[Yazlist] support for solrmarc?

Michael Lackhoff lackhoff at zbmed.de
Tue Mar 18 18:33:14 UTC 2014

On 18.03.2014 18:40 Adam Dickmeiss wrote:

>> &#1D; -> #29;  (record separator)
>> &#1E; => #30;  (field separator)
>> &#1F; => #31;  (subfield separator)
> Boy that's ugly. And how's #29 then handled if it's NOT a control sequence?

It is only a control sequence if followed by a ';'.
In a Perl-application I need just one line to get perfectly valid iso2709:
$in =~ s/#([0-9]{2});/chr($1)/eg;
After this one-liner I can use any MARC-tool on the planet with it.

> IMHO, using an element for that would have been more elegant.

I have no influence on that and though it might not look like it, it
really works well with data in the wild.

>> <marc inputformat="solrmarc" outputformat="marcxml" outputcharset="utf-8"/>
> In the conversion from solrmarc to marcxml there wouldn't be any illegal 
> characters.
> But the stylesheet to convert from solrmarc to marcxml would be a hairy 
> XSL transform indeed.

A stylesheet would indeed be horrible. That is why I would like a
variant of
<marc inputformat="marc" outputformat="marcxml" outputcharset="utf-8"/>
With the only difference that something like the regular expression
above is applied first

> Given such a stylesheet you'd use something like this to offer marcxml:
>           <retrieval syntax="xml" name="marcxml">
>                  <backend syntax="xml">
>                     <xslt stylesheet="solrmarc2marcxml.xsl"/>
>                 </backend>
>          </retrieval>

Because such a stylesheet would be so difficult to write I want <marc>
not <xslt> for the conversion.
As I said it would be an almost trivial change (well I am not a C
programmer but I hope it is). And to avoid any confusion or possible
errors in normal iso2709 I came up with the suggestion of a new allowed
inputformat for <marc>
After all solrmarc is just another variant like 'marc', 'marcxml' or

> You could use record_transform to also offer marc21/usmarc in a marc 
> step (taking marcxml as input).

That is exactly my intention: Offer all standard MARC variants from the
one that is equivalent but is just this one-liner away from being
useable as it is.


More information about the Yazlist mailing list