A transforming proxy for the National Library of Australia's Trove API
The proxy consists of a docker container containing:
Access the proxied Trove API as you would normally access the Trove API, except replacing
https://api.trove.nla.gov.au/ with http://localhost:8080/proxy/ as the base URI of the API service.
The proxied API accepts some additional parameters:
| Parameter name | Values |
|---|---|
| proxy-include-people-australia | If set to true then additional information about people will be added from People Australia |
| proxy-format | Set to tei to return TEI XML, atom to return Atom Syndication XML, csv to return Comma Separated Values, or leave blank for Trove-style XML |
| proxy-metadata-format | Set to ro-crate to return an RO-Crate description of the data instead of the actual data |
| proxy-metadata-license-uri | Set to the URL of a licence such as https://creativecommons.org/licenses/by-nc-sa/3.0/au/ |
| proxy-metadata-name | Provide a title for your dataset, e.g. Scone recipes |
| proxy-metadata-description | Provide a description for your dataset, e.g. A collection of scone recipes |
To build the docker application, naming the image trove-proxy:
docker build -t trove-proxy .To launch the trove-proxy image:
docker run --publish 8080:8080 trove-proxy For convenience while developing the proxy application, you can use the --mount command to
mount the src folder into the container, so that you can edit the XProc and XSLT code and have the changes reflected in the running container
immediately, without having to rebuild the docker image. If accessing the application using a web browser, you can just refresh the browser to
see the effect of any changes to the code.
docker run --publish 8080:8080 --mount type=bind,src=`pwd`/src,dst=/src trove-proxyWith the container running, the proxied API can be accessed as if it were the Trove API, but substituting
http://localhost:8080/proxy/ in place of https://api.trove.nla.gov.au/.
In addition to Trove's own query syntax, the URI should typically contain the parameter proxy-format,
whose value determines the output format of the proxy; values should be either tei, atom, or if the
proxy-format parameter is missing from the URI, or if it's left blank, the proxy will return Trove's XML
without transformation, beyond the rectification of a few errors and infelicities.
XSLT transformations should be placed in the src/xslt/crosswalks folder, and named for the format which they produce, e.g. tei.xsl to output tei. To install a new output
format, it's enough to add the appropriately named XSLT file into that folder.
Before the XML retrieved from the Trove API is supplied to the crosswalk stylesheets, it will have been slightly altered by the fix-trove-response.xsl pre-processing stylesheet:
articleText element of a newspaper article will be replaced
with embedded <p> and <span> elements (i.e. the articleText element will be "mixed content").A second pre-processor rewrite-trove-uris-as-proxy-uris.xsl will also replace URIs within the Trove XML which
are identifiers for individual works with proxied URIs which should resolve to proxied versions of those works.
Crosswalk stylesheets should not need to change these URIs and should be able to simply copy them into the
appropriate place for their output format.
The crosswalk stylesheets will be passed a request-uri parameter, which will be the URI of the proxied resource
(i.e. it's usable as a self reference).