee is planning to both make corrections to the MADS 2.0 XSLT we currently use, and release the XSLT for transforming MARCXML into MADS 2.1 in the very near future.

[Mackenzie Johnson (mtjohnsonupei)]

At the request of the Authorities Working Group, I am attaching my version of the MARC-to-MADS2.0 XSLT with my fixes and recommendations for compatibility with MADS2.1 (though on this latter point I'd wait to see the official new MADS2.1 XSLT that should be published soon). I have commented on all my changes -- they all begin with UPEI's "CaPCU" MARC identifier code for easy identification of what I changed.

I should note, this file does not include the copy-pasted contents of MARC21slimUtils.xsl, which the original XSLT from LoC simply calls, but in EG by default is stored within the fuller MADS XSLT.